Converter valve anomaly detection method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of power system operation and control, in particular to a converter valve anomaly detection method and device, computer equipment and a storage medium. The method comprises the steps of obtaining target operation data of a to-be-detected converter valve at a current moment; performing feature relationship analysis on the historical operation data and the target operation data of the to-be-detected converter valve to obtain a comprehensive feature code corresponding to the to-be-detected converter valve; and performing anomaly detection on the to-be-detected converter valve according to the comprehensive feature code to obtain an anomaly detection result of the to-be-detected converter valve. According to the invention, the method achieves the quick and effective anomaly detection of the converter valve, effectively improves the operation safety and stability of the converter valve, greatly reduces the unplanned shutdown, reduces the economic loss and maintenance cost caused by a fault, and provides a powerful technical guarantee for the efficient operation of a converter station.

Description

Technical Field

[0001] The present application relates to the technical field of power system operation and control, and particularly to an abnormal detection method, device, computer device, and storage medium for a converter valve. Background Art

[0002] UHV power transmission technology is the core technology in the field of modern power transmission, designed specifically for long-distance and large-capacity electric energy transmission. It breaks through the technical bottleneck of traditional power transmission methods, enabling the efficient transmission of abundant energy (such as hydropower, wind energy, and solar energy) in remote areas to load centers with dense populations, providing technical support for the optimal allocation of energy resources.

[0003] In a UHV DC power transmission system, the converter station, as a key node for AC-DC energy conversion, directly determines the operation efficiency and stability of the power transmission system. The converter valve, the core component of the converter station, generates a large amount of heat when operating in a high-voltage and large-current environment. To ensure the safe and stable operation of the converter valve, a valve cooling system has emerged. The valve cooling system effectively avoids equipment failures or performance degradation caused by overheating by precisely controlling the operating temperature of the converter valve, and is a key technical means to ensure the reliable operation of the converter station.

[0004] However, relying solely on the temperature control of the valve cooling system is not sufficient to comprehensively ensure the safe operation of the converter valve. Under complex operating conditions of high voltage and large current, the operating state of the converter valve is affected by various factors, such as ambient temperature fluctuations, load changes, and equipment aging. These factors may cause early abnormalities in the converter valve, and traditional manual monitoring or alarm methods based on fixed thresholds often have difficulty in detecting potential risks in a timely manner, thus increasing the possibility of equipment failures and outages. Summary of the Invention

[0005] Based on this, it is necessary to provide an abnormal detection method, device, computer device, and storage medium for a converter valve that can detect potential risks in a timely manner for the above technical problems.

[0006] In the first aspect, the present application provides an abnormal detection method for a converter valve. The method includes:

[0007] Obtain the target operation data of the converter valve to be detected at the current moment;

[0008] Conduct a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain a comprehensive feature code corresponding to the converter valve to be detected;

[0009] Perform abnormal detection on the converter valve to be detected according to the comprehensive feature code to obtain an abnormal detection result of the converter valve to be detected.

[0010] In one of the embodiments, the analysis of the characteristic relationship between the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive characteristic code corresponding to the converter valve to be detected includes:

[0011] Perform discretization processing on the target operation data to obtain sample data;

[0012] Determine the trend data corresponding to the historical operation data according to the change trend of the historical operation data;

[0013] Perform vector embedding processing on the sample data and the trend data to obtain vector embedding codes; wherein, the vector embedding codes include the first vector embedding code corresponding to the sample data and the second vector embedding code corresponding to the trend data;

[0014] Construct a feature correlation matrix according to the sample data and the trend data to obtain the autocorrelation matrix and the random permutation matrix of the converter valve to be detected; wherein, the autocorrelation matrix includes the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data;

[0015] Perform feature processing on the vector embedding codes, the autocorrelation matrix and the random permutation matrix to obtain a comprehensive feature code.

[0016] In one of the embodiments, the construction of the feature correlation matrix according to the sample data and the trend data to obtain the autocorrelation matrix and the random permutation matrix of the converter valve to be detected includes:

[0017] Perform inner product operations on the sample data and the trend data respectively to obtain the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data;

[0018] Perform position permutation processing on the elements included in the sample data and the trend data to obtain the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data.

[0019] In one of the embodiments, the performance of feature processing on the vector embedding codes, the autocorrelation matrix and the random permutation matrix to obtain a comprehensive feature code includes:

[0020] Perform permutation processing on the autocorrelation matrix according to the random permutation matrix to obtain a target matrix; wherein, the target matrix includes the first target matrix corresponding to the sample data and the second target matrix corresponding to the trend data;

[0021] Perform local feature extraction on the autocorrelation matrix and the target matrix based on a convolutional neural network (CNN) to obtain the autocorrelation encoding corresponding to the autocorrelation matrix and the target encoding corresponding to the target matrix;

[0022] Perform weighted processing on the autocorrelation encoding and the target encoding to obtain a first feature encoding;

[0023] Perform global feature interaction on the vector embedding encoding based on a multi-layer perceptron (MLP) to obtain a second feature encoding;

[0024] Perform weighted processing on the first feature encoding and the second feature encoding to obtain a comprehensive feature encoding.

[0025] In one embodiment, the abnormal detection of the converter valve to be detected according to the comprehensive feature encoding to obtain the abnormal detection result of the converter valve to be detected includes:

[0026] Perform attribute clustering on the comprehensive feature encoding respectively based on at least one unsupervised algorithm to obtain the clustering results corresponding to the respective unsupervised algorithms;

[0027] Determine the abnormal detection result of the converter valve to be detected according to the number of algorithms whose clustering results for the respective unsupervised algorithms are abnormal attributes.

