High-voltage variable-frequency power supply power supply test protection method and system based on intelligent perception

Abstract

The present invention discloses a high-voltage variable-frequency power supply power supply test protection method and system based on intelligent perception, including: obtaining historical power supply test instances of high-voltage variable-frequency power supplies of different specifications based on big data retrieval, classifying historical test protection instances into types, and constructing a power supply test database; monitoring the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, performing abnormality detection according to the real-time power supply test monitoring information, and obtaining abnormality detection information; presetting a trigger threshold, judging whether to perform power supply protection control according to the abnormality detection information, and performing abnormal event prediction if the trigger threshold is not reached; obtaining abnormal power supply test information of abnormal power supply test, analyzing abnormal data change patterns, and updating the power supply test database. The safety of power supply test is improved, and the comprehensiveness of power supply test protection is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply testing and protection, and in particular to a high-voltage variable-frequency power supply testing and protection method and system based on intelligent perception. Background Art

[0002] With the rapid development of modern power technology, high-voltage variable-frequency power supply, as the core equipment of industrial production and power transmission, has attracted more and more attention for its safety and stability. Especially in the fields of industrial automation, power transmission and new energy generation, the performance of high-voltage variable-frequency power supply directly affects the operating efficiency and safety of the entire system. The traditional high-voltage variable-frequency power supply test and protection method mainly relies on manual inspection and regular testing, which has many shortcomings. Therefore, it is particularly important to conduct comprehensive and accurate testing and protection of high-voltage variable-frequency power supply. Summary of the invention

[0003] The present invention overcomes the defects of the prior art and provides a high-voltage variable-frequency power supply power supply test protection method and system based on intelligent perception, the important purpose of which is to improve the safety and comprehensiveness of power supply testing.

[0004] To achieve the above-mentioned purpose, the first aspect of the present invention provides a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception, comprising:

[0005] Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, the historical power supply test instances are classified into types, and a power supply test database is constructed;

[0006] Monitor the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, perform anomaly detection based on the real-time power supply test monitoring information, and obtain anomaly detection information;

[0007] Preset a trigger threshold, determine whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, perform abnormal event prediction;

[0008] Acquire abnormal power test information of power test abnormalities, analyze abnormal data change patterns, and update the power test database.

[0009] In this solution, the historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained based on big data retrieval, the historical power supply test instances are classified into types, and a power supply test database is constructed, which specifically includes:

[0010] Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, and each historical power supply test instance is divided into a normal test instance and an abnormal test instance according to the test protection information corresponding to each historical power supply test instance;

[0011] Extract features of normal test instances and abnormal test instances respectively to obtain first feature information and second feature information, construct a test feature change graph of the normal test instance based on the first feature information, and construct a normal test portrait;

[0012] Calculate the Mahalanobis distance between each abnormal feature of each abnormal test instance and the abnormal test event, and compare it with the preset threshold, remove irrelevant features, and sort the Mahalanobis distances corresponding to the removed abnormal features in descending order;

[0013] According to the sorting results, the root cause of the abnormal features corresponding to each historical abnormal instance is analyzed, and an abnormal test profile is constructed;

[0014] The normal test profile and the abnormal test profile are combined to form a power supply test database.

[0015] In this solution, the abnormality detection is performed according to the real-time power supply test monitoring information to obtain abnormality detection information, specifically including:

[0016] Based on the power supply test database, a normal test image and first feature information are obtained to form a training data set, and an abnormality detection model is constructed using RNN;

[0017] VAE is introduced to optimize the model, the training data set is input into the variational autoencoder to obtain the potential feature distribution of the training data, which is saved in the hidden layer as a latent space representation, and the probability distribution of the latent variables is generated;

[0018] Sampling latent variables according to the probability distribution of the latent variables, generating predicted values ​​according to the sampling results, obtaining true values ​​corresponding to the predicted values ​​to calculate reconstruction errors, calculating gradient values ​​according to the calculated reconstruction errors, and iteratively optimizing according to the gradient values ​​using a back propagation algorithm;

[0019] Based on the training data, the original feature samples are obtained, and the corresponding feature reconstruction samples are obtained through the variational encoder network. The deviation between the feature reconstruction samples and the original samples is calculated, and the deviation is compared with the preset threshold. The data corresponding to the samples greater than the preset threshold are regarded as abnormal data;

[0020] The target high-voltage variable-frequency power supply test is monitored to obtain real-time power supply test monitoring information, and the real-time power supply test monitoring information is preprocessed and input into the anomaly detection model for detection to obtain anomaly detection information.

