A method for detecting short circuit between turns of iron core reactor

Abstract

The present invention relates to the technical field of power equipment detection, and more specifically, to a detection method for turn-to-turn short circuit of an iron core reactor, which is used to solve the problem that the prior art cannot comprehensively evaluate the turn-to-turn short circuit risk of the iron core reactor, cannot capture the equipment state change in real time, and reduces the sensitivity and accuracy of turn-to-turn short circuit risk monitoring; the present invention comprehensively evaluates the turn-to-turn short circuit risk of the iron core reactor by comprehensively considering the operating current, DC resistance, inductance and temperature parameters of key parts, and captures the equipment state change in real time by dynamically calculating the change rate of each parameter, thereby significantly improving the monitoring sensitivity and accuracy; a logistic regression model trained based on historical data can accurately predict the turn-to-turn short circuit probability, realize automatic risk monitoring, reduce manual intervention, and reduce the misjudgment rate.

Description

Technical Field

[0001] The invention relates to the technical field of power equipment detection, and more specifically to a method for detecting inter-turn short circuit of an iron core reactor. Background Art

[0002] Reactors are commonly used devices in power systems, used to limit short-circuit currents, smooth power supply voltage fluctuations, etc. However, during the use of reactors, inter-turn short circuits may occur due to various reasons, which will not only affect the normal operation of the reactor, but may also cause damage to the power system; existing reactor inter-turn short circuit detection methods, such as resistance test method, induction current test method, open circuit test method, etc., although they can detect inter-turn short circuits to a certain extent, have problems such as low detection accuracy and complex operation.

[0003] The patent application with reference publication number CN118962519A discloses a method, system, device and medium for detecting inter-turn short-circuit fault of a reactor, which relates to the technical field of reactors. The detection method uses structural data and operating data to construct a model to generate an adjustable circuit model corresponding to the reactor; coil parameters are identified based on the simulated current corresponding to the adjustable circuit model and the measured current in the operating data to generate target coil parameters; inter-turn short-circuit fault detection is performed based on the target coil parameters and the normal coil impedance matrix corresponding to the reactor to generate inter-turn short-circuit detection data; the present invention takes the target coil parameters and the normal coil impedance matrix corresponding to the reactor as the observation objects, and can effectively sense the operating status of each layer of coils of the reactor, thereby improving the accuracy of the detection results;

[0004] However, the above-mentioned reference patent realizes high-sensitivity and high-reliability turn-to-turn short-circuit fault detection by constructing an adjustable circuit model of the inductor and combining measured data to identify coil parameters, which significantly improves the accuracy of the detection results. However, it cannot fully capture the electrical, mechanical and thermal characteristics of the turn-to-turn short-circuit, which reduces the accuracy and reliability of fault diagnosis. It cannot accurately capture the characteristics of the turn-to-turn short-circuit and cannot realize real-time turn-to-turn short-circuit fault detection, which reduces the efficiency and accuracy of fault diagnosis. At the same time, it cannot comprehensively evaluate the turn-to-turn short-circuit risk of the iron core inductor and cannot capture equipment status changes in real time, which reduces the sensitivity and accuracy of turn-to-turn short-circuit risk monitoring. Summary of the invention

[0005] The purpose of the present invention is to provide a detection method for inter-turn short circuit of an iron core reactor, which solves the problem that the prior art cannot fully capture the electrical, mechanical and thermal characteristics of the inter-turn short circuit, reduces the fault diagnosis accuracy and reliability, and cannot accurately capture the characteristics of the inter-turn short circuit, cannot realize real-time inter-turn short circuit fault detection, reduces the efficiency and accuracy of fault diagnosis, and cannot comprehensively evaluate the inter-turn short circuit risk of the iron core reactor, cannot capture the equipment state change in real time, and reduces the sensitivity and accuracy of the inter-turn short circuit risk monitoring.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A method for detecting an inter-turn short circuit of an iron core reactor comprises the following steps:

[0008] Step 1: collect the operating status parameters of the iron core reactor in real time, and perform preprocessing operations on the collected operating status parameters;

[0009] Step 2: extracting turn-to-turn short-circuit feature data from the preprocessed operating state parameters, and fusing the extracted turn-to-turn short-circuit feature data to generate a turn-to-turn short-circuit comprehensive feature matrix;

[0010] Step 3: Collect historical operating status parameters, build a turn-to-turn short-circuit fault diagnosis model, and use the model to perform real-time turn-to-turn short-circuit detection on the iron core reactor;

[0011] Step 4: Used to monitor the turn-to-turn short-circuit risk parameters of the iron core reactor in real time, and monitor and evaluate the turn-to-turn short-circuit risk of the iron core reactor;

[0012] Step 5: It is used to conduct a comprehensive evaluation and analysis on the operation stability of the iron core reactor according to the short-circuit detection results and the short-circuit risk assessment results, and take corresponding measures according to the analysis results.

[0013] As a preferred embodiment of the present invention, the specific process of extracting the inter-turn short circuit characteristic data and fusing the extracted inter-turn short circuit characteristic data in step 2 is as follows:

[0014] Obtaining pre-processed operating state parameters, the operating state parameters include current signals, voltage signals, vibration signals and key part temperatures, and extracting turn-to-turn short-circuit characteristic data from the pre-processed operating state parameters, the turn-to-turn short-circuit characteristic data include harmonic components, impedance change rate, vibration spectrum, peak frequency and temperature rise rate;

[0015] The steps to extract harmonic components are as follows:

[0016] The current signal is sampled to obtain a discrete signal I[n]. The discrete signal is subjected to a fast Fourier transform and the frequency domain signal I(f) is obtained by the following formula:

[0017] ;

[0018] Where I[n] is a discrete current signal, which indicates the current value at time point n; I(f) is a frequency domain current signal, which indicates the current amplitude at frequency f; f is the frequency, which indicates the position of the signal in the frequency domain; N is the number of sampling points, which indicates the total number of times the signal is sampled in the time domain; j is an imaginary unit, which satisfies j 2 =-1;

[0019] Extract a specific frequency component from the frequency domain signal I(f) and obtain the harmonic component X(f) using the following formula:

[0020] X(f)=I(f)·H(f), where H(f) is the harmonic extraction function, usually a bandpass filter or a function that selects a specific frequency, and X(f) is the harmonic component, which represents the harmonic amplitude at frequency f.

