A sensing method of a multi-modal high-voltage switch and related device

Abstract

The application discloses a kind of multi-modal high-voltage switch perception method and related device, comprising: obtaining the detection data of high-voltage switch state detection sensor and image data, combining detection data and image data and judging the state of high-voltage switch by introducing switch state weight matrix;The binary time sequence and environmental factor weight matrix are obtained by binary encoding switch weight matrix, and the binary time sequence is decomposed into several stationary time subsequences by wavelet decomposition technique;The prediction network model is trained by stationary time subsequence, the binary time sequence is input into the trained prediction network model, and the predicted time subsequence is obtained;Sequence reconstruction is carried out to predicted time subsequence, and predicted time reconstruction sequence is obtained, predicted time reconstruction sequence is introduced into environmental factor weight matrix, and switch state prediction sequence is obtained, so as to solve the problem that the prior art has poor high-voltage switch state detection and prediction accuracy.

Description

Technical Field

[0001] This application relates to the field of high-voltage switch status analysis technology, and in particular to a sensing method and related device for a multi-mode high-voltage switch. Background Technology

[0002] High-voltage switch status prediction is a crucial area for safety in smart grid construction, forming the foundation for a safe and stable smart grid. Accurate prediction of high-voltage switch status allows for forecasting the next on / off state of each switch. Through predictive analysis, staff can observe abnormal data changes, identify the root cause of problems, and then perform maintenance or repairs on the entire or specific parts of the power grid. High-voltage switch status prediction not only enhances the overall safety performance of the smart grid but also improves economic efficiency, reduces workload for staff, minimizes losses in the event of disasters, and has a positive impact on grid planning and construction, significantly contributing to the improvement of power system security.

[0003] In detecting the status of high-voltage switches, the mainstream methods are mainly divided into two categories: one is to use sensors for detection, and the other is to apply hardware circuits, such as voltage sampling circuits. Because the power grid structure is complex and subject to strong electromagnetic interference, sensors, which generate digital signals, are greatly affected, potentially leading to unstable data acquisition. The method of using voltage sampling circuits to convert high-voltage electricity to low-voltage before sampling and identifying the switch's closed state requires multiple stages of conversion, making the circuit more complex, and the sampled data may also be unstable. Therefore, software or hardware filtering is needed to improve detection accuracy.

[0004] In terms of predicting the state of high-voltage switches, there are currently very few papers or patents on switch state prediction. Switch prediction mainly involves predicting time series data. Currently, this prediction method mostly uses LSTM models or their branches. Recently, some scholars have also used Transformer combined with time series for prediction, but Transformer has high time complexity and its accuracy is not particularly high. Summary of the Invention

[0005] This application provides a sensing method and related device for multimodal high-voltage switches to solve the technical problem of poor accuracy in high-voltage switch state detection and prediction in the prior art.

[0006] In view of this, the first aspect of this application provides a sensing method for a multi-mode high-voltage switch, the method comprising:

[0007] The detection data from several high-voltage switch status detection sensors and the image data from high-voltage switch status image recognition are acquired. The detection data and the image data are combined, and the status of the high-voltage switch is determined by introducing a switch status weight matrix.

[0008] The switch weight matrix is ​​binary encoded to obtain a binary time series and an environmental factor weight matrix. The binary time series is then decomposed into several stationary time subsequences using wavelet decomposition technology.

[0009] The prediction network model is trained using the stationary time subsequence, and the binary time series is input into the trained prediction network model to obtain the prediction time subsequence.

[0010] The predicted time subsequence is reconstructed to obtain a predicted time reconstructed sequence. The predicted time reconstructed sequence is then incorporated into the environmental factor weight matrix to obtain a switch state prediction sequence.

[0011] Optionally, training the prediction network model using the stationary time subsequence and inputting the binary time series into the trained prediction network model to obtain the prediction time subsequence specifically includes:

[0012] Each stationary time subsequence is divided into a training set and a test set. The prediction network model constructed by the residual convolutional network and the Informer layer is trained to obtain the trained prediction network model. The binary time series is input into the prediction network model to obtain the predicted time subsequence.

