Early warning method based on power electronic transformer and related product

Abstract

The invention discloses an early warning method based on a power electronic transformer and a related product, and the method comprises the steps: firstly obtaining an original operation data set of the power electronic transformer in a preset time period, then carrying out the dimension reduction processing of the original operation data set, obtaining a target operation data set, and carrying out the fault diagnosis based on the target operation data set, the method comprises the steps that first fault information is obtained, then a fault level corresponding to each fault type in preset fault types is determined based on the first fault information, multiple fault levels are obtained, finally, a target alarm instruction is generated based on the multiple fault levels, and the target alarm instruction is used for giving an alarm to a target user. By adopting the embodiment of the invention, the fault solving efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformers, and in particular to an early warning method based on a power electronic transformer and related products. Background Art

[0002] With the development of society, a high proportion of new energy sources are widely connected in new power systems. In addition to the functions of traditional transformers, power electronic transformers can also adjust input power factor, control voltage and current, compensate for power quality, etc. In generator grid connection, they play the role of voltage change, electrical isolation, system protection and improving grid stability. However, power electronic transformers will have many faults during operation, resulting in the inability of generator grid connection to work safely. At present, when a power electronic transformer fails, the efficiency of solving the fault is not high. Summary of the invention

[0003] The embodiments of the present application provide an early warning method and related products based on a power electronic transformer, which generates a target alarm instruction by determining the fault levels of different fault types corresponding to the power electronic transformer, and alarms the target user, thereby improving the efficiency of fault resolution.

[0004] In a first aspect, an embodiment of the present application provides an early warning method based on a power electronic transformer, which is applied to a control chip of the power electronic transformer, and the method includes:

[0005] Obtaining an original operating data set of the power electronic transformer within a preset time period;

[0006] Perform dimensionality reduction processing on the original running data set to obtain the target running data set;

[0007] Perform fault diagnosis based on the target operation data set to obtain first fault information;

[0008] Determine a fault level corresponding to each fault type in the preset fault types based on the first fault information to obtain multiple fault levels;

[0009] A target alarm instruction is generated based on multiple fault levels; the target alarm instruction is used to alarm a target user.

[0010] In a second aspect, an embodiment of the present application provides an early warning device based on a power electronic transformer, which is applied to a control chip of the power electronic transformer, and the device includes: an acquisition unit and a processing unit;

[0011] An acquisition unit, used for acquiring an original operation data set of the power electronic transformer within a preset time;

[0012] A processing unit, used for performing dimensionality reduction processing on the original running data set to obtain a target running data set;

[0013] Perform fault diagnosis based on the target operation data set to obtain first fault information;

[0014] Determine a fault level corresponding to each fault type in the preset fault types based on the first fault information to obtain multiple fault levels;

[0015] A target alarm instruction is generated based on multiple fault levels; the target alarm instruction is used to alarm a target user.

[0016] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor so that the electronic device executes the method of the first aspect.

[0017] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method of the first aspect.

[0018] In a fifth aspect, an embodiment of the present invention provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, so that a computer executes the method of the first aspect.

[0019] The implementation of the present invention has the following beneficial effects:

[0020] It can be seen that the early warning method based on the power electronic transformer and related products described in the embodiments of the present invention are applied to the control chip of the power electronic transformer. The method includes: first, obtaining the original operating data set of the power electronic transformer within a preset time period, and then performing dimensionality reduction processing on the original operating data set to obtain a target operating data set, and then performing fault diagnosis based on the target operating data set to obtain first fault information, and then determining the fault level corresponding to each fault type in the preset fault type based on the first fault information to obtain multiple fault levels, and finally generating a target alarm instruction based on the multiple fault levels, wherein the target alarm instruction is used to alarm the target user, thereby improving the efficiency of fault resolution. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the implementation methods of the present application or the background technology, the drawings required for use in the implementation methods of the present application or the background technology will be described below.

[0022] Figure 1 is a schematic diagram of a control chip of a power electronic transformer provided in an embodiment of the present application;

[0023] Figure 2It is a flow chart of an early warning method based on a power electronic transformer provided by an embodiment of the present invention;

[0024] Figure 3 It is a structural schematic diagram of an early warning device based on a power electronic transformer provided in an embodiment of the present application;

[0025] Figure 4 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the implementation mode of the present application will be clearly and completely described below in conjunction with the drawings in the implementation mode of the present application. Obviously, the described implementation mode is only a part of the implementation mode of the present application, not all the implementation modes. Based on the implementation mode in the present application, all other implementation modes obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0027] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0028] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0029] See also Figure 1 , Figure 1 1 is a schematic diagram of a control chip of a power electronic transformer provided in an embodiment of the present application. The control chip 10 of the power electronic transformer includes a harmonic compensation unit 101 and a fault prediction and remote diagnosis unit 102 .

[0030] The fault prediction and remote diagnosis unit 102 first obtains the original operating data set of the power electronic transformer within a preset time period, then performs dimensionality reduction processing on the original operating data set to obtain a target operating data set, and then performs fault diagnosis based on the target operating data set to obtain first fault information, and then determines the fault level corresponding to each fault type in the preset fault type based on the first fault information to obtain multiple fault levels, and finally generates a target alarm instruction based on the multiple fault levels, wherein the target alarm instruction is used to alarm a target user. The harmonic compensation unit 101 first determines the fundamental current and p groups of harmonic currents in the original operation data set, where p is a positive integer, determines the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents, and obtains p distortion rates, and then determines m distortion rates greater than the preset distortion rate among the p distortion rates, and pm distortion rates less than or equal to the preset distortion rate among the p distortion rates, where m is a positive integer less than p, and then eliminates pm groups of harmonic currents corresponding to pm distortion rates based on the harmonic compensation unit, and then determines m groups of harmonic currents corresponding to m distortion rates, and finally generates a first alarm instruction based on m groups of harmonic currents, where the first alarm instruction is used to alarm the target user. The target alarm instruction and the first alarm instruction are used to alarm the target user, thereby improving the efficiency of fault resolution.

[0031] See also Figure 2 , Figure 2 1 is a flowchart of an early warning method based on a power electronic transformer provided by an embodiment of the present invention. The method is applied to Figure 1 The control chip of the power electronic transformer shown includes but is not limited to the following steps:

[0032] S201: Acquire an original operating data set of a power electronic transformer within a preset time period.

