Nuclear power plant pipe network state monitoring system and method based on multi-dimensional data fusion and storage medium

Abstract

The invention relates to the technical field of nuclear power plant pipe network state monitoring, in particular to a nuclear power plant pipe network state monitoring system and method based on multi-dimensional data fusion and a storage medium. The system comprises a data acquisition unit; a data synchronization unit; a data preprocessing unit; a feature extraction unit; a multi-dimensional data fusion unit; a deep learning model unit; a state monitoring unit; an alarm unit; and an updating and maintaining unit. The method comprises the steps of deep learning model training: deploying a trained deep learning model to practical application. According to the method, the richness and comprehensiveness of monitoring data are improved, the accuracy and reliability of prediction are improved, and the accuracy and reliability of nuclear power plant pipe network state monitoring are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of nuclear power plant pipe network status monitoring, and in particular to a nuclear power plant pipe network status monitoring system, method and storage medium based on multi-dimensional data fusion. Background Art

[0002] As an important part of nuclear power plants, the safe and stable operation of nuclear power plant pipe network system has a significant impact on the safety and economy of nuclear power plants. However, the traditional nuclear power plant pipe network status monitoring method mainly relies on a single sensor or monitoring means, such as voiceprint sensor, vibration sensor, etc., which has problems such as single monitoring data and limited monitoring range. In addition, traditional monitoring methods often cannot achieve real-time monitoring and prediction of nuclear power plant pipe network status, and it is difficult to meet the high requirements of nuclear power plants for safety and economy.

[0003] The traditional monitoring methods for the status of nuclear power plant pipelines have the following shortcomings: the monitoring data is single. The traditional monitoring methods mainly rely on a single sensor or monitoring method and cannot fully reflect the status of the nuclear power plant pipelines; the monitoring scope is limited. The traditional monitoring methods cannot cover all key parts of the nuclear power plant pipelines, resulting in monitoring blind spots; real-time monitoring and prediction cannot be achieved. The traditional monitoring methods cannot monitor the status of the nuclear power plant pipelines in real time, nor can they predict future status changes, making it difficult to meet the high requirements of nuclear power plants for safety and economy. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a nuclear power plant pipeline network status monitoring system, method and storage medium based on multi-dimensional data fusion, which improves the richness and comprehensiveness of monitoring data, improves the accuracy and reliability of prediction, and improves the accuracy and reliability of nuclear power plant pipeline network status monitoring.

[0005] The present invention provides a nuclear power plant pipe network status monitoring system based on multi-dimensional data fusion, comprising:

[0006] Data acquisition unit, used to collect multi-dimensional data such as monitoring video, voiceprint, and vibration data of the nuclear power plant pipeline network;

[0007] Data synchronization unit ensures that the collection of multi-dimensional data is synchronized in time and the data collection frequency is consistent;

[0008] A data preprocessing unit, which preprocesses the multidimensional data after synchronization processing;

[0009] A feature extraction unit, used to extract visual feature vectors, frequency domain feature vectors and time series feature vectors;

[0010] The multi-dimensional data fusion unit combines the feature vectors of surveillance video data, voiceprint data, and vibration data to form a comprehensive feature vector;

[0011] The deep learning model unit uses the RNN model to train the concatenated feature vectors, learns the differences between the corresponding states of the multi-dimensional data, and classifies the concatenated feature vectors. The network status is defined based on the multi-classification results of the RNN model, and the cross-validation method is used to evaluate the performance of the model.

[0012] The state monitoring unit is used to monitor the state of the power plant network in real time using the deep learning model, process the data collected by the data acquisition unit in real time, and perform state monitoring and abnormal position location according to the output of the deep learning model;

[0013] The alarm unit triggers an alarm when it detects an abnormal pipeline network status, prompting the operator to handle the problem. The update and maintenance unit regularly uses newly collected data to retrain and verify the deep learning model to adapt to changes in the pipeline network status, while ensuring the stable operation of the monitoring system and regularly inspecting and maintaining hardware equipment.

[0014] In a specific embodiment of the present invention, the feature extraction unit,

[0015] For surveillance video data, the CNN network is used to extract features from the preprocessed video frames to generate visual feature vectors of the video frames; for voiceprint data, the CNN network is used to extract features from the preprocessed voiceprint signals to generate frequency domain feature vectors of the voiceprint signals; for vibration data, the LSTM network is used to extract features from the preprocessed vibration signals to generate time series feature vectors of the vibration signals.

