DETAILED ACTION
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 12/31/2025 has been entered.
Claims 11 – 13 and 15 have been presented for examination. Claims 11 – 13, 15 are currently amended. Claims 1 – 10, 14, 16 – 25 are cancelled.
The instant office action relies on Sayyar-Rodsari et al. (US 2015/0185717) which is cited on the IDS.
Response to Written Description Support
Applicant’s arguments regarding the written description support have been fully considered. However, the Office does not consider them to be persuasive.
Applicant argues: “Thus the focus is exclusively on calculating parameter gradients for data fitting and using penalty-based constraints, with no teaching or requirement for process gain calculation present in the description of the invention.”
Applicant argues that the absence of explicit disclosure of “process gain calculation” provides written description support for the limitation “wherein said control system is configured to apply reverse-mode automatic differentiation with respect to the system of differential equations and generate an estimate representing an evaluation of the parameterized process model of the system of pumps or hydropower stations and associated reservoirs in a manner that does not also generate the substantially full set of process gains with respect to process inputs”. Examiner notes that a negative limitation further limits the claim and the mere absence of explicit disclosure does not provide written description support for said negative limitation since the disclosure is broader and could be implemented in a different manner (i.e., with gain calculations).
Applicant argues: “The use of a gradient-descent or ascent procedure without a Hessian matrix is supported by the specification at [896-899]: "This can be used to calculate derivatives on a control objective function and/or a modelling error function and optimize the model and/or control parameters using a gradient descent/ascent-based method." Here the specication [sic] explicitly teaches updating parameters based on a "gradient-descent or ascent procedure." Further support can be found in lines [1048-1068] of the Detailed Description”
Applicant points to Paragraphs 896-899 and 1048 – 1068. However, there is no explicit disclosure of this negative limitation. Therefore, Applicant’s arguments are not persuasive based on the preceding remarks.
Applicant argues: “The system of pumps and/or hydropower plants and the control objectives that include maximum and/or minimum levels are described in the Detailed Description at [1173] - [1233].”
Examiner notes that it is merely disclosed that a control objective includes electricity costs and “high” reservoir levels (see the instant application Paragraph 38 “The company defines a control objective that is punishing the model for keeping water levels in 1180 the reservoirs at very high levels and punishing it for a value proportional to the electricity costs and includes it as a variable in the Modelica model. The control objective variable is designed such that it sums all past costs and high water levels over simulated time.”). Therefore, there does not appear to be written description support for “maximum level”. Further, there does not appear to be explicit disclosure of the recited “minimum levels”.
Response to Claim Interpretation -35 USC § 112(f)
Examiner that the invocation of 112(f) was withdrawn in the previous Office Action. Therefore, Applicant’s arguments are moot.
Response to Claim Rejections -35 USC § 101
Applicant’s arguments have been fully considered. However, the Office does not consider them to be persuasive.
Applicant argues: “The recited operations-simulation of process dynamics, reverse-mode automatic differentiation, parameter updating, and generation of control signals-are not claimed in the abstract. Rather, they are expressly tied to a physical industrial process, operate on technical sensor data representing real-world states, and are used to control the operation of that physical process.
As emphasized by the USPTO's 2019 and 2024 Subject Matter Eligibility Guidance, claims are not "directed to" a judicial exception where the alleged mathematical operations are part of a larger technological process involving physical systems and real-world effects. Here, the claimed mathematical techniques are embedded within a closed-loop industrial control architecture, not claimed as standalone calculations or mental processes.
Accordingly, the claims are directed to a technical control system, not to an abstract idea per se.” (emphasis in bold added)
Applicant argues that the claimed invention is specifically tied to a physical industrial process operating on data representing real-world states, and used for controlling the physical process in a closed-loop industrial control architecture Examiner notes that the “control architecture” is implemented using no more than a general purpose computer (i.e., “a computer-implemented control system”) in combination with generic sensors (i.e., “obtain technical sensor data”), with the controlling is recited at a high-level of generality (i.e., “configured to produce control signals that control”) in combination with the results of the parameterized process model. Examiner further notes that the claimed improvement is to simulating based on the process model itself, as contrasted with the use of the process model for downstream controlling of the process (see Applicant’s arguments below “Applicant's insight-that hybrid models with millions of parameters invert these trade-offs-is non-obvious and enabled by the novel architecture”).
Applicant argues: “Specifically, claim 11 requires that the control system: * operates on technical sensor data obtained from a real system of pumps or hydropower stations and associated reservoirs; * the control system is configured to apply reverse-mode automatic differentiation … * the control system is configured to update … * produces control signals that control the operation of This is not a mere "apply it" scenario. The claimed invention improves the technical operation of industrial control systems by enabling adaptive, model-based control of physical processes using automatically updated process models. The result is improved control accuracy, robustness, and adaptability of real-world industrial systems, which constitutes a practical application under MPEP § 2106.0S(a) and (c).
