A method for measuring the water content of the solid insulation of a transformer

Abstract

The present application relates to a kind of transformer solid insulation water content measurement method, first, different water content transformer is measured in different frequency range using FDS and PDC method respectively, obtains the dielectric spectrum curve based on FDS and the dielectric spectrum curve based on PDC under different water content;Then respectively to the dielectric spectrum curve based on FDS and the dielectric spectrum curve based on PDC in the extraction of characteristic parameter, again respectively with FDS and PDC dielectric spectrum curve extraction characteristic parameter and frequency as input layer, water content is output layer and carries out BP neural network fitting target model function;Finally, further to the target model function respectively fitted in front respectively set weight factor, establish the target model of whole frequency section;Further using particle swarm optimization algorithm to the weight factor in the whole frequency section target model established under different water content is optimized, to obtain the most accurate target model, at this time can accurately measure the water content of transformer solid insulation.

Description

Technical Field

[0001] This invention relates to the field of insulation assessment technology, and in particular to a method for measuring the moisture content of solid insulation in transformers based on frequency domain analysis and time domain current measurement. Background Technology

[0002] Transformers are core equipment in power systems, and their insulation condition directly affects the safe and stable operation of the power grid. The moisture content of the transformer's solid insulation (mainly composed of cellulose insulating paper, insulating paperboard, and support strips) reflects the transformer's insulation condition and aging degree, and is of great significance for predicting the transformer's remaining lifespan and diagnosing faults. The moisture content of a transformer is mainly concentrated in its solid materials; for example, 97% of the moisture in the transformer's insulating medium exists in the solid insulation materials. Therefore, the water content in the solid insulation is a crucial parameter reflecting the insulation condition. Thus, accurately measuring the moisture content in the solid insulation is key to assessing the insulation performance of electrical equipment.

[0003] Currently, the assessment of transformer insulation is limited to routine preventive tests. Due to the limitations of routine preventive tests, such as insulation resistance measurement, partial discharge measurement, dissolved gas chromatography analysis in oil, and dielectric loss factor and capacitance testing, these methods are easily affected by surface leakage current, temperature and humidity. It is difficult to accurately measure the water content of solid insulation materials in transformers, that is, it is impossible to accurately assess the insulation condition of transformers containing solid insulation materials and to perform on-site diagnosis of transformers.

[0004] In the field of insulation measurement, experts and scholars have proposed a measurement method called the frequency domain spectrum method. However, this traditional single frequency domain spectrum method has serious shortcomings when used for insulation condition assessment, namely, the measurement time is too long. From 1kHz to 10μHz, the traditional frequency domain spectrum method requires tens of hours or even longer to complete the measurement. Therefore, there is an urgent need for a new method to measure the moisture content of transformer solid insulation materials that can overcome the limitations of conventional preventive testing, has a shorter measurement time, and provides more accurate results. Summary of the Invention

[0005] The purpose of this invention is to provide a dielectric spectrum method for measuring the water content of transformer solid insulation based on frequency domain analysis and time domain current measurement. Addressing the limitations of existing routine preventative tests for transformers and the excessively long measurement time of traditional frequency domain spectral methods, this invention employs a dielectric spectrum analysis method combining frequency domain spectral analysis (FDS) and polarization-depolarization current (PDC). This fully combines FDS, suitable for high-frequency measurements, with PDC, suitable for low-frequency measurements. Furthermore, a BP neural network and particle swarm optimization algorithm are used to fit and optimize the target model function to obtain the most accurate target model, thereby achieving the most precise measurement of the water content of transformer solid insulation.

[0006] This invention provides a method for measuring the moisture content of solid insulation in a transformer, comprising the following steps:

[0007] Step 1: Based on the FDS method, dielectric spectrum response tests are performed on transformers with different water contents in the frequency range of 5kHz-100mHz to obtain dielectric spectrum curves of transformers with different water contents; Based on the PDC method, dielectric spectrum response tests are performed on transformers with different water contents in the frequency range of 100mHz~10μHz to obtain dielectric spectrum curves of transformers with different water contents.

[0008] Step 2: Extract characteristic parameters from the spectral response curves of each medium in the frequency range of 5kHz to 100mHz based on the FDS method; extract characteristic parameters from the spectral response curves of each medium in the frequency range of 100mHz to 10μHz based on the PDC method.

