66kV alternating current XLPE cable damp degree evaluation method

Abstract

A 66kV AC XLPE cable moisture degree evaluation method belongs to the technical field of electrical engineering, and comprises the following steps: step 1, establishing a cable head end reflection coefficient spectrum containing noise signals according to a cable distribution parameter model, and performing data processing through a Hankle matrix to obtain a cable positioning component and a corresponding amplitude; and step 2, punishing the number of parameters of the cable distribution model with low singular value sudden change value through an improved Bayesian information criterion IBIC to prevent overfitting, and adopting a likelihood function to ensure the goodness of fitting of the model. In order to solve the problem that amplitude information of inverse Fourier transform is difficult to reflect the moisture degree of a power cable at high precision, a frequency domain reflection method is taken as a basis, and a TLS-ESPRIT algorithm is introduced to evaluate the moisture degree of a 66kV cable.

Description

Technical Field

[0001] The invention belongs to the technical field of electrical engineering, and in particular relates to a method for evaluating the moisture degree of a 66kV AC XLPE cable. Background Art

[0002] Whether the power cable is laid directly in the ground or in a pipeline, it is difficult to avoid the intrusion of water or moisture and the water tree effect it induces. The growth process of water trees under the induction of electric fields will accelerate the deterioration and aging of insulation. If the degree of moisture in the cable cannot be accurately assessed and handled in time, it may have a serious impact on the insulation performance of the cable, and even cause insulation breakdown, leading to power supply interruption. Therefore, the local moisture assessment of the power cable body is of great significance to the reliability of the power system.

[0003] At present, the frequency domain reflection method has attracted a lot of attention and research from scholars due to its superior positioning performance. The linear sweep signal injected by it greatly compensates for the deficiency of the time domain reflection method requiring high test frequency, and on this basis, the reflection coefficient spectrum method and broadband impedance spectrum method have been developed. Some technicians used broadband impedance spectrum method to study the location of moisture defects and the range of moisture areas of 10kV power cables, but did not obtain the quantitative relationship between signal amplitude and moisture degree. Some scholars used the integral transformation method to process broadband impedance data to improve the sensitivity of positioning and identification, but the original data of intact cables were required as a comparison, which made practical application difficult. In addition, various window functions such as Gaussian window and Hanning window were introduced into frequency domain reflection analysis to improve the positioning accuracy of cable defects, but the setting of parameters required human intervention and could not be solved adaptively. Some engineers also obtained the positioning spectrum of 10kV three-core cable fault defects by applying the total least squares-rotation vector invariant technology, the improved BIC criterion and the DBSCAN clustering algorithm, but did not involve single-core cables with higher voltage levels and more complex structures. Some literature has proposed a cable fault detection method based on the correlation coefficient of the reflection coefficient spectrum, which determines the fault type by analyzing the impedance state, but does not involve relevant research on the moisture problem. Some literature uses the TLS-ESPRIT algorithm to identify the parameters of the input signal and obtain the real frequency components in the signal, but it has not been applied to the cable. Therefore, the existing technology urgently needs a new technical solution to solve the above problems. Summary of the invention

[0004] The technical problem to be solved by the present invention is: to provide a method for evaluating the moisture degree of a 66kV AC XLPE cable. Aiming at the problem that the amplitude information of the inverse Fourier transform is difficult to reflect the moisture degree of the power cable with high precision, a TLS-ESPRIT algorithm is introduced based on the frequency domain reflection method to evaluate the moisture degree of the 66kV cable.

[0005] A method for evaluating the moisture degree of a 66kV AC XLPE cable comprises the following steps:

[0006] Step 1: According to the cable distributed parameter model, establish the reflection coefficient spectrum of the cable head end containing the noise signal, perform data processing through the Hankle matrix, and obtain the cable positioning component and the corresponding amplitude;

[0007] Step 2: The number of cable distribution model parameters with low singular value mutation values ​​is penalized by the improved Bayesian information criterion IBIC to prevent overfitting, and the likelihood function is used to ensure the goodness of model fitting.