[0028] In one embodiment, the determining the abnormal detection result of the converter valve to be detected according to the number of algorithms whose clustering results for the respective unsupervised algorithms are abnormal attributes includes:

[0029] If the clustering results corresponding to all the unsupervised algorithms are abnormal attributes, determine that the abnormal detection result of the converter valve to be detected is an abnormal condition; if there is at least one normal attribute in the clustering results corresponding to the respective unsupervised algorithms, determine that the abnormal detection result of the converter valve to be detected is a normal condition.

[0030] In a second aspect, the present application also provides an abnormal detection device for a converter valve. The device includes:

[0031] An acquisition module, configured to acquire the target operation data of the converter valve to be detected at the current moment;

[0032] An analysis module, configured to perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature encoding corresponding to the converter valve to be detected;

[0033] A detection module, configured to perform abnormal detection on the converter valve to be detected according to the comprehensive feature encoding to obtain the abnormal detection result of the converter valve to be detected.

[0034] In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0035] Obtain the target operation data of the converter valve to be detected at the current moment;

[0036] Conduct a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected;

[0037] Perform anomaly detection on the converter valve to be detected according to the comprehensive feature code to obtain the anomaly detection result of the converter valve to be detected.

[0038] In a fourth aspect, the present application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0039] Obtain the target operation data of the converter valve to be detected at the current moment;

[0040] Conduct a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected;

[0041] Perform anomaly detection on the converter valve to be detected according to the comprehensive feature code to obtain the anomaly detection result of the converter valve to be detected.

[0042] In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0043] Obtain the target operation data of the converter valve to be detected at the current moment;

[0044] Conduct a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected;

[0045] Perform anomaly detection on the converter valve to be detected according to the comprehensive feature code to obtain the anomaly detection result of the converter valve to be detected.

[0046] The above abnormal detection method, device, computer equipment and storage medium of the converter valve obtain the target operation data of the converter valve to be detected at the current moment, and perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected; furthermore, the abnormal detection of the converter valve to be detected is performed according to the comprehensive feature code, and the abnormal detection result of the converter valve to be detected is obtained. According to the above content, it can be known that in the process of abnormal detection of the converter valve in this application, feature relationship analysis will be combined with historical operation data and target operation data. Since the target operation data can include various types of data such as temperature, water flow of the valve cooling system, conductivity of cooling water, and electrical parameters, it is possible to represent the actual allowable situation of the converter valve to be detected through the comprehensive feature code. Furthermore, the abnormal detection of the converter valve to be detected is performed according to the comprehensive feature code, and the abnormal detection result of the converter valve to be detected is obtained. During the detection process, there is no need to manually label a large amount of data, reducing the manual cost and time cost, realizing fast and effective abnormal detection of the converter valve, effectively improving the safety and stability of the converter valve operation, greatly reducing unplanned outages, and reducing the economic losses and maintenance costs caused by faults, providing strong technical support for the efficient operation of the converter station. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 FIG. is an application environment diagram of an abnormal detection method for a converter valve provided by an embodiment of the present application;

[0048] Figure 2 FIG. is a flowchart of a first abnormal detection method for a converter valve provided by an embodiment of the present application;

[0049] Figure 3 FIG. is a flowchart of a second abnormal detection method for a converter valve provided by an embodiment of the present application;

[0050] Figure 4 FIG. is a flowchart of a third abnormal detection method for a converter valve provided by an embodiment of the present application;

[0051] Figure 5 FIG. is a flowchart of a fourth abnormal detection method for a converter valve provided by an embodiment of the present application;

[0052] Figure 6 FIG. is a structural block diagram of an abnormal detection device for a converter valve provided by an embodiment of the present application;

[0053] Figure 7 FIG. is an internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] To make the objectives, technical solutions, and advantages of this application more clear and understandable, the following further elaborates on this application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely used to explain this application and are not intended to limit this application.

[0055] The abnormal detection method for a converter valve provided by an embodiment of this application can be applied to an application environment as shown in Figure 1 the following. Among them, the terminal 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or placed in the cloud or on other network servers. Obtain the target operation data of the converter valve to be detected at the current moment, and perform a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected; furthermore, perform abnormal detection on the converter valve to be detected according to the comprehensive feature code to obtain the abnormal detection result of the converter valve to be detected. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

[0056] In one embodiment, as shown in Figure 2 the following, a method for detecting abnormalities in a converter valve is provided. Taking the method applied to the server 104 in Figure 1 as an example, the method includes the following steps:

[0057] S201, obtain the target operation data of the converter valve to be detected at the current moment.

[0058] It should be noted that the target operation data can include, but is not limited to, various types of data such as temperature, water flow rate of the valve cooling system, conductivity of the cooling water, and electrical parameters. Therefore, when it is necessary to obtain the target operation data of the converter valve to be detected at the current moment, different types of sensors can be pre-set at at least one detection point corresponding to the converter valve to realize the collection of operation data of the converter valve to be detected through different types of sensors, so as to realize the operation of obtaining the target operation data of the converter valve to be detected at the current moment.

[0059] S202, perform a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected.

[0060] It should be noted that since the target operation data contains a large amount of data and has a high data dimension, in order to ensure that the converter valve to be detected can be accurately detected for abnormalities in the follow-up, it is necessary to fully obtain the data relationships in the target operation data. Moreover, due to the dynamic changes in the operating environment and load conditions of the converter valve to be detected, it will also affect the accuracy of the abnormality detection (for example, factors such as seasonal temperature fluctuations, rainfall changes, or load fluctuations may cause normal fluctuations in the target operation data). Therefore, in order to avoid the impact of the normal fluctuations in the target operation data on the detection accuracy, when obtaining the data relationships in the target operation data, it is also necessary to combine the historical operation data of the converter valve to be detected to realize the dynamic learning of the change trend of the data.

