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 .
Specification
The disclosure is objected to because of the following informalities:
In ¶ [10] and ¶ [51] the standard neural network term “Dropout” is misspelled as both “Drodot” and “Dropot”. Appropriate correction is required.
Claim Objections
Claims 2-3 are objected to because of the following informalities:
Claim 2 recites “Res-long short-term memory (Res-LSTM)”. For the purpose of examination, we consider “Res” as “Residual”.
Claim 3 recites that three of the four LSTM blocks “have a same structure and each comprise an LSTM unit …”. To be grammatically correct the phrase “each comprise” should read “each comprises”.
Appropriate correction is required.
Claim Rejections - 35 USC § 112
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 1-9 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as failing to set forth 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.
Claim 1: The claim recites both “state data” and “acquired ancient building data”, but it is unclear whether these terms refer to the same data or to different types of data. Additionally, the claim recites that the service platform “is of a cluster system” and that the expert module “is of a neural network model.” The phrase “is of” is unclear and fails to specify the structural relationship between the recited elements. For purposes of this examination, the office interprets “state data” as the intended terminology and interprets the phrase “is of” as “comprises”.
Claim 3 recites “Drodot layer”. This appears to be a typographical error. The claim specifies ratios (0.2, 0.2, 0.4) for this Drodot layer. Because this error in the component itself, the ratios and the overall model structure are indefinite. For examination purposes it will be assumed that Drodot layer is “Dropout layer” as shown in Fig. 2.
The phrase “is of” is unclear and fails to specify the structural relationship between the recited elements. For purposes of this examination, the office interprets “Dropout layer” as the intended terminology and interprets the phrase “is of” as “comprises”.
Claim 5: The variables in loss function only use a single subscript i. It is unclear how the inner summation operates since the variables in the summation do not depend on the summation index. For the purpose of our examination, we consider the correct form of this formula:
L
y
ˇ
,
y
=
-
1
n
∑
i
=
1
n
∑
j
=
1
k
y
i
,
j
log
y
ˇ
i
,
j
Claims 2-9 are rejected based on their dependency from claim 1. Claims 4, 5 and 8 are reject also based on their dependency from claim 3. Additionally claim 8 is rejected for dependency from claim 5.
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 1, 7 and 9 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite an abstract idea as discussed below. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception for the reasons discussed below.
Step 1 of the 2019 Guidance requires the examiner to determine if the claims are to one of the statutory categories of invention. Applied to the present application, the claims belong to one of the statutory classes of a process or product as a computer implemented method or a computer system/product.
Claims 1-9 are reproduced below with the abstract idea underlined.
A monitoring and early warning system based on Internet of Things for an ancient building, comprising an information acquisition system, a service platform and a user side, wherein the information acquisition system is configured to acquire state data of the ancient building and upload the state data to the service platform by a 4G/5G gateway; the user side is integrated in a visualization device for a user to manage, analyze and interact with acquired ancient building data, and the user side comprises a monitoring module, a pre-alarm module, a management module and an expert module; the service platform comprises a cluster system integrating a plurality of applications, caches and database servers; and the expert module comprises a neural network model for evaluating a health state of the ancient building
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 1, wherein the information acquisition system comprises a vibration sensor and a crack sensor, and the neural network model is a Res-long short-term memory (Res-LSTM) neural network model for evaluating a crack state of the ancient building.
The monitoring and early warning system based on Internet of Things for the ancient building according to claim 2, wherein the Res-LSTM neural network model comprises an input layer, a first LSTM block, a second LSTM block, a third LSTM block, a fourth LSTM block, a generic average pooling (GAP) layer and a SoftMax classifier, wherein the input layer comprises multidimensional vibration data acquired by a plurality of acceleration sensors; three of the four LSTM blocks have a same structure and each comprise an LSTM unit, a batch normalization (BN) layer, a rectified linear unit (ReLU) layer and a Dropout layer, and ratios of the Dropout layer to the LSTM blocks are 0.2, 0.2 and 0.4 respectively; the fourth LSTM block is comprises of an LSTM unit and a BN layer, and the first LSTM block, the second LSTM block, the third LSTM block and the fourth LSTM block jointly form a residual structure..