Contrary to the Examiner's position, the production of control signals here is not insignificant post-solution activity. The generation of control signals is the functional objective of the claimed system and directly causes physical changes in the controlled industrial process, which is a hallmark of patent-eligible subject matter.” (emphasis added)
Examiner notes that various limitations argued by Applicant are insignificant extra-solution activity or part of the abstract idea itself (see Claim Rejections - 35 USC § 101). Regarding the production of real control signals to effectuate improved control accuracy, etc., Examiner notes that the problem being overcome is related to simulation using the process model itself for updating process model parameters (see Applicant’s arguments below “Applicant's insight-that hybrid models with millions of parameters invert these trade-offs-is non-obvious and enabled by the novel architecture”). Further, there is not recited, nor does there appear to be any explicitly disclosed, improvement in “control accuracy, robustness, and adaptability”. Looking to the disclosure, there are various other improvements not related to the controlling itself: execution of the simulation itself (see the instant application Line 65 – 67 “65 It is a specific object to provide computationally more efficient calculation of sensor-based error gradients on acausal declarative models that contain universal function approximators.”), creation of the process model (see the instant application Line 38 – 41 “At the same time there is a need to model systems using Artificial Intelligence (Al) Methods such as neural networks and other similar universal function approximators in order to reduce the need for expensive manual modelling hours and/or for modelling systems with unknown behaviour”),and data handling in the process model (see the instant application Line 51 – 54 “The industrial tendency to digitalize the operations and the possibility to create more complex models with the improved modelling languages means that modelling systems are facing new magnitudes of data that cannot be solved with current computational systems and methods.”).
Applicant argues: “The claimed control system is not a generic computer performing generic data processing. Instead it includes:
PNG
media_image1.png
492
720
media_image1.png
Greyscale
”
Examiner notes that limiting the claimed invention to a field of use (i.e., a specific industrial context) does not amount to significantly more (see MPEP 2106.05(h)). Further, the “technical sensor” are wholly generic. Further, the “control signals” are merely used to “control the operation” without reciting any specific control action nor form of said control signals, therefore, they amount to reciting the words “apply it”. The remaining argued limitations “process models” and “parameter adaptation” are part of the abstract idea (see Claim Rejections -35 USC § 101).
Applicant argues that the claims “are rooted in industrial process control technology”. Examiner notes that the process model simulation and estimation and updating requires no more than a general purpose computer. Further, the recited “produce control signals that control” does limit the claim to any specific industrial process control technology and merely recites what is being controlled (i.e., “operation of the system of pumps or hydropower stations and associated reservoirs”).
Response to Claim Rejections -35 USC § 103
Applicant’s amendments overcome the prior art rejection. Therefore, it is withdrawn. However, a new rejection is included over Coelho, B. “Energy efficiency of water supply systems using optimisation techniques and micro-hydroturbines” (see Claim Rejections -35 USC § 103).
Applicant argues: “In reference to cited arts, the claimed invention is non-obvious because it requires rejecting the primary teachings of both Andersson and Sayyar-Rodsari across four interlocking design choices: (1) shooting vs. collocation, (2) sensitivity constraints vs. gain-free evaluation, (3) Hessian-based vs. Hessian-free optimization, and (4) forward-mode vs. reverse-mode AD. The references actively would have no motivation to use the claimed combination by promoting methodologies fundamentally incompatible with scalable hybrid model optimization.” (emphasis added)
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., “scalable” hybrid model optimization) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). Nor does the claimed invention appear to explicitly recite limitations that inherently result in this feature.
Applicant argues: “1. Direct collocation methods are both easier and faster to train. This is also the method taught as superior by Andersson in Table 1 and 5 .2 … Therefore, a Person of Ordinary Skill in the Art (POSITA), following the explicit teaching of Andersson, would be dissuaded from selecting a shooting method and would be directed toward the superior collocation method …
The lack of derivatives with process input and the lack of Hessian matrix excludes the direct allocation methods of Andersson and any practical variation of it through multiple claim restrictions.” (emphasis added)
Examiner notes that the disclosure of non-preferred embodiments does not constitute teaching away (see MPEP 2123(II) “Disclosed examples and preferred embodiments do not constitute a teaching away from a broader disclosure or nonpreferred embodiments”). Further, this is the base reference and teaches away is germane to combining a secondary reference with the base reference.
Regarding the lack of a Hessian matrix, Examiner notes that Andersson does not explicitly require the use of a Hessian matrix, nor does the claimed invention explicitly limit the robustness or convergence (see Page 7, Left second order derivatives improve the result but are not required “As described in Section 2.1, one strategy for solvIng large-scale dynamic optimization problems is direct collocation, where the dynamic DAE constraint is replaced by a discrete time approximation … In addition, second order derivatives can often improve robustness and convergence of such algorithms.”). Further, the lack of the Hessian matrix is part of gradient-descent of gradient-ascent which is taught by Baydin.
Applicant argues: “Assume Andersson in light of Sayyar-Rodsari, we arrive at a myriad of possible solutions, with direct collocation being the direct application of the teachings of Andersson. Let use assume instead an attempt at a multiple shooting model together with the PUNDA model.”
Applicant appears to provide a conclusory statement regarding “a myriad of possible solutions”, in combination with a different assumption of multiple shooting model. Therefore, Applicant’s arguments are not persuasive.
Applicant argues: “The preferred embodiment in Sayyar-Rodsari, forward mode, already produces the derivatives w.r.t. inputs (see Sayyar-Rodsari [0144] and [0145], preempting any computational advantage of reverse mode AD. Further, the number of outputs of the models becomes large when sensitivites are calculated, which means reverse-mode AD would be slower, as will be explained below and in Fig 2. and 3. …
Therefore, a POSITA would be dissuaded from abandoning the calculation of process gams, as Sayyar-Rodsari Teaches they are a necessity for a 'successful model.'”