[0013] Optionally, the prediction network model consists of residual convolutional layers, Informer layers, flattening layers, and fully connected layers in LSTM, wherein the residual convolutional layers are composed of multiple convolutional layers transformed multiple times.

[0014] Optionally, the framework of the Informer layer consists of an encoder (Incoder) and a decoder (Decoder), and is composed of multiple convolutional layers, embedding layers, and pooling layers, and has multiple attention mechanisms.

[0015] A second aspect of this application provides a sensing system for a multi-mode high-voltage switch, the system comprising:

[0016] The detection module is used to acquire detection data from several high-voltage switch status detection sensors and image data from high-voltage switch status image recognition, combine the detection data and the image data, and determine the status of the high-voltage switch by introducing a switch status weight matrix.

[0017] The preprocessing module is used to encode the switch weight matrix into binary to obtain a binary time series and an environmental factor weight matrix, and to decompose the binary time series into several stationary time subsequences using wavelet decomposition technology.

[0018] The training module is used to train the prediction network model using the stationary time subsequence and input the binary time series into the trained prediction network model to obtain the prediction time subsequence.

[0019] The prediction module is used to reconstruct the predicted time subsequence to obtain a predicted time reconstructed sequence, and to introduce the predicted time reconstructed sequence into the environmental factor weight matrix to obtain a switch state prediction sequence.

[0020] Optionally, the training module is specifically used for:

[0021] Each stationary time subsequence is divided into a training set and a test set. The prediction network model constructed by the residual convolutional network and the Informer layer is trained to obtain the trained prediction network model. The binary time series is input into the prediction network model to obtain the predicted time subsequence.

[0022] Optionally, the prediction network model consists of residual convolutional layers, Informer layers, flattening layers, and fully connected layers in LSTM, wherein the residual convolutional layers are composed of multiple convolutional layers transformed multiple times.

[0023] Optionally, the framework of the Informer layer consists of an encoder (Incoder) and a decoder (Decoder), and is composed of multiple convolutional layers, embedding layers, and pooling layers, and has multiple attention mechanisms.

[0024] A third aspect of this application provides a sensing device for a multimodal high-voltage switch, the device comprising a processor and a memory:

[0025] The memory is used to store program code and transmit the program code to the processor;

[0026] The processor is configured to execute the steps of the sensing method for the multimodal high-voltage switch as described in the first aspect above, according to the instructions in the program code.

[0027] A fourth aspect of this application provides a computer-readable storage medium for storing program code for executing the sensing method for a multimodal high-voltage switch described in the first aspect.

[0028] As can be seen from the above technical solutions, this application has the following advantages:

[0029] This application provides a sensing method for multimodal high-voltage switches, comprising: acquiring detection data from several high-voltage switch state detection sensors and image data from high-voltage switch state image recognition; combining the detection data and image data and determining the state of the high-voltage switch by introducing a switch state weight matrix; binary encoding the switch weight matrix to obtain a binary time series and an environmental factor weight matrix, and decomposing the binary time series into several stationary time subsequences using wavelet decomposition technology; training a prediction network model using the stationary time subsequences and inputting the binary time series into the trained prediction network model to obtain a prediction time subsequence; reconstructing the prediction time subsequence to obtain a prediction time reconstruction sequence, and introducing the prediction time reconstruction sequence into the environmental factor weight matrix to obtain a switch state prediction sequence.

[0030] Compared with existing technologies:

[0031] 1) In terms of detection, in order to alleviate the problem of accuracy in detection, this invention combines image detection with sensor detection and uses a weighted network to jointly determine the state of the switch, thereby improving the accuracy and efficiency of detection.

[0032] To address the issue of low accuracy in prediction, this invention employs the latest Informer layer for time series prediction processing and combines it with an LSTM residual convolutional network structure for model building, resulting in low time complexity and high accuracy. Furthermore, wavelet decomposition is used in the preprocessing stage to further improve accuracy. Environmental factors are also considered, employing an environmental factor weight matrix to make the prediction results more closely resemble reality. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating an embodiment of a sensing method for a multimodal high-voltage switch provided in this application.

