State monitoring method and device of power transformation equipment, terminal equipment and storage medium

Abstract

The invention discloses a state monitoring method and device for power transformation equipment, terminal equipment and a storage medium. The method comprises the following steps: acquiring test data of a plurality of test instruments on to-be-tested power transformation equipment; inputting the test data into an encoder, and generating to-be-tested potential representation data corresponding to the test data; wherein the encoder generates potential representation data to be tested based on the time sequence characteristics of the test data; inputting the potential representation data to be detected into a decoder, and decoding to obtain reconstructed data; calculating a reconstruction error between the reconstruction data and the test data through a loss function containing a dynamic feature weight, and determining the state of the to-be-tested power transformation equipment based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation degree of each feature corresponding to the test data. According to the invention, automation of state monitoring of the power transformation equipment is realized, and the efficiency of state monitoring of the power transformation equipment is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of substation equipment monitoring, and in particular to a state monitoring method, device, terminal equipment and storage medium for substation equipment. Background Art

[0002] In recent years, with the increasing complexity of power systems, the reliability and stability of substation equipment have become particularly important. The traditional fault monitoring method of substation equipment mainly relies on manual detection, but due to the huge amount of data generated during the operation of the equipment, manual analysis is difficult and inefficient, resulting in the current technical problem of inefficient status monitoring of substation equipment.

[0003] Therefore, there is an urgent need for a state monitoring strategy for substation equipment to solve the problem of low efficiency in state monitoring of substation equipment. Summary of the invention

[0004] The embodiments of the present invention provide a state monitoring method, device, terminal device and storage medium for power substation equipment to solve the problem of low efficiency in state monitoring of power substation equipment.

[0005] In order to solve the above problem, an embodiment of the present invention provides a state monitoring method of a substation, comprising:

[0006] Obtain test data of the substation equipment to be tested by several test instruments;

[0007] Inputting the test data into an encoder to generate potential representation data to be tested corresponding to the test data; wherein the encoder generates the potential representation data to be tested based on the time series characteristics of the test data;

[0008] Inputting the potential representation data to be tested into a decoder, and decoding to obtain reconstructed data;

[0009] The reconstruction error between the reconstructed data and the test data is calculated through a loss function containing dynamic feature weights, and the state of the substation to be tested is determined based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation of each feature corresponding to the test data.

[0010] As an improvement of the above solution, the determining the state of the substation to be tested based on the reconstruction error includes:

[0011] Generating a potential deviation degree to be tested according to the potential representation data to be tested;

[0012] Determining the state of the substation to be tested according to the reconstruction error and the potential deviation to be tested;

[0013] If the reconstruction error is greater than or equal to the error threshold or the potential deviation is greater than or equal to the deviation threshold, the state of the substation to be tested is faulty;

[0014] If the reconstruction error is less than the error threshold and the deviation threshold, the state of the substation to be tested is normal.

[0015] As an improvement of the above solution, before inputting the potential representation data into a decoder, the method further includes:

[0016] Acquire operation data samples of substation equipment in normal status;

[0017] Determining a distribution characteristic interval of normal potential representation data according to the operation data sample;

[0018] Determining whether the potential representative data to be tested is located in the distribution characteristic interval;

[0019] If yes, continue to input the latent representation data into the decoder;

[0020] If not, a fault detection is performed on the test instrument.

[0021] As an improvement of the above solution, generating the potential deviation to be tested according to the potential representation data to be tested includes:

[0022] Inputting the operation data samples of the substation equipment in a normal state into the encoder to generate normal potential representation data;

[0023] The potential representation data to be tested and the normal potential representation data are input into a Euclidean distance deviation calculation formula to calculate the potential deviation to be tested.

[0024] As an improvement of the above solution, the step of calculating the reconstruction error between the reconstructed data and the test data includes:

[0025] Based on the data value of each feature in the test data, the normal operating data corresponding to each feature, and the standard deviation of the normal operating data corresponding to each feature, calculate the abnormal deviation of each feature corresponding to the test data;

[0026] Determine the dynamic feature weight of each feature according to the abnormal deviation of each feature corresponding to the test data;

[0027] The reconstructed data and test data are input into the loss function containing the dynamic feature weight of each feature to calculate the reconstruction error.