[0061] In an embodiment of the present application, when it is necessary to perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature code corresponding to the converter valve to be detected, the target operation data and the historical operation data can be preprocessed to obtain sample data and trend data. Furthermore, feature relationship analysis is performed on the sample data and the trend data to obtain the comprehensive feature code corresponding to the converter valve to be detected, so as to ensure that the comprehensive feature code can more effectively represent the actual operation situation of the converter valve to be detected.

[0062] S203, perform abnormality detection on the converter valve to be detected according to the comprehensive feature code to obtain the abnormality detection result of the converter valve to be detected.

[0063] In an embodiment of the present application, a normal feature code and an abnormal feature code can be pre-constructed. Furthermore, after obtaining the comprehensive feature code, the similarities between the comprehensive feature code and the normal feature code and the abnormal feature code can be verified respectively; if the similarity between the comprehensive feature code and the normal feature code is higher than that of the abnormal feature code, it indicates that the abnormality detection result of the converter valve to be detected is normal; if the similarity between the comprehensive feature code and the abnormal feature code is higher than that of the normal feature code, it indicates that the abnormality detection result of the converter valve to be detected is abnormal.

[0064] The above abnormal detection method for the converter valve obtains the target operation data of the converter valve to be detected at the current moment, analyzes the characteristic relationship between the historical operation data and the target operation data of the converter valve to be detected, and obtains the comprehensive characteristic code corresponding to the converter valve to be detected; furthermore, according to the comprehensive characteristic code, the abnormal detection of the converter valve to be detected is carried out to obtain the abnormal detection result of the converter valve to be detected. According to the above content, it can be seen that in the process of abnormal detection of the converter valve in this application, the characteristic relationship analysis will be combined with the historical operation data and the target operation data. Since the target operation data can include various types of data such as temperature, water flow of the valve cooling system, conductivity of the cooling water, and electrical parameters, in order to realize the actual allowable situation of the converter valve to be detected through the comprehensive characteristic code. Furthermore, according to the comprehensive characteristic code, the abnormal detection of the converter valve to be detected is carried out to obtain the abnormal detection result of the converter valve to be detected. During the detection process, there is no need to manually label a large amount of data, reducing the labor cost and time cost, realizing fast and effective abnormal detection of the converter valve, effectively improving the safety and stability of the converter valve operation, greatly reducing the unplanned outage, reducing the economic loss and maintenance cost caused by the failure, and providing a strong technical guarantee for the efficient operation of the converter station.

[0065] In one embodiment, as Figure 3 shown, when it is necessary to analyze the characteristic relationship between the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive characteristic code corresponding to the converter valve to be detected, the following specific contents may be included:

[0066] S301, discretize the target operation data to obtain sample data.

[0067] It should be noted that when it is necessary to discretize the target operation data, different discretization rules for different types of target operation data can be pre-constructed. Among them, different discretization rules record different discrete levels corresponding to multiple value ranges of different types of target operation data, so as to eliminate the redundancy brought by sparse features and improve the representativeness of the data.

[0068] As an example, the temperature in the target operation data can be divided into five discrete levels: "very high", "high", "medium", "low", and "very low"; among them, each discrete level corresponds to a different value range of temperature.

[0069] Furthermore, although the ambient temperature and the cooling water temperature in the target operation data are both temperatures, they cannot be discretized by the same set of discretization rules. Therefore, the ambient temperature and the cooling water temperature have different influences on the operation state of the converter valve.

[0070] S302, determine the trend data corresponding to the historical operation data according to the change trend of the historical operation data.

[0071] It should be noted that when it is necessary to determine the trend data corresponding to the historical operation data, the rising and falling changes of the historical operation data can be analyzed in advance, and then the rising and falling conditions of the historical operation data are represented in binary, so as to obtain the trend data corresponding to the historical operation data.

[0072] As an example, if a certain historical operation data is: [1, 2, 3, 2, 1]; then the rising and falling changes of this historical operation data are "rising, rising, falling, falling"; it is pre-specified that in binary representation, 1 represents the data rising; 0 represents the data falling. Therefore, the trend data of this historical operation data can be represented as 1100.

[0073] S303. Perform vector embedding processing on the sample data and the trend data to obtain a vector embedding code.

[0074] Among them, the vector embedding code includes a first vector embedding code corresponding to the sample data and a second vector embedding code corresponding to the trend data.

[0075] In an embodiment of the present application, when performing vector embedding processing on the sample data and the trend data, an embedding (vector embedding) model can be used to implement the operation of performing vector embedding processing on the sample data and the trend data to obtain a vector embedding code; among them, the embedding model used for performing vector embedding processing on the sample data and the embedding model used for performing vector embedding processing on the trend data are different models.

[0076] Among them, performing vector embedding processing on the sample data and the trend data to obtain a vector embedding code can be expressed as:

[0077] E = Embedding(X);

[0078] Among them, E refers to the vector embedding code; X refers to the sample data and the trend data.

[0079] Specifically, by inputting the sample data into the embedding model, a first vector embedding code corresponding to the sample data is obtained; by inputting the trend data into the embedding model, a second vector embedding code corresponding to the trend data is obtained.

[0080] S304. Construct a feature correlation matrix according to the sample data and the trend data to obtain an autocorrelation matrix and a random permutation matrix of the converter valve to be detected.

[0081] Among them, the autocorrelation matrix includes a first autocorrelation matrix corresponding to the sample data and a second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes a first random permutation matrix corresponding to the sample data and a second random permutation matrix corresponding to the trend data.

[0082] It should be noted that when constructing a feature correlation matrix based on sample data and trend data to obtain the autocorrelation matrix and random permutation matrix of the converter valve to be detected, the following specific contents may be included: performing inner product operations on the sample data and trend data respectively to obtain the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data; performing position permutation processing on the elements included in the sample data and trend data to obtain the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data.

[0083] In an embodiment of the present application, when performing inner product operations on the sample data and trend data respectively, matrix inner product can be performed on the first vector embedding encoding of the sample data and the second vector embedding encoding of the trend data, thereby obtaining the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data.