Examiner notes that Sayyar-Rodsari merely teaches that process gains must be accurately captured, not necessarily that they are generated. Specifically, process gains can be captured using constraints on the first order derivatives (see Paragraph 46 “It is generally accepted that a successful model for optimization and control must accurately capture both process gains and dynamics”, and Paragraph 47 “In addition to the derivative constraints (the first order of which are commonly referred to as gain constraints in the literature), the training of the neural network block in the PUNDA model can be constrained to ensure desired dynamic behavior for the PUNDA model.”). Further, the mere disclosure of non-preferred embodiments does not amount to a teaching away (see MPEP 2123(II) “Disclosed examples and preferred embodiments do not constitute a teaching away from a broader disclosure or nonpreferred embodiments”).
Applicant argues: “Andersson teaches Hessian-based optimization in IPOPT (Andersson chapters 2.4 "qpOASES", 3.4 "hessian_approximation", 5.1 "both of them using an inexact Hessian approximation" and 5.2 "the Python-based collocation algorithm is clearly superiorm it converges more quickly, which is likely due to the provided exact Hessian.", with Casadi API documentation for qpOASES requiring a Hessian [4]. Sayyar-Rodsari also teaches the use of Hessian, as indicated in [0104] "A preferred embodiment for training of the neural network model. .. This constrained NLP model may then be solved with any appropriate NLP technology (e.g Sequential Quadratic Programming ... )".
The use of Hessian is supported by Sequential Quadratic Programming being repeatedly being referred to as the only named optimizer (Sayyar-Rodsari: [0045],[0104][0164]), possibly in combination with branch-and-bound describe in Sayyar-Rodsari [0152]. ” (emphasis added)
Applicant previously argued against using the shooting method of Andersson. However, the 3.4 and 5.1 hessian_approximation is used in single shooting method. (see Page 6, Right “To demonstrate the tool, we show how to implement a simple, single shooting method for the Van der Pol oscillator used as a benchmark in section 5.1:”, and Page 7, Right this is part of single shooting method
PNG
media_image2.png
134
423
media_image2.png
Greyscale
). Further, merely teaching that using the Hessian improves convergence does not require using said Hessian. Further, the claimed invention does not explicitly limit the robustness or convergence. The reference to qpOASES is part of ACADO toolkit and is not explicitly part CasADi, nor is quadratic programming explicitly relied upon in the collection method. Further, Sayyar-Rodsari merely mentions SQP as one alternative among many and does not require its use.
Applicant argues: “The scalability advantages of current invention are enabled by avoiding a Hessian-based optimization. It would be known to the POSITA that second order methods are generally known to converge faster, as also implied in Andersson 5.2, and he would be dissuaded to abandon these.
Therefore, a POSITA would be dissuaded from abandoning Hessian-based optimization, as both references consistently employ them for their recognized convergence advantages.
Therefore, the POSITA would be dissuaded from abandoning the use of Hessians”
Applicant’s arguments are not persuasive based on the preceding remarks.
Applicant argues: “Therefore, a POSITA, guided by the known computational trade-offs [1][2][3] and the teachings of Sayyar-Rodsari, would be dissuaded from using reverse-mode AD in the context of process optimization”
Applicant’s arguments are not persuasive since prior art reference [1] [2] [3] are not relied upon in the instant Office Action. Further, Applicant refers to them as “known computational trade-offs” as contrasted with teaching that are inherent to Sayyar-Rodsari.
Applicant argues: “The prior art addresses small-to-medium scale problems where forward-mode AD and Hessian methods are optimal. Applicant discovered that for very high-parameter hybrid models (millions of parameters), the computational trade-offs invert. This scale was not contemplated by the references.
This insight is only actionable after making the prior, non-obvious decisions to use a shooting method, forgo strict sensitivity constraints, and employ a Hessian-free optimizer. These choices collectively create the conditions where reverse-mode AD becomes the most efficient choice, but each one individually is taught against by the prior art …
The references demonstrate success with small-scale problems where their methods are optimal. Applicant's insight-that hybrid models with millions of parameters invert these trade-offs-is non-obvious and enabled by the novel architecture.
The jump directly to large networks, or alternatively large amounts of small networks, must take place before any experimental results show any benefit”
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., “shooting method”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). Further, there is no limit on the “scale” of the problem to create the conditions to realize one or more of the argued efficiencies.
Applicant argues: “The system described by the new claim limitations would be non-obvious to the POSITA. A POSITA would have no motivation to combine these teachings in the manner claimed, as doing so would require rejecting the explicit superior methods taught by Andersson while adopting an approach (shooting with reverse-mode AD) that the literature shows is computationally inferior for the problem sizes addressed by the prior art”
Applicant’s arguments are not persuasive based on the preceding remarks.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claim 11 – 13 and 15 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
With regard to claim 11, it recites “a control objective that includes electricity costs and penalties for exceeding minimum and/or maximum reservoir levels”. However, it is merely disclosed that a control objective includes electricity costs and “high” reservoir levels (see the instant application Paragraph 38 “The company defines a control objective that is punishing the model for keeping water levels in 1180 the reservoirs at very high levels and punishing it for a value proportional to the electricity costs and includes it as a variable in the Modelica model. The control objective variable is designed such that it sums all past costs and high water levels over simulated time.”).