[0028] As an improvement of the above solution, the method of determining the dynamic feature weight of each feature according to the abnormal deviation of each feature corresponding to the test data includes:

[0029] Get the correlation matrix and outliers corresponding to each feature in the test data;

[0030] Determine the learnable adjustable parameters of each feature according to the abnormal deviation of each feature corresponding to the test data;

[0031] Generate a relevance weighting factor for each feature based on the correlation matrix and outliers corresponding to each feature of the test data;

[0032] The abnormal deviation, correlation weighting factor and learnable adjustable parameter of each feature in the test data are substituted into the dynamic weight calculation formula to obtain the dynamic feature weight of each feature.

[0033] As an improvement of the above solution, the step of inputting the reconstructed data and the test data into a loss function containing a dynamic feature weight of each feature to calculate the reconstruction error includes:

[0034] The reconstructed data and test data corresponding to each feature are input into a loss function containing the dynamic feature weight corresponding to the current feature to calculate the reconstruction error; wherein the loss function includes:

[0035]

[0036] In the formula, x i is the test data of feature i, is the reconstructed data of feature i, ω i is the dynamic feature weight of feature i.

[0037] Accordingly, an embodiment of the present invention further provides a state monitoring device for a substation, comprising: a data acquisition module, an encoding module, a decoding module and a state acquisition module;

[0038] The data acquisition module is used to acquire test data of the substation to be tested by several test instruments;

[0039] The encoding module is used to input the test data into an encoder to generate potential representation data to be tested corresponding to the test data; wherein the encoder generates the potential representation data to be tested based on the time series characteristics of the test data;

[0040] The decoding module is used to input the potential representation data to be tested into a decoder, and decode to obtain reconstructed data;

[0041] The state acquisition module is used to calculate the reconstruction error between the reconstructed data and the test data through a loss function containing dynamic feature weights, and determine the state of the substation to be tested based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation of each feature corresponding to the test data.

[0042] Correspondingly, an embodiment of the present invention further provides a computer terminal device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a state monitoring method for a substation equipment as described in the present invention is implemented.

[0043] Correspondingly, an embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute a state monitoring method for a substation equipment as described in the present invention.

[0044] As can be seen from the above, the present invention has the following beneficial effects:

[0045] The present invention provides a state monitoring method for substation equipment. The method comprises the following steps: inputting test data collected by a test instrument for the substation equipment to be tested into an encoder for encoding, inputting the test data into a decoder for decoding when obtaining potential representation data to be tested, obtaining reconstructed data, determining a dynamic feature weight by the abnormal deviation of each feature of the test data, and calculating a reconstruction error between the reconstructed data and the test data based on a loss function containing the dynamic feature weight, so that the calculated reconstruction error can reflect the abnormality of the test data, and then analyzing the reconstruction error to determine the state of the substation equipment to be tested. The present invention realizes the automation of state monitoring of the substation equipment and greatly improves the efficiency of state monitoring of the substation equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a method for monitoring the state of a substation provided by an embodiment of the present invention;

[0047] Figure 2 It is a structural schematic diagram of a state monitoring device for a substation provided by an embodiment of the present invention;

[0048] Figure 3 It is a schematic diagram of the structure of a terminal device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Embodiment 1

[0051] See also Figure 1 , Figure 1FIG. 1 is a flow chart of a method for monitoring the state of a substation provided by an embodiment of the present invention. Figure 1 As shown, this embodiment includes steps 101 to 104, and each step is specifically as follows:

[0052] Step 101: Acquire test data of a substation to be tested from a plurality of test instruments.

[0053] In this embodiment, the relationship between the test instruments is specifically as follows: 1. Winding deformation, short-circuit impedance, and DC resistance: When winding deformation and short-circuit impedance are abnormal, it may indicate that the winding is affected by mechanical stress. If the DC resistance is also abnormal, it usually indicates that the winding structure is unstable or severely damaged.

[0054] 2. Insulation resistance and dielectric loss test: If both are abnormal, it may indicate that the insulation material is aging or damp.

[0055] 3. On-load switch, short-circuit impedance and DC resistance: Under certain tap positions, short-circuit impedance and DC resistance are abnormal, indicating that the tap switch may be faulty.

[0056] 4. Turns ratio test, winding deformation and insulation resistance: Abnormal fluctuations in the turns ratio test, abnormal winding deformation test and insulation resistance may indicate internal inter-turn insulation damage in the winding.

[0057] In a specific embodiment, the test instrument and the corresponding test data include:

[0058] 1. DC resistance tester: Collect data: DC resistance value (unit: Ω), reflecting the change of internal winding resistance of the equipment.

[0059] 2. Insulation resistance tester: Collects data: insulation resistance value (unit: MΩ), used to monitor the health status of insulation materials.