[0021] As a preferred embodiment of the present invention, the steps of extracting the impedance change rate are as follows:

[0022] The voltage signal V(t) and the current signal I(t) are sampled, and the impedance Z(t) is obtained by the following formula:

[0023] , where V(t) is the voltage signal, indicating the voltage value at time point t, I(t) is the current signal, indicating the current value at time point t, and Z(t) is the impedance, indicating the impedance value at time point t;

[0024] Analyze the impedance change trend over time and obtain the impedance change rate ∆Z(t) through the following formula:

[0025] ;

[0026] Where Z(t) is the impedance at the current moment, Z(t-1) is the impedance at the previous moment, and ∆t is the time interval;

[0027] The steps to extract the vibration spectrum are as follows:

[0028] The vibration signal is sampled to obtain a discrete signal Y[n]. The discrete signal is subjected to a fast Fourier transform and the frequency domain signal Y(f) is obtained by the following formula:

[0029] ;

[0030] Where Y[n] is a discrete vibration signal, which indicates the vibration value at time point n, and Y(f) is a frequency domain vibration signal, which indicates the vibration amplitude at frequency f;

[0031] The vibration spectrum is the amplitude spectrum of the frequency domain vibration signal Y(f), which is obtained by calculating the modulus of Y(f):

[0032] ;

[0033] Where Re(Y(f)) is the real part of the frequency domain vibration signal, Im(Y(f)) is the imaginary part of the frequency domain vibration signal, |Y(f)| is the amplitude of the frequency domain vibration signal, indicating the vibration intensity at frequency f;

[0034] The steps to extract the peak frequency are as follows:

[0035] Analyze the vibration spectrum |Y(f)| and obtain the frequency f with the maximum amplitude through the following formula: zdf :

[0036] , where f zdf is the peak frequency.

[0037] As a preferred embodiment of the present invention, the steps of extracting the temperature rise rate are as follows:

[0038] The temperature signal is sampled to obtain a discrete signal T[n], and the temperature rise rate ∆T[n] is obtained by the following formula:

[0039] ;

[0040] Where T[n] is the temperature at the current moment, T[n-1] is the temperature at the previous moment, and ∆t is the time interval;

[0041] The extracted harmonic component X(f), impedance change rate ∆Z(t), vibration spectrum |Y(f)|, peak frequency f zdf and the temperature rise rate ∆T[n] are taken as eigenvectors respectively, and these eigenvectors are combined to generate a turn-to-turn short-circuit comprehensive characteristic matrix DF, which is expressed by the following expression:

[0042] ;

[0043] Where X(f i ) is the harmonic component of the i-th frequency, ∆Z j is the jth impedance change rate, |Y(f k )| is the vibration spectrum of the kth frequency, f zdfl is the lth peak frequency, ∆T m is the mth temperature rise rate;

[0044] In order to eliminate the dimensional differences between different features, the turn-to-turn short-circuit comprehensive feature matrix DF is standardized to obtain the turn-to-turn short-circuit standardized matrix DBJ.

[0045] As a preferred implementation of the present invention, the specific process of constructing a turn-to-turn short-circuit fault diagnosis model and performing real-time turn-to-turn short-circuit detection on the iron core reactor through the model in step three is as follows:

[0046] Collect historical operating status parameters, generate a collection cycle, and divide the collection cycle S into multiple collection time periods {s1, s2, …, sn}, that is, S = {s1, s2, …, sn};

[0047] Preprocessing the historical operating state parameters, extracting turn-to-turn short-circuit feature data from the historical operating state parameters in multiple collection periods, and fusing the extracted turn-to-turn short-circuit feature data to generate a turn-to-turn short-circuit comprehensive feature matrix DF. After standardizing the feature matrix DF, multiple groups of turn-to-turn short-circuit standardized matrices DBJ are obtained.

[0048] The turn-to-turn short-circuit standardized matrix DBJ is used as the input of the machine learning model, and the turn-to-turn short-circuit detection evaluation value corresponding to each group of turn-to-turn short-circuit standardized matrices DBJ is used as the output of the machine learning model. The turn-to-turn short-circuit detection evaluation value is used as the prediction target, and the minimum sum of prediction errors of all training data is used as the training target. The machine learning model is trained until the sum of prediction errors converges and the training is stopped to obtain the turn-to-turn short-circuit fault diagnosis model.

[0049] As a preferred embodiment of the present invention, the expression formula of the turn-to-turn short circuit fault diagnosis model is as follows:

[0050] DJP = γ1·X(f i )+γ2·∆Z j +γ3·|Y(f k )|+γ4·f zdfl +γ5·∆T m +ε;

[0051] Where DJP is the turn-to-turn short-circuit detection evaluation value, γ1, γ2, γ3, γ4 and γ5 are all regression coefficients, and ε represents the error term;

[0052] Obtain the real-time operating status parameters, convert them into the corresponding feature matrix and input them into the turn-to-turn short-circuit fault diagnosis model, and obtain the real-time turn-to-turn short-circuit detection evaluation value DJP through the turn-to-turn short-circuit fault diagnosis model 实时 , the real-time turn-to-turn short-circuit detection evaluation value DJP 实时 Compare with the preset turn-to-turn short circuit detection evaluation value threshold:

[0053] If DJP 实时 If it is less than the preset turn-to-turn short-circuit detection evaluation value threshold, it indicates that the iron core reactor operates well and no turn-to-turn short-circuit fault is detected;

[0054] If DJP实时 If it is greater than or equal to the preset turn-to-turn short-circuit detection evaluation value threshold, it indicates that the iron core reactor is operating abnormally and a turn-to-turn short-circuit fault has occurred.