[0034] Figure 2 This is a schematic diagram of the structure of a sensing system embodiment for a multimodal high-voltage switch provided in this application. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0036] Please see Figure 1 The sensing method for a multi-mode high-voltage switch provided in this application includes:

[0037] Step 101: Obtain detection data from several high-voltage switch status detection sensors and image data from high-voltage switch status image recognition. Combine the detection data and image data and determine the status of the high-voltage switch by introducing a switch status weight matrix.

[0038] It should be noted that this embodiment collects data from multiple high-voltage switch status detection sensors and image data from high-voltage switch status image recognition. Image processing methods are used to identify the switch position and determine the high-voltage switch status based on sensor data. Data is then sampled again within a short period to eliminate the influence of noise or jitter. A switch status weight matrix is ​​introduced to obtain the high-voltage switch status.

[0039] Step 102: Encode the switch weight matrix into binary to obtain the binary time series and the environmental factor weight matrix, and decompose the binary time series into several stationary time subsequences using wavelet decomposition technology.

[0040] It should be noted that the switch state data in step 101 is binary encoded to obtain a binary time series and an environmental factor weight matrix, and the above time series is decomposed into multiple stationary time subsequences using wavelet decomposition.

[0041] Step 103: Train the prediction network model using stationary time subsequences, and input the binary time series into the trained prediction network model to obtain the prediction time subsequences;

[0042] It should be noted that the stationary time subsequence is divided into training and test sets and trained through a network constructed by residual convolutional network and Informer layer to obtain a trained network model. The binary time series at each time step is then passed through this network model to obtain the predicted time subsequence.

[0043] The prediction network model consists of residual convolutional layers, Informer layers, flattening layers, and fully connected layers in LSTM. The residual convolutional layers are composed of multiple convolutional layers transformed multiple times.

[0044] The Informer layer framework consists of an encoder (Incoder) and a decoder (Decoder), and comprises multiple convolutional layers, embedding layers, and pooling layers, as well as multiple attention mechanisms.

[0045] Step 104: Reconstruct the predicted time subsequence to obtain the predicted time reconstructed sequence. Introduce the predicted time reconstructed sequence into the environmental factor weight matrix to obtain the switch state prediction sequence.

[0046] Finally, the predicted time subsequence is reconstructed to obtain a new predicted time series, which is then introduced into the previously obtained environmental factor weight matrix to obtain the final switch state prediction sequence.

[0047] The above describes a sensing method for a multi-mode high-voltage switch provided in the embodiments of this application. The following describes a sensing system for a multi-mode high-voltage switch provided in the embodiments of this application.

[0048] Please see Figure 2 The sensing system for a multi-mode high-voltage switch provided in this application embodiment includes:

[0049] The detection module 201 is used to acquire detection data from several high-voltage switch status detection sensors and image data from high-voltage switch status image recognition, combine the detection data and image data, and determine the status of the high-voltage switch by introducing a switch status weight matrix.

[0050] Preprocessing module 202 is used to encode the switch weight matrix into binary to obtain a binary time series and an environmental factor weight matrix, and to decompose the binary time series into several stationary time subsequences using wavelet decomposition technology.

[0051] The training module 203 is used to train the prediction network model using stationary time subsequences and input binary time series into the trained prediction network model to obtain the prediction time subsequence;

[0052] The prediction module 204 is used to reconstruct the prediction time subsequence to obtain the prediction time reconstruction sequence, and to introduce the prediction time reconstruction sequence into the environmental factor weight matrix to obtain the switch state prediction sequence.

[0053] Furthermore, this application embodiment also provides a sensing device for a multi-mode high-voltage switch, the device including a processor and a memory:

[0054] The memory is used to store program code and transmit the program code to the processor;

[0055] The processor is used to execute the steps of the sensing method for multimodal high-voltage switches as described in the above method embodiments, according to the instructions in the program code.