[0084] Specifically, substituting the sample data and trend data into the following formula to complete the matrix inner product processing:

[0085] A = EE T ;

[0086] Among them, A refers to the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data, and E refers to the first vector embedding encoding of the sample data and the second vector embedding encoding of the trend data.

[0087] Among them, the shape matrix A[N, N] is the constructed correlation matrix, and each element in the matrix represents the relationship between two features. Any combination of these elements will expose the corresponding potential feature association patterns. Therefore, all possible relationship patterns between features can be exposed in matrix form through matrix A.

[0088] In an embodiment of the present application, when performing position permutation processing on the elements included in the sample data and trend data, the positions of the elements included in the sample data and trend data can be rearranged respectively to make the originally distant elements closer and the originally close elements farther.

[0089] Taking the N elements included in the sample data as an example, rearrange the positions of the N elements (1, 2, 3, ···, N) to ensure that the originally adjacent elements are no longer adjacent, and the non-adjacent elements become adjacent. Arrange the N elements into an R×L matrix, where the sizes of N, R, and L are approximately equal (N = R×L). Transpose this matrix and then reshape it into an N×1 vector. This newly obtained N×1 vector represents the transformed positions of the original data. Next, create a zero matrix M of size N×n. According to the position relationship in the N×1 vector, place the N elements at the corresponding row and column positions of the zero matrix M to obtain the first random permutation matrix corresponding to the sample data; the process of obtaining the second random permutation matrix corresponding to the trend data is the same as above.

[0090] As an example, if the N elements are 1 - 9 respectively, arrange the numbers 1 to 9 into a 3×3 matrix [[1, 2, 3], [4, 5, 6], [7, 8, 9]]. Perform permutation and reshaping on the matrix to obtain a new sequence [1, 4, 7, 2, 5, 8, 3, 6, 9] (corresponding to the above N×1 vector). Using this sequence, place the nine elements at the corresponding row and column positions of the zero matrix M to obtain the random permutation matrix.

[0091] S305, perform feature processing on the vector embedding encoding, autocorrelation matrix, and random permutation matrix to obtain the comprehensive feature encoding.

[0092] It should be noted that when performing feature processing on the vector embedding encoding, autocorrelation matrix, and random permutation matrix to obtain the comprehensive feature encoding, the following content may be included: perform permutation processing on the autocorrelation matrix according to the random permutation matrix to obtain the target matrix; among them, the target matrix includes the first target matrix corresponding to the sample data and the second target matrix corresponding to the trend data; based on the convolutional neural network CNN, perform local feature extraction on the autocorrelation matrix and the target matrix to obtain the autocorrelation encoding corresponding to the autocorrelation matrix and the target encoding corresponding to the target matrix; perform weighted processing on the autocorrelation encoding and the target encoding to obtain the first feature encoding; based on the multi-layer perceptron MLP, perform global feature interaction on the vector embedding encoding to obtain the second feature encoding; perform weighted processing on the first feature encoding and the second feature encoding to obtain the comprehensive feature encoding.

[0093] In an embodiment of the present application, when performing permutation processing on the autocorrelation matrix according to the random permutation matrix to obtain the target matrix, the operation of performing permutation processing on the autocorrelation matrix according to the random permutation matrix can be realized by substituting the random permutation matrix and the autocorrelation matrix into the following formula:

[0094]

[0095] Among them, P refers to the target matrix; M refers to the random permutation matrix; A refers to the autocorrelation matrix. The target matrix can effectively determine the features of long-range correlation during the convolution operation.

[0096] In an embodiment of the present application, the CNN can effectively focus on the local elements in the matrix. By increasing the kernel size or the number of layers to expand the receptive field of the CNN, different combinations of elements can be effectively sampled from the matrix; moreover, when performing local feature extraction on the autocorrelation matrix and the target matrix based on the CNN, the CNN is only equipped with a lightweight structure (for example, the kernel size is 3×3, and the number of channels is 8 and 16).

[0097] In an embodiment of the present application, during the process of weighting the autocorrelation encoding and the target encoding, the influence weights for the autocorrelation encoding and the target encoding can be preset in advance. Then, according to the influence weights corresponding to the autocorrelation encoding and the target encoding respectively, the autocorrelation encoding and the target encoding are weighted to obtain the first feature encoding. Similarly, when weighting the first feature encoding and the second feature encoding, the influence weights for the first feature encoding and the second feature encoding can also be preset in advance. Then, according to the influence weights corresponding to the first feature encoding and the second feature encoding respectively, the first feature encoding and the second feature encoding are weighted to obtain the comprehensive feature encoding.

[0098] The above abnormal detection method for the converter valve realizes the construction of the vector embedding encoding, the autocorrelation matrix, and the random permutation matrix through the sample data and the trend data obtained by preprocessing. Then, feature processing is performed on the vector embedding encoding, the autocorrelation matrix, and the random permutation matrix to obtain the comprehensive feature encoding, and it is realized that the operation condition of the converter valve to be detected can be effectively reflected according to the comprehensive feature encoding, providing a basic guarantee for accurately detecting the abnormality of the converter valve to be detected subsequently.

[0099] In one embodiment, as Figure 4 shown, when it is necessary to perform abnormal detection on the converter valve to be detected based on the comprehensive feature encoding to obtain the abnormal detection result of the converter valve to be detected, the following specific content may be included:

[0100] S401, perform attribute clustering on the comprehensive feature encoding respectively based on at least one unsupervised algorithm to obtain the clustering results corresponding to each unsupervised algorithm.

[0101] Among them, at least one unsupervised algorithm may include but is not limited to: HBOS unsupervised algorithm, KMeans unsupervised algorithm, EllipticEnvelope unsupervised algorithm, etc.