[0033] In this embodiment, various sensors can be installed on the power electronic transformer to collect data. For example, the voltage sensor is used to measure the input and output voltages, the current sensor is used to monitor the current size and direction in the winding, and the temperature sensor is used to detect the temperature of key parts of the transformer (such as the core, winding, etc.). For some high-precision monitoring needs, optical fiber sensors can also be used to detect fault-related signals such as partial discharge. These sensors can collect data in real time or periodically and transmit the data to the data acquisition system. The data acquisition system integrates and records the data collected by the sensors. It can be a dedicated industrial data acquisition device or a computer-based software system. The data acquisition system needs to ensure the accuracy and integrity of the data and timestamp the collected data so that the time point corresponding to the data can be accurately determined later.

[0034] It should be explained that the original operating data set may include voltage data, current data, power data, temperature data, insulation resistance data, etc., which are not limited here. Among them, voltage data includes information such as input voltage and output voltage amplitude and phase. The fluctuation of input voltage can reflect the power supply quality on the grid side. The change of output voltage may be related to winding fault, capacitor fault, etc. inside the transformer. Current data records the current size and waveform in the transformer winding. By analyzing the current data, the load of the transformer can be understood. In the case of overload, the current will exceed the rated value. At the same time, the change of current harmonic content may also be related to the change of nonlinear characteristics of the transformer, such as core saturation. Power data includes parameters such as input power, output power, and power factor. The change of power factor can reflect the reactive power compensation of the transformer. Abnormal power loss may indicate that there is a fault of increased loss inside the transformer, such as increased core loss or increased winding resistance. Temperature data is an important basis for judging whether the transformer is operating normally. It monitors the temperature at different locations, such as the surface temperature of the core and the hot spot temperature of the winding. If the temperature is too high, it may accelerate the aging of the insulation material and even cause faults such as short circuit. Insulation resistance data is the insulation resistance of the transformer measured regularly, which can evaluate the performance of the insulation material. The decrease in insulation resistance may be caused by moisture, aging or local damage to the insulation.

[0035] S202: Perform dimensionality reduction processing on the original operating data set to obtain a target operating data set.

[0036] In this embodiment, illustratively, the data in the original operation data set is standardized to obtain a standardized data set. Specifically, the original operation data is standardized mainly to eliminate the differences in dimensions and numerical ranges between different features, so that the data are comparable. For example, voltage data may be between hundreds of volts and thousands of volts, while temperature data may be in the range of tens of degrees Celsius. If these data of different dimensions are not processed, some features may dominate the algorithm and affect the diagnosis results, so the data needs to be standardized. The original operation data set may contain multiple features (such as voltage, current, temperature, etc.), and each feature needs to be standardized separately. For the voltage features, current features, and temperature features in the original operation data set, their mean, standard deviation, or minimum and maximum values ​​are calculated respectively, and then processed according to the standardized method to obtain a standardized data set. Common standardized methods include Z-Score standardization, which converts the data into a distribution with a mean of 0 and a standard deviation of 1.

[0037] Exemplarily, the covariance matrix corresponding to the standardized data set is determined. Specifically, the covariance matrix is ​​used to describe the linear relationship between different variables (features) in the data set. For the standardized data set, each element in the covariance matrix represents the covariance between two variables. If the covariance is positive, it means that the two variables are positively correlated. If it is negative, it means that they are negatively correlated. If it is close to 0, it means that the linear correlation between the two variables is weak. For example, in the power electronic transformer data, the voltage and current may have a strong positive correlation, and their corresponding elements in the covariance matrix will be a large positive number.

[0038] Exemplarily, k eigenvalues ​​corresponding to the covariance matrix are determined, where k is an integer. Specifically, for the covariance matrix, the eigenvalue reflects the variance of the data in each direction, and the k eigenvalues ​​correspond to the k dimensions of the covariance matrix. A larger eigenvalue indicates that the variance of the data in the direction corresponding to the eigenvalue is larger, that is, the distribution of the data in this direction is more dispersed, while a smaller eigenvalue indicates that the variance of the data in the corresponding direction is smaller, and the data distribution is more concentrated. For example, in a two-dimensional data set, the covariance matrix has two eigenvalues, and the direction corresponding to the larger eigenvalue may be the direction in which the data changes most dramatically.

[0039] Exemplarily, n eigenvalues ​​whose eigenvalues ​​are greater than a preset threshold are selected from k eigenvalues, where n is a positive integer less than k. Specifically, the preset threshold is a screening criterion set according to actual conditions. By selecting eigenvalues ​​greater than this threshold, those directions that contribute more to data changes can be screened out. This is a dimensionality reduction strategy, because usually only a few directions corresponding to important eigenvalues ​​can retain most of the data information. For example, if the threshold is set to a certain value, directions with eigenvalues ​​greater than the threshold are considered important, and directions with eigenvalues ​​less than the threshold are considered to contain less useful information and can be ignored to a certain extent.

[0040] Exemplarily, n eigenvectors corresponding to n eigenvalues ​​are determined, where each eigenvalue corresponds to an eigenvector. Specifically, the eigenvector represents the change pattern of the data in the direction corresponding to the eigenvalue. For example, in a three-dimensional data space, an eigenvector may represent the change trend of the data in a certain oblique direction. These eigenvectors are determined to construct a projection matrix so as to project the original data to these important directions, thereby achieving dimensionality reduction. These eigenvectors constitute a new space in which the main information of the original data can be represented by fewer dimensions.

[0041] Exemplarily, the target projection matrix is ​​determined based on n eigenvectors. Specifically, the target projection matrix is ​​a matrix composed of n selected eigenvectors. Its function is to project the original high-dimensional data into a low-dimensional space. This low-dimensional space is a space spanned by the selected n eigenvectors. Through the projection operation, each data point in the original data set can be converted to this low-dimensional space, thereby achieving the purpose of dimensionality reduction. For example, if the original data is in three-dimensional space, the data points can be projected onto a two-dimensional plane through this projection matrix. This two-dimensional plane is determined by two selected eigenvectors.

[0042] Exemplarily, the original operating data set is projected onto the target projection matrix to obtain the target operating data set. Specifically, each data point in the original operating data set is multiplied by the target projection matrix, so that the data point can be converted from the original high-dimensional space to the low-dimensional space determined by the target projection matrix. The obtained new data set is the target operating data set, which has a lower dimension than the original operating data set. At the same time, it retains the information in the original data set that contributes more to the data change, which is beneficial to subsequent fault diagnosis and other operations.

[0043] It can be seen that the target operating data set is a data set after dimensionality reduction, and its dimension is lower than that of the original operating data set, which reduces the complexity of the data and reduces the computing resources and time required for subsequent fault diagnosis.

