Transformer cooling system oil pump anomaly detection method, system, equipment and medium

Abstract

The invention discloses a transformer cooling system oil pump anomaly detection method, system and device and a medium. The method comprises the steps that the oil temperature, the environment temperature, the high-use full load, the high-use low load, the medium-use full load, the medium-use low load, the low-use full load and the low-use low load of a cooling system oil pump at the current moment are obtained; inputting the oil temperature, the environment temperature, the high-use full load, the high-use low load, the medium-use full load, the medium-use low load, the low-use full load and the low-use low load of the cooling system oil pump at the current moment into a trained transformer cooling system oil pump anomaly detection model to obtain an anomaly score of the cooling system oil pump at the current moment; according to the method, the system, the equipment and the medium, whether the oil pump of the transformer cooling system is abnormal or not can be accurately judged.

Description

Technical Field

[0001] The present invention belongs to the field of time series anomaly detection and deep learning, and relates to a transformer cooling system oil pump anomaly detection method, system, equipment and medium. Background Art

[0002] Transformers play an irreplaceable role in all kinds of electrical equipment in the power system. They provide strong guarantees for the stable operation and efficient transmission of the power system by realizing functions such as voltage and current conversion, impedance matching, electrical isolation, and voltage stabilization. Failure in the transformer cooling system will cause the transformer to overheat and may cause the windings to burn. The role of the oil pump in the transformer cooling system is to force the transformer oil to circulate in the system. In an oil-immersed transformer, the oil pump draws the oil from the oil tank, passes it through a radiator or other cooling device, and then returns it to the oil tank. The oil pump ensures the continuous flow of oil, thereby effectively transferring the heat generated by the windings and core to the radiator, and then dissipating it into the environment through the radiator. Therefore, in order to ensure the normal operation of power equipment and the safe and stable operation of the power system, it is crucial to detect abnormalities in the transformer cooling system oil pump in a timely and accurate manner.

[0003] The abnormality of the oil pump in the transformer cooling system may be caused by the following reasons: Mechanical failures of the oil pump include bearing damage, sealing airflow excitation, loose parts on the rotor, pump vibration or abnormal noise, bearing heating, excessive shaft clearance, etc. These failures may occur alone or in the form of coupled failures; Low oil flow alarm failure may be caused by oil flow circuit blockage, oil pump failure, oil flow indicator failure, or improper operation or failure of the cooler circuit; Oil pump motor failure may be caused by AC circuit failure of the oil pump motor causing the AC power switch to trip. The above three types of oil pump failures will cause the transformer cooling system to not operate normally and effectively, the oil temperature to rise abnormally, and may even cause the winding to burn. The present invention detects the abnormality of the cooling system oil pump caused by the above three oil pump failures.

[0004] The data of the transformer cooling system mainly includes oil temperature, ambient temperature, and 6 different types of loads. The 6 different types of loads include: high use full load, indicating full load power demand under high load; high use low load, indicating low load power demand under high load; medium use full load, indicating full load power demand under medium load; medium use low load, indicating low load power demand under medium load; low use full load, indicating full load power demand under low load; low use low load, indicating low load power demand under low load. The operation of the oil pump is usually affected by the oil temperature and load of the transformer. When the oil temperature or load of the transformer reaches a certain threshold, the oil pump will automatically start to enhance the cooling effect. Under different ambient temperatures, the fluidity of the oil will be affected. In low temperature environments, the viscosity of the oil increases and the fluidity decreases, which may affect the start-up and operation of the oil pump. The load changes of the transformer will also affect the operation of the oil pump. When the load is high, the oil pump may need to run at a higher speed to provide sufficient cooling. On the contrary, when the load is low, the running speed of the oil pump can be reduced to reduce energy consumption. In summary, there are complex nonlinear dependencies between data of different dimensions, which affect each other. Therefore, in the abnormal detection of oil pump in transformer cooling system, the correlation between different variables must be considered and the potential relationship between different features must be explored.