[0102] S402, determine the abnormal detection result of the converter valve to be detected according to the number of algorithms whose clustering results for each unsupervised algorithm are abnormal attributes.

[0103] In an embodiment of the present application, the correspondence between the number of algorithms with abnormal attributes in the clustering results and the abnormal detection results can be specified in advance, and then the abnormal detection results of the converter valve to be detected can be determined according to the correspondence.

[0104] Specifically, if the clustering results corresponding to each unsupervised algorithm are all abnormal attributes, it is determined that the abnormal detection result of the converter valve to be detected is an abnormal situation; if there is at least one normal attribute in the clustering results corresponding to each unsupervised algorithm, it is determined that the abnormal detection result of the converter valve to be detected is a normal situation.

[0105] The above abnormal detection method for the converter valve respectively performs attribute clustering on the comprehensive feature encoding through at least one unsupervised algorithm, so as to determine the abnormal detection results of the converter valve to be detected according to the number of algorithms with abnormal attributes in the clustering results of each unsupervised algorithm, realizing fast and effective abnormal detection for the converter valve, effectively improving the safety and stability of the converter valve operation, greatly reducing unplanned outages, and reducing the economic losses and maintenance costs caused by faults, providing strong technical support for the efficient operation of the converter station.

[0106] In an embodiment, as Figure 5 shown, when it is necessary to determine the abnormal detection results of the converter valve to be detected, the following contents may be included:

[0107] S501, Obtain the target operation data of the converter valve to be detected at the current moment.

[0108] S502, Discretize the target operation data to obtain sample data.

[0109] S503, Determine the trend data corresponding to the historical operation data according to the change trend of the historical operation data.

[0110] S504, Perform vector embedding processing on the sample data and the trend data to obtain vector embedding encodings; wherein, the vector embedding encodings include the first vector embedding encoding corresponding to the sample data and the second vector embedding encoding corresponding to the trend data.

[0111] S505, Perform inner product operations on the sample data and the trend data respectively to obtain the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data.

[0112] S506, Perform position permutation processing on the elements included in the sample data and the trend data to obtain the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data.

[0113] S507, permute the autocorrelation matrix according to a random permutation matrix to obtain a target matrix; wherein, the target matrix includes a first target matrix corresponding to the sample data and a second target matrix corresponding to the trend data.

[0114] S508, perform local feature extraction on the autocorrelation matrix and the target matrix based on a convolutional neural network CNN to obtain an autocorrelation encoding corresponding to the autocorrelation matrix and a target encoding corresponding to the target matrix.

[0115] S509, perform weighted processing on the autocorrelation encoding and the target encoding to obtain a first feature encoding.

[0116] S510, perform global feature interaction on the vector embedding encoding based on a multi-layer perceptron MLP to obtain a second feature encoding.

[0117] S511, perform weighted processing on the first feature encoding and the second feature encoding to obtain a comprehensive feature encoding.

[0118] S512, perform attribute clustering on the comprehensive feature encoding respectively based on at least one unsupervised algorithm to obtain clustering results corresponding to each unsupervised algorithm.

[0119] S513, if the clustering results corresponding to each unsupervised algorithm are all abnormal attributes, determine that the abnormal detection result of the converter valve to be detected is an abnormal situation; if there is at least one normal attribute in the clustering results corresponding to each unsupervised algorithm, determine that the abnormal detection result of the converter valve to be detected is a normal situation.

[0120] For the above abnormal detection method of the converter valve, obtain the target operation data of the converter valve to be detected at the current moment, and perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain a comprehensive feature encoding corresponding to the converter valve to be detected; furthermore, perform abnormal detection on the converter valve to be detected according to the comprehensive feature encoding to obtain the abnormal detection result of the converter valve to be detected. According to the above content, it can be known that in the process of performing abnormal detection on the converter valve in this application, feature relationship analysis will be combined with historical operation data and target operation data. Since the target operation data can include various types of data such as temperature, water flow rate of the valve cooling system, conductivity of the cooling water, and electrical parameters, it is possible to represent the actual allowable situation of the converter valve to be detected through the comprehensive feature encoding. Furthermore, perform abnormal detection on the converter valve to be detected according to the comprehensive feature encoding to obtain the abnormal detection result of the converter valve to be detected. During the detection process, there is no need to perform manual annotation on a large amount of data, reducing the manual cost and time cost. It realizes fast and effective abnormal detection for the converter valve, effectively improves the safety and stability of the converter valve operation, greatly reduces unplanned outages, and reduces the economic losses and maintenance costs caused by faults, providing a strong technical guarantee for the efficient operation of the converter station.

[0121] It should be understood that although the steps in the flowcharts involved in the above embodiments are shown in sequence according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0122] Based on the same inventive concept, an embodiment of the present application also provides an abnormal detection device for a converter valve for implementing the abnormal detection method of the converter valve involved above. The implementation solution provided by this device to solve the problem is similar to the implementation solution recorded in the above method. Therefore, the specific limitations in one or more embodiments of the abnormal detection device for the converter valve provided below can refer to the limitations on the abnormal detection method of the converter valve in the above text, and will not be repeated here.

[0123] In one embodiment, as Figure 6 shown, an abnormal detection device for a converter valve is provided, including: an acquisition module 10, an analysis module 20, and a detection module 30, where:

[0124] The acquisition module 10 is configured to acquire the target operation data of the converter valve to be detected at the current moment.

[0125] The analysis module 20 is configured to perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected, and obtain the comprehensive feature code corresponding to the converter valve to be detected.

[0126] The detection module 30 is configured to perform abnormal detection on the converter valve to be detected according to the comprehensive feature code, and obtain the abnormal detection result of the converter valve to be detected.

