Converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform

Abstract

The invention discloses a converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform, and relates to the technical field of insulation monitoring of transformer bushing equipment, and the method comprises the steps: obtaining transformer bushing leakage current data, and setting a state space, an action space and a reward function of reinforcement learning; selecting a wavelet basis function, and performing wavelet decomposition on the obtained leakage current data of the transformer bushing; and calculating the energy of each scale of the wavelet coefficient, and judging the energy threshold of each scale. According to the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transformation, wavelet transformation is used when transformer bushing leakage current data is preprocessed, the number of decomposition layers is improved, the optimal number of wavelet decomposition layers is objectively selected by using the advantage of reinforcement learning autonomous decision, and the reliability of the transformer bushing leakage current data is improved. And multi-resolution frequency spectrum information of the signal is provided, noise in the original signal is suppressed, and leakage current data of the transformer bushing are efficiently processed.

Description

Technical Field

[0001] The invention relates to the technical field of insulation monitoring of transformer bushing equipment, and in particular to a method for preprocessing leakage current data of a bushing on a valve side of a converter transformer based on improved wavelet transform. Background Art

[0002] The power industry is one of the basic industries that the whole world is competing for. Among them, the insulation state of the transformer bushing, as an important power equipment, determines the stable operation of the transformer, thus affecting the reasonable operation of the entire power network. Therefore, the monitoring of the leakage current of the transformer bushing plays an important role in the insulation state monitoring of the transformer.

[0003] Compared with the traditional Fourier transform, which is only applicable to the spectrum analysis of stationary signals, the wavelet transform has better non-stationary signal analysis capabilities and can perform different scales of analysis on different time periods of the signal. In addition, the wavelet transform can better capture the local features and mutations in the signal, while the Fourier transform divides the entire signal into frequency domain segments of the same length, making it difficult to extract local features. The wavelet transform can provide multi-resolution spectrum information and can simultaneously obtain the low-frequency and high-frequency components of the signal, while the Fourier transform can only provide single-scale spectrum information. In general, the wavelet transform has better performance than the Fourier transform in processing non-stationary signals, mutation features, and time-frequency representation. Therefore, in many practical applications, the wavelet transform is more suitable for analyzing and processing signals.

[0004] However, wavelet transform also has certain defects. In wavelet transform, it is necessary to select the appropriate number of decomposition layers to extract information of different scales. However, the selection of the number of decomposition layers is usually subjective, and there is a lack of an objective and automated method to determine the best number of decomposition layers.

[0005] Artificial intelligence technologies represented by reinforcement learning have been widely used in finance, autonomous driving and other fields. Reinforcement learning can adapt to changing environments and tasks. Through continuous learning and adjustment, it can make optimal decisions in different situations and has strong flexibility. Therefore, when choosing the number of decomposition layers, the reinforcement learning method can be combined to obtain the optimal number of decomposition layers to avoid the subjective influence of empirical judgment. Summary of the invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is that the existing transformer bushing leakage current monitoring signal method has subjectivity in selecting the number of decomposition layers, low automation, and how to dynamically select the optimal number of wavelet decomposition layers to adapt to different signal characteristics.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for preprocessing the leakage current data of the bushing on the valve side of the converter transformer based on improved wavelet transform, comprising obtaining the leakage current data of the bushing of the transformer, setting the state space, action space and reward function of reinforcement learning. Selecting the wavelet basis function, performing wavelet decomposition on the obtained transformer bushing leakage current data; calculating the energy of each scale of the wavelet coefficient, and judging the energy threshold of each scale.

[0009] As a preferred solution of the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform described in the present invention, wherein: the acquisition of transformer bushing leakage current data includes collecting the current data waveform at any time of the day using a current transformer.

[0010] As a preferred solution of the method for preprocessing leakage current data of the valve side bushing of the converter transformer based on improved wavelet transform described in the present invention, wherein: the setting of the state space, action space and reward function of reinforcement learning includes setting the decomposition number of the state space to 1 to N layers, and the initial decomposition number of layers is 1.

[0011] The action space increases or decreases the number of decomposition levels.

[0012] The reward function is expressed as: , Among them, s represents the current decomposition level, a represents increasing or decreasing the decomposition level, represents the expected return of taking action a in state s, α represents the learning rate, γ represents the discount factor, and r represents the immediate reward after taking action a. Indicates the next state The maximum Q value of all possible actions.

[0013] As a preferred solution of the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform described in the present invention, wherein: the selection of wavelet basis function includes selecting Symlet function as the wavelet basis function.