[0005] At present, the methods for detecting abnormal oil pumps in transformer cooling systems mainly include: using temperature sensors to monitor the oil temperature and winding temperature of the transformer in real time, and regularly checking the operating status of the cooling fan. However, these methods have some shortcomings, such as relying on regular manual inspection and maintenance, which may lead to untimely or missed detection; temperature monitoring is limited by the accuracy and reliability of the sensor, and sometimes there will be some misjudgments; changes in ambient temperature or load will cause changes in oil temperature, but this does not mean that the cooling system oil pump is abnormal. Traditional cooling system oil pump abnormality detection methods do not fully consider the above factors.

[0006] With the development of unsupervised learning and time series anomaly detection technology and the improvement of computer computing power, it is possible to detect anomalies of oil pumps in transformer cooling systems by combining artificial intelligence technology. The current mainstream anomaly detection methods are mainly based on deep learning models, such as DAGMM, LSTM-NDT, LSTM-VAE and other models. The limitation of LSTM-NDT and LSTM-VAE is that they do not explicitly deal with the correlation between multi-dimensional features in the data. DAGMM does not consider time dependency, which affects the accuracy of anomaly detection. Summary of the invention

[0007] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a method, system, device and medium for detecting abnormality of the oil pump of a transformer cooling system. The method, system, device and medium can accurately determine whether the oil pump of the transformer cooling system is abnormal.

[0008] In order to achieve the above object, the present invention adopts the following technical scheme:

[0009] In one aspect, the present invention provides a method for detecting an abnormality of an oil pump in a transformer cooling system, comprising:

[0010] Get the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment;

[0011] The oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment are input into the trained transformer cooling system oil pump anomaly detection model to obtain the anomaly score of the cooling system oil pump at the current moment;

[0012] Whether the transformer cooling system oil pump is abnormal is evaluated based on the abnormal score of the cooling system oil pump at the current moment.

[0013] The transformer cooling system oil pump abnormality detection method of the present invention is further improved in that:

[0014] Furthermore, before inputting the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment into the trained transformer cooling system oil pump abnormality detection model, it also includes:

[0015] Build a dataset;

[0016] Build an abnormality detection model for oil pump in transformer cooling system based on deep learning model;

[0017] The transformer cooling system oil pump anomaly detection model is trained using the data set to obtain a trained transformer cooling system oil pump anomaly detection model.

[0018] Furthermore, the transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module connected in sequence, wherein the feature-oriented graph attention module is used to construct a relationship graph between time series data of different dimensions to capture the correlation between different dimensions; the time-oriented multi-head self-attention module is used to learn the complex dependencies between different time nodes through a multi-head self-attention mechanism; the reconstruction module is used to reconstruct the data processed by the feature-oriented graph attention module and the time-oriented multi-head self-attention module to obtain a reconstructed value, and calculate the anomaly score according to the error between the reconstructed value and the actual value.

[0019] Furthermore, the process of evaluating whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment is:

[0020] When the abnormality score of the cooling system oil pump at the current moment is higher than the preset abnormality score threshold, it indicates that the cooling system oil pump at the current moment is abnormal.

[0021] In a second aspect of the present invention, the present invention provides a transformer cooling system oil pump abnormality detection system, comprising:

[0022] An acquisition module is used to obtain the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment;

[0023] An evaluation module is used to input low load, medium full load, medium low load, low full load and low low load into the trained transformer cooling system oil pump anomaly detection model to obtain the anomaly score of the cooling system oil pump at the current moment;

[0024] The judgment module is used to evaluate whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment.

[0025] The transformer cooling system oil pump abnormality detection system of the present invention is further improved in that:

[0026] Furthermore, it also includes:

[0027] Building modules, used to build data sets;

[0028] Establish a module for building an abnormality detection model for oil pumps in transformer cooling systems based on deep learning models;

[0029] The training module is used to train the transformer cooling system oil pump anomaly detection model using the data set to obtain a trained transformer cooling system oil pump anomaly detection model.

[0030] Furthermore, the transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module connected in sequence, wherein the feature-oriented graph attention module is used to construct a relationship graph between time series data of different dimensions to capture the correlation between different dimensions; the time-oriented multi-head self-attention module is used to learn the complex dependencies between different time nodes through a multi-head self-attention mechanism; the reconstruction module is used to reconstruct the data processed by the feature-oriented graph attention module and the time-oriented multi-head self-attention module to obtain a reconstructed value, and calculate the anomaly score according to the error between the reconstructed value and the actual value.