[0055] As a preferred implementation mode of the present invention, the specific process of monitoring and evaluating the risk of inter-turn short circuit of the iron core reactor in step 4 is as follows:

[0056] Obtain historical turn-to-turn short-circuit risk parameters, including operating current, DC resistance, inductance, and key part temperature, generate a monitoring cycle, and divide the monitoring cycle T into multiple monitoring periods {t1, t2, …, tn}, that is, T={t1, t2, …, tn};

[0057] Obtain the operating current change rate of the iron core reactor in multiple monitoring time periods, where the operating current change rate represents the ratio between the operating current change amount and the length of the corresponding time period, and calculate the arithmetic mean of the multiple operating current change rates obtained, and record the arithmetic mean of the multiple operating current change rates as the average operating current change rate PDL;

[0058] The DC resistance change rate of the iron core reactor in multiple monitoring time periods is obtained. The DC resistance change rate represents the ratio between the DC resistance change and the length of the corresponding time period. The arithmetic mean of the multiple DC resistance change rates obtained is calculated and recorded as the average DC resistance change rate PDZ.

[0059] As a preferred embodiment of the present invention, the inductance change rate of the iron core reactor in multiple monitoring time periods is obtained, the inductance change rate represents the ratio between the inductance change amount and the length of the corresponding time period, and the arithmetic mean of the obtained multiple inductance change rates is calculated, and the arithmetic mean of the multiple inductance change rates is recorded as the average inductance change rate PDG;

[0060] The temperature change rate of the key parts of the iron core reactor in multiple monitoring periods is obtained. The temperature change rate of the key parts represents the ratio between the temperature change of the key parts and the length of the corresponding time period. The arithmetic mean of the multiple key part temperature change rates obtained is calculated, and the arithmetic mean of the multiple key part temperature change rates is recorded as the average key part temperature change rate PBW.

[0061] As a preferred embodiment of the present invention, the average operating current change rate PDL, the average DC resistance change rate PDZ, the average inductance change rate PDG and the average key part temperature change rate PBW are obtained, and the average operating current change rate PDL, the average DC resistance change rate PDZ, the average inductance change rate PDG and the average key part temperature change rate PBW are combined to construct a short circuit risk feature matrix DFJ;

[0062] The constructed short-circuit risk feature matrix DFJ is used as the input of the logistic regression model, and the label vector u is used as the output of the logistic regression model. The label vector u indicates whether there is a risk of inter-turn short circuit in the iron core reactor. The output of the label vector u is 0 or 1, 0 indicates that there is no risk of inter-turn short circuit in the iron core reactor, and 1 indicates that there is a risk of inter-turn short circuit in the iron core reactor. The label vector u is used as the prediction target, and the sum of the prediction errors of all training data is minimized as the training target. The logistic regression model is trained until the sum of the prediction errors reaches convergence and the training is stopped. The logistic regression model that predicts the label vector u is obtained.

[0063] The expression formula of the logistic regression model is:

[0064] ;

[0065] Where P(u=1) represents the probability value of the iron core reactor inter-turn short circuit, v0 represents the intercept term, which represents the baseline probability of the iron core reactor inter-turn short circuit, and v1, v2, v3 and v4 are all regression coefficients;

[0066] The real-time turn-to-turn short-circuit risk parameters are obtained and processed to construct the real-time short-circuit risk feature matrix DFJ. The probability value P of the turn-to-turn short-circuit of the iron core reactor is calculated through the trained logistic regression model. 实时 .

[0067] As a preferred embodiment of the present invention, the specific process of the step 5 of comprehensively evaluating and analyzing the operating stability of the iron core reactor is as follows:

[0068] Get the real-time turn-to-turn short circuit detection evaluation value DJP 实时 The probability value P of the turn-to-turn short circuit of the iron core reactor 实时 , the running stability evaluation coefficient WDP is calculated by the following formula:

[0069] ;

[0070] Where d1 and d2 are both preset proportional factor coefficients, d2>d1>0, and the operation stability evaluation coefficient WDP is compared with the preset operation stability evaluation coefficient threshold:

[0071] If the operation stability evaluation coefficient WDP is less than the preset operation stability evaluation coefficient threshold, it indicates that the operation stability of the iron core reactor is good;

[0072] If the operation stability evaluation coefficient WDP is greater than or equal to the preset operation stability evaluation coefficient threshold, it indicates that the operation stability of the iron core reactor is poor;

[0073] When it is detected that the iron core reactor is running stably, corresponding measures are taken to ensure that the iron core reactor maintains stable operation in the current state;

[0074] When it is detected that the operation stability of the iron core reactor is poor, corresponding measures are taken to process it so that the iron core reactor can resume stable operation.

[0075] Compared with the prior art, the advantages of the present invention are:

[0076] In the present invention, by extracting the characteristics of harmonics, impedance change rate, vibration spectrum, peak frequency and temperature rise rate, a comprehensive feature matrix is ​​generated to fully capture the electrical, mechanical and thermal characteristics of the turn-to-turn short circuit, significantly improving the fault diagnosis accuracy and reliability;

[0077] In the present invention, a large amount of historical data is efficiently processed by using machine learning technology through the turn-to-turn short circuit fault diagnosis model, and real-time detection is achieved, which significantly improves the efficiency and accuracy of fault diagnosis. Through feature extraction and standardization processing, the characteristics of turn-to-turn short circuits can be captured more accurately, thereby improving the prediction accuracy.