Further, it recites “wherein said control system is configured to apply reverse-mode automatic differentiation with respect to the system of differential equations and generate an estimate representing an evaluation of the parameterized process model of the system of pumps or hydropower stations and associated reservoirs in a manner that does not also generate the substantially full set of process gains with respect to process inputs”. However, there is no disclosed algorithm or steps to teach this effect. Examiner notes that a negative limitation further limits the claim and the mere absence of explicit disclosure does not provide written description support since the disclosure is broader and could be implemented in a different manner (i.e., with gain calculations).
Further, it recites “wherein said control system is configured to update at least one process model parameter of the fully or partially acausal modular parameterized process model of the industrial and/or technical process system of pumps or hydropower stations and associated reservoirs based on the generated evaluation estimate and based on a gradient-descent or ascent procedure that is performed without calculating a Hessian matrix”. However, there is no disclosed algorithm or steps to teach this effect. Examiner notes that a negative limitation further limits the claim and the mere absence of explicit disclosure does not provide written description support since the disclosure is broader and could be implemented in another manner (i.e., with gain calculations).
With regard to claims 12 – 13 and 15, they are rejected by virtue of depending from a rejected parent claim, and without reciting additional limitations to overcome the unclarity.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 11 – 13 and 15 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
With regard to claim 11, it recites “substantially” in “wherein said control system is configured to apply reverse-mode automatic differentiation with respect to the system of differential equations and generate an estimate representing an evaluation of the parameterized process model of the system of pumps or hydropower stations and associated reservoirs in a manner that does not also generate the substantially full set of process gains with respect to process inputs”. The metes and bounds of “substantially” are unclear since the term is relative. Further, the disclosure does not clarify the boundaries of “substantially” nor is there provided a special definition (see MPEP 2173.01(I)). The limitation is interpreted for examination purposes as generating a subset of the full set of process gains, or not requiring the generation of process gains.
With regard to claims 12 – 13 and 15, they are rejected by virtue of depending from a rejected parent claim, and without reciting additional limitations to overcome the unclarity.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 11 – 13 and 15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., an abstract idea) without significantly more.
Independent claim 11 recites at Step 1 a statutory category (i.e. a machine) computer-implemented control system for a system of pumps or hydropower stations and associated reservoirs, where said control system is configured to control the system of pumps and/or hydropower stations and associated reservoirs by optimizing against a control obiective that includes electricity costs and penalties for exceeding minimum and/or maximum reservoir levels, wherein said control system is configured to: simulate the dynamics of one or more states of the system of pumps or hydro power stations and associated reservoirs over time based on the fully or partially acausal modular parameterized process model and a corresponding system of differential equations, and apply reverse-mode automatic differentiation with respect to the system of differential equations and generate an estimate representing an evaluation of the parameterized process model of the system of pumps or hydro power stations and associated reservoirs in a manner that does not also generate the substantially full set of process gains with respect to process inputs, and the system is configured to generate the evaluation estimate at least partly based on the obtained technical sensor data, update at least one process model parameter of the fully or partially acausal modular parameterized process model of the system of pumps or hydropower stations and associated reservoirs based on the generated evaluation estimate and based on a gradient-descent or ascent procedure that is performed without calculating a Hessian matrix. At Step 2A, Prong I the recited limitations, alone or in combination, amount to steps that, under its broadest reasonable interpretation, cover mathematical concepts (see MPEP 2106.04(a)(2)(I)). For example, the “simulate” explicitly relies on a parameterized process model (i.e. a function) having corresponding differential equations. The “apply reverse-mode automatic differentiation” recites performing a specific mathematical operation. The “update” amounts to adjusting a parameter in the parameterized process model using a specific updated technique. Accordingly, the claim recites an abstract idea.
At Step 2A, Prong II this judicial exception is not integrated into a practical application since the claimed invention further claims: that the control system is computer-implemented; obtain at least one technical model, being a fully or partially acausal modular parameterized process model of the system of pumps or hydropower stations and associated reservoirs, also referred to as a parameterized process model, including one or more process model parameters of the fully or partially acausal modular parameterized process model, wherein the fully or partially acausal modular parameterized process model is defined such that the system of pumps or hydropower stations and associated reservoirs is at least partly modeled by one or more neural networks used as universal function approximator(s); and obtain technical sensor data representing one or more states of the system of pumps or hydropower stations and associated reservoirs at one or more time instances; and produce control signals that control the operation of the system of pumps or hydropower stations and associated reservoirs based on the updated process model parameter or parameters. The “computer-implemented” is incorporated into the claim at a high-level of generality such that the structure amounts to no more than mere application of the judicial exception using generic computer components which does not amount to an improvement in computer functionality (see MPEP 2106.04(a)(I)). The “obtain” amounts to insignificant data gathering since it is recited at a high-level of generality where the subsequent steps rely on the obtained data in a generic manner (see MPEP 2106.05(g)). The “produce control signals” recites the idea of an outcome (i.e., controlling based on updated process model parameters) and therefore amounts to reciting the words “apply it”. The claim is directed to an abstract idea.