[0021] In this solution, the preset trigger threshold determines whether to perform power protection control according to the abnormal detection information. If the trigger threshold is not reached, abnormal event prediction is performed, specifically including:

[0022] Acquire abnormality detection information, perform feature extraction on the abnormality detection information to obtain each abnormal test data and corresponding test process features, and obtain abnormal test feature information;

[0023] Acquire the test data benchmark corresponding to each test process according to the abnormal test feature information, and calculate the test data benchmark and the abnormal test feature to obtain the deviation between each abnormal test data and the benchmark data;

[0024] A trigger threshold is preset, and the calculated deviation is compared with the trigger threshold. If the abnormal detection information is greater than the trigger threshold, the power protection mechanism is activated;

[0025] If the abnormality detection information is less than the trigger threshold, an abnormal event prediction is performed based on the abnormality detection information, and an early warning prompt is issued.

[0026] In this solution, the abnormal event prediction based on the abnormal detection information and the early warning prompt specifically include:

[0027] Based on big data retrieval, various power supply test abnormal events are obtained, the test monitoring data corresponding to each abnormal event is extracted, and the test monitoring curve corresponding to each abnormal event is constructed;

[0028] Extract abnormal test monitoring features for each power supply test abnormal event, mark the test monitoring curves corresponding to each abnormal event, and form an instance data set;

[0029] Constructing an abnormal event prediction model, introducing a Markov chain to optimize the model, training the abnormal event prediction model based on the instance data set, and forming a state space based on the input training data;

[0030] Preset time steps, define the transfer state according to the test monitoring curve of each abnormal event, and calculate the state transfer probability corresponding to each time step. Construct a state transfer matrix based on the calculated state transfer probability, perform model training and optimization, and obtain an abnormal event prediction model that meets expectations;

[0031] Inputting the abnormality detection information into the abnormal event prediction model for analysis, extracting features from the input abnormality detection information to form a feature sequence;

[0032] Analyze and obtain a state transfer matrix in the constructed state space according to the feature sequence, and predict the next state according to the state transfer matrix to obtain abnormal event prediction information;

[0033] An early warning is issued for the current high-voltage variable-frequency power supply test based on the abnormal event prediction information.

[0034] In this solution, the acquisition of abnormal power supply test information of abnormal power supply test, analysis of abnormal data change mode, and updating of power supply test database specifically include:

[0035] Acquire abnormal power supply test information of a power supply test abnormality, and construct a power supply test data change diagram of a current power supply test abnormality event based on the abnormal power supply test information;

[0036] Calculate the similarity between the power supply test data change graph and the test images in the power supply test database, and make a judgment with a preset threshold;

[0037] If the similarity value is less than the preset threshold, it means that the current power supply abnormality test is an abnormal test instance that does not exist in the database, and an abnormal test portrait is generated according to the power supply test data change diagram of the current power supply test abnormal event, and the power supply test database is updated;

[0038] If the similarity value is greater than the preset threshold, it means that similar abnormal test portraits are stored in the current power supply test database, and the similar portraits are sorted to extract the similar portrait with the highest similarity;

[0039] Perform difference feature analysis on the similar portrait with the highest similarity and the power supply test data change graph of the current power supply test abnormal event, and update the similar portrait with the highest similarity according to the difference features obtained by the analysis.

[0040] A second aspect of the present invention provides a high-voltage variable-frequency power supply power supply test protection system based on intelligent perception, the system comprising: a memory, a processor, the memory containing a high-voltage variable-frequency power supply power supply test protection method program based on intelligent perception, the high-voltage variable-frequency power supply power supply test protection method program based on intelligent perception when executed by the processor implements the following steps:

[0041] Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, the historical power supply test instances are classified into types, and a power supply test database is constructed;

[0042] Monitor the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, perform anomaly detection based on the real-time power supply test monitoring information, and obtain anomaly detection information;

[0043] Preset a trigger threshold, determine whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, perform abnormal event prediction;

[0044] Acquire abnormal power test information of power test abnormalities, analyze abnormal data change patterns, and update the power test database.

