Noise evaluation method based on high-voltage oscillatory wave transformer turn-to-turn insulation detection method

Abstract

The invention discloses a noise evaluation method based on a high-voltage oscillatory wave transformer turn-to-turn insulation detection method, and the method comprises the steps: obtaining the oscillatory wave data of a transformer winding through a high-voltage oscillatory wave detection system, and firstly building a state space model of the process and measurement of a discrete time nonlinear system with non-zero average noise statistics; then, a maps estimator is introduced to estimate noise parameters; combining the measurement model with the state space model, and introducing a gradient algorithm; and finally, calculating the parameters of the noise estimator to provide a reference for denoising the high-voltage oscillation wave detection data, and improving the precision of the detection state.

Description

Technical Field

[0001] The present application relates to the technical field of fault diagnosis of electric power equipment, and in particular to a noise evaluation method based on a high-voltage oscillating wave transformer turn-to-turn insulation detection method. Background Art

[0002] In the field of maintenance and inspection of power equipment, the transformer is a core component in the power grid, and its stability and reliability are of vital importance. In order to timely discover and deal with potential faults of the transformer, the high-voltage oscillation wave detection method is widely used in non-destructive inspection of the transformer. This method injects a high-voltage oscillation wave with specific parameters into the transformer, observes and analyzes the transformer's response to this excitation, and thus determines its internal state and whether there is a fault.

[0003] However, in practical applications, the signal obtained during the high-voltage oscillation wave detection process often contains not only the response information of the transformer itself, but also various noise components. These noises may come from the electronic noise of the detection equipment, environmental interference, the nonlinear characteristics of the transformer itself, and the attenuation and distortion during signal transmission. The presence of noise will seriously affect the clarity and accuracy of the detection signal, thereby reducing the sensitivity and reliability of fault detection.

[0004] In order to accurately evaluate and utilize the signals obtained by the high-voltage oscillation wave detection method, it is particularly important to effectively analyze and quantify the noise contained therein. However, at present, the evaluation method for the noise in the high-voltage oscillation wave detection signal is still imperfect, and there is a lack of a systematic and accurate noise parameter evaluation algorithm. This makes it difficult to accurately distinguish the true response of the transformer from the noise interference in actual operation, which limits the application effect of the high-voltage oscillation wave detection method in transformer fault detection.

[0005] In view of this, the present invention proposes a noise evaluation method based on the high-voltage oscillation wave transformer inter-turn insulation detection method, aiming to develop an efficient and accurate noise parameter evaluation algorithm. This method realizes the accurate extraction and quantitative evaluation of noise components by deeply analyzing the characteristics of the high-voltage oscillation wave detection signal and combining advanced signal processing technology. Through this algorithm, the true response of the transformer under the action of the high-voltage oscillation wave can be more clearly identified, the accuracy of fault detection can be improved, and a more reliable basis can be provided for the transformer's state monitoring and fault diagnosis. Summary of the invention

[0006] The present application provides a noise evaluation method based on the high-voltage oscillation wave transformer turn-to-turn insulation detection method. According to the proposed method, the accuracy of transformer fault detection by the high-voltage oscillation wave method can be improved. The high-voltage oscillation wave transformer detection platform is mainly composed of a high-voltage oscillation wave generator 1, a power transformer 2, an oscillation wave signal receiving device 3, and a signal processing computer 4.

[0007] The present application provides a noise evaluation method based on a high-voltage oscillating wave transformer turn-to-turn insulation detection method, the method comprising:

[0008] 1) Modeling the oscillation wave signal obtained in the high-voltage oscillation wave detection method, where the process and measurement state space model of the discrete-time nonlinear system with non-zero mean noise statistics can be expressed as

[0009] (1)

[0010] (2)

[0011] in is the state vector, is the measurement vector, and are the nonlinear state and measurement functions, respectively; and are the process and measurement means, respectively; and ,in and are process and measurement noise, respectively, generated by two independent zero-mean Gaussian distributions with covariances of and ;

[0012] 2) In Bayesian statistics, MAP (Maximum A Posteriori) estimation can use empirical data to obtain point estimates of unobserved quantities; to use the MAP estimator that maximizes the posterior density, from which the noise parameters are estimated; to maximize the given measurement of ( ,Right now is the process average , process noise covariance , measured average value , measurement noise covariance , composed of a matrix), that is, maximize ; The MAP estimate can be expressed based on Bayes' theorem as follows

[0013] (3)

[0014] Argmax is the function that takes the maximum value point. Represents any , P(Date|M) represents the probability of occurrence of data Date in the M model;

[0015] 3) Expand the initial state By performing logarithmic operations, the process and measurement models in equations (1) and (2) can be combined as follows

[0016] (4)

[0017] Where C is a logarithmic constant;

[0018] 4) Gradient descent is a first-order optimization algorithm. To find the local minimum of a function using gradient descent, iteratively search for points of a specified step length in the opposite direction of the gradient (or approximate gradient) corresponding to the current point on the function; the parameters are obtained using the maximized gradient descent method. , as shown in the following table

[0019] (5)

[0020] According to this equation, the noise estimator is expressed as

[0021] (6)

[0022] (7)

[0023] (8)

[0024] (9)

[0025] (Note: In the above formula represents the model order, To find the lower bound of the sum, and represent the estimated process and measured average, respectively);

[0026] 5) Based on the calculated noise estimator, a filter algorithm corresponding to the oscillation wave test method is designed to achieve noise reduction processing for the high-voltage oscillation wave signal and improve the accuracy of the oscillation wave detection method.

