Non-destructive testing method for insulation performance parameters of distribution transformer based on polarization current

Abstract

The application relates to the technical field of power distribution equipment operation and maintenance, and provides a nondestructive detection method for insulation performance parameters of a power distribution transformer based on polarization current, which adopts a detrended fluctuation analysis algorithm to extract a detrended signal from complete polarization current data of the power distribution transformer, the complete polarization current data being polarization current data with a time length greater than 2000 seconds; an evaluation function is used to calculate the effective value of the detrended signal, a parameter beta is calculated according to the functional relationship between the effective value of the detrended signal and the parameter beta, and the parameter beta represents a scale coefficient of the polarization current curve; and the insulation performance parameters of transformer insulation oil paper grease are calculated according to the parameter beta. The application can improve detection efficiency, shorten detection time, and will not cause damage to the transformer.

Description

Technical Field

[0001] This invention relates to the field of power distribution equipment operation and maintenance technology, and specifically to a method for detecting insulation performance parameters of power distribution transformers. Background Technology

[0002] Currently, the maintenance of distribution transformers is based on the assessment of their insulation system condition. As an important component of the insulation system, transformer oil paper is subject to aging effects from multiple factors such as temperature, electric field, oxygen, moisture and metallic impurities during operation, resulting in a decline in insulation performance and directly threatening the safe operation of the transformer. Furthermore, the lifespan of the insulation system determines the service life of the transformer.

[0003] There are several methods that can be used to assess the health of transformer insulation, including: DP value (degree of polymerization of transformer insulation material), DGA (dissolved gas in oil) measurement, furan analysis, and PDC (depolarization current analysis).

[0004] While DP values ​​provide accurate information about transformer insulation systems, they require samples taken from the transformer, making them unsuitable for transformers already in operation. DGA analysis often fails to determine the condition of solid insulation. The concentration of furan compounds depends on the test cycle, test conditions, and other parameters such as the oil sample. Furthermore, due to varying operating conditions in different transformers, the aging of transformer insulation paper is uneven, making it unreliable to analyze the insulation paper's condition. PDC analysis is widely used in engineering practice due to its reliability in predicting insulation conditions. It typically uses multiple parallel resistor-capacitor series branches to simulate the oil-paper insulation system, thus establishing a simple RC insulation model. The depolarization current is used to calculate the insulation model parameters R and C to characterize the degree of insulation aging. PDC analysis has the following drawbacks: 1) The establishment of the insulation model depends on the number of resistor-capacitor series branches; an excessively large number will lead to excessively long branch parameter identification times. 2) The performance parameters determined using this insulation model are also affected by the geometry of the insulation system and cannot be directly used to compare different transformers. 3) Measuring depolarization current is relatively complicated: it requires polarizing the insulating medium first, and then depolarizing it, which takes a long time. Summary of the Invention

[0005] The purpose of this invention is to solve the problems existing in the prior art and provide a non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current, which can improve testing efficiency, shorten testing time, and will not cause damage to the transformer.

[0006] This invention is achieved through the following technical solution: a non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current. The method employs a detrended fluctuation analysis algorithm to extract detrended signals from complete polarization current data of the distribution transformer, wherein the complete polarization current data refers to polarization current data with a duration greater than 2000 seconds; an evaluation function is used to calculate the effective value of the detrended signal; and the effective value of the detrended signal is then compared with the parameters... Calculation parameters of functional relationship , The scaling factor represents the polarization current curve; based on the parameters. Calculate the insulation performance parameters of the transformer insulating oil paper.

[0007] Furthermore, predicting the complete polarization current based on the measured short-time polarization current within 1000 seconds includes the following steps: constructing a sample dataset, wherein each sample in the sample data includes the actual measured short-time polarization current and the corresponding complete polarization current; using the actual measured short-time polarization current as training input data, iteratively training the feedforward neural network model until the error between the predicted value and the actual measured complete polarization current reaches the requirement; and using the trained feedforward neural network model to predict the complete polarization current.

[0008] Furthermore, a short-term polarization current lasting 600 seconds is input into the trained feedforward neural network model to predict the full polarization current lasting 10,000 seconds.

[0009] Furthermore, the steps for extracting detrended signals using a detrended volatility analysis algorithm include:

[0010] The synthesized signal is constructed using the fully polarized current signal as the input signal:

[0011]

[0012] In the formula, Indicates a synthesized signal; This represents the total number of value points, i.e., the total number of samples. Indicates the first element in the input signal The value of the point; This represents the sample mean;

[0013] Segmenting the synthesized signal: Dividing the synthesized signal into equal parts There are 1 unit window, and the length of each unit window is 1. , Indicates the integer operation;

[0014] From the synthesized signal Synthetic signals with different window lengths extracted from the middle , Indicates the window length. , ;

[0015] Constructing signals with different window lengths: The complete polarization current signal, i.e., the input signal, is combined with the synthesized signal. The constructed signal is obtained after least squares fitting. P w (l) ;

[0016] Calculate the detrending signal for different window lengths: ,when When =0, =0.

