DETAILED ACTIONS
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 12/19/2023. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
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.
Claims 1-13 are rejected under 35 U.S.C. 103 as being unpatentable over Maev et al. (US 2023/0228716 A1, hereinafter Maev) and in view of Zhao Jianqiang (KR 20210078615 A, hereinafter Zhao, an original copy with translated preview is uploaded by the examiner.).
Regarding Claim 1, Maev teaches,
A system comprising:
a data storage system (Maev, Figure 2, Figure 7, [0087] The AI model 129 (a machine learning model) would be stored and operated by computer 20 of FIG. 1”) configured to store a deep neural network (DNN) model (Maev, Figure 9, AI system 144, and 148 [0001], “The present application discloses a process and resultant systems for accurately and comprehensively characterizing ultrasonic signatures from NDE of resistance spot welds in real time, using deep learning. [0134] an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach”) pretrained to:
receive input parameters for a resistance-spot welding (RSW) system (Maev, [0125] the disclosure includes a data management system for storage and manipulation of a collection of ultrasonic signature data from resistance spot welds, corresponding metadata”)
wherein the input parameters have categories comprising base materials, attributes (Maev,[Figure 1, [0052] “ The formation, size and location of the liquid weld nugget 30 is measured by the ultrasound waves and monitored over time by the computer 20.etc”)., coupon geometries, condition, and schedule (Maev,[0128] The training dataset has sufficient coverage of the space of possible weld sheet combinations, weld durations, and weld quality which are observed in practice, so as to yield performant and generalizable mathematical models. Based on observations from industry practices, the training dataset may contain ultrasonic signatures, corresponding metadata, and corresponding weld labels for any number of welds of any conceivable combination of welded sheet thicknesses, number of sheets, weld duration, and weld nugget size. [0134], Some weld information such as welding schedule, welded stack sheet combination, welded materials, etc. may be known a priori. See [0141], and
a computer system configured to: retrieve the pretrained DNN model from the data storage system (Maev [0134] an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach”. Figure 7, [0087] The AI model 129 (a machine learning model) would be stored and operated by computer 20 of FIG. 1”)), access
(i) sets of experimental input parameters (Maev, [0128] Based on observations from industry practices, the training dataset may contain ultrasonic signatures, corresponding metadata”. NOTE: industry practice observation data reads on experimental data) used by the RSW system to produce respective joints (Maev, Figure 1, “sheet 26. 27 and sheet 28”, the three sheets pairs )
and
(ii) sets of experimental joint-performance metrics corresponding to the produced joints (Maev, [0136] an AI system takes as input one or more ultrasonic signatures of a completed resistance spot weld and outputs one or more numerical matrices which contain encoded information relating to the quality of the analyzed weld”)., normalize the experimental input parameters and the experimental joint- performance metrics in a manner expected by the pretrained DNN model retrain the DNN model using the normalized experimental input parameters and the normalized experimental joint-performance metrics (Maev, Figure 8, [[0097] The network takes as input a preprocessed ultrasonic M-scan 136, shown at the bottom of the schematic. The network transforms the input image using a variety of potential operations 137 including but not limited to pooling, convolutions, batch normalization, convolutional attention, and dropout (solid arrows). Figure 9-10 [0134] This preprocessed A-scan 147 from time-step t, which will be used as AI model 148 input, is denoted x,. At some point in time prior to the current weld, an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach.”)