[0045] The present invention discloses a high-voltage variable-frequency power supply power supply test protection method and system based on intelligent perception, including: obtaining historical power supply test instances of high-voltage variable-frequency power supplies of different specifications based on big data retrieval, classifying the historical power supply test instances into types, and constructing a power supply test database; monitoring the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, performing abnormality detection according to the real-time power supply test monitoring information, and obtaining abnormality detection information; presetting a trigger threshold, judging whether to perform power supply protection control according to the abnormality detection information, and performing abnormal event prediction if the trigger threshold is not reached; obtaining abnormal power supply test information of abnormal power supply test, analyzing the abnormal data change mode, and updating the power supply test database. The safety of power supply test is improved, and the comprehensiveness of power supply test protection is enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0047] Figure 1 A flow chart of a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception provided by one embodiment of the present invention;

[0048] Figure 2 A flow chart of a high-voltage variable-frequency power supply test and monitoring provided in accordance with an embodiment of the present invention;

[0049] Figure 3 A block diagram of a high-voltage variable-frequency power supply test and protection system based on intelligent perception provided by an embodiment of the present invention;

[0050] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0053] Figure 1A flow chart of a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception provided by one embodiment of the present invention;

[0054] like Figure 1 As shown, the present invention provides a flow chart of a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception, including:

[0055] S102, based on big data retrieval, obtaining historical power supply test instances of high-voltage variable-frequency power supplies of different specifications, classifying the historical power supply test instances by type, and building a power supply test database;

[0056] S104, monitoring the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, performing anomaly detection according to the real-time power supply test monitoring information to obtain anomaly detection information;

[0057] S106, preset a trigger threshold, determine whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, perform abnormal event prediction;

[0058] S108, obtaining abnormal power supply test information of power supply test abnormality, analyzing abnormal data change mode, and updating the power supply test database.

[0059] It should be noted that the present invention provides a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception, which obtains historical power supply test instances of various specifications of high-voltage variable-frequency power supplies through big data retrieval technology, and classifies these instances into types to build a power supply test database. When the target high-voltage variable-frequency power supply is tested, real-time monitoring and acquisition of real-time power supply test monitoring information are performed, anomaly detection is performed to identify potential abnormal data, and abnormal detection information is generated. Preset trigger thresholds. When the abnormal data corresponding to the abnormal detection information reaches or exceeds these thresholds, the power supply protection control is immediately triggered to ensure the safety of the power supply equipment. However, if the abnormality does not reach the trigger threshold, further abnormal event prediction is performed to predict possible faults or problems in advance. When an abnormality occurs in the actual power supply test, the abnormal power supply test information is recorded, and the change pattern of the abnormal data is analyzed. The power supply test database is updated and optimized to make it more in line with the needs of actual testing, and the accuracy and safety of power supply testing are continuously improved.

[0060] Further, in a preferred embodiment of the present invention, the historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained based on big data retrieval, the historical power supply test instances are classified into types, and a power supply test database is constructed, which specifically includes:

[0061] Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, and each historical power supply test instance is divided into a normal test instance and an abnormal test instance according to the test protection information corresponding to each historical power supply test instance;

[0062] Extract features of normal test instances and abnormal test instances respectively to obtain first feature information and second feature information, construct a test feature change graph of the normal test instance based on the first feature information, and construct a normal test portrait;

[0063] Calculate the Mahalanobis distance between each abnormal feature of each abnormal test instance and the abnormal test event, and compare it with the preset threshold, remove irrelevant features, and sort the Mahalanobis distances corresponding to the removed abnormal features in descending order;

[0064] According to the sorting results, the root cause of the abnormal features corresponding to each historical abnormal instance is analyzed, and an abnormal test profile is constructed;

[0065] The normal test profile and the abnormal test profile are combined to form a power supply test database.