[0078] In the present invention, the risk of inter-turn short circuit of iron core reactor is comprehensively evaluated by comprehensively considering the operating current, DC resistance, inductance and temperature parameters of key parts. The change rate of each parameter is dynamically calculated to capture the change of equipment status in real time, which significantly improves the sensitivity and accuracy of monitoring. The logistic regression model trained based on historical data can accurately predict the probability of inter-turn short circuit, realize automated risk monitoring, reduce manual intervention and reduce the misjudgment rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 It is a flow chart of the detection method of the inter-turn short circuit of the iron core reactor in the present invention;

[0080] Figure 2 This is a logical flow diagram of the first embodiment of the present invention;

[0081] Figure 3 This is a logical flow diagram of the second embodiment of the present invention. DETAILED DESCRIPTION

[0082] The following will combine the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all the embodiments. All other embodiments obtained by ordinary technicians in this field without creative work based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0083] Embodiment 1: Figure 1 and Figure 2As shown, a method for detecting an inter-turn short circuit of an iron core reactor proposed by the present invention comprises the following steps:

[0084] Step 1: collect the operating status parameters of the core reactor in real time, including current signal, voltage signal, vibration signal and temperature of key parts, and perform preprocessing operations on the collected operating status parameters, including data cleaning, denoising, format conversion and standardization;

[0085] Real-time collection of operating status parameters of iron-core reactors and preprocessing operations can significantly improve data quality and availability. Data cleaning can remove outliers and noise to ensure data accuracy and reliability. Format conversion can unify data formats for easy storage and analysis. Standardization processing can eliminate dimensional differences and make data comparable on the same scale. These operations not only improve data quality, but also optimize the efficiency of subsequent analysis and the performance of machine learning models, providing a solid foundation for condition monitoring and fault diagnosis.

[0086] Step 2: Extract turn-to-turn short-circuit characteristic data from the preprocessed operating state parameters. The turn-to-turn short-circuit characteristic data includes harmonic components, impedance change rate, vibration spectrum, peak frequency, and temperature rise rate. The extracted turn-to-turn short-circuit characteristic data are fused to generate a turn-to-turn short-circuit comprehensive characteristic matrix.

[0087] The specific process of step 2 extracting the turn-to-turn short-circuit feature data and fusing the extracted turn-to-turn short-circuit feature data is as follows:

[0088] Obtaining pre-processed operating state parameters, the operating state parameters include current signals, voltage signals, vibration signals and key part temperatures, and extracting turn-to-turn short-circuit characteristic data from the pre-processed operating state parameters, the turn-to-turn short-circuit characteristic data include harmonic components, impedance change rate, vibration spectrum, peak frequency and temperature rise rate;

[0089] The steps to extract harmonic components are as follows:

[0090] The current signal is sampled to obtain a discrete signal I[n]. The discrete signal is subjected to a fast Fourier transform and the frequency domain signal I(f) is obtained by the following formula:

[0091] ;

[0092] Where I[n] is a discrete current signal, which indicates the current value at time point n; I(f) is a frequency domain current signal, which indicates the current amplitude at frequency f; f is the frequency, which indicates the position of the signal in the frequency domain; N is the number of sampling points, which indicates the total number of times the signal is sampled in the time domain; j is an imaginary unit, which satisfies j 2 =-1;

[0093] Extract specific frequency components from the frequency domain signal I(f), usually focusing on the fundamental frequency and its multiples, and obtain the harmonic component X(f) through the following formula:

[0094] X(f)=I(f)·H(f), where H(f) is the harmonic extraction function, usually a bandpass filter or a function that selects a specific frequency, and X(f) is the harmonic component, which represents the harmonic amplitude at frequency f;

[0095] The steps to extract the impedance change rate are as follows:

[0096] The voltage signal V(t) and the current signal I(t) are sampled, and the impedance Z(t) is obtained by the following formula:

[0097] , where V(t) is the voltage signal, indicating the voltage value at time point t, I(t) is the current signal, indicating the current value at time point t, and Z(t) is the impedance, indicating the impedance value at time point t;

[0098] Analyze the impedance change trend over time and obtain the impedance change rate ∆Z(t) through the following formula:

[0099] ;

[0100] Where Z(t) is the impedance at the current moment, Z(t-1) is the impedance at the previous moment, and ∆t is the time interval;

[0101] The steps to extract the vibration spectrum are as follows:

[0102] The vibration signal is sampled to obtain a discrete signal Y[n]. The discrete signal is subjected to a fast Fourier transform and the frequency domain signal Y(f) is obtained by the following formula:

[0103] ;

[0104] Where Y[n] is a discrete vibration signal, which indicates the vibration value at time point n, and Y(f) is a frequency domain vibration signal, which indicates the vibration amplitude at frequency f;

[0105] The vibration spectrum is the amplitude spectrum of the frequency domain vibration signal Y(f), which is obtained by calculating the modulus of Y(f):

[0106] ;

[0107] Where Re(Y(f)) is the real part of the frequency domain vibration signal, Im(Y(f)) is the imaginary part of the frequency domain vibration signal, |Y(f)| is the amplitude of the frequency domain vibration signal, indicating the vibration intensity at frequency f;

[0108] The steps to extract the peak frequency are as follows:

[0109] Analyze the vibration spectrum |Y(f)| and obtain the frequency f with the maximum amplitude through the following formula: zdf :

[0110] , where f zdf is the peak frequency;

[0111] The steps to extract the temperature rise rate are as follows:

[0112] The temperature signal is sampled to obtain a discrete signal T[n], and the temperature rise rate ∆T[n] is obtained by the following formula:

[0113] ;

[0114] Where T[n] is the temperature at the current moment, T[n-1] is the temperature at the previous moment, and ∆t is the time interval;

[0115] The extracted harmonic component X(f), impedance change rate ∆Z(t), vibration spectrum |Y(f)|, peak frequency f zdf and the temperature rise rate ∆T[n] are taken as eigenvectors respectively, and these eigenvectors are combined to generate a turn-to-turn short-circuit comprehensive characteristic matrix DF, which is expressed by the following expression:

[0116] ;

[0117] Where X(f i ) is the harmonic component of the i-th frequency, ∆Z j is the jth impedance change rate, |Y(f k )| is the vibration spectrum of the kth frequency, f zdfl is the lth peak frequency, ∆T m is the mth temperature rise rate;

[0118] In order to eliminate the dimensional differences between different features, the turn-to-turn short-circuit comprehensive feature matrix DF is standardized to obtain the turn-to-turn short-circuit standardized matrix DBJ;

[0119] By extracting the characteristics of harmonics, impedance change rate, vibration spectrum, peak frequency and temperature rise rate, a comprehensive feature matrix is ​​generated to comprehensively capture the electrical, mechanical and thermal characteristics of turn-to-turn short circuits, significantly improving the accuracy and reliability of fault diagnosis. Standardization processing eliminates dimensional differences, ensures data comparability, provides high-quality input for machine learning, enhances the generalization ability of the model, and optimizes the efficiency of fault classification, prediction and status assessment, making it more stable under different working conditions.

[0120] Step 3: Collect historical operating status parameters, build a turn-to-turn short-circuit fault diagnosis model, and use the model to perform real-time turn-to-turn short-circuit detection on the iron core reactor;

[0121] Step 3: Construct a turn-to-turn short-circuit fault diagnosis model and use the model to perform real-time turn-to-turn short-circuit detection on the iron core reactor. The specific process is as follows:

[0122] Collect historical operating status parameters, generate a collection cycle, and divide the collection cycle S into multiple collection time periods {s1, s2, …, sn}, that is, S = {s1, s2, …, sn};

[0123] Preprocessing the historical operating state parameters, extracting turn-to-turn short-circuit feature data from the historical operating state parameters in multiple collection periods, and fusing the extracted turn-to-turn short-circuit feature data to generate a turn-to-turn short-circuit comprehensive feature matrix DF. After standardizing the feature matrix DF, multiple groups of turn-to-turn short-circuit standardized matrices DBJ are obtained.

[0124] The turn-to-turn short-circuit standardized matrix DBJ is used as the input of the machine learning model, and the turn-to-turn short-circuit detection evaluation value corresponding to each group of turn-to-turn short-circuit standardized matrix DBJ is used as the output of the machine learning model. The turn-to-turn short-circuit detection evaluation value is used as the prediction target, and the minimum sum of the prediction errors of all training data is used as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence and the training is stopped to obtain the turn-to-turn short-circuit fault diagnosis model.

[0125] The expression formula of the turn-to-turn short-circuit fault diagnosis model is as follows:

[0126] DJP = γ1·X(f i )+γ2·∆Z j +γ3·|Y(f k )|+γ4·f zdfl +γ5·∆T m +ε;

[0127] Where DJP is the turn-to-turn short-circuit detection evaluation value, γ1, γ2, γ3, γ4 and γ5 are all regression coefficients, and ε represents the error term;

[0128] Obtain the real-time operating status parameters, convert them into the corresponding feature matrix and input them into the turn-to-turn short-circuit fault diagnosis model, and obtain the real-time turn-to-turn short-circuit detection evaluation value DJP through the turn-to-turn short-circuit fault diagnosis model 实时 , the real-time turn-to-turn short-circuit detection evaluation value DJP 实时 Compare with the preset turn-to-turn short circuit detection evaluation value threshold:

[0129] If DJP 实时 If it is less than the preset turn-to-turn short-circuit detection evaluation value threshold, it indicates that the iron core reactor operates well and no turn-to-turn short-circuit fault is detected;

[0130] If DJP 实时If it is greater than or equal to the preset turn-to-turn short-circuit detection evaluation value threshold, it indicates that the iron core reactor is operating abnormally and a turn-to-turn short-circuit fault has occurred;

[0131] The turn-to-turn short-circuit fault diagnosis model uses machine learning technology to efficiently process a large amount of historical data and achieve real-time detection, which significantly improves the efficiency and accuracy of fault diagnosis. Through feature extraction and standardization processing, it can more accurately capture the characteristics of turn-to-turn short circuits, thereby improving prediction accuracy.

[0132] Step 4: Used to monitor the turn-to-turn short-circuit risk parameters of the iron core reactor in real time. The turn-to-turn short-circuit risk parameters include operating current, DC resistance, inductance, and key part temperature, and monitor and evaluate the turn-to-turn short-circuit risk of the iron core reactor;

[0133] Step 4 The specific process of monitoring and evaluating the risk of inter-turn short circuit of iron core reactor is as follows:

[0134] Obtain historical turn-to-turn short-circuit risk parameters, including operating current, DC resistance, inductance, and key part temperature, generate a monitoring cycle, and divide the monitoring cycle T into multiple monitoring periods {t1, t2, …, tn}, that is, T={t1, t2, …, tn};

[0135] Obtain the operating current change rate of the iron core reactor in multiple monitoring time periods, where the operating current change rate represents the ratio between the operating current change amount and the length of the corresponding time period, and calculate the arithmetic mean of the multiple operating current change rates obtained, and record the arithmetic mean of the multiple operating current change rates as the average operating current change rate PDL;

[0136] Obtain the DC resistance change rate of the iron core reactor in multiple monitoring time periods, where the DC resistance change rate represents the ratio between the DC resistance change amount and the length of the corresponding time period, and calculate the arithmetic average of the multiple DC resistance change rates obtained, and record the arithmetic average of the multiple DC resistance change rates as the average DC resistance change rate PDZ;

[0137] Obtain the inductance change rate of the iron core reactor in multiple monitoring time periods, where the inductance change rate represents the ratio between the inductance change amount and the length of the corresponding time period, and calculate the arithmetic average of the multiple inductance change rates obtained, and record the arithmetic average of the multiple inductance change rates as the average inductance change rate PDG;