and instruct the data storage system to store the retrained DNN model (Maev, Figure 1, computer 20, [0050] The computer 20 includes at least one processor and electronic storage (i.e. at least one non-transitory computer readable media) for storing data and instructions which when executed by the at least one processor performs the functions”) and
controller circuitry (Maev, Figure 1, control circuit 16) configured to:
receive one or more new input parameters to be included in the input parameters that, when used by the RSW system to join two dissimilar materials, cause the RSW system to produce a new joint having two or more target joint-performance metrics, retrieve, from the data storage system, the retrained DNN model and use it to determine remaining input parameters to be used by the RSW system in conjunction with the new input parameters to produce the new joint having the target joint-performance metrics, and instruct the RSW system to use as input parameters the new input parameters and the determined input parameters to join the two dissimilar materials. (Maev, Figure 9, [0107] “The inference engine 149, at some point in time, loads a mathematical model 148 which was developed using a machine learning approach. The inference engine 149 takes the preprocessed A-scan 147 as input,
pushes the A-scan through the model 148, and outputs raw model output 150. This model output 150 undergoes a postprocessing step such that weld quality or progress 151 can be determined from it. Here, the ultrasonically-measured
properties identified by the AI, including time-domain process events and feature occurrences/positions, are used to compute geometric measurements of the observed welding process and resultant physical weld nugget. This may, for
example, involve checking if model output values reached critical thresholds, transformation of coordinates from image space to the coordinate system of the welded stack, or checks for presence or absence of key features or events.
Finally, weld quality or progress information is then reported outside of the AI system as necessary ( e.g. to the weld gun to control welding or to an interface which commits the quality information to a database for long-term storage). Figure 9, [0134] “Throughout the weld process, the inference engine 149 continues to receive new preprocessed A-scans (x1 , x2 , x3 , ... , xn) and pushes each one into the loaded model to produce corresponding model outputs (Yu y2 , y3 , ... , Yn). Each model output y, is subject to postprocessing, in this case a simple thresholding mechanism that considers an event at index k to have occurred ify,[k]>0.5. Then, the occurrence (or lack thereof) is reported externally”).
Maev teaches predict two or more joint-performance metrics of a joint (Maev, [0136] an AI system takes as input one or more ultrasonic signatures of a completed resistance spot weld and outputs one or more numerical matrices which contain encoded information relating to the quality of the analyzed weld. The data preprocessing pipeline, inference engine, and model output postprocessingpipeline for an AI system for post-process ultrasonic signature characterization may be similar to those mentioned above for in-process characterization”). Maev teaches multiple sheets are steel sheets see (Maev, [0054], figure 1 (26, 27, and 28, The schematic shows weld formation between three sheets of steel (workpieces 26, 27, 28).)
Maev is silent on the that the joint of two dissimilar materials to be produced by the RSW system using the input parameters; and joints of pair-wise dissimilar materials,
However, Zhao teaches joint of two dissimilar materials to be produced by the RSW system using the input parameters; and joints of pair-wise dissimilar materials,
(Zhao, Page 1, top paragraph, “It is an object of the present invention to provide a dissimilar material spot welding method for an aluminum alloy and a galvanized steel sheet that is simple to implement in an automobile and can be used to perform dissimilar material welding between an aluminum alloy and a galvanized steel sheet”).
It would have been obvious to a person of ordinary skill before the effective filing date to modify Maev system to include the well-known dissimilar materials sheet as a second or third sheet as taught by Zhao in order to obtain resistance spot welding parameters for the dissimilar material sheets with the benefits of high quality, highly efficient production, and low cost of spot-welded joints. (Zhao, Page 6, Top paragraph). Using dissimilar material and measuring resistance spot welding parameters for the dissimilar joint is a well know design choice not an inventive step.
Regarding Claim 2, Combination of Maev and Zhao teaches the system of claim 1, Maev further teaches wherein the two materials to be joined by the RSW system comprise one of: an Al alloy and a steel, or a first steel and a second steel (Maev, [0054], figure 1 (26, 27, and 28, The schematic shows weld formation between three sheets of steel (workpieces 26, 27, 28). or a first Al alloy and a second Al alloy.