[0044] It should be noted that the control chip includes a harmonic compensation unit. Before the original operating data set is subjected to dimensionality reduction processing to obtain the target operating data set, it is also necessary to determine the fundamental current and p groups of harmonic currents in the original operating data set based on the harmonic compensation unit, wherein p is a positive integer. Specifically, in the original operating data set of the power electronic transformer, the current signal contains the fundamental wave and various harmonic components. The fundamental current is the main current component having the same frequency as the power supply frequency, while the harmonic current is the current component having a frequency that is an integer multiple of the fundamental frequency. Through the detection function of the harmonic compensation unit, the fundamental current and p groups of harmonic currents can be separated from the original operating data set.

[0045] Exemplarily, based on the fundamental current and p groups of harmonic currents, the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents is determined to obtain p distortion rates. Specifically, the fundamental frequency and fundamental peak value corresponding to the fundamental current, and the harmonic frequency and harmonic peak value corresponding to the first harmonic current are first determined, wherein the first harmonic current is any group of harmonic currents in the p groups of harmonic currents. Then, the frequency difference between the fundamental frequency and the harmonic frequency, and the peak difference between the fundamental peak value and the harmonic peak value are determined, and the frequency deviation value is determined based on the fundamental frequency and the frequency difference, and the peak deviation value is determined based on the fundamental peak value and the peak difference. The frequency deviation value is an indicator for measuring the degree of frequency deviation determined based on the fundamental frequency and the frequency difference, and can be a ratio obtained by dividing the frequency difference by the fundamental frequency. The peak deviation value can be a ratio obtained by dividing the peak difference by the fundamental peak value. This ratio can intuitively represent the degree of influence of the harmonic peak value on the fundamental peak value. Exemplarily, a first weight corresponding to the frequency deviation value and a second weight corresponding to the peak deviation value are determined. Specifically, in the application scenario of power equipment that is sensitive to frequency, it may be considered that the influence of frequency deviation on current distortion is more important, so a larger first weight is assigned to the frequency deviation value, while in other cases where the change in current amplitude is concerned, a relatively large second weight may be assigned to the peak deviation value. Exemplarily, a weighted operation is performed based on the first weight and the second weight of the frequency deviation value and the peak deviation value to obtain the distortion rate corresponding to the first harmonic current. Specifically, the distortion rate corresponding to the first harmonic current is calculated according to the following formula: distortion rate = first weight × frequency deviation value + second weight × peak deviation value. This distortion rate can more comprehensively reflect the influence of harmonic current on fundamental current. It combines the deviation factors of frequency and peak value. Compared with the method that only considers a single factor, it can more accurately evaluate the distortion of current.

[0046] Exemplarily, m distortion rates greater than a preset distortion rate among p distortion rates, and pm distortion rates less than or equal to the preset distortion rate among the p distortion rates are determined, wherein m is a positive integer less than p. Specifically, the preset distortion rate is a preset threshold value used to distinguish between acceptable and unacceptable harmonic distortion levels. The calculated p distortion rates are compared with the preset distortion rate, and the distortion rates greater than the preset distortion rate are selected, which are m in number. At the same time, the distortion rates less than or equal to the preset distortion rate are determined, which are pm in number.

[0047] Exemplarily, based on the harmonic compensation unit, pm groups of harmonic currents corresponding to pm distortion rates are eliminated. Specifically, the harmonic compensation unit can use active filtering or passive filtering methods to offset the harmonic currents. For example, a compensation current equal to the harmonic current and opposite in direction can be generated and injected into the system so that these harmonic currents offset each other, thereby making the current waveform in the system closer to a sine wave and reducing the adverse effects of harmonics on the power system. A harmonic current loop can also be added to the harmonic compensation unit to feed back the harmonic current to the control chip of the power electronic transformer, so as to achieve the purpose of eliminating the harmonic components in the generator grid.

[0048] Exemplarily, m groups of harmonic currents corresponding to m distortion rates are determined. Specifically, the m groups of harmonic currents are parts that have not been processed and may cause great harm to the generator grid connection.

[0049] Exemplarily, a first alarm instruction is generated based on m groups of harmonic currents, wherein the first alarm instruction is used to alert a target user. Specifically, the first alarm instruction may include detailed information about the harmonic current, such as the harmonic order, the distortion rate, etc. The purpose of this first alarm instruction is to notify the target user that there is a more serious harmonic current problem in the power electronic transformer.

[0050] It can be seen that the power quality of the system can be improved by using the harmonic compensation unit to eliminate the harmonic currents of the PM group with a smaller distortion rate. Even if the distortion rate of these harmonic currents is not particularly high, their existence will still have a certain adverse effect on the system. By eliminating these harmonic currents, the operation of the system can be made more stable and efficient. For the harmonic currents of the M group with a larger distortion rate, the first alarm instruction is generated to alarm the target user, which can promptly remind relevant personnel to take measures, which helps to avoid further deterioration of the harmonic problem and reduce possible equipment damage and failure risks. Timely processing can ensure the reliable operation of the power system and improve the overall performance of the system.

[0051] S203: Perform fault diagnosis based on the target operation data set to obtain first fault information.

[0052] In this embodiment, exemplarily, a preset operating data set and second fault information corresponding to the preset operating data set are obtained. Specifically, it is first necessary to collect data for training and testing the convolutional neural network model, namely, the preset operating data set. This data set contains operating parameter records of the power electronic transformer under various conditions of normal operation and possible failures, such as data on changes in parameters such as voltage, current, temperature, and power over time. At the same time, it is also necessary to obtain second fault information corresponding to these operating data. These fault information are known and confirmed records of real fault conditions, which are used as standards for subsequent convolutional neural network model training and evaluation.

[0053] Exemplarily, the preset operating data set is divided into a training set and a test set according to a preset ratio. Specifically, in order to effectively train and evaluate the model, the obtained preset operating data set needs to be divided into two parts. The preset ratio is determined based on experience or specific model training requirements. Common division ratios are 70%-30%, 80%-20%, etc. Most of the data (training set) is used to train the model to allow the model to learn the relationship between operating data and fault information. For example, by continuously adjusting the parameters of the model, it can accurately predict fault conditions on the training set data. The remaining small part of the data (test set) is used to independently test the performance of the model to verify whether the model can accurately diagnose faults on data that has not been seen.