[0056] Furthermore, this application embodiment also provides a computer-readable storage medium for storing program code for executing the multimodal high-voltage switch sensing method described in the above method embodiment.

[0057] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0058] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0059] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0060] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0061] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0062] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0063] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0064] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A sensing method for a multi-mode high-voltage switch, characterized in that, include: The detection data from several high-voltage switch status detection sensors and the image data from high-voltage switch status image recognition are acquired. The detection data and the image data are combined, and the status of the high-voltage switch is determined by introducing a switch status weight matrix. The switch weight matrix is ​​binary encoded to obtain a binary time series and an environmental factor weight matrix. The binary time series is then decomposed into several stationary time subsequences using wavelet decomposition technology. The prediction network model is trained using the stationary time subsequence, and the binary time series is input into the trained prediction network model to obtain the prediction time subsequence. The predicted time subsequence is reconstructed to obtain a predicted time reconstructed sequence. The predicted time reconstructed sequence is then incorporated into the environmental factor weight matrix to obtain a switch state prediction sequence.

2. The sensing method for multi-mode high-voltage switches according to claim 1, characterized in that, The step of training the prediction network model using the stationary time series and inputting the binary time series into the trained prediction network model to obtain the prediction time series specifically includes: Each stationary time subsequence is divided into a training set and a test set. The prediction network model constructed by the residual convolutional network and the Informer layer is trained to obtain the trained prediction network model. The binary time series is input into the prediction network model to obtain the predicted time subsequence.

3. The sensing method for multi-mode high-voltage switches according to claim 2, characterized in that, The prediction network model consists of residual convolutional layers, Informer layers, flattening layers, and fully connected layers in LSTM. The residual convolutional layers are composed of multiple convolutional layers transformed multiple times.

4. The sensing method for multi-mode high-voltage switches according to claim 3, characterized in that, The Informer layer framework consists of an encoder (Incoder) and a decoder (Decoder), and is composed of multiple convolutional layers, embedding layers, and pooling layers, and has multiple attention mechanisms.

5. A sensing system for a multi-mode high-voltage switch, characterized in that, include: The detection module is used to acquire detection data from several high-voltage switch status detection sensors and image data from high-voltage switch status image recognition, combine the detection data and the image data, and determine the status of the high-voltage switch by introducing a switch status weight matrix. The preprocessing module is used to encode the switch weight matrix into binary to obtain a binary time series and an environmental factor weight matrix, and to decompose the binary time series into several stationary time subsequences using wavelet decomposition technology. The training module is used to train the prediction network model using the stationary time subsequence and input the binary time series into the trained prediction network model to obtain the prediction time subsequence. The prediction module is used to reconstruct the predicted time subsequence to obtain a predicted time reconstructed sequence, and to introduce the predicted time reconstructed sequence into the environmental factor weight matrix to obtain a switch state prediction sequence.

6. The sensing system for the multi-mode high-voltage switch according to claim 5, characterized in that, The training module is specifically used for: Each stationary time subsequence is divided into a training set and a test set. The prediction network model constructed by the residual convolutional network and the Informer layer is trained to obtain the trained prediction network model. The binary time series is input into the prediction network model to obtain the predicted time subsequence.

7. The sensing system for the multi-mode high-voltage switch according to claim 6, characterized in that, The prediction network model consists of residual convolutional layers, Informer layers, flattening layers, and fully connected layers in LSTM. The residual convolutional layers are composed of multiple convolutional layers transformed multiple times.

8. The sensing system for the multi-mode high-voltage switch according to claim 7, characterized in that, The Informer layer framework consists of an encoder (Incoder) and a decoder (Decoder), and is composed of multiple convolutional layers, embedding layers, and pooling layers, and has multiple attention mechanisms.

9. A sensing device for a multi-mode high-voltage switch, characterized in that, The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the sensing method of the multimodal high-voltage switch according to any one of the claims 1-4 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program code for executing the sensing method of the multimodal high-voltage switch according to any one of claims 1-4.

Images