Regarding Claim 3, Combination of Maev and Zhao teaches the system of claim 1,
Maev wherein the base materials category comprises one or more of thickness parameters material type parameters (Maev, [0055] The corresponding metadata may include but is not limited to: a unique weld signature ID within the data management system, a weld ID assigned by the welding system, the time at which the welding process began, the thickness of each individual sheet involved in the welded stack”), base, or coating parameters,
the attributes category comprises one or more of button size parameters, nugget size parameters IMC parameters, hardness parameters, indentation parameters, or expulsion parameters (Maev, [0141], “can identify five patterns in the ultrasonic images: nugget growth, nugget solidification, whole nugget, nugget top, and nugget bottom. The method and system disclosed herein for post-process characterization can identify the same five patterns, as well as many others including but not limited to discontinuities in the outer interfaces indicative of expulsions or other process non-conformities”), the coupon geometries category comprises dimensions of coupon (Maev, [0107] Here, the ultrasonically-measured properties identified by the AI, including time-domain process events and feature occurrences/positions, are used to compute geometric measurements of the observed welding process and resultant physical weld nugget”).
the condition category comprises one or more of adhesive parameters, baking parameters, aging parameters, or ELPO parameters (Maev, [ 0126] the positions of
the top and bottom of the molten nugget in the ultrasonic signature at any point in time (105, 106), rate of liquid nugget thickness growth (derived from 105, 106, 109, 112), rate of liquid nugget solidification (derived from 105, 1066, 113, 114), exact moment(s) of expulsion event(s), and the degree of electrode indentation into the welded sheets.”)
the schedule (Maev, [0134]”Some weld information such as welding schedule”) category comprises one or more of pre-heating parameters, phase parameters, electrode cap parameters(Maev, [0143] The present disclosure has seen extremely successful production use and is able to identify and measure the desired ultrasonic characteristics accurately and under an extremely wide variety of conditions (weld quality, weld geometry, weld cap quality), , or clamp load parameters (Maev, [0051] clamping force).
Regarding Claim 4, Combination of Maev and Zhao teaches the system of claim 1,
Maev further teaches, wherein the joint-performance metrics comprise a measured peak load (Maev, [0051] clamping force)., a measured extension at break (Maev, [0051] clamping force), and a total energy. (Maev, [0001] An ultrasonic A-scan is a set of voltage measurements of the ultrasonic transducer output to represent the amount of received ultrasonic energy as a function of time”. ([0107] computer 20 of FIG. 1. A data preprocessing system 146 takes as input the ultrasonic A-scan 145 data on which it performs some set of operations (cropping, rescaling, filtering, etc.) such that it is more conducive to AI-based characterization. It outputs a preprocessed version 147 of the input A-scan on which inference is conducted using some inference engine 149”)
Regarding Claim 5, Combination of Maev and Zhao teaches the system of claim 1,
Maev further teaches, wherein the computer system comprises one or more of a personal computer, or a supercomputer system. (Maev, Figure 1, A computer 20)
Regarding Claim 6, Combination of Maev and Zhao teaches the system of claim 1,
Maev further teaches wherein the DNN model is a physics-driven, unified, expandable architecture including an input layer, three or more hidden layers, and an output layer. (Maev, [0134] This preprocessed A-scan 147 from time-step t, which will be used as AI model 148 input, is denoted x,. At some point in time prior to the current weld, an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach”. Figure 7, [0132] “The AI model 129 may take the form of a recurrent neural network, which is a specialized neural architecture for processing sequential inputs because each recurrent layer of the network holds an
internal state which is potentially modified as each timeslice of the input sequence is processed. Initial hidden states 131 of the model's recurrent layers 132 may be initialized to an arbitrary value e.g. zero. The recurrent layers 132 of the network may take the form of ID convolutional long shortterm memory units (LSTM), which are specialized gated units capable of preserving and understanding the spatial relationships within each time-slice of the input sequence”)
Regarding Claim 7, Combination of Maev and Zhao teaches the system of claim 1,