At Step 2B the claim does not recite additional elements that, alone or in an ordered combination, are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract idea into a practical application, the recited “general purpose computer system” amounts to no more than mere instructions to apply the judicial exception using generic computer components. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Further, the recited “obtain” amount(s) to well-understood, routine, and convention activity since it reasonably encompasses using any electronic means to obtain models or sensor data (see MPEP 2106.05(d)(II) “i. Receiving or transmitting data over a network”). The “produce control signals” amounts to reciting the words “apply it”. Considering the additional elements in combination does not add anything more than when considering them individually since the “obtain” and “produce control signals” requires no more than generic computer functions. For at least these reasons, the claim is not patent eligible.
Dependent claim 12 – 13 and 15 recite(s) at Step 1 the same statutory category as the parent claim(s), and further recite(s) that the system is configured to: In claim 12 generate an estimate of a gradient related to one or more simulated states with respect to at least one loss function representing an error between i) the model-based simulated system of pumps or hydropower stations and associated reservoirs and ii) a real-world representation of the system of pumps or hydropower stations and associated reservoirs at least partly based on the technical sensor data; In claim 13 simulate the control process performed by said control system based on the fully or partially acausal modular parameterized control model and a corresponding system of differential equations; apply reverse-mode automatic differentiation with respect to the system of differential equations to generate an evaluation estimate of the parameterized control model; and update at least one control parameter of the fully or partially acausal modular parameterized control model based on the evaluation estimate of the parameterized control model. At Step 2A, Prong I the recited limitations, alone or in combination, amount to steps that, under its broadest reasonable interpretation, cover mathematical concepts (see MPEP 2106.04(a)(2)(I)). For example, the “generating an estimate of a gradient” uses a loss function. The “simulate” explicitly relies on a parameterized process model (i.e. a function) having corresponding differential equations. The “apply reverse-mode automatic differentiation” recites performing a specific mathematical operation. The “update” amounts to adjusting a parameter in the parameterized process model using a specific update technique. Accordingly, the claim(s) recite(s) an abstract idea.
At Step 2A, Prong II this judicial exception is not integrated into a practical application since the claimed invention further claims: In claim 13 obtain a fully or partially acausal modular parameterized control model, also referred to as a parameterized control model, of a control process performed by said control system controlling at least part of the system of pumps or hydropower stations and associated reservoirs, including one or more control parameters of the fully or partially acausal modular parameterized control model, wherein the fully or partially acausal modular parameterized control model is defined for interaction with at least part of the parameterized process model of the system of pumps or hydropower stations and associated reservoirs as optimized by said at least one updated process model parameter, wherein the control process is at least partly modeled by one or more neural networks used as universal function approximator(s); wherein said control system is configured to use at least part of the updated control parameter or parameters of the parameterized control model as a basis for controlling the system of pumps or hydropower stations and associated reservoirs; and In claim 15 wherein the system comprises processing circuitry and memory, wherein the memory comprises instructions, which, when executed by the processing circuitry, causes the processing circuitry to evaluate and/or adapt said at least one technical model related to the system of pumps or hydropower stations and associated reservoirs of said system of pumps or hydropower stations and associated reservoirs. The “obtain” amounts to insignificant data gathering since it is recited at a high-level of generality where the subsequent steps rely on the obtained data in a generic manner (see MPEP 2106.05(g)). The “control system is configured to use” amounts to reciting the words “apply it” since it recites the idea of an outcome (i.e., a control system configured to control using specific parameters as a basis for controlling). The “processing circuitry and memory” is recited at a high-level of generality such that it amounts to no more than mere application of the judicial exception using generic computer components which does not amount to an improvement in computer functionality (see MPEP 2106.04(a)(I)). The claim is directed to an abstract idea.
At Step 2B the claim(s) do not recite additional elements that, alone or in an ordered combination, are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract idea into a practical application, the “control system is configured to use” amount to reciting the words “apply it”. The recited “processors” amounts to no more than mere instructions to apply the judicial exception using generic computer components. The additional elements do not amount to a particular machine (see MPEP 2106.05(b)(I)). Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The recited “obtain” amount(s) to well-understood, routine, and convention activity since it reasonably encompasses using any electronic means to obtain models or sensor data (see MPEP 2106.05(d)(II) “i. Receiving or transmitting data over a network”). Considering the additional elements in combination does not add anything more than when considering them individually since the “transfer” and “control system is configured to use” and “obtain” requires no more than generic computer functions. For at least these reasons, the claim is not patent eligible.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
Determining the scope and contents of the prior art.
Ascertaining the differences between the prior art and the claims at issue.
Resolving the level of ordinary skill in the pertinent art.
Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 11 – 13, 15 and 22 are rejected under 35 U.S.C. 103 as being unpatentable over Andersson et al. “Integration of CasADi and JModelica.org” (henceforth “Andersson”) in view of Sayyar-Rodsari et al. (US 2015/0185717) (henceforth “Sayyar-Rodsari (717)”), and further in view of Baydin et al. “Automatic Differentiation in Machine Learning: a Survey” (henceforth “Baydin”), and further in view of Coelho, B. “Energy Efficiency of Water Supply Systems Using Optimisation Techniques and Micro-Hydroturbines” (henceforth “Coelho (Thesis)”). Andersson and Sayyar-Rodsari (717) and Baydin and Coelho (Thesis) are analogous art because they solve the same problem of simulating a control model, and because they are from the same field of endeavor of control simulation.