[0031] Furthermore, the process of evaluating whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment is:

[0032] When the abnormality score of the cooling system oil pump at the current moment is higher than the preset abnormality score threshold, it indicates that the cooling system oil pump at the current moment is abnormal.

[0033] In a third aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for detecting anomaly of an oil pump of a transformer cooling system when executing the computer program.

[0034] In a fourth aspect of the present invention, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the transformer cooling system oil pump abnormality detection method are implemented.

[0035] The present invention has the following beneficial effects:

[0036] During specific operation, the transformer cooling system oil pump abnormality detection method, system, equipment and medium of the present invention input the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment into the trained transformer cooling system oil pump abnormality detection model, obtain the abnormality score of the cooling system oil pump at the current moment, and use it to judge whether the cooling system oil pump is abnormal. The operation is simple and convenient, and the accuracy is high.

[0037] Furthermore, the transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module connected in sequence. The graph attention network and the multi-head self-attention mechanism are used to better learn the correlation between the features of different dimensions of the transformer cooling equipment and the complex time dependency within each dimension, thereby realizing accurate detection of the transformer cooling system oil pump anomaly. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0039] Figure 1 is a flow chart of the method of the present invention;

[0040] Figure 2 It is a schematic diagram of the time series data after preprocessing;

[0041] Figure 3 This is the architecture diagram of the oil pump anomaly detection model for the transformer cooling system;

[0042] Figure 4 This is the transformer anomaly score calculation result diagram;

[0043] Figure 5 This is a comparison chart between transformer anomaly detection results and actual results. DETAILED DESCRIPTION

[0044] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only an embodiment of a part of the present invention, not all embodiments, and is not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concepts disclosed in the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0045] The accompanying drawings show schematic diagrams of structures according to embodiments disclosed in the present invention. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0046] Embodiment 1

[0047] refer to Figure 1 The transformer cooling system oil pump abnormality detection method of the present invention comprises the following steps:

[0048] 1) Build a dataset;

[0049] A data set is constructed based on the historical parameters of the oil pump of the transformer cooling system. The data set includes the data recording date and the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load, low use low load and the score of the oil pump of the cooling system. Considering the influence of transmission equipment abnormality and other factors during the data collection process, some time node data are missing. The regional mean filling method is used to fill the missing values ​​to provide data support for subsequent model training. Some working data of the cooling system of the transformer are as follows: Figure 2 As shown:

[0050] 2) Construct an abnormality detection model for oil pump in transformer cooling system based on deep learning model;

[0051] It should be noted that the reference Figure 3 The core of the task of detecting abnormalities in the oil pump of the transformer cooling system lies in the in-depth analysis and processing of the time series data containing multi-dimensional features generated by the transformer, capturing the correlation between different features and the time dependency within each feature, thereby identifying the abnormal state of the oil pump of the transformer cooling system, effectively detecting whether there is an abnormality in the transformer cooling system, and providing guarantee for the normal operation of the power equipment. By introducing the attention mechanism and linear transformation, the graph attention network has a more powerful feature expression ability, can better capture the relationship between nodes and the changes in feature information, and fully explore the correlation between different dimensional features in time series data with multi-dimensional features. The multi-head self-attention mechanism obtains the attention distribution of different subspaces of the input sequence by running multiple independent attention mechanisms in parallel. This mechanism significantly enhances the model's expression ability and ability to capture complex time dependencies.

[0052] The transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module which are connected in sequence, wherein the feature-oriented graph attention module is used to construct a relationship graph between time series data of different dimensions and capture the correlation between different dimensions; the time-oriented multi-head self-attention module is used to learn the complex dependency between different time nodes through a multi-head self-attention mechanism, which helps to detect anomalies more accurately; the reconstruction module is used to reconstruct the data processed by the feature-oriented graph attention module and the time-oriented multi-head self-attention module, calculate the anomaly score according to the error between the reconstructed value and the actual value, and obtain the anomaly detection result.

[0053] The specific operations of the feature-oriented graph attention module are:

[0054] 11) First, use the preprocessed data to construct an undirected graph G, where each node in the undirected graph G represents a feature of a dimension in the transformer cooling system data, and the feature vector of each node is the value of the time series data of the dimension, which is given by the formula Calculate the similarity between two nodes, where x i and x j is the feature vector of the node, that is, the value of the time series represented by the node. Select e from all nodes except node i ij The highest first k nodes are considered to have edge connections with node i, thereby constructing an undirected graph G, which is a relationship graph between different features in the transformer cooling system time series data.