[0016] In a specific embodiment of the present invention, the pipe network status includes: normal status and abnormal status.

[0017] The abnormal state includes pipeline fault location positioning and leakage point positioning.

[0018] In a specific embodiment of the present invention, the preprocessing method of the data preprocessing unit is:

[0019] For surveillance video data, the preprocessing methods are cropping, scaling, and normalization.

[0020] For voiceprint or vibration data, the preprocessing method is data cleaning and noise removal.

[0021] In a specific embodiment of the present invention, the data synchronization unit uses a timestamp to ensure that the collection of multi-dimensional data is synchronized in time, sets a unified collection frequency, and ensures that different types of data are aligned on the time axis.

[0022] The present invention provides a method for monitoring the status of a nuclear power plant pipeline network based on multi-dimensional data fusion, comprising the following steps:

[0023] Step 1: Deep learning model training:

[0024] Step 1-1: Collect multi-dimensional data such as monitoring video, voiceprint, and vibration data of the nuclear power plant pipeline network;

[0025] Step 1-2: Perform time synchronization processing on the collected multi-dimensional data;

[0026] Step 1-3: Crop, scale and normalize the collected video frames, and clean and remove noise from the voiceprint and vibration data;

[0027] Step 1-4: Extract visual feature vectors, frequency domain feature vectors and time series feature vectors;

[0028] Step 1-5: Concatenate the feature vectors of video, voiceprint and vibration data;

[0029] Step 1-6: Use the RNN model to train the concatenated feature vectors, learn the differences between the corresponding states of the multi-dimensional data, classify the concatenated feature vectors, define the network status according to the results of the RNN model multi-classification, and use the cross-validation method to evaluate the performance of the model;

[0030] Step 2: Deploy the trained deep learning model to actual applications, including:

[0031] Step 2-1: Real-time collection of multi-dimensional data including surveillance video, voiceprint, and vibration data of the nuclear power plant pipeline network;

[0032] Step 2-2: Perform time synchronization processing on the collected multi-dimensional data;

[0033] Step 2-3: Crop, scale and normalize the collected video frames, and clean and remove noise from the voiceprint and vibration data;

[0034] Step 2-4: Extract visual feature vectors, frequency domain feature vectors and time series feature vectors; concatenate the feature vectors of video, voiceprint and vibration data;

[0035] Step 2-5: Load the trained deep learning model and perform status monitoring and abnormal location positioning based on the model output;

[0036] When the pipe network status is abnormal or the leakage point is located, the alarm is triggered immediately and the operator is notified by means of sound and light alarms;

[0037] Step 2-6: Regularly collect new data, retrain and validate the model, and regularly inspect and maintain hardware equipment.

[0038] In a specific embodiment of the present invention, in steps 1-4, for monitoring video data, a CNN network is used to perform feature extraction on the preprocessed video frames to generate visual feature vectors of the video frames; for voiceprint data, a CNN network is used to perform feature extraction on the preprocessed voiceprint signals to generate frequency domain feature vectors of the voiceprint signals; for vibration data, an LSTM network is used to perform feature extraction on the preprocessed vibration signals to generate time series feature vectors of the vibration signals.

[0039] In a specific embodiment of the present invention, in step 1-2 and step 2-2, the collection of multidimensional data is synchronized in time using a timestamp, and a uniform collection frequency is set to ensure that different types of data are aligned on the time axis.

[0040] The present invention provides an electronic device, including a processor and a memory;

[0041] The memory is used to store programs;

[0042] The processor executes the program to implement the method described in the above technical solution.

[0043] The present invention provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the method described in the above technical solution.