With regard to claim 11, Andersson teaches a computer-implemented system for an industrial and/or technical system, where said system is configured to model a control problem of an industrial and/or technical process to be performed by said industrial and/or technical system, wherein said system is configured to (Andersson Page 9, Left a general purpose computer with memory “All the calculations have been performed on an Dell Latitude E6400 laptop with an Intel Core Duo processor of 2.4 GHz, 4 GB of RAM, 3072 KB of L2 Cache and 128 kB if L1 cache, running Linux”)
obtain said at least one technical model, being a fully or partially acausal modular parameterized process model of the industrial and/or technical process, also referred to as a parameterized process model, including one or more process model parameters of the fully or partially acausal modular parameterized process model, (Andersson Page 4, Right “In addition, Modelica supports acausal modeling, enabling explicit modeling of physical interfaces … Optimica adds to Modelica a small number of constructs for expressing cost functions, constraints, and what parameters and controls to optimize.”)
simulate the dynamics of one or more states of the industrial and/or technical process over time based on the fully or partially acausal modular parameterized process model and a corresponding system of differential equations, and (Andersson Page 4, Right “It is worth noticing that both differential and algebraic equations are supported and that there is no need, for the user, to solve the model equations for the derivatives, which is common in block-based modeling frameworks”, and Figure 3 show time-domain simulation results, and Andersson Page 2, Left “High-level modeling frameworks such as Modelica are becoming increasingly used in industrial applications. Existing modeling languages enable users to rapidly develop complex large-scale models”)
apply reverse-mode automatic differentiation with respect to the system of differential equations and generate an estimate representing an evaluation of the parameterized process model of the industrial and/or technical process in a manner that does not also generate the substantially full set of process gains with respect to process inputs, (Andersson Page 5, Left “CasADi is a minimalistic computer algebra system implementing automatic differentiation, AD (see [22]) in forward and adjoint modes by means of a hybrid symbolic/numeric approach,”, and Page 5, Right an automatic differentiation on a computation graph would directly compute the derivatives of any output versus any input for said output (generate an evaluate estimate representing an evaluation) “Whereas conventional AD tools are designed to be applied black-box to C or Fortran code, CasADi allows the user to build up symbolic representations of functions in the form of computational graphs, and then apply the automatic differentiation code to the graph directly.”)
and the said system is configured to generate the evaluation estimate at least partly based on obtained data, (Andersson Page 5, Right the computation graph desirably computes the change in any output from any input(s), and would include all the related values in the model)
Andersson does not appear to explicitly disclose: that the system is a computer-implemented control system for an industrial and/or technical system, wherein said control system is configured to control an industrial and/or technical process to be performed by said industrial and/or technical system; wherein the fully or partially acausal modular parameterized process model is defined such that the industrial and/or technical process is at least partly modeled by one or more neural networks used as universal function approximator(s); wherein said control system is configured to obtain technical sensor data representing one or more states of the industrial and/or technical process at one or more time instances; wherein said control system is configured to update at least one process model parameter of the fully or partially acausal modular parameterized process model of the industrial and/or technical process based on the generated evaluation estimate and based on a gradient-descent or ascent procedure, and wherein said control system is configured to produce controls signals that control the operation of the industrial and/or technical process based on the updated process model parameter or parameters.
However, Sayyar-Rodsari (717) teaches:
a computer-implemented control system for an industrial and/or technical system, wherein said control system is configured to control an industrial and/or technical process to be performed by said industrial and/or technical system; (Sayyar-Rodsari (717) Paragraph 9 “the computer system 102 executes software programs (including computer based predictive models) that receive process data 106 from the process 104 and generate optimized decisions and/or actions, which may then be applied to the process 104 to improve operations based on specified goals and objectives.”)
a modular parameterized process model is defined such that an industrial and/or technical process is at least partly modeled by one or more neural networks used as universal function approximator(s) (Sayyar-Rodsari (717) Figure 5A and Paragraph 97 a physical process is modeled using a parameterized sub-block (physical sub-model) and a non-linear approximator (neural network sub-model)
PNG
media_image3.png
411
578
media_image3.png
Greyscale
)
wherein said control system is configured to obtain technical sensor data representing one or more states of the industrial and/or technical process at one or more time instances, (Sayyar-Rodsari (717) Paragraph 58 “In various embodiments, the initialization may be performed by a human expert, and expert system, or via a systematic methodology of identifying the initial conditions of the model given available current and past measurements from the physical process, among others”)
wherein said control system is configured to update at least one process model parameter of a modular parameterized process model of the industrial and/or technical process, wherein said control system is configured to produce controls signals that control the operation of the industrial and/or technical process based on the updated process model parameter or parameters (Sayyar-Rodsari (717) Figure 5B and Paragraph 109 the process model is used inside of a controller for determining control actions with the optimized process model)
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process disclosed by Andersson with the universal approximator model coupled to a parameterized model as part of a control system disclosed by Sayyar-Rodsari (717). One of ordinary skill in the art would have been motivated to make this modification in order improve the modeling process of industrial automation system (Sayyar-Rodsari (717) Paragraph 43).
Andersson in view of Sayyar-Rodsari (717) does not appear to explicitly disclose: update at least one process model parameter of the fully or partially acausal modular parameterized process model of the industrial and/or technical process based on the generated evaluation estimate and based on a gradient-descent or ascent procedure that is performed without calculating a Hessian matrix.