[0055] 12) Model the correlation between oil temperature, ambient temperature, and six loads in the transformer cooling system.

[0056] Specifically, on an undirected graph G with n nodes, the feature vectors of these n nodes are {x1,x2…,x n}, calculate the attention score α in the undirected graph G ij :

[0057]

[0058] in, represents the concatenation of two node representations, w∈R 2m Represents a column vector of learnable parameters, and LeakyReLU is a nonlinear activation function.

[0059] 13) Calculate the feature vector of each node after aggregating the features of neighboring nodes, where the feature vector h of node i after aggregating neighboring nodes is i for:

[0060]

[0061] Among them, h i With input x i The shape is the same, σ represents the sigmoid activation function, α ij is the attention score of node j to node i, where j is one of i’s neighbor nodes.

[0062] 2) The operation process of the time-oriented multi-head self-attention layer is:

[0063] 21) Concatenate the outputs of all nodes into a matrix h feature ∈R n×k ,h feature The time series data of each dimension in the dataset aggregates the features of other dimensions, and then passes the data to the time-oriented multi-head self-attention layer. In the time-oriented multi-head self-attention layer, there are h heads and the input is h feature , for each head i (i=1,……,h), according to the formula For h featurePerform linear projection and obtain three projection matrices, among which, K i ∈R n×n ,V i ∈R n×n is the corresponding weight matrix.

[0064] 22) Passing Calculate the attention of the i-th head through the formula

[0065]

[0066] The calculation results of the h-head Attention function are concatenated and further linearly projected to obtain the final output h of the time-oriented multi-head self-attention layer. time , where W O is the linear projection matrix of the learnable parameters.

[0067] 3) The operation process of the reconstruction module is:

[0068] 31) The input of the reconstruction module is h time ∈R n×k , according to the formula X t =AvgPool(Padding(h time )) and X s =h time -X t Divide the input data into trend parts X t and the remaining part X s ;

[0069] 32) According to formula H s =W s X s and H t =W t X t , the decomposed trend part and cycle part sequences are processed by single-layer linear networks respectively, where W s ∈R n×n and W t ∈R n×n is a learnable weight matrix, and then according to the formula X ′ =H s +H t , and get the final output result X ′ ;

[0070] 33) During the model training process, update the model parameters according to the formula Get the loss function of the model, where x i ∈R nrepresents the value of the i-th dimension feature in the actual time series data of the transformer cooling system, X i i ∈R n Represents the value of the i-th dimension feature in the final output sequence reconstruction result.

[0071] 34) According to the formula Calculate the anomaly score for each time node, where X i ′ t Indicates the value of the i-th dimension feature at the t-th time node in the output reconstruction result, x it represents the value of the i-th dimension feature at the t-th time node in the actual time series data of the transformer cooling equipment, (X i ′ t -x it ) is the error between the reconstructed value and the actual value, indicating the degree of deviation between the actual value and the reconstructed value of feature i. The anomaly score of each dimensional feature is calculated, and the combined anomaly scores of all dimensional features are taken as the final anomaly score of the oil pump of the transformer cooling system.

[0072] 35) Set the anomaly score threshold to 0.6. When the anomaly score corresponding to a certain time node is greater than 0.6, there is an anomaly in the oil pump of the transformer cooling system at that time node.

[0073] Confirmatory test

[0074] As shown in Table 1, the performance of the present invention is compared with other mainstream anomaly detection methods on cooling equipment datasets on different transformers. It can be found that the anomaly detection performance of the present invention is significantly better than other comparison methods, showing excellent generalization ability, and achieving the best performance on all three datasets.

[0075] Table 1

[0076]

[0077] As shown in Table 2, the actual gains of the model anomaly detection performance by the feature-oriented graph attention module, the time-oriented multi-head self-attention module, and the anomaly detection module are compared. The experiment adopts the control variable method and uses the F1 score as the evaluation indicator to verify the gain effect of each module. The experimental results show that compared with the time-oriented multi-head self-attention module and the reconstruction module, the feature-oriented graph attention module has the greatest effect on improving the model anomaly detection performance. This is because for the multi-dimensional time series anomaly detection task, the graph attention network can better extract the implicit correlation information between dimensions in the multi-dimensional time series data.