Maev further teaches wherein the DNN model is configured to include one or more input layer neurons (Figure 7, in put layer 130) corresponding to input parameters from each of the following five resistance spot weld input parameter categories (Maev, Figure 7, 0088] Inputs 130 are shown at the bottom of the schematic- preprocessed ultrasonic A-scans x1 ... xn -and data flows through the model bottom-to-top (with subsequent model layers) and left-to-right (with each time-slice of the input sequence). Circles represent model states for the various layers of the model, and arrows represent various operations on the model state data which transform the internal states, ultimately influencing the model's outputs. The final outputs 134 of the model are used to infer weld quality while the weld is unfolding”. : weld schedule (Maev, [0134]”Some weld information such as welding schedule”), weld attributes, base materials(Maev, [0141], “can identify five patterns in the ultrasonic images: nugget growth, nugget solidification, whole nugget, nugget top, and nugget bottom. The method and system disclosed herein for post-process characterization can identify the same five patterns, as well as many others including but not limited to discontinuities in the outer interfaces indicative of expulsions or other process non-conformities”), the coupon geometries (Maev, [0107] Here, the ultrasonically-measured properties identified by the AI, including time-domain process events and feature occurrences/positions, are used to compute geometric measurements of the observed welding process and resultant physical weld nugget”).and weld condition, and one or more output layer neurons corresponding to output parameters from each of the following three resistance spot weld output parameter categories: peak load, extension at break, and total energy. peak load (Maev, [0051] clamping force)., a measured extension at break (Maev, [0051] clamping force). ,and a total energy. (Maev, [0001] An ultrasonic A-scan is a set of voltage measurements of the ultrasonic transducer output to represent the amount of received ultrasonic energy as a function of time”. ([0107] computer 20 of FIG. 1. A data preprocessing system 146 takes as input the ultrasonic A-scan 145 data on which it performs some set of operations ( cropping, rescaling, filtering, etc.) such that it is more conducive to AI-based characterization. It outputs a preprocessed version 147 of the input A-scan on which inference is conducted using some inference engine 149”)
Regarding Claim 8, Combination of Maev and Zhao teaches the system of claim 7,
Maev further teaches wherein the DNN model further includes three or more hidden layers between the input layer and output layer (Maev, Figure 7,”layers a recurrent neural network 129 using convolutional long short-term memory layers with a decision-making layer 133 that produces a corresponding output 134 for every time-step of the sequential input”), and wherein the DNN model further includes a rectified linear unit layer and a dropout layer between each of the hidden layers and before the output layer( Maev, [0097], . The network transforms the input image using a variety of potential operations 137 including but not limited to pooling, convolutions, batch normalization, convolutional attention, and dropout (solid arrows).
Regarding Claim 9, Combination of Maev and Zhao teaches the system of claim 1,
Maev further teaches , wherein the DNN model is configured to include one or more input layer neurons corresponding to input parameters from each of the following four resistance spot weld input parameter categories: weld attributes, base materials, coupon geometry, and weld condition, (Maev, Figure 9-10, (Maev, [0125] the disclosure includes a data management system for storage and manipulation of a collection of ultrasonic signature data from resistance spot welds, corresponding metadata” [Figure 1, [0052] “ The formation, size and location of the liquid weld nugget 30 is measured by the ultrasound waves and monitored over time by the computer 20.etc”) and one or more output layer neurons corresponding to output parameters from each of the following three resistance spot weld output parameter categories: peak load, extension at break, and total energy. (Maev, [0097] The network may contain upsampling operations 139 with concatenation operations 140 ( +) to combine outputs of different layers at different scales, a technique which has generally been shown to improve performance in convolutional neural networks. Output channels 141 (thick dotted arrows) of the network consist of convolutional neural subnetworks, Finally, some potential network output postprocessing 142 may occur (thin dotted arrows), e.g. non-maximum suppression, aggregation of bounding boxes (e.g. nugget boxes and stack outer boxes), bounding box rescaling, etc. Final outputs 143 are used to infer weld quality. Maev, [0051] clamping force)., a measured extension at break (Maev, [0051] clamping force). ,and a total energy. (Maev, [0001] An ultrasonic A-scan is a set of voltage measurements of the ultrasonic transducer output to represent the amount of received ultrasonic energy as a function of time”).