[0127] In one embodiment, the target operation data is discretized to obtain sample data;

[0128] According to the change trend of the historical operation data, the trend data corresponding to the historical operation data is determined;

[0129] Perform vector embedding processing on the sample data and the trend data to obtain vector embedding codes; where the vector embedding codes include a first vector embedding code corresponding to the sample data and a second vector embedding code corresponding to the trend data;

[0130] Construct a feature correlation matrix based on the sample data and trend data to obtain the autocorrelation matrix and random permutation matrix of the converter valve to be detected; among them, the autocorrelation matrix includes the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data;

[0131] Perform feature processing on the vector embedding encoding, autocorrelation matrix and random permutation matrix to obtain the comprehensive feature encoding.

[0132] In one embodiment, perform an inner product operation on the sample data and trend data respectively to obtain the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data;

[0133] Perform a position permutation process on the elements included in the sample data and trend data to obtain the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data.

[0134] In one embodiment, perform a permutation process on the autocorrelation matrix according to the random permutation matrix to obtain the target matrix; among them, the target matrix includes the first target matrix corresponding to the sample data and the second target matrix corresponding to the trend data;

[0135] Based on the convolutional neural network CNN, perform local feature extraction on the autocorrelation matrix and the target matrix to obtain the autocorrelation encoding corresponding to the autocorrelation matrix and the target encoding corresponding to the target matrix;

[0136] Perform a weighted process on the autocorrelation encoding and the target encoding to obtain the first feature encoding;

[0137] Based on the multi-layer perceptron MLP, perform global feature interaction on the vector embedding encoding to obtain the second feature encoding;

[0138] Perform a weighted process on the first feature encoding and the second feature encoding to obtain the comprehensive feature encoding.

[0139] In one embodiment, based on at least one unsupervised algorithm, perform attribute clustering on the comprehensive feature encoding respectively to obtain the clustering results corresponding to each unsupervised algorithm;

[0140] Determine the anomaly detection result of the converter valve to be detected according to the number of algorithms with abnormal attributes in the clustering results of each unsupervised algorithm.

[0141] In one embodiment, if the clustering results corresponding to each unsupervised algorithm are all abnormal attributes, determine that the anomaly detection result of the converter valve to be detected is an abnormal situation; if there is at least one normal attribute in the clustering results corresponding to each unsupervised algorithm, determine that the anomaly detection result of the converter valve to be detected is a normal situation.

[0142] The abnormal detection device of the commutation valve obtains the target operation data of the commutation valve to be detected at the current moment, analyzes the characteristic relationship between the historical operation data and the target operation data of the commutation valve to be detected, and obtains the comprehensive characteristic code corresponding to the commutation valve to be detected; furthermore, the commutation valve to be detected is subjected to abnormal detection according to the comprehensive characteristic code, and the abnormal detection result of the commutation valve to be detected is obtained. According to the above content, it can be seen that in the process of abnormal detection of the commutation valve in this application, the characteristic relationship analysis will be combined with the historical operation data and the target operation data. Since the target operation data can include various types of data such as temperature, water flow of the valve cooling system, conductivity of the cooling water, and electrical parameters, the actual allowable situation of the commutation valve to be detected can be characterized by the comprehensive characteristic code. Furthermore, the commutation valve to be detected is subjected to abnormal detection according to the comprehensive characteristic code, and the abnormal detection result of the commutation valve to be detected is obtained. During the detection process, there is no need to manually annotate a large amount of data, which reduces the labor cost and time cost, realizes fast and effective abnormal detection of the commutation valve, effectively improves the safety and stability of the commutation valve operation, greatly reduces the unplanned outage, reduces the economic loss and maintenance cost caused by the failure, and provides a strong technical guarantee for the efficient operation of the converter station.

[0143] Each module in the abnormal detection device of the commutation valve can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in the processor of the computer device in the form of hardware or be independent of it, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0144] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 7As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it realizes an abnormal detection method for a converter valve. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the computer device housing, or an external keyboard, a touchpad, or a mouse, etc.

[0145] Those skilled in the art can understand that Figure 7 the structure shown in

[0146] is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0147] Obtain the target operation data of the converter valve to be detected at the current moment;

[0148] Perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature coding corresponding to the converter valve to be detected;

[0149] Perform abnormal detection on the converter valve to be detected according to the comprehensive feature coding to obtain the abnormal detection result of the converter valve to be detected.

[0150] In one embodiment, when the processor executes the computer program, the following steps are also implemented:

[0151] Perform discretization processing on the target operation data to obtain sample data;

[0152] Determine the trend data corresponding to the historical operation data according to the change trend of the historical operation data;

[0153] Perform vector embedding processing on the sample data and the trend data to obtain vector embedding codes; wherein, the vector embedding codes include a first vector embedding code corresponding to the sample data and a second vector embedding code corresponding to the trend data;

[0154] Construct a feature correlation matrix according to the sample data and the trend data to obtain an autocorrelation matrix and a random permutation matrix of the converter valve to be detected; wherein, the autocorrelation matrix includes a first autocorrelation matrix corresponding to the sample data and a second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes a first random permutation matrix corresponding to the sample data and a second random permutation matrix corresponding to the trend data;

[0155] Perform feature processing on the vector embedding codes, the autocorrelation matrix and the random permutation matrix to obtain a comprehensive feature code.

[0156] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0157] Perform inner product operations on the sample data and the trend data respectively to obtain a first autocorrelation matrix corresponding to the sample data and a second autocorrelation matrix corresponding to the trend data;

[0158] Perform position permutation processing on the elements included in the sample data and the trend data to obtain a first random permutation matrix corresponding to the sample data and a second random permutation matrix corresponding to the trend data.