[0027] The beneficial effect of the present invention is that after obtaining the oscillation wave data of the transformer winding through the high-voltage oscillation wave detection system, a process and measurement state space model of a discrete-time nonlinear system with non-zero average noise statistics is first established, and then a maps estimator is introduced to estimate the noise parameters; then the measurement model is combined with the state space model, and a gradient algorithm is introduced; finally, the parameters of the noise estimator are calculated to provide a reference for denoising the high-voltage oscillation wave detection data, thereby improving the accuracy of the detection state. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is a flow chart of the method of the present invention;

[0029] Figure 2 The invention relates to a high-voltage oscillation wave detection system of the method of the invention. DETAILED DESCRIPTION

[0030] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0031] like Figure 1 As shown, a noise evaluation method based on a high-voltage oscillating wave transformer turn-to-turn insulation detection method comprises:

[0032] 1) Build as Figure 2 The high-voltage oscillation wave transformer detection platform shown in the figure mainly consists of a high-voltage oscillation wave generator 1, a power transformer 2, an oscillation wave signal receiving device 3, and a signal processing computer 4. When the system is working, the high-voltage oscillation wave generator 1 generates an oscillation wave excitation signal, which is injected into the transformer winding through the bushing of the power transformer 2, and then output to the oscillation wave signal receiving device 3 through the bushing after the internal winding of the transformer oscillates; the oscillation wave signal receiving device transmits the input oscillation wave signal to the signal processing computer 4 for algorithm model calculation.

[0033] 2) Modeling the oscillation wave signal obtained in the high-voltage oscillation wave detection method, where the process and measurement state space model of the discrete-time nonlinear system with non-zero mean noise statistics can be expressed as

[0034] (1)

[0035] (2)

[0036] in is the state vector, is the measurement vector, and are the nonlinear state and measurement functions, respectively; and are the process and measurement means, respectively; and ,in and are process and measurement noise, respectively, generated by two independent zero-mean Gaussian distributions with covariances of and ;

[0037] 3) In Bayesian statistics, MAP (Maximum A Posteriori) estimation can use empirical data to obtain point estimates of unobserved quantities; to use the MAP estimator that maximizes the posterior density, from which the noise parameters are evaluated; it is necessary to maximize the given measurement of ( ,Right now is the process average , process noise covariance , measured average value , measurement noise covariance , composed of a matrix), that is, maximize ; The MAP estimate can be expressed based on Bayes' theorem as follows

[0038] (3)

[0039] Argmax is the function that takes the maximum value point. Represents any , P(Date|M) represents the probability of occurrence of data Date in the M model;

[0040] 4) Expand the initial state By performing logarithmic operations, the process and measurement models in equations (1) and (2) can be combined as follows

[0041] (4)

[0042] Where C is a logarithmic constant;

[0043] 5) Gradient descent is a first-order optimization algorithm. To find the local minimum of a function using the gradient descent method, iteratively search for points of a specified step length in the opposite direction of the gradient (or approximate gradient) corresponding to the current point on the function; the parameters are obtained using the maximized gradient descent method. , as shown in the following table

[0044] (5)

[0045] According to this equation, the noise estimator is expressed as

[0046] (6)

[0047] (7)

[0048] (8)

[0049] (9)

[0050] (Note: In the above formula represents the model order, To find the lower bound of the sum, and represent the estimated process and measured average, respectively);

[0051] 6) Based on the calculated noise estimator, a filter algorithm corresponding to the oscillation wave test method is designed to achieve noise reduction processing for the high-voltage oscillation wave signal and improve the accuracy of the oscillation wave detection method.

Claims

1. A noise evaluation method based on a high voltage oscillating wave transformer turn-to-turn insulation detection method, characterized in that: 1) Modeling the oscillation wave signal obtained in the high-voltage oscillation wave detection method, where the process and measurement state space model of the discrete-time nonlinear system with non-zero mean noise statistics can be expressed as (1) (2) in is the state vector, is the measurement vector, and are the nonlinear state and measurement functions, respectively; and are the process and measurement means, respectively; and ,in and are process and measurement noise, respectively, generated by two independent zero-mean Gaussian distributions with covariances of and ; 2) In Bayesian statistics, MAP (Maximum A Posteriori) estimation can use empirical data to obtain point estimates of unobserved quantities; to use the MAP estimator that maximizes the posterior density, from which the noise parameters are estimated; to maximize the given measurement of ( ,Right now is the process average , process noise covariance , measured average value , measurement noise covariance , composed of a matrix), that is, maximize ; The MAP estimate can be expressed based on Bayes' theorem as follows (3) Argmax is the function that takes the maximum value point. Represents any , P(Date|M) represents the probability of occurrence of data Date in the M model; 3) Expand the initial state By performing logarithmic operations, the process and measurement models in equations (1) and (2) can be combined as follows (4) Where C is a logarithmic constant; 4) Gradient descent is a first-order optimization algorithm. To find the local minimum of a function using gradient descent, iteratively search for points of a specified step length in the opposite direction of the gradient (or approximate gradient) corresponding to the current point on the function; the parameters are obtained using the maximized gradient descent method. , as shown in the following table (5) According to this equation, the noise estimator is expressed as (6) (7) (8) (9) (Note: In the above formula represents the model order, To find the lower bound of the sum, and represent the estimated process and measured average, respectively); 5) Based on the calculated noise estimator, a filter algorithm corresponding to the oscillation wave test method is designed to achieve noise reduction processing for the high-voltage oscillation wave signal and improve the accuracy of the oscillation wave detection method.

Images