[0054] Exemplarily, a preset convolutional neural network model is trained based on a training set to obtain a target convolutional neural network model. Specifically, a preset convolutional neural network model is trained using a divided training set. The convolutional neural network has the ability to automatically extract data features and performs well in processing data with time series or spatial structures (such as operating parameter sequences of power electronic transformers). During the training process, the model calculates the predicted fault information through forward propagation, and then compares the predicted result with the actual second fault information in the training set, calculates the loss function (such as cross entropy loss, mean square error, etc.), and then adjusts the model parameters (such as convolution kernel weights, biases, etc.) according to the value of the loss function through a back propagation algorithm. This process is continuously iterated until the performance of the model on the training set reaches a satisfactory level, and finally a trained target convolutional neural network model is obtained.

[0055] Exemplarily, the test set is input into the target convolutional neural network model multiple times for testing to obtain multiple sets of test fault information. Specifically, in order to more comprehensively evaluate the performance and stability of the target convolutional neural network model, the test set data is input into the preset convolutional neural network model multiple times for testing. Each time the test set is input, the model will output a set of prediction information about the fault situation. This prediction information is called test fault information. Since the data input each time is the same but there may be some randomness inside the model (such as parameter initialization, random optimization during training, etc.), multiple tests can observe the different performances of the model under the same data, thereby more accurately measuring the stability and reliability of the model.

[0056] Exemplarily, the similarity between the second fault information and each group of test fault information in multiple groups of test fault information is determined to obtain multiple similarities. Specifically, for each group of test fault information, it is necessary to compare it with the known second fault information to determine the similarity between them. For each test fault information obtained, a similarity value with the actual second fault information can be obtained, thereby obtaining multiple similarities.

[0057] Exemplarily, the mean square error corresponding to multiple similarities is determined. The mean square error is a statistic that measures the degree of dispersion of a set of data. After obtaining multiple similarities, the mean square error of these similarities is calculated. The mean square error can reflect the fluctuation of these similarity values, that is, the stability of the model's performance in multiple tests. If the mean square error is small, it means that the similarity between the model's prediction of fault information and the actual fault information under different test conditions is relatively stable, and the performance of the model is relatively reliable. If the mean square error is large, it means that the performance of the model varies greatly in different tests and may not be stable enough.

[0058] Exemplarily, determining the stability value corresponding to the mean square error may be a preset mapping relationship between the mean square error and the stability value, and the stability value corresponding to the mean square error may be determined according to the mapping relationship.

[0059] Exemplarily, when the stability value is greater than the preset stability value, the target operating data set is input into the target convolutional neural network model for fault diagnosis to obtain the first fault information. Specifically, the preset stability value is a pre-set standard used to determine whether the model meets the stability requirements for practical application. When the model stability value calculated by the previous steps is greater than the preset stability value, the model is considered to be stable enough to be used for actual fault diagnosis. At this time, the target operating data set is input into the target convolutional neural network model, and the model will perform fault diagnosis on these data, and the output fault diagnosis result is the first fault information.

[0060] It can be seen that by comparing the similarity between the test fault information and the real second fault information, the accuracy of each test of the convolutional neural network model can be intuitively understood. The mean square error can reflect the fluctuation of the similarity, that is, the stability of the performance of the convolutional neural network model. A smaller mean square error means that the convolutional neural network model is relatively stable in the accuracy of fault diagnosis in multiple tests. Determining the stability value based on the mean square error can provide a clear standard for whether the convolutional neural network model can be used for actual fault diagnosis. When the stability of the convolutional neural network model meets the requirements, the first fault information can be obtained by using the target operation data set for fault diagnosis. This enables us to use a trained and stable convolutional neural network model to process the target operation data, detect faults in the equipment in a timely manner, and take corresponding measures.

[0061] S204: Determine a fault level corresponding to each fault type in the preset fault types based on the first fault information to obtain multiple fault levels.

[0062] In this embodiment, the preset fault types include voltage fault types and temperature fault types. When the preset fault types include voltage fault types and temperature fault types, fault information corresponding to each fault type in the preset fault types is determined based on the first fault information to obtain voltage fault information and temperature fault information. Specifically, by analyzing and extracting the first fault information, the fault details related to the voltage are separated to obtain the voltage fault information. Similarly, the fault details related to the temperature are extracted to obtain the temperature fault information.

[0063] Exemplarily, the maximum voltage value and the minimum voltage value in the voltage fault information are obtained. Specifically, in the voltage fault scenario of the power electronic transformer, the voltage may fluctuate abnormally due to certain fault reasons. The maximum voltage value and the minimum voltage value can help us understand the amplitude of this fluctuation.

[0064] Exemplarily, the voltage difference between the maximum voltage value and the minimum voltage value is determined to obtain the target voltage difference, and the voltage fault probability corresponding to the voltage fault type is determined based on the target voltage difference. Specifically, the reference voltage fault probability corresponding to the target voltage difference is first determined, which may be a pre-established mapping relationship between the voltage difference and the voltage fault probability, and the reference voltage fault probability corresponding to the target voltage difference can be determined according to the mapping relationship. Exemplarily, multiple voltage values ​​in the voltage fault information are obtained, and the target mean square deviation of the multiple voltage values ​​is determined. The mean square deviation is a statistic that measures the degree of discreteness of data, which can reflect the dispersion of voltage values ​​around the average value. If the target mean square deviation is large, it means that the discreteness of the voltage value is high, that is, the voltage fluctuation is relatively chaotic, which may imply a more complex or more serious voltage fault situation. Exemplarily, the target adjustment coefficient corresponding to the target mean square deviation is determined, which may be a pre-established mapping relationship between the mean square deviation and the adjustment coefficient, and the target adjustment coefficient corresponding to the target mean square deviation can be determined according to the mapping relationship. Exemplarily, the reference voltage failure probability is adjusted based on the target adjustment coefficient to obtain the voltage failure probability. The reference voltage failure probability can be adjusted according to the target adjustment coefficient. The specific calculation formula is as follows: voltage failure probability = reference voltage failure probability × (1 + target adjustment coefficient); The voltage failure probability can be obtained according to the above formula.

[0065] Exemplarily, multiple temperature values ​​and multiple temperature value determination moments are obtained in the temperature fault information, wherein each temperature value corresponds to a temperature value determination moment, and fitting is performed based on the multiple temperature values ​​and the multiple temperature value determination moments to obtain a temperature change straight line. Specifically, after obtaining the multiple temperature values ​​and the multiple temperature value determination moments, a suitable fitting method is selected, such as the least squares method for fitting, so as to obtain the temperature change straight line.