[0060] 3. Dielectric loss tester: Collects data: dielectric loss factor or dielectric loss angle (unit: % or angle), indicating the aging or moisture condition of the medium (such as oil or solid insulating material).

[0061] 4. Winding deformation tester: Collect data: winding geometric dimensions or deformation indicators (such as millimeter-level changes), reflecting the changes in the winding after being affected by mechanical stress.

[0062] 5. Transformation ratio tester: collects data: transformation ratio value (for example, the ratio of rated transformation ratio to actual transformation ratio), reflecting whether the status of transformer core and winding is normal.

[0063] 6. On-load tap changer tester: Collects data: switch contact status and contact resistance value (unit: mΩ), which is used to determine whether the tap changer has poor contact or other faults.

[0064] 7. Short-circuit impedance tester: Collects data: short-circuit impedance value (unit: Ω), which is used to evaluate the performance changes of the transformer under short-circuit conditions.

[0065] Step 102: Input the test data into an encoder to generate potential representation data to be tested corresponding to the test data; wherein the encoder generates the potential representation data to be tested based on the time series characteristics of the test data.

[0066] In this embodiment, the sequence data of the test data is passed through an LSTM encoder to extract the temporal features therein and generate a latent representation;

[0067] Through adversarial discriminator training, it is ensured that the latent space satisfies the prior distribution, so that the latent representation of normal data is concentrated in a specific area, while the abnormal data deviates from this area.

[0068] In a specific embodiment, the operation of the encoder is specifically as follows:

[0069] 1. Time series data construction: The data obtained by each test instrument are recorded in the form of time series, and data is collected every 15 minutes to form a multidimensional time series. Assume that within a fixed time window, we obtain a T×7 data matrix, where T represents the number of time steps and 7 represents the characteristics of seven different test data.

[0070] 2. Use LSTM encoder to extract time series features: Input layer: Input the above T×7 data into the LSTM encoder. Hidden layer: LSTM captures the dynamic changes and long-term dependencies of data in the time dimension through its gating mechanism (input gate, forget gate, output gate).

[0071] Generate latent representation: The hidden state of the last time step is mapped to a low-dimensional latent representation z∈R d , where d is the dimension of the latent space.

[0072] Step 103: input the potential representation data to be tested into a decoder, and decode to obtain reconstructed data.

[0073] In a specific embodiment, the latent representation is input to a decoder, which restores it to reconstructed data similar to the original input.

[0074] In this embodiment, before inputting the potential representation data into a decoder, the method further includes:

[0075] Acquire operation data samples of substation equipment in normal status;

[0076] Determining a distribution characteristic interval of normal potential representation data according to the operation data sample;

[0077] Determining whether the potential representative data to be tested is located in the distribution characteristic interval;

[0078] If yes, continue to input the latent representation data into the decoder;

[0079] If not, a fault detection is performed on the test instrument.

[0080] In this embodiment, generating the potential deviation to be tested according to the potential representation data to be tested includes:

[0081] Inputting the operation data samples of the substation equipment in a normal state into the encoder to generate normal potential representation data;

[0082] The potential representation data to be tested and the normal potential representation data are input into a Euclidean distance deviation calculation formula to calculate the potential deviation to be tested.

[0083] In view of the advantages of the adversarial encoder, the Euclidean distance deviation calculation formula is additionally introduced to jointly determine whether there is a fault with the reconstruction error. The Euclidean distance deviation calculation formula is as follows:

[0084] D=||z-μ||#(2)

[0085] Among them, D is the potential deviation to be tested, z represents the potential representation to be tested generated by the test data through the LSTM encoder, and μ represents the mean of the normal potential representation data.

[0086] Step 104: Calculate the reconstruction error between the reconstructed data and the test data through a loss function containing dynamic feature weights, and determine the state of the substation to be tested based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation of each feature corresponding to the test data.

[0087] In this embodiment, determining the state of the substation to be tested based on the reconstruction error includes:

[0088] Generating a potential deviation degree to be tested according to the potential representation data to be tested;

[0089] Determining the state of the substation to be tested according to the reconstruction error and the potential deviation to be tested;

[0090] If the reconstruction error is greater than or equal to the error threshold or the potential deviation is greater than or equal to the deviation threshold, the state of the substation to be tested is faulty;

[0091] If the reconstruction error is less than the error threshold and the deviation threshold, the state of the substation to be tested is normal.