[0078] Table 2

[0079]

[0080] In summary, reference Figure 4 and Figure 5 , the present invention has the following characteristics:

[0081] In the feature-oriented graph attention module, the present invention performs graph structure learning to learn the relationship between the feature data of different dimensions of the transformer cooling equipment. After establishing the relationship graph, the graph attention network is used to capture the correlation between the different features of the transformer cooling equipment data. Secondly, in the time-oriented multi-head self-attention module, the present invention uses the multi-head self-attention mechanism to capture the complex time dependency within each dimension. Finally, in the reconstruction module, a sequence reconstruction method is proposed to learn the potential representation of the entire time series and improve the performance of oil pump anomaly detection in the transformer cooling system.

[0082] The traditional methods have problems such as insufficient anomaly detection accuracy and reliability, low automation, and failure to fully consider complex external environment conditions. In contrast, the present invention has the following features:

[0083] High anomaly detection accuracy: The present invention adopts graph attention network and multi-head self-attention mechanism, which can well learn the correlation between the features of different dimensions of transformer cooling equipment and the complex time dependence within each dimension, and realize accurate detection of oil pump anomalies in transformer cooling system.

[0084] Strong generalization ability: The present invention makes full use of multiple different types of time series data generated by the transformer cooling equipment, can cope with complex external environmental changes, and improves the versatility of the transformer cooling system oil pump abnormality detection method.

[0085] Intelligent early warning: The present invention adopts an unsupervised learning method to improve the degree of automation, realize intelligent early warning, and maintain the safe and stable operation of the power system.

[0086] Embodiment 2

[0087] The transformer cooling system oil pump abnormality detection system of the present invention comprises:

[0088] An acquisition module is used to obtain the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment;

[0089] An evaluation module is used to input low load, medium full load, medium low load, low full load and low low load into the trained transformer cooling system oil pump anomaly detection model to obtain the anomaly score of the cooling system oil pump at the current moment;

[0090] The judgment module is used to evaluate whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment.

[0091] As an embodiment of the present invention, it also includes:

[0092] Building modules, used to build data sets;

[0093] Establish a module for building an abnormality detection model for oil pumps in transformer cooling systems based on deep learning models;

[0094] The training module is used to train the transformer cooling system oil pump anomaly detection model using the data set to obtain a trained transformer cooling system oil pump anomaly detection model.

[0095] As an embodiment of the present invention, the transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module connected in sequence, wherein the feature-oriented graph attention module is used to construct a relationship graph between time series data of different dimensions to capture the correlation between different dimensions; the time-oriented multi-head self-attention module is used to learn the complex dependencies between different time nodes through a multi-head self-attention mechanism; the reconstruction module is used to reconstruct the data processed by the feature-oriented graph attention module and the time-oriented multi-head self-attention module to obtain a reconstructed value, and calculate the anomaly score according to the error between the reconstructed value and the actual value.

[0096] As an implementation mode of the present invention, the process of evaluating whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment is:

[0097] When the abnormality score of the cooling system oil pump at the current moment is higher than the preset abnormality score threshold, it indicates that the cooling system oil pump at the current moment is abnormal.

[0098] The division of modules in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present application may be integrated into a processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0099] Embodiment 3

[0100] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the transformer cooling system oil pump abnormality detection method when executing the computer program, for example, comprising: obtaining the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment; inputting the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment into a trained transformer cooling system oil pump abnormality detection model to obtain the abnormality score of the cooling system oil pump at the current moment; and evaluating whether the transformer cooling system oil pump is abnormal according to the abnormality score of the cooling system oil pump at the current moment. The memory may include a memory, such as a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk memory, etc. The processor, network interface, and memory are interconnected through an internal bus, and the internal bus may be an industrial standard architecture bus, a peripheral component interconnection standard bus, an extended industrial standard architecture bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program may include a program code, and the program code includes computer operation instructions. The memory may include a memory and a non-volatile memory, and provide instructions and data to the processor.