[0138] The temperature change rate of the key parts of the core reactor in multiple monitoring periods is obtained. The temperature change rate of the key parts represents the ratio between the temperature change of the key parts and the length of the corresponding time period. The arithmetic mean of the multiple key parts temperature change rates obtained is calculated, and the arithmetic mean of the multiple key parts temperature change rates is recorded as the average key parts temperature change rate PBW;

[0139] The average operating current change rate PDL, the average DC resistance change rate PDZ, the average inductance change rate PDG and the average key part temperature change rate PBW are obtained, and the average operating current change rate PDL, the average DC resistance change rate PDZ, the average inductance change rate PDG and the average key part temperature change rate PBW are combined to construct a short circuit risk feature matrix DFJ;

[0140] The constructed short-circuit risk feature matrix DFJ is used as the input of the logistic regression model, and the label vector u is used as the output of the logistic regression model. The label vector u indicates whether there is a risk of inter-turn short circuit in the iron core reactor. The output of the label vector u is 0 or 1, 0 indicates that there is no risk of inter-turn short circuit in the iron core reactor, and 1 indicates that there is a risk of inter-turn short circuit in the iron core reactor. The label vector u is used as the prediction target, and the sum of the prediction errors of all training data is minimized as the training target. The logistic regression model is trained until the sum of the prediction errors reaches convergence and the training is stopped. The logistic regression model that predicts the label vector u is obtained.

[0141] The expression formula of the logistic regression model is:

[0142] ;

[0143] Where P(u=1) represents the probability value of the iron core reactor inter-turn short circuit, v0 represents the intercept term, which represents the baseline probability of the iron core reactor inter-turn short circuit, and v1, v2, v3 and v4 are all regression coefficients;

[0144] The real-time turn-to-turn short-circuit risk parameters are obtained and processed to construct the real-time short-circuit risk feature matrix DFJ. The probability value P of the turn-to-turn short-circuit of the iron core reactor is calculated through the trained logistic regression model. 实时 ;

[0145] By comprehensively evaluating the operating current, DC resistance, inductance and temperature parameters of key parts, the risk of inter-turn short circuit of iron core reactors can be comprehensively evaluated. By dynamically calculating the change rate of each parameter, the change of equipment status can be captured in real time, which significantly improves the sensitivity and accuracy of monitoring. The logistic regression model trained based on historical data can accurately predict the probability of inter-turn short circuit, realize automated risk monitoring, reduce manual intervention, and reduce the rate of misjudgment.

[0146] Embodiment 2: The technical solution of the embodiment of the present invention is different from that of embodiment 1 in that:

[0147] like Figure 1 and Figure 3 As shown, step five: for comprehensively evaluating and analyzing the operation stability of the iron core reactor according to the short circuit detection result and the short circuit risk assessment result, and taking corresponding measures to deal with it according to the analysis result;

[0148] Get the real-time turn-to-turn short circuit detection evaluation value DJP 实时 The probability value P of the turn-to-turn short circuit of the iron core reactor 实时 , the running stability evaluation coefficient WDP is calculated by the following formula:

[0149] ;

[0150] Where d1 and d2 are both preset proportional factor coefficients, d2>d1>0, and the operation stability evaluation coefficient WDP is compared with the preset operation stability evaluation coefficient threshold:

[0151] If the operation stability evaluation coefficient WDP is less than the preset operation stability evaluation coefficient threshold, it indicates that the operation stability of the iron core reactor is good;

[0152] If the operation stability evaluation coefficient WDP is greater than or equal to the preset operation stability evaluation coefficient threshold, it indicates that the operation stability of the iron core reactor is poor;

[0153] When it is detected that the iron core reactor is running stably, take corresponding measures to ensure that the iron core reactor maintains stable operation in the current state. The specific processing measures are as follows:

[0154] Maintain real-time data collection and provide timely warnings based on trend analysis;

[0155] Strictly implement maintenance plans and replace aging parts predictably;

[0156] Ensure that the equipment operates within a safe range and can be flexibly adjusted according to demand;

[0157] When it is detected that the operation stability of the iron core reactor is poor, take corresponding measures to restore the iron core reactor to stable operation. The specific treatment measures are as follows:

[0158] Quickly locate and fix specific problems to ensure rapid recovery of normal operations;

[0159] Increase monitoring frequency and use additional testing methods to ensure a comprehensive understanding of equipment status;

[0160] Flexibly adjust operating parameters and load distribution to reduce pressure on reactors;

[0161] Regularly evaluate maintenance strategies and plan technology upgrades to improve system reliability;

[0162] By combining the real-time turn-to-turn short-circuit detection evaluation value and the probability value of turn-to-turn short-circuit in the iron-core reactor, the operation stability evaluation coefficient is calculated to fully reflect the operating status of the iron-core reactor. Based on the preset threshold, the equipment stability is quickly judged to provide a basis for decision-making. When the stability is good, real-time monitoring, trend analysis and preventive maintenance are used to ensure the continuous and stable operation of the equipment. When the stability is poor, rapid repairs, increased monitoring frequency and adjustment of operating parameters are used to quickly restore the stability of the equipment.