[0092] In this embodiment, calculating the reconstruction error between the reconstructed data and the test data includes:

[0093] Based on the data value of each feature in the test data, the normal operating data corresponding to each feature, and the standard deviation of the normal operating data corresponding to each feature, calculate the abnormal deviation of each feature corresponding to the test data;

[0094] Determine the dynamic feature weight of each feature according to the abnormal deviation of each feature corresponding to the test data;

[0095] The reconstructed data and test data are input into the loss function containing the dynamic feature weight of each feature to calculate the reconstruction error.

[0096] In this embodiment, determining the dynamic feature weight of each feature according to the abnormal deviation of each feature corresponding to the test data includes:

[0097] Get the correlation matrix and outliers corresponding to each feature in the test data;

[0098] Determine the learnable adjustable parameters of each feature according to the abnormal deviation of each feature corresponding to the test data;

[0099] Generate a relevance weighting factor for each feature based on the correlation matrix and outliers corresponding to each feature of the test data;

[0100] The abnormal deviation, correlation weighting factor and learnable adjustable parameter of each feature in the test data are substituted into the dynamic weight calculation formula to obtain the dynamic feature weight of each feature.

[0101] In a specific embodiment, based on historical data and test experience, the Pearson correlation between the features in the test data is calculated to obtain a correlation matrix C; wherein C ij It represents the correlation between feature i and feature j.

[0102] For the i-th feature, its dynamic weight ω i The construction of the feature takes into account the abnormal deviation z i and the abnormality of the features associated with it. i The calculation formula is as follows:

[0103]

[0104] Among them, x i represents the data of the ith feature collected in real time, u i represents the mean of the data when the i-th feature operates normally, σ i represents the standard deviation of the normal operation of the i-th feature, z iIndicates the degree of abnormal deviation of the data. When the abnormal deviation exceeds a certain threshold (this threshold needs to be combined with the experience of experts in the field of substation equipment to ensure that the statistical results and judgment criteria are consistent with the actual equipment operation characteristics), the feature is considered abnormal.

[0105] Dynamic weight ω i The specific steps of construction are as follows:

[0106] 1) Construct relevance weighting factor: For feature i, define its relevance weighting factor A i For all strongly associated features (let the set be s i ) is the sum of the weighted abnormal deviations.

[0107]

[0108] Among them, C ij The larger the value (indicating a stronger correlation), when feature j is abnormal Z j (larger), the more obvious the impact on feature i.

[0109] 2) Construct the final dynamic weight: Combine the degree of abnormality and the correlation weighting factor to define the dynamic weight ω i as follows:

[0110] ω i =1+λ·(z i +A i )

[0111] Among them, λ is a learnable adjustable parameter used to balance the ratio between basic loss and abnormal amplification. During the training process of the model, the optimal parameter will be automatically learned. The "1" here ensures that the weight is 1 when there is no abnormality, which does not affect the original loss; when the data is abnormal, especially when multiple related features are abnormal at the same time, ω i will be significantly greater than 1, thus amplifying the corresponding reconstruction error.

[0112] λ is the setting process of the learnable adjustable parameter:

[0113] First, set the initial value λ = 0.1. This is done to maintain the stability of weight adjustment in the early stages of training, to prevent the model from amplifying noise too early due to an initial value that is too large when multiple associated features are abnormal, or from failing to effectively learn abnormal patterns due to an initial value that is too small. By starting with a smaller value, the model can gradually explore the appropriate amplification ratio during training.

[0114] Secondly, λ is directly embedded into the loss function for end-to-end learning. When calculating the reconstruction error, the dynamic weight ω i It will automatically adjust as λ changes. i and A iAt the same time, the model automatically increases the value of λ through back propagation, thereby significantly increasing the loss weight of these samples. On the contrary, in the case of normal data or isolated anomalies, λ tends to a smaller value.

[0115] At the same time, constraints are imposed on the value range of λ. After each parameter update, λ is forced to be truncated within the interval [0,1]. First, it prevents the weight from being less than 1 when λ is negative, which interferes with the reconstruction error of normal samples; second, it limits the maximum value of λ to avoid infinite weight amplification when multiple strongly correlated features are abnormal, which leads to unstable training.

[0116] In actual training, λ shares an optimizer with other model parameters, and the parameters are directly added to the optimizer's parameter list.