[0009] Step 3: Using the feature parameters extracted from the FDS method of each medium spectrum curve in Step 2 as the input layer and the water content corresponding to each medium spectrum curve as the output layer, train the target model function y using a BP neural network model. （FDS） =f( (n∈1~4); The feature parameters extracted from the media spectrum curves based on the PDC method in step 2 are used as the input layer, and the water content corresponding to each media spectrum curve is used as the output layer. The target model function y is trained using a BP neural network model. （PDC） =f( ), m∈5~8; where, A n A m a is the conductance amplitude; n、 a m X is the conductance slope; n X m This represents the polarization amplitude. , The polarization time constant;

[0010] Step 4, convert the target model function y （FDS） =f( ), n∈1~4, y （PDC） =f( For m∈5~8, set weight factors Q1 and Q2 respectively to establish the target model y for the entire frequency band. （FDS，PDC） =Q1×f( )+Q2×f( );

[0011] Step 5: Use the particle swarm optimization algorithm to analyze the target model y across the entire frequency band under different water contents. （FDS，PDC） =Q1×f( )+Q2×f( The weighting factors Q1 and Q2 in the model are optimized to obtain the most accurate target model.

[0012] Step 6: Obtain the accurate measurement model y for the moisture content of the dielectric spectroscopy transformer solid material insulation based on FDS and PDC methods. （FDS，PDC） =Q1×f( )+Q2×f( Based on this model y （FDS，PDC） The obtained water content is the specific water content of the transformer being tested.

[0013] Furthermore, step 2, which involves extracting characteristic parameters from the spectral response curves of each medium in the frequency range of 5 kHz to 100 mHz based on the FDS method, includes:

[0014] The frequency range of 5kHz to 100mHz is divided into four bands: band ① 5kHz to 100Hz, band ② 100Hz to 10Hz, band ③ 10Hz to 1Hz, and band ④ 1Hz to 100mHz. The characteristic parameter is the conductance amplitude A within each of the four bands. n Conductivity slope a n Polarization amplitude X n Polarization time constant n=1~4, that is, within the frequency band ① 5kHz~100Hz, respectively denoted as: conductance amplitude A1, conductance slope a1, polarization amplitude X1, polarization time constant. Similarly, the naming of the characteristic parameters for the remaining frequency bands ②, ③, and ④ is carried out in the same manner.

[0015] Furthermore, step 2, which involves extracting characteristic parameters from the spectral response curves of each medium within the frequency range of 100 mHz to 10 μHz based on the PDC method, includes:

[0016] The frequency range of 100mHz to 10μHz is divided into four bands: band ⑤ (100mHz to 10mHz), band ⑥ (10mHz to 1mHz), band ⑦ (1000μHz to 100μHz), and band ⑧ (100μHz to 10μHz). The characteristic parameter is taken as the conductance amplitude A within each of the four bands. m Conductivity slope a m Polarization amplitude X m Polarization time constant m=5-8, that is, within the frequency band ⑤ 100mHz~10mHz, respectively denoted as: conductivity amplitude A5, conductivity slope a5, polarization amplitude X5, and polarization time constant. Similarly, the naming of the characteristic parameters for the remaining frequency bands ⑥, ⑦, and ⑧ is carried out in the same manner.

[0017] The above scheme, through the method of measuring the water content of solid insulation in transformers based on frequency domain analysis and time domain current measurement, has the following technical effects:

[0018] Overcoming the limitations of conventional preventive tests, such as insulation resistance measurement, partial discharge measurement, and dissolved gas chromatography analysis in oil, these methods are easily affected by surface leakage current, temperature, and humidity during measurement, which can easily lead to deviations and cannot accurately evaluate the insulation performance of transformers.

[0019] Overcoming the limitations of conventional preventive testing, conventional testing methods, whether single or comprehensive, are limited by their underlying principles and cannot accurately measure the water content in the solid insulation material inside the transformer.