[0014] As a preferred solution of the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform described in the present invention, wherein: the wavelet decomposition of the acquired transformer bushing leakage current data includes obtaining wavelet coefficients at different scales and performing wavelet decomposition, which is expressed as: , in, is the signal to be analyzed, j is the translation scale, is the wavelet coefficient, is the selected Symlet wavelet basis function, t represents time, k represents the translation of the wavelet basis function along the time axis, and Z represents an integer set.

[0015] As a preferred solution of the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform described in the present invention, wherein: the calculation of each scale energy of the wavelet coefficient includes calculating each scale energy spectrum of the wavelet coefficient and squaring the absolute value of the coefficient.

[0016] As a preferred solution of the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform described in the present invention, wherein: the energy threshold judgment of each scale includes outputting the decomposition result if the energy of each scale is higher than the threshold.

[0017] If the energy of each scale is lower than the threshold, the Q-value function is used to guide the action selection, that is, to increase or decrease the number of decomposition layers.

[0018] Another object of the present invention is to provide a converter transformer valve side bushing leakage current data preprocessing system based on improved wavelet transform, which can select the optimal number of wavelet decomposition layers through autonomous decision-making, thereby solving the problem that the current transformer bushing leakage current monitoring signal contains the inability to dynamically select the optimal number of wavelet decomposition layers to adapt to different signal characteristics.

[0019] As a preferred solution of the converter transformer valve side bushing leakage current data preprocessing system based on improved wavelet transform described in the present invention, it includes: a data acquisition module, a data processing module, and a scale energy analysis module.

[0020] The data acquisition module is used to obtain transformer bushing leakage current data and set the state space, action space and reward function of reinforcement learning.

[0021] The data processing module is used to select a wavelet basis function and perform wavelet decomposition on the acquired transformer bushing leakage current data.

[0022] The scale energy analysis module is used to calculate the energy of each scale of the wavelet coefficients and to determine the energy threshold of each scale.

[0023] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of a method for preprocessing leakage current data of a converter transformer valve side bushing based on improved wavelet transform.

[0024] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for preprocessing leakage current data of a converter transformer valve-side bushing based on improved wavelet transform.

[0025] Beneficial effects of the present invention: The converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform provided by the present invention introduces reinforcement learning technology, so that the selection process of the decomposition layer number can be adaptively optimized and dynamically adjusted to the optimal decomposition level, so that the multi-resolution analysis of the signal is more accurate, which not only retains the effective signal characteristics, but also avoids excessive calculation and improves the efficiency of data processing. Through the reinforcement learning autonomous decision-making mechanism, the system can select the best wavelet mother function and decomposition layer under different signal characteristic conditions, thereby improving the accuracy of signal processing, better adapting to different working conditions and noise characteristics, and improving the accuracy of leakage current data analysis. The use of intelligent optimization strategies reduces unnecessary computing overhead, significantly improves the computational efficiency of wavelet transform, and meets the needs of real-time monitoring and analysis. The present invention achieves better results in both processing efficiency and noise suppression. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0027] Figure 1 The overall flow chart of a converter transformer valve-side bushing leakage current data preprocessing method based on improved wavelet transform is provided for the first embodiment of the present invention.

[0028] Figure 2 A comparison diagram of experimental data of a converter transformer valve-side bushing leakage current data preprocessing method based on improved wavelet transform provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0030] Example 1.

[0031] Reference Figure 1 , which is an embodiment of the present invention, provides a method for preprocessing leakage current data of a converter transformer valve side bushing based on improved wavelet transform, comprising: S1: Obtain transformer bushing leakage current data and set the state space, action space and reward function of reinforcement learning.

[0032] Furthermore, obtaining the transformer bushing leakage current data includes collecting the current data waveform at any time of the day using a current transformer.

[0033] It should be noted that setting the state space, action space and reward function of reinforcement learning includes setting the number of decomposition layers of the state space to 1 to N layers, and the initial number of decomposition layers to 1 layer.

[0034] The action space increases or decreases the number of decomposition levels.

[0035] The reward function is expressed as: , Among them, s represents the current decomposition level, a represents increasing or decreasing the decomposition level, represents the expected return of taking action a in state s, α represents the learning rate, γ represents the discount factor, and r represents the immediate reward after taking action a. Indicates the next state The maximum Q value of all possible actions under this reward function is referred to as the Q-value function.