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 3, wherein when the Res-LSTM neural network model is trained, a ReLU is used as an activation function, and a formula of the activation function is: f(x)=max(0, x).
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 4, wherein in a training process of the Res-LSTM neural network model, a classification cross entropy is used as a loss function of the Res-LSTM neural network model, and a calculation formula of the loss function is
L
y
ˇ
,
y
=
-
1
n
∑
i
=
1
n
∑
j
=
1
k
y
i
log
y
ˇ
i
wherein
y
i
j
and
y
ˇ
i
j
are a real output and a predicted output respectively, and n and k are a total number of training samples and a number of classes respectively.
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 1, wherein the information acquisition system further comprises a camera, an acceleration sensor, a tilt sensor, a temperature and humidity sensor, a displacement sensor, and a wind speed and direction sensor.
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 1, wherein the management module comprises an ancient building information management portion, an information acquisition system management portion and a data pre-alarm management portion, wherein the ancient building information management portion is used for adding or modifying basic information of the ancient building, the information acquisition system management portion is used for adding or modifying information of the information acquisition system, and the data pre-alarm management portion is used for counting all pre-alarm data of the information acquisition system, wherein the pre-alarm data comprises time and a frequency.
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 5, wherein the crack state of the ancient building is evaluated by using the Res-LSTM neural network model through steps of: S1, data query: querying original monitoring data of the crack sensor from the service platform by a user; S2, data preprocessing: performing abnormal data elimination, missing data complementation and data smoothing processing on the original monitoring data obtained in step S1 in sequence to obtain a data set; and S3, neural network model evaluation: inputting the data set obtained in step S2 into the Res-LSTM neural network model to obtain a final evaluation value.
The monitoring and early warning system based on the Internet of Things for the ancient building according to claim 1, wherein the expert module uses an independent server to run background services of the expert module.
Claims 1-9 are directed to a system, which falls within one of the statutory categories of invention (machine) under 35 U.S.C. §101.
Claims 1–9 recite limitations directed to collecting monitoring data relating to an ancient building, analyzing the data using a neural network model to evaluate a crack state, and generating or managing warning information. These limitations involve observing, evaluating, and processing information, which constitute mental processes.
Additionally, several claims (e.g., Claims 2–6) recite the use of a Res-long short-term memory (Res-LSTM) neural network model to analyze the monitoring data. A neural network model represents a mathematical algorithm used to process data sequences, and therefore recites a mathematical concept.
Accordingly, claims 1–9 recite a judicial exception in the form of mental processes and mathematical concepts.
In Step 2A prong 2: examiner needs to determine if the claim(s) recite additional elements that integrate the exception into a practical application of the exception.
The additional elements in the claim have been left in normal font. Claims 1, 7, and 9 do not integrate the judicial exception into a practical application because of the following reasons:
Claim 1: the additional elements represent generic computing and networking components used for their normal functions of collecting, transmitting, processing, and displaying data. The claim does not recite a specific technological improvement to the functioning of the computer, networking components, or monitoring technology. Accordingly, the additional elements merely implement the abstract idea using generic components.
Claim 7: the additional elements in this claim represent data management and record-keeping operations performed on collected monitoring data. Such operations only represent the automation of human activity on a computer and do not provide a technological improvement to the computer’s functionality.
Claim 9: Using an independent server for evaluation process is a well-understood computing technique. Accordingly, the additional elements merely implement the abstract idea using generic components.
Similarly, claims 1, 7 and 9 also fail Step 2B analysis. Claims 1, 7 and 9 lack an inventive concept that is significantly more than abstract idea. They implement the abstract idea of structural data analysis using routine computer components, such as 4G/5G gateways and cluster servers, performing their conventional functions of data transport and storage. Because the administrative tasks of Claim 7 follow well understood workflows, the limitation fails to provide a technical improvement beyond a mere field-of-use restriction to ancient buildings.
Note regarding claims 2-6 and 8. Claims 2 and 6: The additional elements recited in these claims are used in conjunction with the recited NN model to collect specific real-world structural and environmental data related to the health state of the ancient building that is processed to evaluate the condition of a physical structure. Accordingly, the additional elements impose a meaningful limit on the judicial exception and integrate it into a practical application. Moreover, claim 3, which depends from claim 2, claim 4, which depends from claim 3, claim 5, which depends from claim 2, and claim 8, which depends from claim 5, each incorporate all the limitations of their respective base claims. Accordingly, claims 3–5 and 8 also recite the additional elements that integrate the judicial exception into a practical application.