[0163] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and improved concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for detecting inter-turn short circuit of an iron core reactor, characterized in that: The following steps are involved: Step 1: collect the operating status parameters of the iron core reactor in real time, and perform preprocessing operations on the collected operating status parameters; Step 2: extracting turn-to-turn short-circuit feature data from the preprocessed operating state parameters, and fusing the extracted turn-to-turn short-circuit feature data to generate a turn-to-turn short-circuit comprehensive feature matrix; The specific process of extracting the turn-to-turn short-circuit feature data and fusing the extracted turn-to-turn short-circuit feature data is as follows: Obtaining pre-processed operating state parameters, the operating state parameters include current signals, voltage signals, vibration signals and key part temperatures, and extracting turn-to-turn short-circuit characteristic data from the pre-processed operating state parameters, the turn-to-turn short-circuit characteristic data include harmonic components, impedance change rate, vibration spectrum, peak frequency and temperature rise rate; The steps to extract harmonic components are as follows: The current signal is sampled to obtain a discrete signal I[n]. The discrete signal is subjected to a fast Fourier transform and the frequency domain signal I(f) is obtained by the following formula: ； Where I[n] is a discrete current signal, which indicates the current value at time point n; I(f) is a frequency domain current signal, which indicates the current amplitude at frequency f; f is the frequency, which indicates the position of the signal in the frequency domain; N is the number of sampling points, which indicates the total number of times the signal is sampled in the time domain; j is an imaginary unit, which satisfies j 2 =-1; Extract a specific frequency component from the frequency domain signal I(f) and obtain the harmonic component X(f) using the following formula: X(f)=I(f)·H(f), where H(f) is the harmonic extraction function, usually a bandpass filter or a function that selects a specific frequency, and X(f) is the harmonic component, which represents the harmonic amplitude at frequency f; The steps to extract the vibration spectrum are as follows: The vibration signal is sampled to obtain a discrete signal Y[n]. The discrete signal is subjected to a fast Fourier transform and the frequency domain signal Y(f) is obtained by the following formula: ； Where Y[n] is a discrete vibration signal, which indicates the vibration value at time point n, and Y(f) is a frequency domain vibration signal, which indicates the vibration amplitude at frequency f; The vibration spectrum is the amplitude spectrum of the frequency domain vibration signal Y(f), which is obtained by calculating the modulus of Y(f): ； Where Re(Y(f)) is the real part of the frequency domain vibration signal, Im(Y(f)) is the imaginary part of the frequency domain vibration signal, |Y(f)| is the amplitude of the frequency domain vibration signal, indicating the vibration intensity at frequency f; The steps to extract the peak frequency are as follows: Analyze the vibration spectrum |Y(f)| and obtain the frequency f with the maximum amplitude through the following formula: zdf : , where f zdf is the peak frequency; The extracted harmonic component X(f), impedance change rate ∆Z(t), vibration spectrum |Y(f)|, peak frequency f zdf and the temperature rise rate ∆T[n] are taken as eigenvectors respectively, and these eigenvectors are combined to generate a turn-to-turn short-circuit comprehensive characteristic matrix DF, which is expressed by the following expression: ； Where X(f i ) is the harmonic component of the i-th frequency, ∆Z j is the jth impedance change rate, |Y(f k )| is the vibration spectrum of the kth frequency, f zdfl is the lth peak frequency, ∆T m is the mth temperature rise rate; In order to eliminate the dimensional differences between different features, the turn-to-turn short-circuit comprehensive feature matrix DF is standardized to obtain the turn-to-turn short-circuit standardized matrix DBJ; Step 3: Collect historical operating status parameters, build a turn-to-turn short-circuit fault diagnosis model, and use the model to perform real-time turn-to-turn short-circuit detection on the iron core reactor; The specific process of constructing the turn-to-turn short-circuit fault diagnosis model is as follows: Collect historical operating status parameters, generate a collection cycle, and divide the collection cycle S into multiple collection time periods {s1, s2, …, sn}, that is, S = {s1, s2, …, sn}; Preprocessing the historical operating state parameters, extracting turn-to-turn short-circuit feature data from the historical operating state parameters in multiple collection periods, and fusing the extracted turn-to-turn short-circuit feature data to generate a turn-to-turn short-circuit comprehensive feature matrix DF. After standardizing the feature matrix DF, multiple groups of turn-to-turn short-circuit standardized matrices DBJ are obtained. The turn-to-turn short-circuit standardized matrix DBJ is used as the input of the machine learning model, and the turn-to-turn short-circuit detection evaluation value corresponding to each group of turn-to-turn short-circuit standardized matrix DBJ is used as the output of the machine learning model. The turn-to-turn short-circuit detection evaluation value is used as the prediction target, and the minimum sum of the prediction errors of all training data is used as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence and the training is stopped to obtain the turn-to-turn short-circuit fault diagnosis model. The expression formula of the turn-to-turn short-circuit fault diagnosis model is as follows: DJP=γ1·X(f i )+γ2·∆Z j +γ3·∣Y(f k )∣+γ4·f zdfl +γ5·∆T m +e; Where DJP is the turn-to-turn short-circuit detection evaluation value, γ1, γ2, γ3, γ4 and γ5 are all regression coefficients, and ε represents the error term; Step 4: Used to monitor the turn-to-turn short-circuit risk parameters of the iron core reactor in real time, and monitor and evaluate the turn-to-turn short-circuit risk of the iron core reactor; Step 5: It is used to conduct a comprehensive evaluation and analysis on the operation stability of the iron core reactor according to the short-circuit detection results and the short-circuit risk assessment results, and take corresponding measures according to the analysis results.

2. The method for detecting turn-to-turn short circuit of an iron core reactor according to claim 1, characterized in that: The specific process of extracting the impedance change rate and temperature rise rate in step 2 is as follows: The steps to extract the impedance change rate are as follows: The voltage signal V(t) and the current signal I(t) are sampled, and the impedance Z(t) is obtained by the following formula: , where V(t) is the voltage signal, indicating the voltage value at time point t, I(t) is the current signal, indicating the current value at time point t, and Z(t) is the impedance, indicating the impedance value at time point t; Analyze the impedance change trend over time and obtain the impedance change rate ∆Z(t) through the following formula: ； Where Z(t) is the impedance at the current moment, Z(t-1) is the impedance at the previous moment, and ∆t is the time interval; The steps to extract the temperature rise rate are as follows: The temperature signal is sampled to obtain a discrete signal T[n], and the temperature rise rate ∆T[n] is obtained by the following formula: ； Where T[n] is the temperature at the current moment, T[n-1] is the temperature at the previous moment, and ∆t is the time interval.