However, Baydin teaches:
update at least one process model parameter based on the generated evaluation estimate and based on a gradient-descent or ascent procedure that is performed without calculating a Hessian matrix (Baydin Page 20 Paragraph 71 reverse-mode automatic differentiation is used to perform gradient descent “performing training on a corpus of pre-tracked video using an adaptive step size gradient descent with reverse mode AD.”, and Figure 1 gradient-descent only requires first derivative, therefore, it would not require Hessian matrix which represents the second derivative
PNG
media_image4.png
136
682
media_image4.png
Greyscale
)
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process having a universal approximator disclosed by Andersson in view of Sayyar-Rodsari (717) with the use of reverse-mode differentiation and gradient descent to update parameters in a model disclosed by Baydin. One of ordinary skill in the art would have been motivated to make this modification in order to decrease the cost of computing the gradient (Baydin Page 12 “An important advantage of the reverse mode is that it is significantly less costly to evaluate (in terms of operation count) than the forward mode for functions with a large number of inputs.”)
Andersson in view of Sayyar-Rodsari (717), and further in view of Baydin does not appear to explicitly disclose: that the control system is for a system of pumps or hydropower stations and associated reservoirs, wherein said control system is configured to control the system of pumps and/or hydropower stations and associated reservoirs by optimizing against a control obiective that includes electricity costs and penalties for exceeding minimum and/or maximum reservoir levels; that the industrial and/or technical system/process is system of pumps or hydropower stations and associated reservoirs.
However, Coelho (Thesis) teaches:
a control system of pumps or hydropower stations and associated reservoirs; (Coelho (Thesis) Page 42, Middle “In these cases, pumps are only switched on when the reservoirs responsible for supply certain populations are empty (or in the minimum level) and switched off when the same reservoirs reach the maximum level allowable”)
wherein said control the system of pumps and/or hydropower stations and associated reservoirs by optimizing against a control obiective that includes electricity costs and penalties for exceeding minimum and/or maximum reservoir levels (Coelho (Thesis) Page 43, Bottom constraints are modeled using penalties “State and boundary constraints were considered by the authors and penalty functions were used in the case of constraints violation.”, and Page 46, Bottom to Page 47, Top “The developed methodology considers a single objective function which includes not only electricity cost of pumping but also the water production cost. Constraints to the tanks water levels, continuity, velocity limits of the variable-speed pumps and also to the number of pump switches were considered”)
industrial and/or technical process model comprises a system of pumps or hydropower stations and associated reservoirs (Coelho (Thesis) Page 79, Bottom water network is explicitly modeled and then optimized as part of a control application, where Andersson teaches a modeling tool usable for rapidly developing models for industrial applications (see Andersson Page 2, Left) “Water supply and distribution networks are characterised by interconnected hydraulic elements including pipes, junctions, tanks, reservoirs, pumps and valves. … According to the computational models representation)
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process disclosed by Andersson in view of Sayyar-Rodsari (717), and further in view of Baydin with the computational model and control application for pumps and tanks in a water distribution system disclosed by Coelho (Thesis). One of ordinary skill in the art would have been motivated to make this modification in order optimally control a water distribution system (see Coelho (Thesis) Page 4, Top).
With regard to claim 12, Andersson in view of Sayyar-Rodsari (717), and further in view of Baydin, and further in view of Coelho (Thesis) teaches all the elements of the parent claim 11, and further teaches: wherein said control system is configured to (Andersson Page 9, Left a general purpose computer with memory)
generate an estimate of a gradient related to one or more simulated states (Andersson Page 5, Left “CasADi is a minimalistic computer algebra system implementing automatic differentiation, AD (see [22]) in forward and adjoint modes by means of a hybrid symbolic/numeric approach,”)
with respect to at least one loss function representing an error between i) the model-based simulated system of pumps or hydropower stations and associated reservoirs and ii) a real-world representation of the system of pumps or hydropower stations and associated reservoirs at least partly based on the technical sensor data. (Sayyar-Rodsari (717) Figure 5A and Paragraph 97 model error should be reduced, which could be added to the modeling in Andersson having wholly predictable results
PNG
media_image3.png
411
578
media_image3.png
Greyscale
)
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process disclosed by Andersson with the universal approximator model coupled to a parameterized model as part of a control system disclosed by Sayyar-Rodsari (717). One of ordinary skill in the art would have been motivated to make this modification in order improve the modeling process of industrial automation system (Sayyar-Rodsari (717) Paragraph 43).
With regard to claim 13, Andersson in view of Sayyar-Rodsari (717), and further in view of Baydin, and further in view of Coelho (Thesis) teaches all the elements of the parent claim 11, and further teaches: wherein said control system is configured to (Andersson Page 9, Left a general purpose computer with memory)
obtain a fully or partially acausal modular parameterized control model, also referred to as a parameterized control model, of a control process performed by said control system controlling at least part of the system of pumps or hydropower stations and associated reservoirs, including one or more control parameters of the fully or partially acausal modular parameterized control model, (Andersson Figure 5 a combined-cycle plant model comprising an PI controller (a parameterized control model), and Page 3, Left a controller can be simulated (including one or more control parameters) “A common formulation is the optimal control problem (OCP) based on differential-algebraic equations (DAE) on the form”)
wherein the fully or partially acausal modular parameterized control model is defined for interaction with at least part of the parameterized process model of the system of pumps or hydropower stations and associated reservoirs as optimized by said at least one updated process model parameter, (Sayyar-Rodsari (717) Figure 5B and Paragraph 109 the process model is used inside of a controller for determining control actions with the optimized process model “As is well known in the art of optimization, the optimizer 506 and PUNDA model 506 operate in an iterative manner to generate an optimal set of MVs as controller output 515. In other words, in a preferred embodiment, the controller output 515 is the final iterate of the trial model input 508.”