[0066] It should be noted that, first, historical power supply test instances of various specifications of high-voltage variable-frequency power supplies are obtained by retrieving historical data, and the historical power supply test instances include power supply test schemes, power supply test monitoring information, and test protection information. According to the test protection information corresponding to each historical power supply test instance, the historical power supply test instance is divided into a normal test instance and an abnormal test instance. Feature extraction is performed for these two types of instances. After feature extraction, the normal test instance obtains the first feature information characterizing the normal operation state of the power supply, and a test feature change diagram of the normal test instance is constructed to form a normal test portrait. A comprehensive view of the operation data during normal power supply testing is provided. For the abnormal test instance, feature extraction is also performed to obtain the second feature information characterizing the abnormal state of the power supply. The Mahalanobis distance between each abnormal feature and the abnormal test event in each abnormal test instance is calculated. By comparing with the preset threshold, the features irrelevant to the abnormal event are eliminated. Then, the Mahalanobis distance corresponding to the abnormal features after eliminating the irrelevant features is sorted in descending order, and the abnormal features with the greatest impact are analyzed. Based on the sorting results, the root abnormal feature analysis is performed on the abnormal features corresponding to each historical abnormal instance, and an abnormal test portrait is constructed to reveal the inherent law of abnormal operation of the power supply. Finally, the power supply test database is constructed based on the normal test portrait and the abnormal test portrait, which records the normal operating status of the power supply test and also contains detailed records and analysis results of various abnormal situations, providing strong support for power supply testing, maintenance and troubleshooting.

[0067] Further, in a preferred embodiment of the present invention, performing abnormality detection according to the real-time power supply test monitoring information to obtain abnormality detection information specifically includes:

[0068] Based on the power supply test database, a normal test image and first feature information are obtained to form a training data set, and an abnormality detection model is constructed using RNN;

[0069] VAE is introduced to optimize the model, the training data set is input into the variational autoencoder to obtain the potential feature distribution of the training data, which is saved in the hidden layer as a latent space representation, and the probability distribution of the latent variables is generated;

[0070] Sampling latent variables according to the probability distribution of the latent variables, generating predicted values ​​according to the sampling results, obtaining true values ​​corresponding to the predicted values ​​to calculate reconstruction errors, calculating gradient values ​​according to the calculated reconstruction errors, and iteratively optimizing according to the gradient values ​​using a back propagation algorithm;

[0071] Based on the training data, the original feature samples are obtained, and the corresponding feature reconstruction samples are obtained through the variational encoder network. The deviation between the feature reconstruction samples and the original samples is calculated, and the deviation is compared with the preset threshold. The data corresponding to the samples greater than the preset threshold are regarded as abnormal data;

[0072] The target high-voltage variable-frequency power supply test is monitored to obtain real-time power supply test monitoring information, and the real-time power supply test monitoring information is preprocessed and input into the anomaly detection model for detection to obtain anomaly detection information.

[0073] It should be noted that, first, the normal test portrait and the related first feature information are extracted from the power test database to form a training data set. A recurrent neural network (RNN) is used as the basic architecture to build an anomaly detection model. RNN is particularly suitable for processing power test data that changes over time due to its ability to process sequence data. A variational autoencoder (VAE) is introduced for model optimization. VAE is a powerful generative model that generates new data samples by learning the potential feature distribution of the data. The training data set is input into the VAE, and the potential feature distribution of the training data is obtained through its encoding part, and these features are saved in the hidden layer as a latent space representation. At the same time, the decoding part of the VAE generates the probability distribution of the latent variables, which is the basis for the subsequent generation of predicted values. Next, sampling is performed according to the probability distribution of the latent variables generated by the VAE, and the predicted values ​​are generated according to the sampling results. The true value corresponding to the predicted value is obtained, and the reconstruction error between them is calculated. The reconstruction error reflects the degree of fit of the model to the original data. By calculating the gradient value of the reconstruction error, the model is iteratively optimized using the back propagation algorithm to continuously improve its performance. The encoding part of the VAE is used to reconstruct the original feature samples, and the deviation between the reconstructed samples and the original samples is calculated. If the deviation is greater than the preset threshold, the corresponding sample is abnormal data. Finally, the target high-voltage variable-frequency power supply is monitored in real time to obtain real-time power supply test monitoring information. The information is input into the trained anomaly detection model for detection to obtain anomaly detection information, thereby timely discovering and locating abnormal conditions in the power supply operation, and taking corresponding measures to deal with them.