Claim Rejections - 35 USC § 102
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 1 and 7 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by CN 115146230A, hereinafter “Feng”, see English translation.
Regarding claim 1 Feng discloses a monitoring and early warning system (abstract, page 1 and “early warning subsystem” page 4) based on Internet of Things (page 13, last ¶) for an ancient building (abstract, page 1), comprising an information acquisition system (sensing subsystem 110 and data acquisition and transmission subsystem 120, fig 1, page 7), a service platform (health monitoring platform including block 130 fig 1 and page 9) and a user side (health monitoring platform (block 130 and 140 fig. 1 and page 13) (Browser/Server (BS) architecture), page 12), wherein the information acquisition system (block 110 and 120, fig 1, page 7) is configured to acquire state data of the ancient building and upload the state data to the service platform (last ¶ page 8) by a 4G/5G gateway (using wireless communication technology (4G, NB-IOT/5G) to finish the data exchange, page 8); the user side is integrated in a visualization device (data display module and WebGIS technology, items 2 and 3, page 13) for a user to manage, analyze and interact with acquired ancient building data (page 13) , and the user side comprises a monitoring module (data display module, item (3) page 13), a pre-alarm module (pre-warning management module, item (4) page 13), a management module (system management module and device management module, items (1) & (2) page 13) and an expert module (block 140 fig. 1); the service platform comprises a cluster system (The server may be cloud server of a distributed system or a server combined with a block chain, 2nd ¶ page 17) integrating a plurality of applications (system management, device management, data management, pre-warning management and report management, page 13), caches (data storage and management subsystem (block 130 fig.1 and 5th ¶ page 9) (the term “cache” is interpreted as a type of memory or storage), the system also has RAM and ROM memory page 15) and database servers (the conventional systems have data storage service on physical server 4th ¶ page 9 and cloud database on cloud server 5th ¶ page 9); and the expert module (block 140 fig. 1 (damage positioning and early warning subsystem), ¶ 6 page 3) comprises a neural network model (CNN page 2) for evaluating a health state of the ancient building (¶ 6 page 3) .
Regarding claim 7 Feng further teaches wherein the management module comprises an ancient building information management portion (data management and system management page 4), an information acquisition system management portion (device management, page 12) and a data pre-alarm management portion(pre-warning management, page 12), wherein the ancient building information management portion is used for adding or modifying basic information of the ancient building (page 4), the information acquisition system management portion is used for adding or modifying information of the information acquisition system (device management is configured for managing the sensors, item 2, page 13), and the data pre-alarm management portion is used for counting all pre-alarm data of the information acquisition system (item 4, page 13), including time and a frequency (s203 and s204 fig 2 and page 14).
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 for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or non-obviousness.
Claim 2 is rejected under 35 U.S.C. §103 as being unpatentable over Feng in view of CN 110601193 A, hereinafter “Chen” (see English translation).
Regarding Claim 2 Feng further teaches wherein the information acquisition system (block 110 and 120, page 7) comprises a vibration sensor (MEMS acceleration sensors for collecting structural monitoring data, where acceleration sensors measure structural vibration information, page 7) and a crack sensor (page 7).
Feng discloses that the expert module utilizes a convolutional neural network model to analyze the collected monitoring data and determine structural health conditions of the building.
However, Feng does not teach the neural network model is a Residual-long short-term memory (Res-LSTM) neural network model for evaluating a crack state of the ancient building.
Chen teaches a Res-LSTM prediction module for processing multi-sensor monitoring data, where residual connections are introduced to improve training stability and mitigate gradient vanishing (page 8) when analyzing sequential data.
It would have been obvious to a person having ordinary skill in the art before the
effective filing date of the claimed invention to substitute or upgrade the expert module to the Res-LSTM architecture taught by Chen. Chen explains that its Res-LSTM model increases residual operations specifically to avoid the gradient vanishing problem in processing time-sequence sensor data and this shows substituting Res-LSTM into Feng NN can provide a more robust evaluation of crack states over time.