[0066] Exemplarily, the slope corresponding to the temperature change line can be determined to obtain the temperature change slope. The temperature change slope can be obtained by selecting two clear data points on the temperature change line and using the ratio of the change in the ordinate to the change in the abscissa between the two points to calculate the slope corresponding to the temperature change line.

[0067] Exemplarily, the temperature fault probability corresponding to the temperature fault type is determined based on the temperature change slope. Specifically, it can be a pre-established mapping relationship between the temperature change slope and the temperature fault probability. According to the mapping relationship, the temperature fault probability corresponding to the temperature fault type can be determined based on the temperature change slope.

[0068] Exemplarily, the fault levels corresponding to the voltage fault probability and the temperature fault probability are determined to obtain the voltage fault level and the temperature fault level. In order to more intuitively represent the severity of the fault, the fault levels are divided according to the voltage fault probability and the temperature fault probability. The multiple fault levels include the voltage fault level and the temperature fault level. When the fault probability corresponding to the voltage fault probability or the temperature fault probability is less than or equal to the first preset fault probability, the fault level corresponding to the fault probability is level one. When the fault probability is greater than the first preset fault probability and less than the second preset fault probability, the fault level corresponding to the fault probability is level two. When the fault probability is greater than or equal to the second preset fault probability, the fault level corresponding to the fault probability is level three. When the voltage fault probability or the temperature fault probability is less than or equal to the first preset fault probability, it is classified as a level one fault, which indicates that the possibility of the fault is low. When the fault probability is greater than the first preset fault probability and less than the second preset fault probability, it is classified as a level two fault, which means that the possibility of the fault is at a medium level. When the fault probability is greater than or equal to the second preset fault probability, it is determined to be a level three fault, indicating that the possibility of the fault is very high.

[0069] It can be seen that by determining the voltage fault probability based on the target voltage difference, the possibility of voltage fault occurrence can be understood in advance. The temperature fault probability determined based on the temperature change slope can be combined with the voltage fault probability to achieve a comprehensive assessment of equipment failure risks. The voltage fault probability and temperature fault probability are divided into different fault levels, which can provide priority sorting for fault handling. The first-level fault probability is low and routine monitoring can be carried out. The second-level fault probability is moderate and requires increased attention and further inspection. The third-level fault probability is high and immediate measures need to be taken to repair or handle it. This grading method helps to allocate resources reasonably and ensure that when facing multiple potential faults, the most urgent and serious problems can be dealt with first, thereby improving the reliability and safety of equipment operation.

[0070] S205: Generate a target alarm instruction based on multiple fault levels.

[0071] In this embodiment, the target alarm instruction is used to alarm the target user. Specifically, according to different fault levels, the corresponding target alarm instruction is generated. The alarm instruction can be a specific signal or message format, which is used to convey the fault information and urgency to the target user. For example, if the fault level is level one, the alarm instruction can be a low-priority reminder, which may be a yellow indicator light flashing or a relatively mild SMS notification, indicating that there may be some potential problems with the power electronic transformer, but it is not urgent at present, and regular monitoring and inspection are required. When the fault level is level two, the urgency of the alarm instruction will increase, which may be an orange indicator light that is continuously on and a relatively obvious sound alarm is issued. At the same time, a detailed SMS or email is sent to the target user, indicating the specific type of the fault and the possible impact, reminding the target user that it is necessary to pay more attention and arrange further inspection and maintenance. If the fault level is level three, the target alarm instruction will be very urgent, which may be a red indicator light flashing quickly, and a loud alarm sound is issued, and an emergency notification is immediately sent to the target user, requiring the user to take immediate measures to repair or handle it to avoid more serious faults or even damage to the equipment.

[0072] It can be seen that the purpose of sending a target alarm command to the target user is to promptly notify them of equipment failures so that they can take appropriate measures to ensure the safe operation of the equipment and reduce losses. Through timely alarms, the target user can diagnose and handle the failure as soon as possible to avoid further expansion of the failure. The target alarm command can also help the target user to reasonably arrange resources and priorities. According to the severity of the failure, the target user can decide whether to stop the equipment operation immediately, arrange emergency maintenance personnel, prepare spare equipment, etc.

[0073] In summary, the implementation of the present invention has the following beneficial effects:

[0074] It can be seen that the early warning method based on the power electronic transformer and related products described in the embodiments of the present invention are applied to the control chip of the power electronic transformer. The method includes: first, obtaining the original operating data set of the power electronic transformer within a preset time period, and then performing dimensionality reduction processing on the original operating data set to obtain a target operating data set, and then performing fault diagnosis based on the target operating data set to obtain first fault information, and then determining the fault level corresponding to each fault type in the preset fault type based on the first fault information to obtain multiple fault levels, and finally generating a target alarm instruction based on the multiple fault levels, wherein the target alarm instruction is used to alarm the target user and improve the efficiency of fault resolution.

[0075] See also Figure 3 , Figure 3It is a structural schematic diagram of an early warning device based on a power electronic transformer provided in an embodiment of the present application, which is applied to a control chip of a power electronic transformer. The early warning device 300 based on a power electronic transformer includes: an acquisition unit 301 and a processing unit 302;

[0076] An acquisition unit 301 is used to acquire an original operation data set of the power electronic transformer within a preset time;

[0077] The processing unit 302 is used to perform dimensionality reduction processing on the original operation data set to obtain a target operation data set;

[0078] Perform fault diagnosis based on the target operation data set to obtain first fault information;

[0079] Determine a fault level corresponding to each fault type in the preset fault types based on the first fault information to obtain multiple fault levels;

[0080] A target alarm instruction is generated based on multiple fault levels; the target alarm instruction is used to alarm a target user.

[0081] In some possible implementations, in performing dimensionality reduction processing on the original operating data set to obtain the target operating data set, the processing unit 302 is specifically configured to:

[0082] Standardize the data in the original running data set to obtain a standardized data set;

[0083] Determine the covariance matrix corresponding to the standardized data set;

[0084] Determine the k eigenvalues ​​corresponding to the covariance matrix; k is an integer;

[0085] Select n eigenvalues ​​whose eigenvalues ​​are greater than a preset threshold value from among k eigenvalues; n is a positive integer less than k;

[0086] Determine n eigenvectors corresponding to n eigenvalues; each eigenvalue corresponds to one eigenvector;

[0087] Determine the target projection matrix based on the n eigenvectors;

[0088] Project the original running data set onto the target projection matrix to obtain the target running data set.