[0159] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0160] Perform permutation processing on the autocorrelation matrix according to the random permutation matrix to obtain a target matrix; wherein, the target matrix includes a first target matrix corresponding to the sample data and a second target matrix corresponding to the trend data;

[0161] Perform local feature extraction on the autocorrelation matrix and the target matrix based on the convolutional neural network CNN to obtain an autocorrelation code corresponding to the autocorrelation matrix and a target code corresponding to the target matrix;

[0162] Perform weighted processing on the autocorrelation code and the target code to obtain a first feature code;

[0163] Perform global feature interaction on the vector embedding code based on the multi-layer perceptron MLP to obtain a second feature code;

[0164] Perform weighted processing on the first feature code and the second feature code to obtain a comprehensive feature code.

[0165] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0166] Perform attribute clustering on the comprehensive feature encoding respectively based on at least one unsupervised algorithm to obtain the clustering results corresponding to each unsupervised algorithm;

[0167] Determine the anomaly detection result of the converter valve to be detected according to the number of algorithms with abnormal attributes in the clustering results of each unsupervised algorithm.

[0168] In one embodiment, when the processor executes the computer program, the following steps are further implemented:

[0169] If the clustering results corresponding to each unsupervised algorithm are all abnormal attributes, determine that the anomaly detection result of the converter valve to be detected is an abnormal situation; if there is at least one normal attribute in the clustering results corresponding to each unsupervised algorithm, determine that the anomaly detection result of the converter valve to be detected is a normal situation.

[0170] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0171] Obtain the target operation data of the converter valve to be detected at the current moment;

[0172] Perform feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature encoding corresponding to the converter valve to be detected;

[0173] Perform anomaly detection on the converter valve to be detected according to the comprehensive feature encoding to obtain the anomaly detection result of the converter valve to be detected.

[0174] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0175] Perform discretization processing on the target operation data to obtain sample data;

[0176] Determine the trend data corresponding to the historical operation data according to the change trend of the historical operation data;

[0177] Perform vector embedding processing on the sample data and the trend data to obtain vector embedding encoding; wherein, the vector embedding encoding includes the first vector embedding encoding corresponding to the sample data and the second vector embedding encoding corresponding to the trend data;

[0178] Construct a feature correlation matrix based on the sample data and trend data to obtain the autocorrelation matrix and random permutation matrix of the converter valve to be detected; among them, the autocorrelation matrix includes the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data;

[0179] Perform feature processing on the vector embedding encoding, autocorrelation matrix, and random permutation matrix to obtain the comprehensive feature encoding.

[0180] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0181] Perform inner product operations on the sample data and trend data respectively to obtain the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data;

[0182] Perform position permutation processing on the elements included in the sample data and trend data to obtain the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data.

[0183] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0184] Perform permutation processing on the autocorrelation matrix according to the random permutation matrix to obtain the target matrix; among them, the target matrix includes the first target matrix corresponding to the sample data and the second target matrix corresponding to the trend data;

[0185] Based on the convolutional neural network CNN, perform local feature extraction on the autocorrelation matrix and the target matrix to obtain the autocorrelation encoding corresponding to the autocorrelation matrix and the target encoding corresponding to the target matrix;

[0186] Perform weighted processing on the autocorrelation encoding and the target encoding to obtain the first feature encoding;

[0187] Based on the multi-layer perceptron MLP, perform global feature interaction on the vector embedding encoding to obtain the second feature encoding;

[0188] Perform weighted processing on the first feature encoding and the second feature encoding to obtain the comprehensive feature encoding.

[0189] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented:

[0190] Based on at least one unsupervised algorithm, perform attribute clustering on the comprehensive feature encoding respectively to obtain the clustering results corresponding to each unsupervised algorithm;

[0191] Determine the anomaly detection result of the converter valve to be detected according to the number of algorithms with abnormal attributes in the clustering results of each unsupervised algorithm.

[0192] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0193] If the clustering results corresponding to each unsupervised algorithm are all abnormal attributes, determine that the abnormal detection result of the converter valve to be detected is an abnormal situation; if there is at least one normal attribute in the clustering results corresponding to each unsupervised algorithm, determine that the abnormal detection result of the converter valve to be detected is a normal situation.

[0194] In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0195] Obtain the target operation data of the converter valve to be detected at the current moment;

[0196] Conduct a feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected to obtain the comprehensive feature coding corresponding to the converter valve to be detected;

[0197] Conduct an abnormal detection on the converter valve to be detected according to the comprehensive feature coding to obtain the abnormal detection result of the converter valve to be detected.

[0198] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0199] Perform discretization processing on the target operation data to obtain sample data;

[0200] Determine the trend data corresponding to the historical operation data according to the change trend of the historical operation data;

[0201] Perform vector embedding processing on the sample data and the trend data to obtain vector embedding coding; wherein, the vector embedding coding includes the first vector embedding coding corresponding to the sample data and the second vector embedding coding corresponding to the trend data;

[0202] Construct a feature correlation matrix according to the sample data and the trend data to obtain the autocorrelation matrix and the random permutation matrix of the converter valve to be detected; wherein, the autocorrelation matrix includes the first autocorrelation matrix corresponding to the sample data and the second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes the first random permutation matrix corresponding to the sample data and the second random permutation matrix corresponding to the trend data;

[0203] Perform feature processing on the vector embedding coding, the autocorrelation matrix and the random permutation matrix to obtain the comprehensive feature coding.

[0204] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0205] Perform an inner product operation on the sample data and the trend data respectively to obtain a first autocorrelation matrix corresponding to the sample data and a second autocorrelation matrix corresponding to the trend data;

[0206] Perform a position permutation process on the elements included in the sample data and the trend data to obtain a first random permutation matrix corresponding to the sample data and a second random permutation matrix corresponding to the trend data.