Regarding Claim 10, Combination of Maev and Zhao teaches the system of
claim 1,
Maev further teaches wherein the computer system includes a deep neural network (DNN) training component configured to receive a DNN spot resistance welding test dataset and a DNN spot resistance welding validation dataset, model (Maev, Figure 9, AI system 144, and 148 [0001], “The present application discloses a process and resultant systems for accurately and comprehensively characterizing ultrasonic signatures from NDE of resistance spot welds in real time, using deep learning. the DNN training component configured to train a spot resistance welding DNN machine learning model as a function of the DNN spot resistance welding test dataset, the DNN training component includes a DNN validation component configured to validate the spot resistance welding DNN machine learning model as a function of the DNN validation dataset (Maev [0134] an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach”); and a DNN processing component configured to receive a new spot resistance weld dataset representing spot resistance parameters for generating a spot resistance weld with a spot resistance welding machine, and to process the new spot resistance weld dataset to predict weld quality associated using the validated spot resistance weld DNN machine learning model. (Maev, [0134] During model training, the ultrasonic signature dataset would necessarily also be labelled in this manner so as to allow the model to learn the relationship between some particular patterns in a sequence of A-scans and the occurrence of a given event, using a supervised deep learning approach. Throughout the weld process, the inference engine 149 continues to receive new preprocessed A-scans (x1 , x2 , x3 , ... , xn) and pushes each one into the loaded model to produce corresponding model outputs (Yu y2 , y3 , ... , Yn).
Regarding Claim 11, Combination of Maev and Zhao teaches the system of
claim 1,
Maev further teaches wherein the controller circuitry is configured to use the retrained DNN model to determine remaining input parameters to be used by the RSW system (Maev, [0125] the disclosure includes a data management system for storage and manipulation of a collection of ultrasonic signature data from resistance spot welds, corresponding metadata”)
in conjunction with the new input parameters to produce the new joint having the target joint-performance metrics. (Maev, [0087] the machine learning model is a recurrent neural network 129 using convolutional long short-term memory layers with a decision-making layer 133 that produces a corresponding output 134 for every time-step of the sequential input”).
Regarding Claim 12, Combination of Maev and Zhao teaches the system of
claim 1,
Maev further teaches comprising an RSW system that includes the controller circuitry. (Maev, Figure 1, Weld control 16).
Regarding Claim 13, Combination of Maev and Zhao teaches the system of
claim 1,
Maev further teaches comprising an RSW system that includes the data storage system. (Maev, Figure 1, computer 20)
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 14-20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Maev.
Regarding Claim 14, Maev teaches
A method for determining weld quality, the method including the steps of accessing, in memory with a deep neural network (DNN) processing component, a pretrained resistance spot welding deep neural network (DNN) model, the pretrained resistance spot welding DNN model being configured to predict weld quality of a weld joint between two base materials produced by an RSW system based on a set of resistance-spot welding input parameters; [0005] a method for resistance spot weld fabrication for collection of ultrasonic signature data, corresponding weld metadata, and corresponding ideal evaluations with which to develop mathematical models for characterization of the ultrasonic signatures, and a method for computationally preprocessing ultrasonic signature data such that the data are more conducive to development of an artificial intelligence for automated characterization” (Maev, Figure 9, AI system 144, and 148 [0001], “The present application discloses a process and resultant systems for accurately and comprehensively characterizing ultrasonic signatures from NDE of resistance spot welds in real time, using deep learning):
receiving, from a user interface, one or more target weld performance metrics associated with weld quality; receiving, from a user interface, values for a subset of the set of resistance-spot welding input parameters (Maev, [0125] the disclosure includes a data management system for storage and manipulation of a collection of ultrasonic signature data from resistance spot welds, corresponding metadata”);
iteratively predicting weld quality of a weld joint between two base materials produced by an RSW system with the pretrained DNN model, (Maev [0134] an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach”. Figure 7, [0087] The AI model 129 (a machine learning model) would be stored and operated by computer 20 of FIG. 1”)), using the DNN processing component, by using the received values for the subset of the set of resistance-spot welding input parameters and different values of one or more remaining resistance-spot welding input parameters (Maev, [0136] an AI system takes as input one or more ultrasonic signatures of a completed resistance spot weld and outputs one or more numerical matrices which contain encoded information relating to the quality of the analyzed weld. The data preprocessing pipeline, inference engine, and model output postprocessing pipeline for an AI system for post-process ultrasonic signature characterization may be similar to those mentioned above for in-process characterization”).