[0074] Further, in a preferred embodiment of the present invention, the preset trigger threshold determines whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, abnormal event prediction is performed, specifically including:

[0075] Acquire abnormality detection information, perform feature extraction on the abnormality detection information to obtain each abnormal test data and corresponding test process features, and obtain abnormal test feature information;

[0076] Acquire the test data benchmark corresponding to each test process according to the abnormal test feature information, and calculate the test data benchmark and the abnormal test feature to obtain the deviation between each abnormal test data and the benchmark data;

[0077] A trigger threshold is preset, and the calculated deviation is compared with the trigger threshold. If the abnormal detection information is greater than the trigger threshold, the power protection mechanism is activated;

[0078] If the abnormality detection information is less than the trigger threshold, an abnormal event prediction is performed based on the abnormality detection information, and an early warning prompt is issued.

[0079] Furthermore, in a preferred embodiment of the present invention, the abnormal event prediction based on the abnormal detection information and the early warning prompt specifically include:

[0080] Based on big data retrieval, various power supply test abnormal events are obtained, the test monitoring data corresponding to each abnormal event is extracted, and the test monitoring curve corresponding to each abnormal event is constructed;

[0081] Extract abnormal test monitoring features for each power supply test abnormal event, mark the test monitoring curves corresponding to each abnormal event, and form an instance data set;

[0082] Constructing an abnormal event prediction model, introducing a Markov chain to optimize the model, training the abnormal event prediction model based on the instance data set, and forming a state space based on the input training data;

[0083] Preset time steps, define the transfer state according to the test monitoring curve of each abnormal event, and calculate the state transfer probability corresponding to each time step. Construct a state transfer matrix based on the calculated state transfer probability, perform model training and optimization, and obtain an abnormal event prediction model that meets expectations;

[0084] Inputting the abnormality detection information into the abnormal event prediction model for analysis, extracting features from the input abnormality detection information to form a feature sequence;

[0085] Analyze and obtain a state transfer matrix in the constructed state space according to the feature sequence, and predict the next state according to the state transfer matrix to obtain abnormal event prediction information;

[0086] An early warning is issued for the current high-voltage variable-frequency power supply test based on the abnormal event prediction information.

[0087] It should be noted that after detecting abnormal data, it is judged that the abnormal data has not reached the level of affecting the power supply test. At this time, it is necessary to predict abnormal events based on the detected abnormal data to analyze its potential dangers. First, various power supply test abnormal events are obtained, and the test monitoring data corresponding to the abnormal events are extracted, including key indicators such as voltage, current, and temperature. Their changes over time constitute the test monitoring curves corresponding to each abnormal event. Next, the abnormal test monitoring features of each power supply test abnormal event are extracted. Reflecting the test data changes of the power supply when the abnormal event occurs, the test monitoring curves corresponding to each abnormal event are marked to construct an instance data set. An abnormal event prediction model is constructed, and a Markov chain is introduced for model optimization. Markov chain is a random process model based on state transition probability. Based on the instance data set, the abnormal event prediction model is trained. During the training process, the state space is constructed according to the input training data, the state transition is defined, the time step is preset, and the transition state is determined according to the test monitoring curves of each abnormal event. Then, the state transition probability corresponding to each time step is calculated to reflect the possibility of transitioning from one state to another. Based on the calculated state transition probability, a state transition matrix is ​​constructed. Through continuous model training and optimization, an abnormal event prediction model that meets expectations is obtained. The abnormal detection information is input into the abnormal event prediction model for analysis. Feature extraction is performed on the input abnormal detection information to form a feature sequence. According to the feature sequence, the analysis is performed in the constructed state space to obtain the corresponding state transition matrix. The next state is predicted based on the state transition matrix to obtain abnormal event prediction information. According to the abnormal event prediction information, an early warning is issued for the power supply test of the current high-voltage variable frequency power supply. The early warning mechanism can help to timely discover potential abnormal events and take corresponding measures to prevent and deal with them, thereby ensuring the stable operation of the power supply test.