[0020] A dielectric spectral analysis method combining frequency domain spectrum analysis (FDS) and polarization-depolarization current analysis (PDC) is adopted. This method fully combines FDS, which is suitable for high-frequency measurements, with PDC, which is suitable for low-frequency measurements. Furthermore, a BP neural network and particle swarm optimization algorithm are used to fit and optimize the target model function to obtain the most accurate target model, thereby achieving the goal of most accurately measuring the water content of transformer solid insulation.

[0021] By fully combining the FDS, which is suitable for high-frequency measurements, with the PDC, which is suitable for low-frequency measurements, and by using BP neural networks and particle swarm optimization algorithms to fit and optimize the target model function, the measurement time of the traditional single frequency domain spectrum method is shortened from more than ten hours to less than two hours.

[0022] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0023] Figure 1 This is a flowchart of the method for measuring the water content of solid insulation in a dielectric transformer based on frequency domain analysis and time domain current measurement, according to the present invention.

[0024] Figure 2 This is a circuit diagram for testing the moisture content of solid insulation in a transformer based on the FDS method in one embodiment of the present invention;

[0025] Figure 3 This is a circuit diagram for testing the moisture content of solid insulation in a transformer based on the PDC method in one embodiment of the present invention;

[0026] Figure 4 This is a diagram of the BP neural network structure of the present invention;

[0027] Figure 5 This is a flowchart of the particle swarm optimization algorithm of the present invention for optimizing weight factors Q1 and Q2. Detailed Implementation

[0028] The following will further describe in detail the specific implementation manners of the present invention in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but not to limit the scope of the present invention.

[0029] Refer Figures 1 to 3 As shown, this embodiment provides a method for measuring the moisture content of the solid insulation of a dielectric spectrum transformer based on frequency-domain analysis and time-domain current measurement, including the following steps:

[0030] S1.1. Based on the FDS method, perform dielectric spectrum response tests on transformers with different moisture contents within the frequency range of 5 kHz - 100 mHz to obtain the dielectric spectrum curves of the transformers under different moisture contents.

[0031] Further, for the convenience of explanation, the specific wiring schematic diagram is as Figure 2 shown, where 1 is a dielectric spectrum analyzer based on FDS; 2 is a simplified schematic diagram of the transformer equivalent; 3 is the 2L4 line of the current input terminal of the dielectric spectrum analyzer; 4 is the L3 line of the working grounding terminal of the dielectric spectrum analyzer; 5 is the 1L2 line of the current input terminal of the dielectric spectrum analyzer; 6 is the L1 line of the voltage output line of the dielectric spectrum analyzer; 7 is the high-voltage bushing of the transformer; 8 is the medium-voltage bushing of the transformer; 9 is the low-voltage bushing of the transformer. The specific wiring is that the L1 line of the voltage output terminal is connected to the high-voltage bushing; the 1L2 line of the current input terminal is connected to the medium-voltage bushing; the L3 line of the working grounding terminal is connected to the transformer casing; the 2L4 line of the current input terminal is connected to the low-voltage bushing. It should be noted that: the schematic diagram in the figure shows a transformer with low, medium, and high voltage levels. If it is a transformer with low and high voltage levels, then 5 and 8 are removed correspondingly in the schematic diagram, and the others remain unchanged.

[0032] S1.2. Based on the PDC method, perform dielectric spectrum response tests on transformers with different moisture contents within the frequency range of 100 mHz - 10 μHz to obtain the dielectric spectrum curves of the transformers under different moisture contents.

[0033] Further, for the convenience of explanation, the specific wiring schematic diagram is as Figure 3 shown, where 10 is a dielectric spectrum analyzer based on PDC; 11 is a simplified schematic diagram of the transformer equivalent; 12 is the 2L8 line of the current input terminal of the dielectric spectrum analyzer; 13 is the L7 line of the working grounding terminal of the dielectric spectrum analyzer; 14 is the 1L6 line of the current input terminal of the dielectric spectrum analyzer; 15 is the L5 line of the voltage output line of the dielectric spectrum analyzer; 16 is the high-voltage bushing of the transformer; 17 is the medium-voltage bushing of the transformer; 18 is the low-voltage bushing of the transformer. The specific wiring is that the L5 line of the voltage output terminal is connected to the high-voltage bushing; the 1L6 line of the current input terminal is connected to the medium-voltage bushing; the L7 line of the working grounding terminal is connected to the transformer casing; the 2L8 line of the current input terminal is connected to the low-voltage bushing. It should be noted that: the schematic diagram in the figure shows a transformer with low, medium, and high voltage levels. If it is a transformer with low and high voltage levels, then 14 and 17 are removed correspondingly in the schematic diagram, and the others remain unchanged.