[0044] Compared with the prior art, the nuclear power plant pipeline network status monitoring system, method and storage medium based on multi-dimensional data fusion of the present invention have the following beneficial effects:

[0045] (1) By collecting and processing multi-dimensional data such as surveillance video, voiceprint data, and vibration data, the system can provide more comprehensive and three-dimensional monitoring results, thereby improving the richness and comprehensiveness of monitoring data;

[0046] (2) Using deep learning models, it is possible to automatically learn useful features from complex multidimensional data, thereby improving the accuracy and reliability of predictions;

[0047] (3) It can process sensor data in real time and perform status monitoring and abnormal location positioning based on model output, which helps to timely discover and deal with problems in pipeline network operation and improve the safety and economy of nuclear power plants;

[0048] (4) By regularly retraining and validating the model using newly collected data, the system can adapt to changes in the status of the pipeline network and improve the adaptability and practicality of the model;

[0049] (5) The nuclear power plant pipeline network status monitoring system based on multi-dimensional data fusion can effectively improve the accuracy and reliability of nuclear power plant pipeline network status monitoring, provide strong support for the safe operation of nuclear power plants, and thus improve the safety and economy of nuclear power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Schematic diagram of nuclear power plant pipeline network status monitoring system based on multi-dimensional data fusion. DETAILED DESCRIPTION

[0051] In order to further understand the present invention, the embodiments of the present invention are described below in conjunction with examples, but it should be understood that these descriptions are only for further illustrating the features and advantages of the present invention, rather than limiting the present invention.

[0052] The embodiment of the present invention discloses a nuclear power plant pipe network status monitoring system based on multi-dimensional data fusion, comprising:

[0053] Data acquisition unit, used to collect multi-dimensional data such as monitoring video, voiceprint, and vibration data of the nuclear power plant pipeline network;

[0054] In order to ensure that the data can be collected and transmitted to the data synchronization unit, it is preferred to integrate the data acquisition unit into the monitoring system of the nuclear power plant;

[0055] Data synchronization unit ensures that the collection of multi-dimensional data is synchronized in time and the data collection frequency is consistent;

[0056] The data synchronization unit uses a timestamp to ensure that the collection of multi-dimensional data is synchronized in time, sets a unified collection frequency, and ensures that different types of data are aligned on the time axis.

[0057] A data preprocessing unit, which preprocesses the multidimensional data after synchronization processing;

[0058] For surveillance video data, the preprocessing methods are cropping, scaling, and normalization.

[0059] For voiceprint or vibration data, the preprocessing method is data cleaning and noise removal.

[0060] A feature extraction unit, used to extract visual feature vectors, frequency domain feature vectors and time series feature vectors;

[0061] For surveillance video data, the CNN network is used to extract features from the preprocessed video frames to generate visual feature vectors of the video frames; for voiceprint data, the CNN network is used to extract features from the preprocessed voiceprint signals to generate frequency domain feature vectors of the voiceprint signals; for vibration data, the LSTM network is used to extract features from the preprocessed vibration signals to generate time series feature vectors of the vibration signals.

[0062] The multi-dimensional data fusion unit combines the feature vectors of surveillance video data, voiceprint data, and vibration data to form a comprehensive feature vector;

[0063] The deep learning model unit uses the RNN model to train the concatenated feature vectors, learns the differences between the corresponding states of the multi-dimensional data, and classifies the concatenated feature vectors. The network status is defined based on the multi-classification results of the RNN model, and the cross-validation method is used to evaluate the performance of the model.

[0064] The pipe network status includes: normal status and abnormal status.

[0065] The abnormal state includes pipeline fault location positioning and leakage point positioning.

[0066] The state monitoring unit is used to monitor the state of the power plant network in real time using the deep learning model, process the data collected by the data acquisition unit in real time, and perform state monitoring and abnormal position location according to the output of the deep learning model;

[0067] The alarm unit triggers an alarm when it detects an abnormal pipe network status, prompting operators to handle the problem;

[0068] Update and maintenance unit, regularly use newly collected data to retrain and verify the deep learning model to adapt to changes in the status of the pipeline network, while ensuring the stable operation of the monitoring system and regularly inspecting and maintaining hardware equipment.

[0069] The embodiment of the present invention discloses a method for monitoring the status of a nuclear power plant pipeline network based on multi-dimensional data fusion, comprising the following steps:

[0070] Step 1: Deep learning model training:

[0071] Step 1-1: Collect multi-dimensional data such as monitoring video, voiceprint, and vibration data of the nuclear power plant pipeline network;

[0072] Step 1-2: Perform time synchronization processing on the collected multi-dimensional data;

[0073] Specifically, timestamps are used to synchronize the collection of multidimensional data in time, and a unified collection frequency is set to ensure that different types of data are aligned on the time axis.