[0117] In this embodiment, the step of inputting the reconstructed data and the test data into a loss function including the dynamic feature weight of each feature to calculate the reconstruction error includes:

[0118] The reconstructed data and test data corresponding to each feature are input into a loss function containing the dynamic feature weight corresponding to the current feature to calculate the reconstruction error; wherein the loss function includes:

[0119]

[0120] In the formula, x i is the test data of feature i, is the reconstructed data of feature i, ω i is the dynamic feature weight of feature i.

[0121] In a specific embodiment, the application of the discriminator in adversarial training includes:

[0122] (1) Adversarial Training Framework:

[0123] Encoder-Decoder: The adversarial autoencoder first generates a latent representation through the encoder and then restores it to the original input data through the decoder.

[0124] Reconstruction error: The reconstruction error is calculated during the reconstruction process, reflecting the difference between the input data and the reconstructed data.

[0125] Discriminator: The discriminator is additionally introduced, whose main task is to determine whether the representation in the latent space belongs to the normal data distribution.

[0126] (2) Normal and abnormal data discrimination of the discriminator:

[0127] Normal data distribution: During the training phase, using a large amount of data labeled as “normal”, the discriminator learns the distribution characteristics of the latent representation so that the latent representation of normal data is concentrated in a specific area.

[0128] Abnormal data representation: When there are abnormalities in the device, the potential representation generated after the encoder often deviates from the distribution area of ​​normal data.

[0129] Euclidean distance deviation: The discriminator can calculate the Euclidean distance between the input potential representation z and the mean of the potential representation of normal data: when D(z) exceeds the preset threshold, it is determined that the data may be abnormal.

[0130] Joint discrimination: The discriminator makes a comprehensive judgment based on the reconstruction error and the Euclidean distance deviation. If either of them exceeds the normal fluctuation range, it is confirmed that the device is faulty.

[0131] It is understandable that during the training process, adversarial training is performed between the discriminator and the encoder / decoder. At the same time, according to the calculation method of dynamic feature weights, when multiple strongly correlated features in the data are abnormal at the same time, the reconstruction error of the corresponding features is dynamically amplified in the loss function, further enhancing the sensitivity of abnormal discrimination.

[0132] See also Figure 2 , Figure 2 2 is a schematic diagram of a state monitoring device for a substation provided by an embodiment of the present invention, comprising: a data acquisition module 201, an encoding module 202, a decoding module 203 and a state acquisition module 204;

[0133] The data acquisition module is used to acquire test data of the substation to be tested by several test instruments;

[0134] The encoding module is used to input the test data into an encoder to generate potential representation data to be tested corresponding to the test data; wherein the encoder generates the potential representation data to be tested based on the time series characteristics of the test data;

[0135] The decoding module is used to input the potential representation data to be tested into a decoder, and decode to obtain reconstructed data;

[0136] The state acquisition module is used to calculate the reconstruction error between the reconstructed data and the test data through a loss function containing dynamic feature weights, and determine the state of the substation to be tested based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation of each feature corresponding to the test data.

[0137] It can be understood that the above-mentioned system item embodiment corresponds to the method item embodiment of the present invention, which can implement the state monitoring method of the substation equipment provided by any one of the above-mentioned method item embodiments of the present invention.

[0138] In this embodiment, the test data collected by the test instrument for the substation to be tested is input into the encoder for encoding, and when the potential representation data to be tested is obtained, it is input into the decoder for decoding to obtain reconstructed data, the dynamic feature weight is determined by the abnormal deviation of each feature of the test data, and the reconstruction error between the reconstructed data and the test data is calculated based on the loss function containing the dynamic feature weight, so that the calculated reconstruction error can reflect the abnormality of the test data, and then the reconstruction error can be analyzed to determine the state of the substation to be tested.

[0139] Embodiment 2

[0140] See also Figure 3 , Figure 3 It is a schematic diagram of the structure of a terminal device provided in one embodiment of the present invention.

[0141] A terminal device of this embodiment includes: a processor 301, a memory 302, and a computer program stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program, the steps of the state monitoring method of each substation device described above in the embodiment are implemented, for example: Figure 1 Alternatively, when the processor executes the computer program, the functions of each module in the above-mentioned device embodiments are realized, for example: Figure 2 All modules of the condition monitoring device for substation equipment are shown.

[0142] In addition, an embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the state monitoring method of the substation equipment described in any of the above embodiments.

[0143] Those skilled in the art will understand that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. The terminal device may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the terminal device may also include input and output devices, network access devices, buses, etc.