[0088] Further, in a preferred embodiment of the present invention, the obtaining of abnormal power supply test information of abnormal power supply test, analyzing the abnormal data change pattern, and updating the power supply test database specifically includes:

[0089] Acquire abnormal power supply test information of a power supply test abnormality, and construct a power supply test data change diagram of a current power supply test abnormality event based on the abnormal power supply test information;

[0090] Calculate the similarity between the power supply test data change graph and the test images in the power supply test database, and make a judgment with a preset threshold;

[0091] If the similarity value is less than the preset threshold, it means that the current power supply abnormality test is an abnormal test instance that does not exist in the database, and an abnormal test portrait is generated according to the power supply test data change diagram of the current power supply test abnormal event, and the power supply test database is updated;

[0092] If the similarity value is greater than the preset threshold, it means that similar abnormal test portraits are stored in the current power supply test database, and the similar portraits are sorted to extract the similar portrait with the highest similarity;

[0093] Perform difference feature analysis on the similar portrait with the highest similarity and the power supply test data change graph of the current power supply test abnormal event, and update the similar portrait with the highest similarity according to the difference features obtained by the analysis.

[0094] It should be noted that after an abnormal power test accident occurs, abnormal power test information is obtained. Including specific data of key parameters such as voltage, current, and temperature when the abnormality occurs, the power test data change graph of the current power test abnormal event is formed to show the change trend and characteristics of the power test data when the abnormal event occurs. Next, the current power test data change graph is similar to the test image stored in the power test database. The test image in the database is a record of previous abnormal test events, which contains data change graphs of various abnormal events. Through similarity calculation, the similarity between the current abnormal event and the known abnormal events in the database is quantified. After the similarity calculation, the obtained similarity value is compared with the preset threshold. Determine whether the current abnormal event is a known abnormal test instance in the database. If the similarity value is less than the preset threshold, it means that the current power abnormal test is not similar to any known abnormal test instance in the database, which means that it is a new and unknown abnormal test instance. For new abnormal test instances, a new abnormal test portrait is generated according to the power test data change graph of the current power test abnormal event, and added to the power test database to enrich the content of the database and provide a reference for future abnormal event prediction. However, if the similarity value is greater than the preset threshold, it means that the current power supply test database has stored abnormal test portraits similar to the current abnormal event. In this case, all similar portraits are extracted from the database and sorted by similarity to find the portrait most similar to the current abnormal event. The similar portrait with the highest similarity is analyzed with the power supply test data change graph of the current power supply test abnormal event for difference features. Find the difference between the two portraits. Based on the difference features obtained from the analysis, the similar portrait with the highest similarity is updated to make it more accurately reflect the characteristics of the current abnormal event, so that the portraits in the power supply test database will become more and more perfect, providing more accurate and reliable support for the prediction and early warning of abnormal events.

[0095] Figure 2 A flow chart of a high-voltage variable-frequency power supply test and monitoring provided in accordance with an embodiment of the present invention;

[0096] like Figure 2 As shown, the present invention provides a high-voltage variable frequency power supply power supply test monitoring flow chart, including:

[0097] S202, connecting the target high-voltage variable-frequency power supply to the test equipment for testing, and obtaining real-time power supply test monitoring information;

[0098] S204, inputting the real-time power supply test monitoring information into the abnormality detection model to detect abnormal test data and obtain abnormality detection information;

[0099] S206, judging whether to execute the power protection mechanism according to the abnormality detection information, and if not, performing abnormal event prediction;

[0100] S208, inputting the abnormality detection information into the abnormal event prediction model for analysis to obtain abnormal event prediction information;

[0101] S210: Providing a potential danger warning prompt based on the abnormal event prediction information.

[0102] It should be noted that when a high-voltage variable-frequency power supply is tested, a power supply test accident may occur, which may cause the power supply to be short-circuited during the experiment. A current transformer is used to detect the output current of the power supply and combined with the voltage on the high-voltage side. When the current suddenly changes, the value of the voltage output drop is detected, and the trigger threshold is set to control the triggering of the protection mechanism. At the same time, according to the detected abnormal data, if the corresponding trigger threshold is not reached, an abnormal prediction is performed to remind the existence of abnormal test risks and improve the safety and comprehensiveness of the test.

[0103] Figure 3 A high-voltage variable-frequency power supply power supply test protection system 3 based on intelligent perception is provided in one embodiment of the present invention. The system includes: a memory 31 and a processor 32. The memory 31 contains a high-voltage variable-frequency power supply power supply test protection method program based on intelligent perception. When the high-voltage variable-frequency power supply power supply test protection method program based on intelligent perception is executed by the processor 32, the following steps are implemented:

[0104] Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, the historical power supply test instances are classified into types, and a power supply test database is constructed;

[0105] Monitor the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, perform anomaly detection based on the real-time power supply test monitoring information, and obtain anomaly detection information;

[0106] Preset a trigger threshold, determine whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, perform abnormal event prediction;

[0107] Acquire abnormal power test information of power test abnormalities, analyze abnormal data change patterns, and update the power test database.