[0034] It should be noted that there is no fixed order between steps S1.1 and S1.2, and they can be interchanged.

[0035] S2.1 Extract characteristic parameters from the spectral response curves of each medium in the frequency range of 5kHz~100mHz based on the FDS method. First, the 5kHz~100mHz range is divided into four frequency bands: band ① 5kHz~100Hz, band ② 100Hz~10Hz, band ③ 10Hz~1Hz, and band ④ 1Hz~100mHz. Further, the characteristic parameters are taken as the conductance amplitude A within each of the four frequency bands. n Conductivity slope a n Polarization amplitude X n Polarization time constant n=1~4, that is, within the frequency band ① 5kHz~100Hz, respectively denoted as: conductance amplitude A1, conductance slope a1, polarization amplitude X1, polarization time constant. Similarly, the naming of the characteristic parameters for the remaining frequency bands ②, ③, and ④ is carried out in the same manner.

[0036] S2.2 The characteristic parameters of the dielectric spectral response curves in the frequency range of 100mHz to 10μHz based on the PDC method are extracted. First, the 100mHz to 10μHz range is divided into four frequency bands: band ⑤ 100mHz to 10mHz, band ⑥ 10mHz to 1mHz, band ⑦ 1000μHz to 100μHz, and band ⑧ 100μHz to 10μHz. Further, the characteristic parameters are taken as the conductance amplitude A within each of the four frequency bands. m Conductivity slope a m Polarization amplitude X m Polarization time constant m=5-8, that is, within the frequency band ⑤ 100mHz~10mHz, respectively denoted as: conductivity amplitude A5, conductivity slope a5, polarization amplitude X5, and polarization time constant. Similarly, the naming of the characteristic parameters for the remaining frequency bands ⑥, ⑦, and ⑧ is carried out in the same manner.

[0037] It should be noted that steps S2.1 and S2.2 have no fixed order and can be interchanged.

[0038] S3.1. Using the feature parameters extracted from the FDS method of each medium spectrum curve in S2.1 as the input layer and the water content corresponding to each medium spectrum curve as the output layer, the target model function y is trained using a BP neural network model. （FDS） =f( ), n∈1~4.

[0039] S3.2. Using the feature parameters extracted from the PDC method of each medium spectrum curve in S2.2 as the input layer and the water content corresponding to each medium spectrum curve as the output layer, the target model function y is trained using a BP neural network model. （PDC） =f( m∈5~8.

[0040] Furthermore, the neural network function described in steps S3.1 and S3.2 is the error feedback neural network algorithm. Structurally, a BP neural network consists of two modules: a forward propagation network for information and a backward propagation network for error. The basic structure of a BP neural network is as follows: Figure 4 As shown.

[0041] As shown in the diagram, a backpropagation (BP) neural network mainly consists of three layers: the input layer, the hidden layers, and the output layer. Various information from the outside is transmitted through the input layer to the hidden layers for processing, and then output through the output layer to obtain the final result. When the error between the output and the pre-set input value is large, the network enters the backpropagation phase, updating the network weights until the error between the output and the expected result meets certain conditions.

[0042] The main steps in the forward propagation of the signal are as follows:

[0043] First, the input variable net of the i-th node in the hidden layer of the neural network. i :

[0044]

[0045] Second, the output variable y of the i-th node in the hidden layer of the neural network. i :

[0046]

[0047] Third, the input variable net of the k-th node in the output layer of the neural network. k :

[0048]

[0049] Fourth, the output variable O of the k-th node in the output layer of the neural network. k :

[0050]