[0144] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 301 is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0145] The memory 302 can be used to store the computer program and / or module. The processor 301 implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory 302. The memory 302 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0146] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0147] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0148] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for monitoring the state of a power substation, characterized in that: include: Obtain test data of the substation equipment to be tested by several test instruments; Inputting the test data into an encoder to generate potential representation data to be tested corresponding to the test data; wherein the encoder generates the potential representation data to be tested based on the time series characteristics of the test data; Inputting the potential representation data to be tested into a decoder, and decoding to obtain reconstructed data; The reconstruction error between the reconstructed data and the test data is calculated through a loss function containing dynamic feature weights, and the state of the substation to be tested is determined based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation of each feature corresponding to the test data.

2. The state monitoring method of power substation according to claim 1, characterized in that: The determining the state of the power transformation device to be tested based on the reconstruction error includes: Generating a potential deviation degree to be tested according to the potential representation data to be tested; Determining the state of the substation to be tested according to the reconstruction error and the potential deviation to be tested; If the reconstruction error is greater than or equal to the error threshold or the potential deviation is greater than or equal to the deviation threshold, the state of the substation to be tested is faulty; If the reconstruction error is less than the error threshold and the deviation threshold, the state of the substation to be tested is normal.

3. The state monitoring method of power transformation equipment according to claim 2, characterized in that: Before inputting the potential representation data into a decoder, the method further comprises: Acquire operation data samples of substation equipment in normal status; Determining a distribution characteristic interval of normal potential representation data according to the operation data sample; Determining whether the potential representative data to be tested is located in the distribution characteristic interval; If yes, continue to input the latent representation data into the decoder; If not, a fault detection is performed on the test instrument.

4. The state monitoring method of power substation according to claim 3, characterized in that: The step of generating a potential deviation degree to be tested according to the potential representation data to be tested comprises: Inputting the operation data samples of the substation equipment in a normal state into the encoder to generate normal potential representation data; The potential representation data to be tested and the normal potential representation data are input into a Euclidean distance deviation calculation formula to calculate the potential deviation to be tested.

5. The state monitoring method of power transformation equipment according to claim 4, characterized in that: The calculating the reconstruction error between the reconstructed data and the test data comprises: Based on the data value of each feature in the test data, the normal operating data corresponding to each feature, and the standard deviation of the normal operating data corresponding to each feature, calculate the abnormal deviation of each feature corresponding to the test data; Determine the dynamic feature weight of each feature according to the abnormal deviation of each feature corresponding to the test data; The reconstructed data and test data are input into the loss function containing the dynamic feature weight of each feature to calculate the reconstruction error.

6. The state monitoring method of power substation according to claim 5, characterized in that: Determining the dynamic feature weight of each feature according to the abnormal deviation of each feature corresponding to the test data includes: Get the correlation matrix and outliers corresponding to each feature in the test data; Determine the learnable adjustable parameters of each feature according to the abnormal deviation of each feature corresponding to the test data; Generate a relevance weighting factor for each feature based on the correlation matrix and outliers corresponding to each feature of the test data; The abnormal deviation, correlation weighting factor and learnable adjustable parameter of each feature in the test data are substituted into the dynamic weight calculation formula to obtain the dynamic feature weight of each feature.

7. The state monitoring method of power substation according to claim 6, characterized in that: The step of inputting the reconstructed data and the test data into a loss function containing the dynamic feature weight of each feature and calculating the reconstruction error includes: The reconstructed data and test data corresponding to each feature are input into a loss function containing the dynamic feature weight corresponding to the current feature to calculate the reconstruction error; wherein the loss function includes: In the formula, x i is the test data of feature i, is the reconstructed data of feature i, ω i is the dynamic feature weight of feature i.

8. A state monitoring device for a power substation, characterized in that: include: Data acquisition module, encoding module, decoding module and status acquisition module; The data acquisition module is used to acquire test data of the substation to be tested by several test instruments; The encoding module is used to input the test data into an encoder to generate potential representation data to be tested corresponding to the test data; wherein the encoder generates the potential representation data to be tested based on the time series characteristics of the test data; The decoding module is used to input the potential representation data to be tested into a decoder, and decode to obtain reconstructed data; The state acquisition module is used to calculate the reconstruction error between the reconstructed data and the test data through a loss function containing dynamic feature weights, and determine the state of the substation to be tested based on the reconstruction error; wherein the dynamic feature weight is determined based on the abnormal deviation of each feature corresponding to the test data.

9. A computer terminal device, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a state monitoring method for a substation equipment as claimed in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute a state monitoring method for substation equipment according to any one of claims 1 to 7.

Images