[0108] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0109] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0110] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0111] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0112] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0113] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A high-voltage variable-frequency power supply power supply test protection method based on intelligent perception, characterized in that: include: Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, the historical power supply test instances are classified into types, and a power supply test database is constructed; Monitor the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, perform anomaly detection based on the real-time power supply test monitoring information, and obtain anomaly detection information; Preset a trigger threshold, determine whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, perform abnormal event prediction; Obtain abnormal power test information of abnormal power test, analyze abnormal data change pattern, and update the power test database; The performing of abnormality detection according to the real-time power supply test monitoring information to obtain abnormality detection information specifically includes: Based on the power supply test database, a normal test image and first feature information are obtained to form a training data set, and an abnormality detection model is constructed using RNN; VAE is introduced to optimize the model, the training data set is input into the variational autoencoder to obtain the potential feature distribution of the training data, which is saved in the hidden layer as a latent space representation, and the probability distribution of the latent variables is generated; Sampling latent variables according to the probability distribution of the latent variables, generating predicted values ​​according to the sampling results, obtaining true values ​​corresponding to the predicted values ​​to calculate reconstruction errors, calculating gradient values ​​according to the calculated reconstruction errors, and iteratively optimizing according to the gradient values ​​using a back propagation algorithm; Based on the training data, the original feature samples are obtained, and the corresponding feature reconstruction samples are obtained through the variational encoder network. The deviation between the feature reconstruction samples and the original samples is calculated, and the deviation is compared with the preset threshold. The data corresponding to the samples greater than the preset threshold are regarded as abnormal data; The target high-voltage variable-frequency power supply test is monitored to obtain real-time power supply test monitoring information, and the real-time power supply test monitoring information is preprocessed and input into the anomaly detection model for detection to obtain anomaly detection information.

2. According to a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception according to claim 1, it is characterized in that: The method of obtaining historical power supply test instances of high-voltage variable-frequency power supplies of different specifications based on big data retrieval, classifying the historical power supply test instances by type, and constructing a power supply test database specifically includes: Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, and each historical power supply test instance is divided into a normal test instance and an abnormal test instance according to the test protection information corresponding to each historical power supply test instance; Extract features of normal test instances and abnormal test instances respectively to obtain first feature information and second feature information, construct a test feature change graph of the normal test instance based on the first feature information, and construct a normal test portrait; Calculate the Mahalanobis distance between each abnormal feature of each abnormal test instance and the abnormal test event, and compare it with the preset threshold, remove irrelevant features, and sort the Mahalanobis distances corresponding to the removed abnormal features in descending order; According to the sorting results, the root cause of the abnormal features corresponding to each historical abnormal instance is analyzed, and an abnormal test profile is constructed; The normal test profile and the abnormal test profile are combined to form a power supply test database.

3. According to a high-voltage variable-frequency power supply power supply test protection method based on intelligent perception as described in claim 1, it is characterized in that: The preset trigger threshold determines whether to perform power protection control according to the abnormality detection information. If the trigger threshold is not reached, abnormal event prediction is performed, specifically including: Acquire abnormality detection information, perform feature extraction on the abnormality detection information to obtain each abnormal test data and corresponding test process features, and obtain abnormal test feature information; Acquire the test data benchmark corresponding to each test process according to the abnormal test feature information, and calculate the test data benchmark and the abnormal test feature to obtain the deviation between each abnormal test data and the benchmark data; A trigger threshold is preset, and the calculated deviation is compared with the trigger threshold. If the abnormal detection information is greater than the trigger threshold, the power protection mechanism is activated; If the abnormality detection information is less than the trigger threshold, an abnormal event prediction is performed based on the abnormality detection information, and an early warning prompt is issued.