to determine values for the one or more remaining resistance-spot welding input parameters where the DNN model predicts weld quality that meets the one or more target weld performance metrics (Maev, Figure 8, [[0097] The network takes as input a preprocessed ultrasonic M-scan 136, shown at the bottom of the schematic. The network transforms the input image using a variety of potential operations 137 including but not limited to pooling, convolutions, batch normalization, convolutional attention, and dropout (solid arrows). Figure 9-10 [0134] This preprocessed A-scan 147 from time-step t, which will be used as AI model 148 input, is denoted x,. At some point in time prior to the current weld, an inference engine loaded a mathematical model 148 which has already been trained for a particular task using a supervised deep learning approach.”);
instructing the RSW system to use as input parameters the received values of the subset of the set of resistance-spot welding input parameters and one of the determined values for the one or more remaining resistance-spot welding input parameters to weld the two base materials. model (Maev, Figure 1, computer 20, [0050] The computer 20 includes at least one processor and electronic storage (i.e. at least one non-transitory computer readable media) for storing data and instructions which when executed by the at least one processor performs the functions”).
Regarding Claim 15, Maev teaches the method of claim 14, Maev further teaches wherein the two materials to be joined by the RSW system comprise one of: an Al alloy and a steel, or a first steel and a second steel (Maev, [0054], figure 1 (26, 27, and 28, The schematic shows weld formation between three sheets of steel (workpieces 26, 27, 28). or a first Al alloy and a second Al alloy.
Regarding Claim 16, Maev teaches the method of claim 14
Maev further wherein the base materials category comprises one or more of thickness parameters material type parameters (Maev, [0055] The corresponding metadata may include but is not limited to: a unique weld signature ID within the data management system, a weld ID assigned by the welding system, the time at which the welding process began, the thickness of each individual sheet involved in the welded stack”), base, or coating parameters
Regarding Claim 17, Maev teaches the method of claim 14
Maev further teaches including receiving, from a user interface, customized ranges for the one or more remaining resistance-spot welding input parameters. (Maev, figure 1)
Regarding Claim 18, Maev teaches the method of claim 14
Maev further teaches including identifying a combination of values of input parameters (Figure 7, in put layer 130) including the received values of the subset of the set of resistance-spot welding input parameters and a range of values for the one or more remaining resistance-spot welding input parameters that the DNN model predicts will cause an RSW system to produce a weld joint with the target weld performance metrics. (Maev, Figure 7, 0088] Inputs 130 are shown at the bottom of the schematic- preprocessed ultrasonic A-scans x1 ... xn -and data flows through the model bottom-to-top (with subsequent model layers) and left-to-right (with each time-slice of the input sequence). Circles represent model states for the various layers of the model, and arrows represent various operations on the model state data which transform the internal states, ultimately influencing the model's outputs. The final outputs 134 of the model are used to infer weld quality while the weld is unfolding” : weld schedule (Maev, [0134]”Some weld information such as welding schedule”), weld attributes, base materials(Maev, [0141], “can identify five patterns in the ultrasonic images: nugget growth, nugget solidification, whole nugget, nugget top, and nugget bottom. The method and system disclosed herein for post-process characterization can identify the same five patterns, as well as many others including but not limited to discontinuities in the outer interfaces indicative of expulsions or other process non-conformities”),
Regarding Claim 19, Maev teaches the method of claim 14
Maev further teaches wherein the DNN model further includes three or more hidden layers between the input layer and output layer (Maev, Figure 7,”layers a recurrent neural network 129 using convolutional long short-term memory layers with a decision-making layer 133 that produces a corresponding output 134 for every time-step of the sequential input”), and wherein the DNN model further includes a rectified linear unit layer and a dropout layer between each of the hidden layers and before the output layer( Maev, [0097], . The network transforms the input image using a variety of potential operations 137 including but not limited to pooling, convolutions, batch normalization, convolutional attention, and dropout (solid arrows