3. The method for detecting turn-to-turn short circuit of an iron core reactor according to claim 1, characterized in that: The specific process of performing real-time turn-to-turn short-circuit detection on the iron core reactor through the model in step 3 is as follows: Obtain the real-time operating status parameters, convert them into the corresponding feature matrix and input them into the turn-to-turn short-circuit fault diagnosis model. The real-time turn-to-turn short-circuit detection evaluation value DJP is obtained through the turn-to-turn short-circuit fault diagnosis model. 实时 , the real-time turn-to-turn short-circuit detection evaluation value DJP 实时 Compare with the preset turn-to-turn short circuit detection evaluation value threshold: If DJP 实时 If it is less than the preset turn-to-turn short-circuit detection evaluation value threshold, it indicates that the iron core reactor operates well and no turn-to-turn short-circuit fault is detected; If DJP 实时 If it is greater than or equal to the preset turn-to-turn short-circuit detection evaluation value threshold, it indicates that the iron core reactor is operating abnormally and a turn-to-turn short-circuit fault has occurred.

4. The method for detecting inter-turn short circuit of an iron core reactor according to claim 1, characterized in that: The specific process of monitoring and evaluating the risk of inter-turn short circuit of the iron core reactor in step 4 is as follows: Obtain historical turn-to-turn short-circuit risk parameters, including operating current, DC resistance, inductance, and key part temperature, generate a monitoring cycle, and divide the monitoring cycle T into multiple monitoring periods {t1, t2, …, tn}, that is, T={t1, t2, …, tn}; Obtain the operating current change rate of the iron core reactor in multiple monitoring time periods, where the operating current change rate represents the ratio between the operating current change amount and the length of the corresponding time period, and calculate the arithmetic mean of the multiple operating current change rates obtained, and record the arithmetic mean of the multiple operating current change rates as the average operating current change rate PDL; The DC resistance change rate of the iron core reactor in multiple monitoring time periods is obtained. The DC resistance change rate represents the ratio between the DC resistance change and the length of the corresponding time period. The arithmetic mean of the multiple DC resistance change rates obtained is calculated and recorded as the average DC resistance change rate PDZ.

5. The method for detecting inter-turn short circuit of an iron core reactor according to claim 4, characterized in that: Obtain the inductance change rate of the iron core reactor in multiple monitoring time periods, where the inductance change rate represents the ratio between the inductance change amount and the length of the corresponding time period, and calculate the arithmetic average of the multiple inductance change rates obtained, and record the arithmetic average of the multiple inductance change rates as the average inductance change rate PDG; The temperature change rate of the key parts of the iron core reactor in multiple monitoring periods is obtained. The temperature change rate of the key parts represents the ratio between the temperature change of the key parts and the length of the corresponding time period. The arithmetic mean of the multiple key part temperature change rates obtained is calculated, and the arithmetic mean of the multiple key part temperature change rates is recorded as the average key part temperature change rate PBW.

6. The method for detecting inter-turn short circuit of an iron core reactor according to claim 5, characterized in that: The average operating current change rate PDL, the average DC resistance change rate PDZ, the average inductance change rate PDG and the average key part temperature change rate PBW are obtained, and the average operating current change rate PDL, the average DC resistance change rate PDZ, the average inductance change rate PDG and the average key part temperature change rate PBW are combined to construct a short circuit risk feature matrix DFJ; The constructed short-circuit risk feature matrix DFJ is used as the input of the logistic regression model, and the label vector u is used as the output of the logistic regression model. The label vector u indicates whether there is a risk of inter-turn short circuit in the iron core reactor. The output of the label vector u is 0 or 1, 0 indicates that there is no risk of inter-turn short circuit in the iron core reactor, and 1 indicates that there is a risk of inter-turn short circuit in the iron core reactor. The label vector u is used as the prediction target, and the sum of the prediction errors of all training data is minimized as the training target. The logistic regression model is trained until the sum of the prediction errors reaches convergence and the training is stopped. The logistic regression model that predicts the label vector u is obtained. The expression formula of the logistic regression model is: ； Where P(u=1) represents the probability value of the iron core reactor inter-turn short circuit, v0 represents the intercept term, which represents the baseline probability of the iron core reactor inter-turn short circuit, and v1, v2, v3 and v4 are all regression coefficients; The real-time turn-to-turn short-circuit risk parameters are obtained and processed to construct the real-time short-circuit risk feature matrix DFJ. The probability value P of the turn-to-turn short-circuit of the iron core reactor is calculated through the trained logistic regression model. 实时 .

7. The method for detecting inter-turn short circuit of an iron core reactor according to claim 1, characterized in that: The specific process of the step 5 for comprehensively evaluating and analyzing the operating stability of the iron core reactor is as follows: Get the real-time turn-to-turn short circuit detection evaluation value DJP 实时 The probability value P of the turn-to-turn short circuit of the iron core reactor 实时 , the running stability evaluation coefficient WDP is calculated by the following formula: ； Where d1 and d2 are both preset proportional factor coefficients, d2>d1>0, and the operation stability evaluation coefficient WDP is compared with the preset operation stability evaluation coefficient threshold: If the operation stability evaluation coefficient WDP is less than the preset operation stability evaluation coefficient threshold, it indicates that the operation stability of the iron core reactor is good; If the operation stability evaluation coefficient WDP is greater than or equal to the preset operation stability evaluation coefficient threshold, it indicates that the operation stability of the iron core reactor is poor; When it is detected that the iron core reactor is running stably, corresponding measures are taken to ensure that the iron core reactor maintains stable operation in the current state; When it is detected that the operation stability of the iron core reactor is poor, corresponding measures are taken to process it so that the iron core reactor can resume stable operation.

Images