[0089] In some possible implementations, in performing fault diagnosis based on the target operating data set to obtain the first fault information, the processing unit 302 is specifically configured to:

[0090] Acquire a preset operation data set and second fault information corresponding to the preset operation data set;

[0091] Divide the preset running data set into a training set and a test set according to the preset ratio;

[0092] Training a preset convolutional neural network model based on the training set to obtain a target convolutional neural network model;

[0093] The test set is input into the target convolutional neural network model for multiple times to obtain multiple groups of test failure information;

[0094] Determine the similarity between the second fault information and each set of test fault information in the plurality of sets of test fault information to obtain a plurality of similarities;

[0095] Determine the mean square error corresponding to multiple similarities;

[0096] Determine the stability value corresponding to the mean square error;

[0097] When the stability value is greater than a preset stability value, the target operating data set is input into the target convolutional neural network model for fault diagnosis to obtain first fault information.

[0098] In some possible implementations, in determining the fault level corresponding to each fault type in the preset fault type based on the first fault information to obtain multiple fault levels, the processing unit 302 is specifically configured to:

[0099] Determine the fault information corresponding to each fault type in the preset fault type based on the first fault information, and obtain voltage fault information and temperature fault information;

[0100] Obtain the maximum voltage value and the minimum voltage value in the voltage fault information;

[0101] Determine a voltage difference between a maximum voltage value and a minimum voltage value to obtain a target voltage difference;

[0102] Determining a voltage fault probability corresponding to a voltage fault type based on a target voltage difference;

[0103] Acquire multiple temperature values ​​and multiple temperature value determination times in the temperature fault information; each temperature value corresponds to a temperature value determination time;

[0104] Fitting is performed based on multiple temperature values ​​and multiple temperature value determination moments to obtain a temperature change straight line;

[0105] Determine the slope corresponding to the temperature change straight line to obtain the temperature change slope;

[0106] Determine the temperature fault probability corresponding to the temperature fault type based on the temperature change slope;

[0107] Determine the fault levels corresponding to the voltage fault probability and the temperature fault probability to obtain the voltage fault level and the temperature fault level; the multiple fault levels include the voltage fault level and the temperature fault level; when the fault probability corresponding to the voltage fault probability or the temperature fault probability is less than or equal to the first preset fault probability, the fault level corresponding to the fault probability is level one, when the fault probability is greater than the first preset fault probability and less than the second preset fault probability, the fault level corresponding to the fault probability is level two, and when the fault probability is greater than or equal to the second preset fault probability, the fault level corresponding to the fault probability is level three.

[0108] In some possible implementations, in determining the voltage fault probability corresponding to the voltage fault type based on the target voltage difference, the processing unit 302 is specifically configured to:

[0109] Determine a reference voltage failure probability corresponding to the target voltage difference;

[0110] Acquire multiple voltage values ​​in voltage fault information;

[0111] determining a target mean square error for a plurality of voltage values;

[0112] Determine a target adjustment coefficient corresponding to the target mean square error;

[0113] The reference voltage failure probability is adjusted based on the target adjustment coefficient to obtain the voltage failure probability.

[0114] In some possible implementations, before performing dimensionality reduction processing on the original operating data set to obtain the target operating data set, the processing unit 302 is specifically configured to:

[0115] Determine the fundamental current and p groups of harmonic currents in the original operating data set based on the harmonic compensation unit; p is a positive integer;

[0116] Determine the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents, and obtain p distortion rates;

[0117] Determine m distortion rates greater than a preset distortion rate among the p distortion rates, and pm distortion rates less than or equal to the preset distortion rate among the p distortion rates; m is a positive integer less than p;

[0118] Eliminate the pm group harmonic currents corresponding to pm distortion rates based on the harmonic compensation unit;

[0119] Determine m groups of harmonic currents corresponding to m distortion rates;

[0120] A first alarm instruction is generated based on the m groups of harmonic currents; the first alarm instruction is used to alarm a target user.

[0121] In some possible implementations, in determining the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents to obtain p distortion rates, the processing unit 302 is specifically used to:

[0122] Determine the fundamental frequency and fundamental peak value corresponding to the fundamental current, and the harmonic frequency and harmonic peak value corresponding to the first harmonic current; the first harmonic current is any group of harmonic currents in the p groups of harmonic currents;

[0123] Determine the frequency difference between the fundamental frequency and the harmonic frequency, and the peak value difference between the fundamental peak value and the harmonic peak value;

[0124] Determine a frequency deviation value based on the fundamental frequency and the frequency difference, and determine a peak deviation value based on the fundamental peak value and the peak difference;

[0125] Determine a first weight corresponding to the frequency deviation value and a second weight corresponding to the peak deviation value;

[0126] A weighted operation is performed based on the frequency deviation value, the first weight value, and the second weight value of the peak deviation value to obtain a distortion rate corresponding to the first harmonic current.

[0127] See also Figure 4 , Figure 4 Schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 4 As shown, the electronic device 400 includes a transceiver 401, a processor 402 and a memory 403. They are connected via a bus 404. The memory 403 is used to store computer programs and data, and the transceiver 401 can transmit the data stored in the memory 403 to the processor 402. The above program includes instructions for executing the following steps:

[0128] Obtaining an original operating data set of the power electronic transformer within a preset time period;

[0129] Perform dimensionality reduction processing on the original running data set to obtain the target running data set;

[0130] Perform fault diagnosis based on the target operation data set to obtain first fault information;

[0131] Determine a fault level corresponding to each fault type in the preset fault types based on the first fault information to obtain multiple fault levels;

[0132] A target alarm instruction is generated based on multiple fault levels; the target alarm instruction is used to alarm a target user.

[0133] In some possible implementations, in terms of performing dimensionality reduction processing on the original operating data set to obtain the target operating data set, the above program includes instructions for executing the following steps:

[0134] Standardize the data in the original running data set to obtain a standardized data set;

[0135] Determine the covariance matrix corresponding to the standardized data set;

[0136] Determine the k eigenvalues ​​corresponding to the covariance matrix; k is an integer;

[0137] Select n eigenvalues ​​whose eigenvalues ​​are greater than a preset threshold value from among k eigenvalues; n is a positive integer less than k;

[0138] Determine n eigenvectors corresponding to n eigenvalues; each eigenvalue corresponds to one eigenvector;

[0139] Determine the target projection matrix based on the n eigenvectors;

[0140] Project the original running data set onto the target projection matrix to obtain the target running data set.