[0017] Furthermore, the evaluation function calculates the effective value of the detrended signal based on the detrended signal with different window lengths using the following formula:

[0018] ;

[0019] In the formula, Indicates length is The effective value of the detrending signal in the window;

[0020] according to ,get .

[0021] Furthermore, the insulation performance parameters include at least one of the following: oil paper humidity, oil paper conductivity, dissipation factor, and activation performance.

[0022] Furthermore, the formula for calculating the moisture content of oiled paper is as follows:

[0023]

[0024] In the formula, Indicates the moisture content of the oil paper.

[0025] Furthermore, the formula for calculating the conductivity of oil paper is as follows:

[0026]

[0027] In the formula, This indicates the conductivity of the oil paper.

[0028] Furthermore, the formula for calculating the dissipation factor is as follows:

[0029]

[0030] In the formula, This represents the dissipation factor.

[0031] Furthermore, the formula for calculating activation performance is as follows:

[0032]

[0033] In the formula, Indicates activation performance.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] 1. This invention is the first to propose using polarization current as the data basis for insulation performance evaluation. Compared with depolarization current, this significantly reduces measurement time and improves detection efficiency. Parameters Due to the degree of insulation system degradation, this invention utilizes a detrended fluctuation analysis (DFA) algorithm to extract detrended signals, and then calculates key parameters based on these detrended signals. This enables parameter control. The calculation.

[0036] 2. The DFA algorithm does not require noise reduction processing of the input data, which reduces the probability of losing valuable information when denoising the measured data.

[0037] 3. In the commonly used PDC analysis method, the establishment of the insulation model depends on the number of branches. If the number of branches is too large, the identification time of branch parameters will be too long, affecting the accuracy of the measurement. The existing PDC analysis method requires the use of a simple RC (resistance-capacitance) based insulation model. Different models using several series and parallel resistors and capacitors will affect the modeling of the oil-paper insulation system, thus affecting the determination of the final insulation parameters.

[0038] The DFA technique proposed in this invention does not require the establishment of a model to solve for insulation parameters, thus decoupling the influence of insulation system modeling on insulation performance parameter estimation. Attached Figure Description

[0039] Figure 1 This is a flowchart of a non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current.

[0040] Figure 2 A schematic diagram of the feedforward neural network model. Detailed Implementation

[0041] The present invention will now be described in further detail with reference to the accompanying drawings:

[0042] refer to Figure 1 As shown, a non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current includes three main steps, which are described in detail below.

[0043] 1) Predict complete polarization current data

[0044] Complete polarization current data refers to polarization current data with a duration of more than 2000 seconds. Predicting the complete polarization current based on short-term polarization current measurements within 1000 seconds (which can be measured using an electrometer) involves the following steps: constructing a sample dataset, where each sample includes the actual measured short-term polarization current and its corresponding complete polarization current; iteratively training a feedforward neural network model using the actual measured short-term polarization current as training input data until the error between the predicted value and the actual measured complete polarization current meets the requirements; and using the trained feedforward neural network model to predict the complete polarization current.

[0045] refer to Figure 2 As shown, the feedforward neural network model contains a single intermediate layer, which takes subsequent time series data as the input for prediction. The relationship between its output function y(m) and the lagged inputs (ym-1, ym-2, ..., ym-k) is as follows:

[0046]

[0047]

[0048] in, α j ( j =0,1,2,… q )and ( i =0,1,2,… p ) are the connection weights, also known as the model parameters determined iteratively; p and q represent the number of input nodes and the number of intermediate hidden nodes, respectively; ε m This represents the random error in the output generated at time m. The output function y can be obtained from the above equation. (m) At time m, the hysteresis polarization current data (y m-1 y m-2 , ..., y m-p The relationship is:

[0049]

[0050] Where w is the parameter vector and f is the weight estimation function. By experimentally measuring polarization current values ​​within different time ranges (i.e., 400s, 500s, 600s, 700s, etc.), the polarization current for 10000s was predicted. It was found that the predicted polarization current data at 600s was more consistent with the actual data. Therefore, the actual polarization current data from the first 600s was used to predict the complete polarization current. Taking the prediction of the 1000s complete polarization current from the 600s short-time polarization current as an example, m=10000, Data at 600-second intervals

[0051] II) Using DFA to obtain numerical parameters

[0052] A detrended fluctuation analysis algorithm is used to extract detrended signals from the complete polarization current data of the distribution transformer:

[0053] The synthesized signal is constructed using the fully polarized current signal as the input signal:

[0054]

[0055] In the formula, Indicates a synthesized signal; This represents the value at any point in the input signal; This represents the total number of value points, i.e., the total number of samples. This represents the sample mean;

[0056] Segmenting the synthesized signal: Dividing the synthesized signal into equal parts There are 10 windows, each with a length of 1. , Indicates the integer operation;

[0057] From the synthesized signal Synthetic signals with different window lengths extracted from the middle , Indicates the window length. , ;

[0058] Constructing signals with different window lengths: The complete polarization current signal, i.e., the input signal, is combined with the synthesized signal. The constructed signal is obtained after least squares fitting. P w (l) ;

[0059] Calculate the detrending signal for different window lengths: ,when When =0, =0.