[0051] Where the variable x jThe meaning of represents the input parameters of the j-th node in the input layer of the BP neural network. It should be noted that here, x j The conductivity amplitude A is respectively n A m Conductivity slope a n , a m Polarization amplitude X n , X m Polarization time constant , n=1~4, m=5~8; w ij The variable θ represents the neural network weight parameters between the i-th node of the hidden layer and the j-th node of the input layer in a BP neural network. i The meaning of represents the threshold parameter of the i-th node in the hidden layer of the BP neural network; the meaning of variable Ø(x) represents the activation function of the hidden layer of the BP neural network; the meaning of variable w ki The meaning represents the weight parameters between the k-th node of the output layer and the i-th node of the hidden layer in a BP neural network, where i = 1 to q; variable α k The meaning of represents the threshold parameter of the k-th node in the output layer of the BP neural network, where k = 1 to L; variable (x) represents the activation function of the output layer of the BP neural network; variable o k The meaning of represents the output of the k-th node in the output layer of the BP neural network. It should be noted that the output variable o here k This refers to the water content.

[0052] It should be noted that steps S3.1 and S3.2 have no fixed order and can be interchanged.

[0053] S4. The target model function y （FDS） =f(A n ,a n ,X n ,τ n ), n∈1~4, y （PDC） =f(A m ,a m ,X m ,τ m For each frequency band, m∈5~8, set weighting factors to establish the target model y. （FDS，PDC） =Q 1× f( )+Q 2× f( ).

[0054] S5. Using the Particle Swarm Optimization (PSO) algorithm to analyze the target model y across the entire frequency band under different water contents. （FDS，PDC） =Q 1×f( )+Q 2× f( The weighting factors Q1 and Q2 are optimized to obtain the most accurate target model.

[0055] Furthermore, the particle swarm optimization algorithm flow described in step S5 is as follows: Figure 5 As shown:

[0056] S6. Obtain a precise measurement model for the moisture content of transformer solid material insulation based on the dielectric spectrum method using FDS and PDC. （FDS，PDC） =Q 1× f( )+Q 2× f( Based on this model y （FDS，PDC） The obtained water content is the specific water content of the transformer being tested.

[0057] The proposed method for measuring the moisture content of transformer solid insulation first employs both the FDS (Frequency Spectrum Distributed) and PDC (Programmable Density Spectrum Distributed) methods to measure the dielectric response curves of transformers with different moisture contents within different frequency ranges. The FDS method operates in the frequency range of 5kHz-100mHz, while the PDC method operates in the frequency range of 100mHz-10μHz, yielding dielectric response curves based on both FDS and PDC methods for different moisture contents. Then, feature parameters are extracted from both FDS and PDC dielectric response curves. Finally, using the extracted feature parameters and frequencies as input layers and the moisture content as the output layer, a backpropagation neural network is used to fit (train) the target model function, denoted as y. （FDS） and y （PDC） Finally, the target model function y that was previously fitted is further... （FDS） and y （PDC） By setting weighting factors separately, a target model y is established for the entire frequency band. （FDS，PDC） Furthermore, the particle swarm optimization algorithm is used to optimize the weighting factors in the full-frequency target model established under different moisture contents to obtain the most accurate target model. This allows for precise measurement of the moisture content of the transformer's solid insulation. Specifically, this method offers the following technical advantages:

[0058] Overcoming the limitations of conventional preventive tests, such as insulation resistance measurement, partial discharge measurement, and dissolved gas chromatography analysis in oil, these methods are easily affected by surface leakage current, temperature, and humidity during measurement, which can easily lead to deviations and cannot accurately evaluate the insulation performance of transformers.

[0059] Overcoming the limitations of conventional preventive testing, conventional testing methods, whether single or comprehensive, are limited by their underlying principles and cannot accurately measure the water content in the solid insulation material inside the transformer.

[0060] A dielectric spectral analysis method combining frequency domain spectrum analysis (FDS) and polarization-depolarization current analysis (PDC) is adopted. This method fully combines FDS, which is suitable for high-frequency measurements, with PDC, which is suitable for low-frequency measurements. Furthermore, a BP neural network and particle swarm optimization algorithm are used to fit and optimize the target model function to obtain the most accurate target model, thereby achieving the goal of most accurately measuring the water content of transformer solid insulation.