PNG
media_image5.png
416
629
media_image5.png
Greyscale
)
wherein the control process is at least partly modeled by one or more neural networks used as universal function approximator(s); (Sayyar-Rodsari (717) Paragraph 109 the process model is explicitly part of the control process and contains the neural network)
simulate the control process performed by said control system based on the fully or partially acausal modular parameterized control model and a corresponding system of differential equations; (Andersson Figure 5 shows a complete plant model having both the process components and a control component, all of which are simulated in CasADi)
apply reverse-mode automatic differentiation with respect to the system of differential equations to generate an evaluation estimate of the parameterized control model; and (Andersson Page 5 a computational graph relates all the parameters to the variables they depend on which includes the control objective “CasADi is a minimalistic computer algebra system implementing automatic differentiation, AD (see [22]) in forward and adjoint modes by means of a hybrid symbolic/numeric approach, … Whereas conventional AD tools are designed to be applied black-box to C or Fortran code, CasADi allows the user to build up symbolic representations of functions in the form of computational graphs, and then apply the automatic differentiation code to the graph directly.”)
update at least one control parameter of the fully or partially acausal modular parameterized control model based on the evaluation estimate of the parameterized control model. (Baydin Page 20 Paragraph 71 reverse-mode automatic differentiation is used to perform gradient descent “performing training on a corpus of pre-tracked video using an adaptive step size gradient descent with reverse mode AD.”)
use at least part of the updated control parameter or parameters of the parameterized control model as a basis for controlling the system of pumps or hydropower stations and associated reservoirs. (Sayyar-Rodsari (717) Figure 5B and Paragraph 109 the process model is used inside of a controller for determining control actions with the optimized process model (updated control parameters as a basis for controlling))
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process disclosed by Andersson with the universal approximator model coupled to a parameterized model as part of a control system disclosed by Sayyar-Rodsari (717). One of ordinary skill in the art would have been motivated to make this modification in order improve the modeling process of industrial automation system (Sayyar-Rodsari (717) Paragraph 43).
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process having a universal approximator disclosed by Andersson in view of Sayyar-Rodsari (717) with the use of reverse-mode differentiation and gradient descent to update parameters in a model disclosed by Baydin. One of ordinary skill in the art would have been motivated to make this modification in order to decrease the cost of computing the gradient (Baydin Page 12 “An important advantage of the reverse mode is that it is significantly less costly to evaluate (in terms of operation count) than the forward mode for functions with a large number of inputs.”)
With regard to claim 15, Andersson in view of Sayyar-Rodsari (717), and further in view of Baydin, and further in view of Coelho (Thesis) teaches all the elements of the parent claim 11, and further teaches:
wherein said control system comprises processing circuitry and memory, wherein the memory comprises instructions, which, when executed by the processing circuitry, causes the processing circuitry to control the system of pumps or hydropower stations and associated reservoirs of said system of pumps or hydropower stations and associated reservoirs. (Sayyar-Rodsari (717) Paragraph 9 “the computer system 102 executes software programs (including computer based predictive models) that receive process data 106 from the process 104 and generate optimized decisions and/or actions, which may then be applied to the process 104 to improve operations based on specified goals and objectives.”)
It would have been obvious to one of ordinary skill in the art to combine the software for modeling an industrial of technical process disclosed by Andersson with the universal approximator model coupled to a parameterized model as part of a control system disclosed by Sayyar-Rodsari (717). One of ordinary skill in the art would have been motivated to make this modification in order improve the modeling process of industrial automation system (Sayyar-Rodsari (717) Paragraph 43).
Examiner General Comments
With regard to the prior art rejection(s), any cited portion of the relied upon reference(s), either by pointing to specific sections or as quotations, is intended to be interpreted in the context of the reference(s) as a whole as would be understood by one of ordinary skill in the art. Although the specified citations are representative of the teachings in the at and are applied to the specific limitations within the individual claim, other passages and figures may apply as well. It is respectfully requested that, in preparing responses, the applicant fully consider the references in their entirety as potentially teaching all or part of the claimed invention since the entire reference is considered to provide disclosure relating to the cited portions. Further, the claims and only the claims form the metes and bounds of the invention. Office personnel are to give the claims their broadest reasonable interpretation in light of the supporting disclosure. Unclaimed limitations appearing in the specification are not read into the claim. Prior art was referenced using terminology familiar to one of ordinary skill in the art. Such an approach is broad in concept and can be either explicit or implicit in meaning. xaminer's Notes are provided with the cited references to assist the applicant to better understand how the examiner interprets the applied prior art. Such comments are entirely consistent with the intent and spirit of compact prosecution.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ALFRED H. WECHSELBERGER whose telephone number is (571)272-8988. The examiner can normally be reached M - F, 10am to 6pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Emerson Puente can be reached at 571-272-3652. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/ALFRED H. WECHSELBERGER/ExaminerArt Unit 2187
/EMERSON C PUENTE/Supervisory Patent Examiner, Art Unit 2187