Therefore, Feng in view of Chen further teaches: the neural network model is a Residual-long short-term memory Res-LSTM neural network model (Chen abstract, page 1) for evaluating a crack state of the ancient building (evaluating a crack state is directly performed in Feng by the crack sensors providing data to the CNN-based damage localization sub-system to generate health monitoring results (page2)).
Claims 3-5 are rejected under 35 U.S.C. §103 as being unpatentable over Feng in view of Chen, US 20210334656 hereinafter Sjogren, US 20210401376 hereinafter Roveda, (Journal of Biomedical Informatics, 113, 103638) hereinafter Ozturk, CN 114778115 A hereinafter Zhang (see English translation).
Regarding claim 3 Feng in view of Chen teaches the monitoring and early warning system based on the Internet of Things for the ancient building according to claim 2;
Chen teaches a Res-LSTM model for processing sequential monitoring data but does not disclose its internal structure. Feng teaches acquiring vibration data from acceleration sensors, which constitutes time-series data. While Feng in view of Chen establishes the high-level functional framework, they do not disclose the detailed internal architecture and layer-wise configuration recited in claim 3.
Specifically, the combination of Feng and Chen does not teach: wherein the Res-LSTM neural network model comprises an input layer, a first LSTM block, a second LSTM block, a third LSTM block, a fourth LSTM block, a generic average pooling (GAP) layer and a SoftMax classifier, wherein the input layer comprises multidimensional vibration data acquired by a plurality of acceleration sensors; three of the four LSTM blocks have a same structure and each comprise an LSTM unit, a batch normalization (BN) layer, a rectified linear unit (ReLU) layer and a Dropout layer, and ratios of the Dropout layer to the LSTM blocks are 0.2, 0.2 and 0.4 respectively; and the fourth LSTM block comprises of an LSTM unit and a BN layer, and the four LSTM blocks jointly form a residual structure.
Sjogren (US 20210334656) teaches a plurality of LSTM blocks, a generic average pooling layer and a SoftMax classifier (an input layer (input sequence fig 2 and fig7 and ¶ [111]), fig4 and fig 7 and ¶ [304], (GAP) layer (¶ [141]) and a SoftMax classifier (¶ [215])). Sjogren also teaches that its neural network model processes any type of data such as sensor data from one or more sensors and specifies that the collected data are multidimensional vectors.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to implement the Res-LSTM model of Chen using known LSTM architectures such as Sjogren and incorporate the multidimensional vector sensor data of Sjogren. Using multidimensional data improves the accuracy and robustness of sensor data analysis, since multidimensional data captures information along multiple axes and provides more comprehensive monitoring. Additionally, using a plurality of sensors allows collection of more comprehensive diverse data which improves monitoring accuracy and reliability. The Sjogren LSTM architecture provides a predictable and reliable framework for processing the data, improving the ability of the model to capture temporal dependencies and generate accurate evaluation.
Feng in view of Chen and Sjogren teaches wherein the Res-LSTM neural network model comprises an input layer (input sequence fig 2 and fig7 and ¶ [111]), a first LSTM block, a second LSTM block, a third LSTM block, a fourth LSTM block (fig4 and fig 7 and ¶ [304]), a generic average pooling (GAP) layer (¶ [141]) and a SoftMax classifier (¶ [215]), wherein the input layer comprises multidimensional vibration data acquired by a plurality of acceleration sensors (Sjogren, the neural network 100 processes any type of data such as sensor data from one or more sensors ¶ [108]. The collected data are multidimensional vectors (¶ [256]).);
Feng in view of Chen and Sjogren discloses that an LSTM layer is comprises of plurality of LSTM blocks but it doesn’t explicitly describe the specific sequence of “BN+ReLU+Dropout” layers inside each block to form the same structure for the three of the four blocks.
Feng in view of Chen and Sjogren doesn’t teach three of the four LSTM blocks have a same structure and each comprise an LSTM unit, a batch normalization (BN) layer, a rectified linear unit (ReLU) layer and a Dropout layer, and ratios of the Dropout layer to the LSTM blocks are 0.2, 0.2 and 0.4 respectively; and the fourth LSTM block comprises of an LSTM unit and a BN layer, and the four LSTM blocks jointly form a residual structure.