[0061] By fully combining the FDS, which is suitable for high-frequency measurements, with the PDC, which is suitable for low-frequency measurements, and by using BP neural networks and particle swarm optimization algorithms to fit and optimize the target model function, the measurement time of the traditional single frequency domain spectrum method is shortened from more than ten hours to less than two hours.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for measuring the moisture content of solid insulation in a transformer, characterized in that, Includes the following steps: Step 1: Based on the FDS method, dielectric spectrum response tests were conducted on transformers with different water contents in the frequency range of 5kHz-100mHz to obtain dielectric spectrum curves of transformers with different water contents. Based on the PDC method, dielectric spectrum response tests were conducted on transformers with different water contents in the frequency range of 100mHz~10μHz to obtain dielectric spectrum curves of transformers with different water contents. Step 2: Extract characteristic parameters from the spectral response curves of each medium in the frequency range of 5kHz to 100mHz based on the FDS method; extract characteristic parameters from the spectral response curves of each medium in the frequency range of 100mHz to 10μHz based on the PDC method. Step 3: Using the feature parameters extracted from the FDS method of each medium spectrum curve in Step 2 as the input layer and the water content corresponding to each medium spectrum curve as the output layer, train the target model function y using a BP neural network model. （FDS） =f( (n∈1~4); The feature parameters extracted from the media spectrum curves based on the PDC method in step 2 are used as the input layer, and the water content corresponding to each media spectrum curve is used as the output layer. The target model function y is trained using a BP neural network model. （PDC） =f( ), m∈5~8; Among them, A n A m a is the conductance amplitude; n a m X is the conductance slope; n X m This represents the polarization amplitude. , The polarization time constant; Step 4, convert the target model function y （FDS） =f( ), n∈1~4, y （PDC） =f( For m∈5~8, set weight factors Q1 and Q2 respectively to establish the target model y for the entire frequency band. （FDS，PDC） =Q1×f( )+Q2×f( ); Step 5: Use the particle swarm optimization algorithm to analyze the target model y across the entire frequency band under different water contents. （FDS，PDC） =Q1×f( )+Q2×f( The weighting factors Q1 and Q2 in the model are optimized to obtain the most accurate target model. Step 6: Obtain the accurate measurement model y for the moisture content of the dielectric spectroscopy transformer solid material insulation based on FDS and PDC methods. （FDS，PDC） =Q1×f( )+Q2×f( Based on this model y （FDS，PDC） The obtained water content is the specific water content of the transformer being tested.

2. The method for measuring the moisture content of solid insulation in a transformer according to claim 1, characterized in that, Step 2, which involves extracting characteristic parameters from the spectral response curves of various media in the frequency range of 5 kHz to 100 mHz based on the FDS method, includes: The frequency range of 5kHz to 100mHz is divided into four bands: band ① 5kHz to 100Hz, band ② 100Hz to 10Hz, band ③ 10Hz to 1Hz, and band ④ 1Hz to 100mHz. The characteristic parameter is the conductance amplitude A within each of the four bands. n Conductivity slope a n Polarization amplitude X n Polarization time constant n=1~4, that is, within the frequency band ① 5kHz~100Hz, respectively denoted as: conductance amplitude A1, conductance slope a1, polarization amplitude X1, polarization time constant. Similarly, the naming of the characteristic parameters for the remaining frequency bands ②, ③, and ④ is carried out in the same manner.

3. The method for measuring the moisture content of solid insulation in a transformer according to claim 1, characterized in that, Step 2, which involves extracting characteristic parameters from the spectral response curves of various media in the frequency range of 100 mHz to 10 μHz based on the PDC method, includes: The frequency range of 100mHz to 10μHz is divided into four bands: band ⑤ (100mHz to 10mHz), band ⑥ (10mHz to 1mHz), band ⑦ (1000μHz to 100μHz), and band ⑧ (100μHz to 10μHz). The characteristic parameter is taken as the conductance amplitude A within each of the four bands. m Conductivity slope a m Polarization amplitude X m Polarization time constant m=5-8, that is, within the frequency band ⑤ 100mHz~10mHz, respectively denoted as: conductivity amplitude A5, conductivity slope a5, polarization amplitude X5, and polarization time constant. Similarly, the naming of the characteristic parameters for the remaining frequency bands ⑥, ⑦, and ⑧ is carried out in the same manner.

Images