4. The high-voltage variable-frequency power supply test protection method based on intelligent perception according to claim 3 is characterized in that: The abnormal event prediction based on the abnormal detection information and the early warning prompt specifically include: Based on big data retrieval, various power supply test abnormal events are obtained, the test monitoring data corresponding to each abnormal event is extracted, and the test monitoring curve corresponding to each abnormal event is constructed; Extract abnormal test monitoring features for each power supply test abnormal event, mark the test monitoring curves corresponding to each abnormal event, and form an instance data set; Constructing an abnormal event prediction model, introducing a Markov chain to optimize the model, training the abnormal event prediction model based on the instance data set, and forming a state space based on the input training data; Preset time steps, define the transfer state according to the test monitoring curve of each abnormal event, and calculate the state transfer probability corresponding to each time step. Construct a state transfer matrix based on the calculated state transfer probability, perform model training and optimization, and obtain an abnormal event prediction model that meets expectations; The abnormality detection information is input into the abnormal event prediction model for analysis, and features are extracted from the input abnormality detection information to form a feature sequence; Analyze and obtain a state transfer matrix in the constructed state space according to the feature sequence, and predict the next state according to the state transfer matrix to obtain abnormal event prediction information; An early warning is issued for the current high-voltage variable-frequency power supply test based on the abnormal event prediction information.

5. The high-voltage variable-frequency power supply power supply test protection method based on intelligent perception according to claim 1 is characterized in that: The obtaining of abnormal power supply test information of abnormal power supply test, analyzing the abnormal data change mode, and updating the power supply test database specifically includes: Acquire abnormal power supply test information of a power supply test abnormality, and construct a power supply test data change diagram of a current power supply test abnormality event based on the abnormal power supply test information; Calculate the similarity between the power supply test data change graph and the test images in the power supply test database, and make a judgment with a preset threshold; If the similarity value is less than the preset threshold, it means that the current power supply abnormality test is an abnormal test instance that does not exist in the database, and an abnormal test portrait is generated according to the power supply test data change diagram of the current power supply test abnormal event, and the power supply test database is updated; If the similarity value is greater than the preset threshold, it means that similar abnormal test portraits are stored in the current power supply test database, and the similar portraits are sorted to extract the similar portrait with the highest similarity; Perform difference feature analysis on the similar portrait with the highest similarity and the power supply test data change graph of the current power supply test abnormal event, and update the similar portrait with the highest similarity according to the difference features obtained by the analysis.

6. A high-voltage variable-frequency power supply test protection system based on intelligent perception, characterized in that: The system includes: a memory and a processor, wherein the memory contains a high-voltage variable-frequency power supply power supply test protection method program based on intelligent perception, and the high-voltage variable-frequency power supply power supply test protection method program based on intelligent perception is executed by the processor to implement the following steps: Based on big data retrieval, historical power supply test instances of high-voltage variable-frequency power supplies of different specifications are obtained, the historical power supply test instances are classified into types, and a power supply test database is constructed; Monitor the target high-voltage variable-frequency power supply test to obtain real-time power supply test monitoring information, perform anomaly detection based on the real-time power supply test monitoring information, and obtain anomaly detection information; Preset a trigger threshold, determine whether to perform power protection control according to the abnormality detection information, and if the trigger threshold is not reached, perform abnormal event prediction; Obtain abnormal power test information of abnormal power test, analyze abnormal data change pattern, and update the power test database; The performing of abnormality detection according to the real-time power supply test monitoring information to obtain abnormality detection information specifically includes: Based on the power supply test database, a normal test image and first feature information are obtained to form a training data set, and an abnormality detection model is constructed using RNN; VAE is introduced to optimize the model, the training data set is input into the variational autoencoder to obtain the potential feature distribution of the training data, which is saved in the hidden layer as a latent space representation, and the probability distribution of the latent variables is generated; Sampling latent variables according to the probability distribution of the latent variables, generating predicted values ​​according to the sampling results, obtaining true values ​​corresponding to the predicted values ​​to calculate reconstruction errors, calculating gradient values ​​according to the calculated reconstruction errors, and iteratively optimizing according to the gradient values ​​using a back propagation algorithm; Based on the training data, the original feature samples are obtained, and the corresponding feature reconstruction samples are obtained through the variational encoder network. The deviation between the feature reconstruction samples and the original samples is calculated, and the deviation is compared with the preset threshold. The data corresponding to the samples greater than the preset threshold are regarded as abnormal data; The target high-voltage variable-frequency power supply test is monitored to obtain real-time power supply test monitoring information, and the real-time power supply test monitoring information is preprocessed and input into the anomaly detection model for detection to obtain anomaly detection information.

Images