[0060] The effective value of the detrending signal is calculated using an evaluation function, and the effective value of the detrending signal is then compared with the parameters. Calculation parameters of functional relationship , The scaling factor represents the polarization current curve;

[0061] The evaluation function calculates the effective value of the detrended signal based on different window lengths using the following formula:

[0062] ;

[0063] In the formula, Indicates length is The effective value of the detrending signal of the window, that is, the effective value of the detrending signal of the cell window;

[0064] according to ,get .

[0065] III) Calculation of insulation performance parameters

[0066] According to parameters Calculate the insulation performance parameters of the transformer insulating oil paper, and evaluate the insulation performance of the distribution transformer based on the insulation performance parameters.

[0067] Water is one of the byproducts of aging in distribution transformers, making the monitoring of the oil paper humidity crucial. The oil paper humidity corresponding to the polarization current prediction data of distribution transformers is also important. )and β There is a corresponding relationship between them. The humidity of the oil paper can be obtained by actual measurement and least squares curve fitting. Table) and β The relationship is cubic:

[0068] (1)

[0069] Over time, the dielectric properties of the paper insulation of operating transformers deteriorate due to moisture absorption and aging, leading to reduced insulation performance. This increases the amplitude of the response current, ultimately increasing the dissipation factor or loss angle (tan φ). δ Experiments revealed that the dissipation factor ( )and β There is a relatively stable functional relationship as shown in equation (2).

[0070] (2)

[0071] Oil paper conductivity The conductivity of the oil-paper insulation is a parameter that measures the insulation health of a system, and its value is affected by moisture content and the degree of aging. Studies have found that the conductivity of the oil-paper insulation in operational power transformers... The same as given in equation (3) β Following the cubic relationship, that is:

[0072] (3)

[0073] Activation performance E a It is an important parameter for assessing the remaining service life of a power transformer. It can be used to... E a and βThe relationship between them is modeled using Equation (4), and then the coefficients in Equation (4) are identified using a curve fitting method based on least squares.

[0074] (4)

[0075] In summary, the proposed method is a technique that can estimate the insulation performance parameters of any operational power transformer using short-term, non-invasive measurements.

[0076] The above technical solution is only one embodiment of the present invention. For those skilled in the art, based on the principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the technical solutions described in the specific embodiments of the present invention. Therefore, the foregoing description is only a preferred option and is not restrictive.

Claims

1. A non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current, characterized in that, A detrended fluctuation analysis algorithm is used to extract the detrended signal from the complete polarized current data of the distribution transformer. The complete polarized current data refers to polarized current data with a duration greater than 2000 seconds. An evaluation function is used to calculate the effective value of the detrended signal, and the effective value of the detrended signal is then compared with the parameters. Calculation parameters of functional relationship , The scaling factor represents the polarization current curve; based on the parameters. Calculate the insulation performance parameters of the transformer insulating oil paper; The insulation performance parameters include at least one of the following: oil paper humidity, oil paper conductivity, dissipation factor, and activation performance. The formula for calculating the moisture content of the oil paper is as follows: In the formula, Indicates the moisture content of the oil paper; The formula for calculating the conductivity of the oil paper is as follows: In the formula, Indicates the electrical conductivity of the oil paper; The formula for calculating the dissipation factor is as follows: In the formula, Indicates the dissipation factor; The formula for calculating the activation performance is as follows: In the formula, Indicates activation performance.

2. The non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current according to claim 1, characterized in that, Predicting the complete polarization current based on the measured short-time polarization current within 1000 seconds includes the following steps: constructing a sample dataset, wherein each sample in the sample data includes the actual measured short-time polarization current and the corresponding complete polarization current; using the actual measured short-time polarization current as training input data, iteratively training a feedforward neural network model until the error between the predicted value and the actual measured complete polarization current reaches the requirement; and using the trained feedforward neural network model to predict the complete polarization current.

3. The non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current according to claim 2, characterized in that, Input a short-term polarization current of 600 seconds into the trained feedforward neural network model to predict the full polarization current of 10,000 seconds.

4. The non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current according to claim 1, characterized in that, The steps for extracting detrended signals using a detrended volatility analysis algorithm include: The synthesized signal is constructed using the fully polarized current signal as the input signal: In the formula, Indicates a synthesized signal; This represents the total number of value points, i.e., the total number of samples. Indicates the first element in the input signal The value of the point; This represents the sample mean; Segmenting the synthesized signal: Dividing the synthesized signal into equal parts There are 1 unit window, and the length of each unit window is 1. , Indicates the integer operation; From the synthesized signal Synthetic signals with different window lengths extracted from the middle , Indicates the window length. , ; Constructing signals with different window lengths: The complete polarization current signal, i.e., the input signal, is combined with the synthesized signal. The constructed signal is obtained after least squares fitting. P w (l) ; Calculate the detrending signal for different window lengths: ,when When =0, =0.

5. The non-destructive testing method for insulation performance parameters of distribution transformers based on polarization current according to claim 4, characterized in that, The evaluation function calculates the effective value of the detrended signal based on different window lengths using the following formula: ； In the formula, Indicates length is The effective value of the detrending signal in the window; according to ,get .

Images