Roveda teaches the specific internal architecture and layer stacking of Res-LSTM expert module including the fourth LSTM block.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to arrange the LSTM blocks in a residual structure as taught by Roveda to improve gradient flow and training stability in Feng in view of Chen and Sjogren system.
Feng in view of Chen, Sjogren and Roveda teaches three of the four LSTM blocks have a same structure and each comprise an LSTM unit, a batch normalization (BN) layer, a rectified linear unit (ReLU) layer and a Dropout layer and the fourth LSTM block comprises of an LSTM unit and a BN layer (Roveda teaches a sequence where the model utilizes "batch normalization" followed by a "rectified linear unit (ReLU) activation function" after each layer, and a "dropout layer" to reduce overfitting, fig10A-10B and ¶ [22-23]),
Feng in view of Chen, Sjogren and Roveda further teaches: and ratios of the Dropout layer to the LSTM blocks are 0.2, 0.2 respectively (Sjogren, fig 4).
Feng in view of Chen, Sjogren and Roveda does not teach the dropout ratio of 0.4.
Ozturk discloses a dropout parameter of 0.35 (page 5 section 3.3) specifically for each LSTM block to protect against overfitting and Zhang teaches using a dropout loss rate of 0.5 (page 9) after the BN layer to prevent over-training.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to demonstrate that dropout ratios within a range of 0.2–0.5 are commonly used in LSTM-based neural networks. Therefore, selecting a dropout ratio of 0.4 would have been an obvious matter of routine optimization of a result-effective variable.
Feng in view of Chen, Sjogren, Roveda, Ozturk and Zhang further teaches the ratios of the dropout layer to the LSTM blocks are 0.2, 0.2, 04 respectively.
Feng in view of Chen, Sjogren, Ozturk and Zhang does not teach: and the four LSTM blocks jointly form a residual structure.
Roveda further teaches the four LSTM blocks jointly form a residual structure (Roveda explicitly teaches a residual neural network 220 composed of a sequence of residual units 222 where an adder pointwise adds the outputs of the pathways ¶ [41-42]).
Accordingly, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to implement the residual structure taught by Roveda to stabilize training and reduce overfitting in neural networks and achieve improved training performance.
Feng in view of Chen, Sjogren, Roveda, Ozturk and Zhang further teaches and the first LSTM block, the second LSTM block, the third LSTM block and the fourth LSTM block jointly form a residual structure (Roveda explicitly teaches a residual neural network 220 composed of a sequence of residual units 222 where an adder pointwise adds the outputs of the pathways ¶ [41-42]).
Regarding claim 4, Feng in view of Chen, Sjogren, Roveda, Ozturk and Zhang further teaches wherein when the Res-LSTM neural network model is trained, a ReLU is used as an activation function, and a formula of the activation function is: f(x)=max (0, x). (Ozturk, section 3.1. page 3, 2nd column).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to implement the activation function using standard f(x) = max(0,x), as this is the well-known mathematical definition of ReLU and is routinely used in neural network models to improve training stability and mitigate vanishing gradient.
Regarding claim 5, Feng in view of Chen, Sjogren, Roveda, Ozturk and Zhang further teaches, wherein in a training process of the Res-LSTM neural network model, a classification cross entropy is used as a loss function of the Res-LSTM neural network model, and a calculation formula of the loss function is
L
y
ˇ
,
y
=
-
1
n
∑
i
=
1
n
∑
j
=
1
k
y
i
,
j
log
y
ˇ
i
,
j
wherein
y
i
j
(p(
x
i
j
)) and
y
ˇ
i
j
(q(
Ө
,
x
i
j
)) are a real output and a predicted output respectively, and n (m) and k (K) are a total number of training samples and a number of classes respectively. (Feng page 11 for explanation and ¶ [15] in Chinese document for formula).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to use the classification cross-entropy loss function formula as taught by Feng because classification cross-entropy formula improves optimization and prediction accuracy by providing a more effective measure of prediction error and facilitating efficient gradient-based optimization.
Claim 6 is rejected under 35 U.S.C. §103 as being unpatentable over Feng in view of (IEEE Access, vol. 8, pp. 50131-50135, 2020), hereinafter “Bacco”
Feng further teaches, wherein the information acquisition system further comprises an acceleration sensor (page 7), a tilt sensor (inclination sensor page 7), a temperature and humidity sensor (page 7), a displacement sensor (page 7), and a wind speed and direction sensor (page 7).