[0036] The current decomposition level is 1 to N, indicating the level of wavelet decomposition currently used. The initial decomposition level is fixed at 1, indicating the level of wavelet decomposition before the algorithm runs.

[0037] Increasing the number of decomposition layers will increase the current wavelet decomposition layer by one unit, and decreasing the number of decomposition layers will decrease the current wavelet decomposition layer by one unit. Assuming the number of decomposition layers is 3, increasing the number of decomposition layers will increase the number of decomposition layers to 4, and decreasing the number of decomposition layers will reduce the number of decomposition layers to 2.

[0038] When the energy retention is above the set threshold, the reward is positive, and when the energy retention is below the set threshold, the reward is negative. Positive values ​​indicate how much the energy retention is above the threshold, while negative values ​​indicate how much the energy retention is below the threshold.

[0039] S2: Select wavelet basis function and perform wavelet decomposition on the obtained transformer bushing leakage current data.

[0040] Furthermore, selecting the wavelet basis function includes selecting the Symlet function as the wavelet basis function. The Symlet function is a commonly used wavelet basis function, and wavelet basis functions of different lengths and frequencies can be obtained by adjusting its parameters. The Symlet function has a set of specific properties such as orthogonality and compact support, and is suitable for signal analysis and feature extraction.

[0041] It should be noted that the wavelet decomposition of the obtained transformer bushing leakage current data includes obtaining wavelet coefficients at different scales and performing wavelet decomposition, which is expressed as: , in, is the signal to be analyzed, j is the translation scale, is the wavelet coefficient, is the selected Symlet wavelet basis function, t represents time, k represents the translation of the wavelet basis function along the time axis, and Z represents an integer set.

[0042] S3: Calculate the energy of each scale of the wavelet coefficients and make energy threshold judgments at each scale.

[0043] Furthermore, calculating the energy of each scale of the wavelet coefficient includes calculating the energy spectrum of each scale of the wavelet coefficient and squaring the absolute value of the coefficient. The energy value of each wavelet coefficient can be divided by the total energy value or the maximum energy value, that is, energy normalization processing is performed to make the energy at different scales comparable.

[0044] It should be noted that performing energy threshold determination at each scale includes outputting a decomposition result if the energy at each scale is higher than the threshold.

[0045] If the energy of each scale is lower than the threshold, the Q-value function is used to guide the action selection, that is, to increase or decrease the number of decomposition layers.

[0046] Specifically, the energy of the wavelet coefficients at each scale is compared with the set threshold to determine whether it is higher than the set threshold. If the energy is higher than the threshold, it means that the signal component at this scale has important information, and the corresponding wavelet decomposition result can be output. If the energy is lower than the threshold, the Q value function is needed to evaluate the effect of different decomposition layers, where the Q value function can be evaluated based on indicators such as signal noise reduction effect or signal-to-noise ratio. The threshold can be obtained by statistical methods, such as calculating the mean or standard deviation to set the threshold.

[0047] Example 2.

[0048] Reference Figure 2 , which is an embodiment of the present invention, provides a converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0049] First, the original data containing noise and useful signals are collected. Secondly, the original signal is input into the traditional signal processing method (the traditional Kalman filtering method is used here) to obtain the signal processed by the existing method. Secondly, the same original signal is input into the signal processing method of the present invention, and the steps of the method of the present invention are as follows: the state_vector function is used to set the reinforcement learning state space to the decomposition layer number, and the initial decomposition layer number is 1. Then the action_space function is used to set the reinforcement learning action space to increase or decrease the decomposition layer number. Then the reward function is set to the Q value function, and the Q value function is used to guide the action selection, and the Q value function is the reward function. Secondly, the symlet function is used as the wavelet basis function, and the original signal is wavelet decomposed, and the energy of the wavelet coefficient is calculated. If the energy is higher than the threshold, it is retained. If it is lower than the threshold, the Q value function is used to judge the action selection until the threshold requirement is met.

[0050] Reference Figure 2 It can be seen that the traditional Fourier transform removes high-frequency noise through a low-pass filter, but due to its global frequency domain characteristics, the recovery effect is poor when there are mutations or local changes in the signal. In the experiment, the original signal is a sine wave with a frequency of 50Hz, with Gaussian noise and slight mutations added. The signal processed by Fourier transform has smooth distortion in the local fluctuation and mutation parts. Through experiments, the MSE value after Fourier transform is 0.12529, indicating that the gap between the restored signal and the original signal is large, especially when processing the mutation part, the loss of signal details is more serious. In contrast, the method of the present invention can adaptively denoise through multi-scale decomposition, retaining the local characteristics and mutation components of the signal. Experiments show that the difference between the signal processed by wavelet transform and the original signal is small, and the denoising effect is better. The MSE value of wavelet transform is 0.0388, indicating that the signal recovered after denoising is closer to the effective sinusoidal component in the original signal, and the restored signal details are clearer in the mutation and local change parts.