[0207] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0208] Perform a permutation process on the autocorrelation matrix according to the random permutation matrix to obtain a target matrix; wherein, the target matrix includes a first target matrix corresponding to the sample data and a second target matrix corresponding to the trend data;

[0209] Based on the convolutional neural network CNN, perform local feature extraction on the autocorrelation matrix and the target matrix to obtain an autocorrelation encoding corresponding to the autocorrelation matrix and a target encoding corresponding to the target matrix;

[0210] Perform a weighting process on the autocorrelation encoding and the target encoding to obtain a first feature encoding;

[0211] Based on the multi-layer perceptron MLP, perform global feature interaction on the vector embedding encoding to obtain a second feature encoding;

[0212] Perform a weighting process on the first feature encoding and the second feature encoding to obtain a comprehensive feature encoding.

[0213] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0214] Based on at least one unsupervised algorithm, perform attribute clustering on the comprehensive feature encoding respectively to obtain a clustering result corresponding to each unsupervised algorithm;

[0215] Determine the anomaly detection result of the converter valve to be detected according to the number of algorithms with abnormal attributes in the clustering results of each unsupervised algorithm.

[0216] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0217] If the clustering results corresponding to each unsupervised algorithm are all abnormal attributes, determine that the anomaly detection result of the converter valve to be detected is an abnormal situation; if there is at least one normal attribute in the clustering results corresponding to each unsupervised algorithm, determine that the anomaly detection result of the converter valve to be detected is a normal situation.

[0218] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties. Moreover, the collection, use, and processing of relevant data need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0219] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., and are not limited thereto. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., and are not limited thereto.

[0220] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope described in this specification.

[0221] The above embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. An abnormal detection method for a commutation valve, characterized in that, The method includes: Obtaining the target operation data of the converter valve to be detected at the current moment; Performing feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected, and obtaining the comprehensive feature code corresponding to the converter valve to be detected; Performing anomaly detection on the converter valve to be detected according to the comprehensive feature code, and obtaining the anomaly detection result of the converter valve to be detected.

2. The method according to claim 1, wherein The performing feature relationship analysis on the historical operation data and the target operation data of the converter valve to be detected, and obtaining the comprehensive feature code corresponding to the converter valve to be detected includes: Performing discretization processing on the target operation data to obtain sample data; Determining the trend data corresponding to the historical operation data according to the change trend of the historical operation data; Performing vector embedding processing on the sample data and the trend data to obtain a vector embedding code; wherein, the vector embedding code includes a first vector embedding code corresponding to the sample data and a second vector embedding code corresponding to the trend data; Constructing a feature correlation matrix according to the sample data and the trend data, and obtaining the autocorrelation matrix and the random permutation matrix of the converter valve to be detected; wherein, the autocorrelation matrix includes a first autocorrelation matrix corresponding to the sample data and a second autocorrelation matrix corresponding to the trend data; the random permutation matrix includes a first random permutation matrix corresponding to the sample data and a second random permutation matrix corresponding to the trend data; Performing feature processing on the vector embedding code, the autocorrelation matrix and the random permutation matrix to obtain a comprehensive feature code.

3. The method according to claim 2, wherein The constructing a feature correlation matrix according to the sample data and the trend data, and obtaining the autocorrelation matrix and the random permutation matrix of the converter valve to be detected includes: Performing inner product operations on the sample data and the trend data respectively, and obtaining a first autocorrelation matrix corresponding to the sample data and a second autocorrelation matrix corresponding to the trend data; Performing position permutation processing on the elements included in the sample data and the trend data, and obtaining a first random permutation matrix corresponding to the sample data and a second random permutation matrix corresponding to the trend data.

4. The method according to claim 2, characterized in that, The performing feature processing on the vector embedding code, the autocorrelation matrix and the random permutation matrix to obtain a comprehensive feature code includes: Performing permutation processing on the autocorrelation matrix according to the random permutation matrix to obtain a target matrix; wherein, the target matrix includes a first target matrix corresponding to the sample data and a second target matrix corresponding to the trend data; Performing local feature extraction on the autocorrelation matrix and the target matrix based on the convolutional neural network CNN, and obtaining an autocorrelation code corresponding to the autocorrelation matrix and a target code corresponding to the target matrix; Performing weighted processing on the autocorrelation code and the target code to obtain a first feature code; Performing global feature interaction on the vector embedding code based on the multi-layer perceptron MLP to obtain a second feature code; Performing weighted processing on the first feature code and the second feature code to obtain a comprehensive feature code.

5. The method according to claim 1, wherein Performing anomaly detection on the to-be-detected converter valve according to the comprehensive feature encoding to obtain the anomaly detection result of the to-be-detected converter valve, including: Based on at least one unsupervised algorithm, respectively performing attribute clustering on the comprehensive feature encoding to obtain the clustering results corresponding to the respective unsupervised algorithms; Determining the anomaly detection result of the to-be-detected converter valve according to the number of algorithms for which the clustering results of the respective unsupervised algorithms are anomaly attributes.

6. The method according to claim 5, wherein The determining the anomaly detection result of the to-be-detected converter valve according to the number of algorithms for which the clustering results of the respective unsupervised algorithms are anomaly attributes includes: If the clustering results corresponding to the respective unsupervised algorithms are all anomaly attributes, determining that the anomaly detection result of the to-be-detected converter valve is an abnormal condition; if there is at least one normal attribute in the clustering results corresponding to the respective unsupervised algorithms, determining that the anomaly detection result of the to-be-detected converter valve is a normal condition.

7. An abnormal detection device for a commutation valve, characterized in that, The device includes: An acquisition module, configured to acquire the target operation data of the to-be-detected converter valve at the current moment; An analysis module, configured to perform feature relationship analysis on the historical operation data and the target operation data of the to-be-detected converter valve to obtain the comprehensive feature encoding corresponding to the to-be-detected converter valve; A detection module, configured to perform anomaly detection on the to-be-detected converter valve according to the comprehensive feature encoding to obtain the anomaly detection result of the to-be-detected converter valve.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images