[0074] Step 1-3: Crop, scale and normalize the collected video frames, and clean and remove noise from the voiceprint and vibration data;

[0075] Step 1-4: Extract visual feature vectors, frequency domain feature vectors and time series feature vectors;

[0076] Specifically, for surveillance video data, the CNN network is used to extract features from the preprocessed video frames to generate visual feature vectors of the video frames; for voiceprint data, the CNN network is used to extract features from the preprocessed voiceprint signals to generate frequency domain feature vectors of the voiceprint signals; for vibration data, the LSTM network is used to extract features from the preprocessed vibration signals to generate time series feature vectors of the vibration signals.

[0077] Step 1-5: Concatenate the feature vectors of video, voiceprint and vibration data;

[0078] Specifically, the steady state and abnormality identified in the video and the stable or sudden changes in the voiceprint and vibration data are combined to form a new set of data;

[0079] Step 1-6: Use the RNN model to train the concatenated feature vectors, learn the differences between the corresponding states of the multi-dimensional data, classify the concatenated feature vectors, define the network status according to the results of the RNN model multi-classification, and use the cross-validation method to evaluate the performance of the model;

[0080] The pipe network status includes: normal status and abnormal status.

[0081] The abnormal state includes pipeline fault location positioning and leakage point positioning.

[0082] Step 2: Deploy the trained deep learning model to actual applications, including:

[0083] Step 2-1: Real-time collection of multi-dimensional data including surveillance video, voiceprint, and vibration data of the nuclear power plant pipeline network;

[0084] Step 2-2: Perform time synchronization processing on the collected multi-dimensional data;

[0085] Step 2-3: Crop, scale and normalize the collected video frames, and clean and remove noise from the voiceprint and vibration data;

[0086] Step 2-4: Extract visual feature vectors, frequency domain feature vectors and time series feature vectors; concatenate the feature vectors of video, voiceprint and vibration data;

[0087] Step 2-:5: Load the trained deep learning model, process sensor data in real time, and perform status monitoring and abnormal location positioning based on the model output;

[0088] When the pipe network status is abnormal or the leakage point is located, the alarm is triggered immediately and the operator is notified by means of sound and light alarms;

[0089] Step 2-6: Regularly collect new data, retrain and validate the model, and regularly inspect and maintain hardware equipment.

[0090] Another aspect of an embodiment of the present invention further provides an electronic device, including a processor and a memory;

[0091] The memory is used to store programs;

[0092] The processor executes the program to implement the method described above.

[0093] The contents of the method embodiments of the present invention are all applicable to the computer-readable storage medium embodiments. The functions specifically implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0094] The present invention also provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the method described above.

[0095] The above embodiments are only used to help understand the method and core idea of ​​the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made to the present invention without departing from the principles of the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

[0096] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A nuclear power plant pipeline network status monitoring system based on multi-dimensional data fusion, characterized in that: include: Data acquisition unit, used to collect multi-dimensional data such as monitoring video, voiceprint, and vibration data of the nuclear power plant pipeline network; Data synchronization unit ensures that the collection of multi-dimensional data is synchronized in time and the data collection frequency is consistent; A data preprocessing unit, which preprocesses the multidimensional data after synchronization processing; A feature extraction unit, used to extract visual feature vectors, frequency domain feature vectors and time series feature vectors; The multi-dimensional data fusion unit combines the feature vectors of surveillance video data, voiceprint data, and vibration data to form a comprehensive feature vector; The deep learning model unit uses the RNN model to train the concatenated feature vectors, learns the differences between the corresponding states of multi-dimensional data, and classifies the concatenated feature vectors. The network status is defined based on the multi-classification results of the RNN model, and the cross-validation method is used to evaluate the performance of the model. The state monitoring unit is used to monitor the state of the power plant network in real time using the deep learning model, process the data collected by the data acquisition unit in real time, and perform state monitoring and abnormal position location according to the output of the deep learning model; The alarm unit triggers an alarm when it detects an abnormal pipe network status, prompting operators to handle the problem; Update and maintenance unit, regularly use newly collected data to retrain and verify the deep learning model to adapt to changes in the status of the pipeline network, while ensuring the stable operation of the monitoring system and regularly inspecting and maintaining hardware equipment.