[0141] In some possible implementations, in terms of performing fault diagnosis based on the target operating data set to obtain first fault information, the above program includes instructions for executing the following steps:

[0142] Acquire a preset operation data set and second fault information corresponding to the preset operation data set;

[0143] Divide the preset running data set into a training set and a test set according to the preset ratio;

[0144] Training a preset convolutional neural network model based on the training set to obtain a target convolutional neural network model;

[0145] The test set is input into the target convolutional neural network model for multiple times to obtain multiple groups of test failure information;

[0146] Determine the similarity between the second fault information and each set of test fault information in the plurality of sets of test fault information to obtain a plurality of similarities;

[0147] Determine the mean square error corresponding to multiple similarities;

[0148] Determine the stability value corresponding to the mean square error;

[0149] When the stability value is greater than a preset stability value, the target operating data set is input into the target convolutional neural network model for fault diagnosis to obtain first fault information.

[0150] In some possible implementations, in terms of determining the fault level corresponding to each fault type in the preset fault type based on the first fault information to obtain multiple fault levels, the program includes instructions for executing the following steps:

[0151] Determine the fault information corresponding to each fault type in the preset fault type based on the first fault information, and obtain voltage fault information and temperature fault information;

[0152] Obtain the maximum voltage value and the minimum voltage value in the voltage fault information;

[0153] Determine a voltage difference between a maximum voltage value and a minimum voltage value to obtain a target voltage difference;

[0154] Determining a voltage fault probability corresponding to a voltage fault type based on a target voltage difference;

[0155] Acquire multiple temperature values ​​and multiple temperature value determination times in the temperature fault information; each temperature value corresponds to a temperature value determination time;

[0156] Fitting is performed based on multiple temperature values ​​and multiple temperature value determination moments to obtain a temperature change straight line;

[0157] Determine the slope corresponding to the temperature change straight line to obtain the temperature change slope;

[0158] Determine the temperature fault probability corresponding to the temperature fault type based on the temperature change slope;

[0159] Determine the fault levels corresponding to the voltage fault probability and the temperature fault probability to obtain the voltage fault level and the temperature fault level; the multiple fault levels include the voltage fault level and the temperature fault level; when the fault probability corresponding to the voltage fault probability or the temperature fault probability is less than or equal to the first preset fault probability, the fault level corresponding to the fault probability is level one, when the fault probability is greater than the first preset fault probability and less than the second preset fault probability, the fault level corresponding to the fault probability is level two, and when the fault probability is greater than or equal to the second preset fault probability, the fault level corresponding to the fault probability is level three.

[0160] In some possible implementations, in terms of determining the voltage fault probability corresponding to the voltage fault type based on the target voltage difference, the program includes instructions for performing the following steps:

[0161] Determine a reference voltage failure probability corresponding to the target voltage difference;

[0162] Acquire multiple voltage values ​​in voltage fault information;

[0163] determining a target mean square error for a plurality of voltage values;

[0164] Determine a target adjustment coefficient corresponding to the target mean square error;

[0165] The reference voltage failure probability is adjusted based on the target adjustment coefficient to obtain the voltage failure probability.

[0166] In some possible implementations, before performing dimensionality reduction processing on the original operating data set to obtain the target operating data set, the above program includes instructions for executing the following steps:

[0167] Determine the fundamental current and p groups of harmonic currents in the original operating data set based on the harmonic compensation unit; p is a positive integer;

[0168] Determine the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents, and obtain p distortion rates;

[0169] Determine m distortion rates greater than a preset distortion rate among the p distortion rates, and pm distortion rates less than or equal to the preset distortion rate among the p distortion rates; m is a positive integer less than p;

[0170] Eliminate the pm group harmonic currents corresponding to pm distortion rates based on the harmonic compensation unit;

[0171] Determine m groups of harmonic currents corresponding to m distortion rates;

[0172] A first alarm instruction is generated based on the m groups of harmonic currents; the first alarm instruction is used to alarm a target user.

[0173] In some possible implementations, in terms of determining the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents to obtain p distortion rates, the program includes instructions for performing the following steps:

[0174] Determine the fundamental frequency and fundamental peak value corresponding to the fundamental current, and the harmonic frequency and harmonic peak value corresponding to the first harmonic current; the first harmonic current is any group of harmonic currents in the p groups of harmonic currents;

[0175] Determine the frequency difference between the fundamental frequency and the harmonic frequency, and the peak value difference between the fundamental peak value and the harmonic peak value;

[0176] Determine a frequency deviation value based on the fundamental frequency and the frequency difference, and determine a peak deviation value based on the fundamental peak value and the peak difference;

[0177] Determine a first weight corresponding to the frequency deviation value and a second weight corresponding to the peak deviation value;

[0178] A weighted operation is performed based on the frequency deviation value, the first weight value, and the second weight value of the peak deviation value to obtain a distortion rate corresponding to the first harmonic current.

[0179] It should be understood that the electronic devices in this application may include smart phones (such as Android phones, iOS phones, Windows Phone phones, etc.), tablet computers, PDAs, laptops, mobile Internet devices MID (Mobile Internet Devices, MID for short) or wearable devices or servers, edge computing nodes, etc. The above electronic devices are only examples, not exhaustive, and include but are not limited to the above electronic devices.

[0180] The embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. The computer program is executed by a processor to implement part or all of the steps of any one of the early warning methods based on power electronic transformers as recorded in the above method embodiments.

[0181] The embodiment of the present application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program can be operated to cause a computer to execute part or all of the steps of any one of the early warning methods based on power electronic transformers as recorded in the above method implementation.

[0182] It should be noted that, for the above-mentioned various method implementations, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the implementations described in the specification are all optional implementations, and the actions and modules involved are not necessarily required by this application.

[0183] In the above-mentioned embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0184] In the several embodiments provided in this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device implementation described above is only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0185] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0186] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of a software program module.

[0187] If the integrated unit is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of each implementation method of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0188] A person skilled in the art can understand that all or part of the steps in the various methods of the above-mentioned embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, and the memory can include: a flash drive, a read-only memory (English: Read-Only Memory, abbreviated as: ROM), a random access memory (English: Random Access Memory, abbreviated as: RAM), a magnetic disk or an optical disk, etc.