Regarding Claim 20, Maev teaches the method of claim 14, wherein the DNN Maev further teaches , wherein the DNN model is configured to include one or more input layer neurons corresponding to input parameters from each of the following four resistance spot weld input parameter categories: weld attributes, base materials, coupon geometry, and weld condition, (Maev, Figure 9-10, (Maev, [0125] the disclosure includes a data management system for storage and manipulation of a collection of ultrasonic signature data from resistance spot welds, corresponding metadata” [Figure 1, [0052] “ The formation, size and location of the liquid weld nugget 30 is measured by the ultrasound waves and monitored over time by the computer 20.etc”) and one or more output layer neurons corresponding to output parameters from each of the following three resistance spot weld output parameter categories: peak load, extension at break, and total energy. (Maev, [0097] The network may contain up-sampling operations 139 with concatenation operations 140 ( +) to combine outputs of different layers at different scales, a technique which has generally been shown to improve performance in convolutional neural networks. Output channels 141 (thick dotted arrows) of the network consist of convolutional neural subnetworks, Finally, some potential network output postprocessing 142 may occur (thin dotted arrows), e.g. non-maximum suppression, aggregation of bounding boxes (e.g. nugget boxes and stack outer boxes), bounding box rescaling, etc. Final outputs 143 are used to infer weld quality. Maev, [0051] clamping force)., a measured extension at break (Maev, [0051] clamping force). ,and a total energy. (Maev, [0001] An ultrasonic A-scan is a set of voltage measurements of the ultrasonic transducer output to represent the amount of received ultrasonic energy as a function of time”).
Conclusion
Citation of Pertinent Prior Art
The prior art made of record and not relied upon is considered pertinent to
applicant's disclosure.
KURPIEWSKI et al. (US 2022/0324060 A1)recites “System and methods for using analytics and algorithms to predict weld quality are provided and include a computer having a processor and memory configured to receive weld parameter data generated during a welding process by a welder to join at least two parts with a weld, input the received weld parameter data to a data analytics model to generate at least one predicted weld quality parameter, compare the predicted weld quality parameter with a weld quality parameter threshold, and generate output indicating at least one of: the at least one predicted weld quality parameter and a result of the comparison between the at least one predicted weld quality parameter and the weld quality parameter threshold”(Abstract)
KITCHEN et al. (US 2021/0312604 A1) recites “A method inspects weld quality in-situ. The method obtains a plurality of sequenced images of an in-progress welding process and generates a multi-dimensional data input based on the plurality of sequenced images and/or one or more weld process control parameters. The parameters may include: (i) shield gas flow rate, temperature, and pressure; (ii) voltage, amperage, wire feed rate and temperature (if applicable); (iii) part preheat/inter-pass temperature; and (iv) part and weld torch relative velocity). The method generates
defect probability and analytics information by applying one or more computer vision techniques on the multi-dimensional data input. The analytics information includes predictive insights on quality features of the in-progress welding process. The method then generates a 3-D visualization of one or more as-welded regions, based on the analytics information, and the plurality of sequenced images. The 3-D visualization displays the quality features for virtual inspection and/or for determining weld quality (abstract).
Any inquiry concerning this communication or earlier communications from the
examiner should be directed to DILARA SUL TANA whose telephone number is
(571 )272-3861. The examiner can normally be reached Mon-Fri, 9 AM-5:30 PM.
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, EMAN ALKAFAWI can be reached on (571) 272-4448. 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/patentcenter 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.
/DILARA SULTANA/Examiner, Art Unit 2858
05/24/2026
/EMAN A ALKAFAWI/Supervisory Patent Examiner, Art Unit 2858 5/29/2026
Prosecution Insights