2. The nuclear power plant pipeline network status monitoring system based on multi-dimensional data fusion according to claim 1 is characterized in that: The feature extraction unit, For surveillance video data, the CNN network is used to extract features from the preprocessed video frames to generate visual feature vectors of the video frames; for voiceprint data, the CNN network is used to extract features from the preprocessed voiceprint signals to generate frequency domain feature vectors of the voiceprint signals; for vibration data, the LSTM network is used to extract features from the preprocessed vibration signals to generate time series feature vectors of the vibration signals.

3. The nuclear power plant pipe network status monitoring system based on multi-dimensional data fusion according to claim 2 is characterized in that: The pipe network status includes: normal status and abnormal status. The abnormal state includes pipeline fault location positioning and leakage point positioning.

4. The nuclear power plant pipe network status monitoring system based on multi-dimensional data fusion according to claim 1, characterized in that: The preprocessing method of the data preprocessing unit is: For surveillance video data, the preprocessing methods are cropping, scaling, and normalization. For voiceprint or vibration data, the preprocessing method is data cleaning and noise removal.

5. The nuclear power plant pipe network status monitoring system based on multi-dimensional data fusion according to claim 1, characterized in that: The data synchronization unit uses a timestamp to ensure that the collection of multi-dimensional data is synchronized in time, sets a unified collection frequency, and ensures that different types of data are aligned on the time axis.

6. A method for monitoring the status of a nuclear power plant pipeline network based on multi-dimensional data fusion, characterized in that: The following steps are involved: Step 1: Deep learning model training: Step 1-1: Collect multi-dimensional data such as monitoring video, voiceprint, and vibration data of the nuclear power plant pipeline network; Step 1-2: Perform time synchronization processing on the collected multi-dimensional data; Step 1-3: Crop, scale and normalize the collected video frames, and clean and remove noise from the voiceprint and vibration data; Step 1-4: Extract visual feature vectors, frequency domain feature vectors and time series feature vectors; Step 1-5: Concatenate the feature vectors of video, voiceprint and vibration data; Step 1-6: Use the RNN model to train the concatenated feature vectors, learn the differences between the corresponding states of the multi-dimensional data, classify the concatenated feature vectors, define the network status according to the multi-classification results of the RNN model, and use the cross-validation method to evaluate the performance of the model; Step 2: Deploy the trained deep learning model to actual applications, including: Step 2-1: Real-time collection of multi-dimensional data including surveillance video, voiceprint, and vibration data of the nuclear power plant pipeline network; Step 2-2: Perform time synchronization processing on the collected multi-dimensional data; Step 2-3: Crop, scale and normalize the collected video frames, and clean and remove noise from the voiceprint and vibration data; Step 2-4: Extract visual feature vectors, frequency domain feature vectors and time series feature vectors; concatenate the feature vectors of video, voiceprint and vibration data; Step 2-5: Load the trained deep learning model and perform status monitoring and abnormal location positioning based on the model output; When the pipe network status is abnormal or the leakage point is located, the alarm is triggered immediately and the operator is notified by means of sound and light alarms; Step 2-6: Regularly collect new data, retrain and validate the model, and regularly inspect and maintain hardware equipment.

7. The method for monitoring the status of a nuclear power plant pipe network based on multi-dimensional data fusion according to claim 6, characterized in that: In the steps 1-4, for the monitoring video data, the CNN network is used to extract features from the preprocessed video frames to generate visual feature vectors of the video frames; for the voiceprint data, the CNN network is used to extract features from the preprocessed voiceprint signals to generate frequency domain feature vectors of the voiceprint signals; for the vibration data, the LSTM network is used to extract features from the preprocessed vibration signals to generate time series feature vectors of the vibration signals.

8. The method for monitoring the status of a nuclear power plant pipe network based on multi-dimensional data fusion according to claim 6, characterized in that: In step 1-2 and step 2-2, the time stamp is used to synchronize the collection of multi-dimensional data in time, and a uniform collection frequency is set to ensure that different types of data are aligned on the time axis.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 6 to 8.

10. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the method according to any one of claims 6 to 8.

Images