[0051] Example 3.

[0052] An embodiment of the present invention provides a converter transformer valve-side bushing leakage current data preprocessing system based on improved wavelet transform, comprising a data acquisition module, a data processing module, and a scale energy analysis module.

[0053] Among them, the data acquisition module is used to obtain the transformer bushing leakage current data and set the state space, action space and reward function of reinforcement learning.

[0054] The data processing module is used to select wavelet basis functions and perform wavelet decomposition on the acquired transformer bushing leakage current data.

[0055] The scale energy analysis module is used to calculate the energy of each scale of the wavelet coefficients and to determine the energy threshold of each scale.

[0056] If the function 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 this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0057] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0058] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk case (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0059] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logical function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0060] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for preprocessing leakage current data of converter transformer valve side bushing based on improved wavelet transform, characterized in that: include: Obtain transformer bushing leakage current data and set the state space, action space, and reward function of reinforcement learning; Select wavelet basis function to perform wavelet decomposition on the acquired transformer bushing leakage current data; Calculate the energy of each scale of the wavelet coefficients and make energy threshold judgments at each scale.

2. The method for preprocessing converter transformer valve side bushing leakage current data based on improved wavelet transform according to claim 1, characterized in that: The method of obtaining the transformer bushing leakage current data includes collecting the current data waveform at any time of the day using a current transformer.

3. The method for preprocessing the leakage current data of the converter transformer valve side bushing based on improved wavelet transform according to claim 2, characterized in that: The setting of the state space, action space and reward function of the reinforcement learning includes setting the number of decomposition layers of the state space to be 1 to N layers, and the initial number of decomposition layers to be 1 layer; The action space increases or decreases the number of decomposition levels; The reward function is expressed as: ， Among them, s represents the current decomposition level, a represents increasing or decreasing the decomposition level, represents the expected return of taking action a in state s, α represents the learning rate, γ represents the discount factor, and r represents the immediate reward after taking action a. Indicates the next state The maximum Q value of all possible actions.

4. The method for preprocessing the leakage current data of the converter transformer valve side bushing based on improved wavelet transform according to claim 3, characterized in that: The selecting of the wavelet basis function includes selecting the Symlet function as the wavelet basis function.

5. The method for preprocessing converter transformer valve-side bushing leakage current data based on improved wavelet transform according to claim 4, characterized in that: The wavelet decomposition of the obtained transformer bushing leakage current data includes obtaining wavelet coefficients at different scales and performing wavelet decomposition, which is expressed as: ， in, is the signal to be analyzed, j is the translation scale, is the wavelet coefficient, is the selected Symlet wavelet basis function, t represents time, k represents the translation of the wavelet basis function along the time axis, and Z represents an integer set.

6. The method for preprocessing converter transformer valve-side bushing leakage current data based on improved wavelet transform according to claim 5, characterized in that: The calculating the energy of each scale of the wavelet coefficient includes calculating the energy spectrum of each scale of the wavelet coefficient and squaring the absolute value of the coefficient.

7. The method for preprocessing the leakage current data of the converter transformer valve side bushing based on improved wavelet transform according to claim 6, characterized in that: The energy threshold determination of each scale includes outputting a decomposition result if the energy of each scale is higher than the threshold; If the energy of each scale is lower than the threshold, the Q-value function is used to guide the action selection, that is, to increase or decrease the number of decomposition layers.

8. A system using the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform as claimed in any one of claims 1 to 7, characterized in that: Including data acquisition module, data processing module, scale energy analysis module; The data acquisition module is used to obtain transformer bushing leakage current data and set the state space, action space and reward function of reinforcement learning; The data processing module is used to select a wavelet basis function and perform wavelet decomposition on the acquired transformer bushing leakage current data; The scale energy analysis module is used to calculate the energy of each scale of the wavelet coefficients and to determine the energy threshold of each scale.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the converter transformer valve-side bushing leakage current data preprocessing method based on improved wavelet transform as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the converter transformer valve side bushing leakage current data preprocessing method based on improved wavelet transform as described in any one of claims 1 to 7 are implemented.

Images