[0189] The above is a detailed introduction to the implementation methods of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above implementation methods is only used to help understand the method and core idea of ​​the present application. At the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. An early warning method based on power electronic transformer, characterized in that: A control chip applied to a power electronic transformer, the method comprising: Acquiring an original operating data set of the power electronic transformer within a preset time period; Performing dimensionality reduction processing on the original operating data set to obtain a target operating data set; Performing fault diagnosis based on the target operation data set to obtain first fault information; Determine a fault level corresponding to each fault type in a preset fault type based on the first fault information to obtain multiple fault levels; A target alarm instruction is generated based on the multiple fault levels; the target alarm instruction is used to alarm a target user.

2. The method according to claim 1, characterized in that The performing dimensionality reduction processing on the original operation data set to obtain the target operation data set includes: Standardizing the data in the original running data set to obtain a standardized data set; Determining a covariance matrix corresponding to the standardized data set; Determine k eigenvalues ​​corresponding to the covariance matrix; k is an integer; Selecting n eigenvalues ​​whose eigenvalues ​​are greater than a preset threshold value from the k eigenvalues; n is a positive integer less than k; Determine n eigenvectors corresponding to the n eigenvalues; each eigenvalue corresponds to one eigenvector; Determine a target projection matrix based on the n eigenvectors; The original running data set is projected onto the target projection matrix to obtain the target running data set.

3. The method according to claim 1, characterized in that The performing fault diagnosis based on the target operation data set to obtain first fault information includes: Acquire a preset operation data set and second fault information corresponding to the preset operation data set; Dividing the preset running data set into a training set and a test set according to a preset ratio; Training a preset convolutional neural network model based on the training set to obtain a target convolutional neural network model; Inputting the test set into the target convolutional neural network model for multiple times to perform testing, thereby obtaining multiple groups of test failure information; Determine the similarity between the second fault information and each set of test fault information in the multiple sets of test fault information to obtain multiple similarities; Determine mean square errors corresponding to the multiple similarities; Determine a stability value corresponding to the mean square error; When the stability value is greater than a preset stability value, the target operation data set is input into the target convolutional neural network model for fault diagnosis to obtain the first fault information.

4. The method according to any one of claims 1 to 3, characterized in that: When the preset fault type includes a voltage fault type and a temperature fault type, the fault level corresponding to each fault type in the preset fault type is determined based on the first fault information to obtain multiple fault levels, including: Determine the fault information corresponding to each fault type in the preset fault type based on the first fault information, and obtain voltage fault information and temperature fault information; Obtaining a maximum voltage value and a minimum voltage value in the voltage fault information; Determine a voltage difference between a maximum voltage value and a minimum voltage value to obtain a target voltage difference; Determining a voltage fault probability corresponding to the voltage fault type based on the target voltage difference; Acquire multiple temperature values ​​and multiple temperature value determination times in the temperature fault information; each temperature value corresponds to a temperature value determination time; Perform fitting based on the multiple temperature values ​​and the multiple temperature value determination times to obtain a temperature change straight line; Determine the slope corresponding to the temperature change straight line to obtain the temperature change slope; Determining a temperature fault probability corresponding to the temperature fault type based on the temperature change slope; Determine the fault levels corresponding to the voltage fault probability and the temperature fault probability to obtain the voltage fault level and the temperature fault level; the multiple fault levels include the voltage fault level and the temperature fault level; when the fault probability corresponding to the voltage fault probability or the temperature fault probability is less than or equal to a first preset fault probability, the fault level corresponding to the fault probability is level one, when the fault probability is greater than the first preset fault probability and less than a second preset fault probability, the fault level corresponding to the fault probability is level two, and when the fault probability is greater than or equal to the second preset fault probability, the fault level corresponding to the fault probability is level three.

5. The method according to claim 4, characterized in that The determining, based on the target voltage difference, the voltage fault probability corresponding to the voltage fault type includes: Determine a reference voltage failure probability corresponding to the target voltage difference; Acquire multiple voltage values ​​in the voltage fault information; determining a target mean square error of the plurality of voltage values; determining a target adjustment coefficient corresponding to the target mean square error; The reference voltage failure probability is adjusted based on the target adjustment coefficient to obtain the voltage failure probability.

6. The method according to claim 1, characterized in that The control chip includes a harmonic compensation unit. Before performing dimensionality reduction processing on the original operation data set to obtain a target operation data set, the method further includes: Determine the fundamental current and p groups of harmonic currents in the original operation data set based on the harmonic compensation unit; p is a positive integer; Determining the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents to obtain p distortion rates; Determine m distortion rates greater than a preset distortion rate among the p distortion rates, and pm distortion rates less than or equal to the preset distortion rate among the p distortion rates; m is a positive integer less than p; Eliminating the pm group of harmonic currents corresponding to the pm distortion rates based on the harmonic compensation unit; Determining m groups of harmonic currents corresponding to the m distortion rates; A first alarm instruction is generated based on the m groups of harmonic currents; the first alarm instruction is used to alarm the target user.

7. The method according to claim 6, characterized in that The step of determining the distortion rate corresponding to each group of harmonic currents in the p groups of harmonic currents based on the fundamental current and the p groups of harmonic currents to obtain p distortion rates includes: Determine the fundamental frequency and fundamental peak value corresponding to the fundamental current, and the harmonic frequency and harmonic peak value corresponding to the first harmonic current; the first harmonic current is any one group of harmonic currents in the p groups of harmonic currents; Determine a frequency difference between the fundamental frequency and the harmonic frequency, and a peak difference between the fundamental peak and the harmonic peak; Determine a frequency deviation value based on the fundamental frequency and the frequency difference, and determine a peak deviation value based on the fundamental peak value and the peak difference; Determine a first weight corresponding to the frequency deviation value and a second weight corresponding to the peak deviation value; A weighted operation is performed based on the frequency deviation value, the peak deviation value, the first weight, and the second weight to obtain a distortion rate corresponding to the first harmonic current.

8. An early warning device based on a power electronic transformer, characterized in that: A control chip applied to a power electronic transformer, the device comprising: an acquisition unit and a processing unit; The acquisition unit is used to acquire an original operation data set of the power electronic transformer within a preset time; The processing unit is used to perform dimensionality reduction processing on the original operation data set to obtain a target operation data set; Performing fault diagnosis based on the target operation data set to obtain first fault information; Determine a fault level corresponding to each fault type in a preset fault type based on the first fault information to obtain multiple fault levels; A target alarm instruction is generated based on the multiple fault levels; the target alarm instruction is used to alarm a target user.

9. An electronic device, characterized in that: The method comprises a processor, a memory, a communication interface and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the one or more programs include instructions for executing the steps in the method described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method according to any one of claims 1 to 7.

Images