[0101] Embodiment 4

[0102] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the transformer cooling system oil pump abnormality detection method are implemented, for example, including: obtaining the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment; inputting the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment into the trained transformer cooling system oil pump abnormality detection model to obtain the abnormality score of the cooling system oil pump at the current moment; evaluating whether the transformer cooling system oil pump is abnormal according to the abnormality score of the cooling system oil pump at the current moment. Specifically, the computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may include random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.

[0103] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0104] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting abnormality of an oil pump in a transformer cooling system, characterized in that: include: Get the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment; The oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment are input into the trained transformer cooling system oil pump anomaly detection model to obtain the anomaly score of the cooling system oil pump at the current moment; Whether the transformer cooling system oil pump is abnormal is evaluated based on the abnormal score of the cooling system oil pump at the current moment.

2. The transformer cooling system oil pump abnormality detection method according to claim 1 is characterized in that: The oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment are input into the trained transformer cooling system oil pump abnormality detection model, and the method further includes: Build a dataset; Build an abnormality detection model for oil pump in transformer cooling system based on deep learning model; The transformer cooling system oil pump anomaly detection model is trained using the data set to obtain a trained transformer cooling system oil pump anomaly detection model.

3. The transformer cooling system oil pump abnormality detection method according to claim 1, characterized in that: The transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module which are connected in sequence, wherein the feature-oriented graph attention module is used to construct a relationship graph between time series data of different dimensions and capture the correlation between different dimensions; the time-oriented multi-head self-attention module is used to learn the complex dependency between different time nodes through a multi-head self-attention mechanism; the reconstruction module is used to reconstruct the data processed by the feature-oriented graph attention module and the time-oriented multi-head self-attention module to obtain a reconstructed value, and calculate the anomaly score according to the error between the reconstructed value and the actual value.

4. The transformer cooling system oil pump abnormality detection method according to claim 1, characterized in that: The process of evaluating whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment is as follows: When the abnormality score of the cooling system oil pump at the current moment is higher than the preset abnormality score threshold, it indicates that the cooling system oil pump at the current moment is abnormal.

5. A transformer cooling system oil pump abnormality detection system, characterized in that: include: An acquisition module is used to obtain the oil temperature, ambient temperature, high use full load, high use low load, medium use full load, medium use low load, low use full load and low use low load of the cooling system oil pump at the current moment; An evaluation module is used to input low load, medium full load, medium low load, low full load and low low load into the trained transformer cooling system oil pump anomaly detection model to obtain the anomaly score of the cooling system oil pump at the current moment; The judgment module is used to evaluate whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment.

6. The transformer cooling system oil pump abnormality detection system according to claim 5, characterized in that: Also includes: Building modules, used to build data sets; Establish a module for building an abnormality detection model for oil pumps in transformer cooling systems based on deep learning models; The training module is used to train the transformer cooling system oil pump anomaly detection model using the data set to obtain a trained transformer cooling system oil pump anomaly detection model.

7. The transformer cooling system oil pump abnormality detection system according to claim 5, characterized in that: The transformer cooling system oil pump anomaly detection model includes a feature-oriented graph attention module, a time-oriented multi-head self-attention module and a reconstruction module which are connected in sequence, wherein the feature-oriented graph attention module is used to construct a relationship graph between time series data of different dimensions and capture the correlation between different dimensions; the time-oriented multi-head self-attention module is used to learn the complex dependency between different time nodes through a multi-head self-attention mechanism; the reconstruction module is used to reconstruct the data processed by the feature-oriented graph attention module and the time-oriented multi-head self-attention module to obtain a reconstructed value, and calculate the anomaly score according to the error between the reconstructed value and the actual value.

8. The transformer cooling system oil pump abnormality detection system according to claim 5, characterized in that: The process of evaluating whether the oil pump of the cooling system of the transformer is abnormal according to the abnormal score of the oil pump of the cooling system at the current moment is as follows: When the abnormality score of the cooling system oil pump at the current moment is higher than the preset abnormality score threshold, it indicates that the cooling system oil pump at the current moment is abnormal.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for detecting abnormality of an oil pump in a transformer cooling system according to any one of claims 1 to 4 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for detecting abnormality of an oil pump in a transformer cooling system according to any one of claims 1 to 4 are implemented.

Images