Feng does not teach a camera.
Bacco explicitly discloses a monitoring system for ancient buildings that integrates mechanical data with "images and context information acquired by an Unmanned Aerial Vehicle (UAV)". It specifies the use of a "phone camera" or onboard UAV units like a "Canon EOS M" to capture patterns of cracks (page 50132).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to combine the base IoT monitoring framework of Feng with a camera in view of Bacco et al., which teaches that visual data from cameras (often mounted on UAVs) allows operators to "promptly inspect critical structural damage" that fixed sensors might miss. Integrating cameras with mechanical sensors provides a "comprehensive central facility database" that improves diagnostic accuracy.
Claim 8 is rejected under 35 U.S.C. §103 as being unpatentable over Feng in view of Chen, Sjogren, Roveda, Ozturk, Zhang, M. Yu (IEEE Access, vol. 9, 2021, pp. 137406-137411) hereinafter M. Yu and She Tao (CN 113988452A) hereinafter She Tao.
Regarding claim 8, Feng in view of Chen, Sjogren, Roveda, Ozturk, Zhang further teaches, wherein the crack state of the ancient building is evaluated by using the Res-LSTM neural network model through steps of: S1, data query: querying original monitoring data of the crack sensor from the service platform by a user (Feng, a B/S-based monitoring platform, S201 fig2 and page 14); S2, data preprocessing (Feng, S202 and S203 fig2, page14): performing data smoothing processing (Zhang teaches preprocessing collected signal data by performing wavelet threshold denoising, which removes noise components and smooths the signal while preserving essential features, abstract) on the original monitoring data obtained in step S1 in sequence to obtain a data set;
Feng in view of Chen, Sjogren, Roveda, Ozturk, Zhang does not teach performing abnormal data elimination, missing data complementation on the original monitoring data obtained in step S1 in sequence to obtain a data set;
M. Yu teaches preprocessing senor data to eliminate outliers and noise prior to input into a neural network.
She Tao teaches filling time-series data using interpolation (Newton interpolation), which is a known technique for reconstructing incomplete sequential data.
M. Yu teaches performing abnormal data elimination (page 137408), She Tao teaches missing data complementation (abstract) on the original monitoring data obtained in step S1 in sequence to obtain a data set.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to incorporate the preprocessing techniques of M. Yu, She Tao and Zhang into Feng’s system in sequence in order to improve data quality, enhance feature extraction, and increase the accuracy of subsequent neural network analysis, since handling missing data and reducing noise are well-known and routine steps in time-series sensor data processing.
Feng in view of Chen, Sjogren, Roveda, Ozturk, Zhang, M.Yu and She. Tao further teaches: and S3, neural network model evaluation: inputting the data set obtained in step S2 into the Res-LSTM neural network model to obtain a final evaluation value (Feng, S204 fig. 2 page 14).
Claim 9 is rejected under 35 U.S.C. §103 as being unpatentable over Feng in view of Sjogren.
Regarding claim 9, Feng teaches the monitoring and early warning system based on the Internet of Things for the ancient building. Feng does not teach wherein the expert module uses an independent server to run background services of the expert module.
Sjogren teaches implementing a deep neural network on a computational server, where the server performs processing of input data and generates predictions, and it communicates the results to a user client. Sjogren teaches wherein the expert module uses an independent server (¶ [180] and ¶ [227]) to run background services of the expert module.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to implement the expert module of Feng on an independent server as taught by Sjogren in order to enable centralized processing of NN analysis, and efficient communication with user-side interfaces.
Relevant Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Wang (Structural Health Monitoring, 20(4), 1443–1461) discloses a deep residual network framework for structural health monitoring that utilizes identity mappings and skip connections to overcome the vanishing gradient problem common in deep architectures. The model identifies structural damages.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SAEEDE NAFOOSHE whose telephone number is (571)272-8629. The examiner can normally be reached Monday-Friday 8:00 am -5:00pm.
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, Andrew Schecter can be reached at 571-272-2302. 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.
/SAEEDE NAFOOSHE/Examiner, Art Unit 2857
/ANDREW SCHECHTER/Supervisory Patent Examiner, Art Unit 2857