[0008] Step 1 The Hankle matrix processes the head-end reflection coefficient spectrum data x(k) (k=1, 2, ..., N), and the obtained M×L order Hankle matrix H is

[0009]

[0010] Where: N is the number of sampling points; M is the number of rows in the matrix; L is the number of columns in the matrix, where L+M-1=N; perform singular value decomposition on the matrix H to obtain the eigenvalue γ i Construct the N×m-order Vandermonde matrix G as

[0011]

[0012] The column vector X is constructed using the collected head-end reflection coefficient spectrum data:

[0013] X=[x(1),x(2),…,x(N)] T

[0014] According to the least squares method, the signal amplitude K corresponding to each frequency component can be obtained i ,have

[0015] K=[K1,K2,…,K m ]

[0016] K=(G T G) -1 G T X;

[0017] The improved Bayesian Information Criterion IBIC described in step 2 is:

[0018]

[0019] Where: N is the number of sampling points; M is the number of rows in the matrix; L is the number of columns in the matrix, where L+M-1=N; y is the number of main signals; λ j (j=1,2,....y) are the singular values ​​in the matrix H arranged in descending order; is the variance corresponding to each y;

[0020] Through the above design scheme, the present invention can bring the following beneficial effects: a method for evaluating the moisture degree of a 66kV AC XLPE cable, based on a cable distributed parameter model, constructs a cable head-end reflection coefficient spectrum containing a noise signal, arranges the obtained data into a Hankle matrix, obtains the number of main signals through singular value decomposition and the IBIC criterion, and then uses the least squares-rotational invariant method to obtain the defect location component and the moisture degree. DETAILED DESCRIPTION

[0022] A method for evaluating the moisture degree of 66kV AC XLPE cable is firstly used to process the head-end reflection coefficient spectrum data x(k) (k=1,2,...,N) with the help of Hankle matrix, and the obtained M×L order Hankle matrix H is

[0023]

[0024] Wherein: N is the number of sampling points; M is the number of rows of the matrix; L is the number of columns of the matrix. The present invention takes L=N / 3, where L+M-1=N.

[0025] Perform singular value decomposition on the matrix H, and we have

[0026] H=USV T

[0027] Where: U is the left feature matrix of M×M dimensions; V is the right feature matrix of L×L dimensions; S is the diagonal matrix of M×L dimensions, and its diagonal element λ n (n=1, 2, ..., L) are the singular values ​​in the matrix H arranged in descending order. The singular values ​​corresponding to the main signal (the number is m) are much larger than the noise space.

[0028] Take the first m columns of matrix V to form matrix V S , respectively V S Remove the last row and the first row of , and get the matrices V1 and V2. Perform singular value decomposition on the merged matrix V3 = [V1 V2], and we have

[0029] V3=AOB T

[0030] Where A is the left feature matrix of M×M dimensions; B is the right feature matrix of 2m×2m dimensions; O is the singular value feature matrix of M×2m ​​dimensions.

[0031] Split the matrix B into 4 m×m dimensional sub-matrices as

[0032]

[0033] Calculate the rotation matrix Ω = -B 12 B 22 -1 The eigenvalue of γ i (i=1,2,…,m), get the position d of each defect i for

[0034]

[0035] Where: f s is the sampling interval; v is the propagation velocity of electromagnetic waves in the cable.

[0036] Using the eigenvalue γ obtained above i Construct the N×m-order Vandermonde matrix G as

[0037]

[0038] The column vector X is constructed using the collected head-end reflection coefficient spectrum data:

[0039] X=[x(1),x(2),…,x(N)] T

[0040] According to the least squares method, the signal amplitude K corresponding to each frequency component can be obtained i ,have

[0041] K=[K1,K2,…,K m ]

[0042] K=(G T G) -1 G T X;

[0043] Through the TLS-ESPRIT algorithm, the main signal can be separated from the noise signal to accurately obtain the cable positioning component x i And the corresponding amplitude K i , thereby realizing the assessment of the moisture degree of the cable.

[0044] The number of singular values ​​m of the main signal needs to be set manually. When observing the matrix S, the position where the singular value size changes suddenly is the value of m. However, when the cable is less damp or there are more noise signals, the singular value mutation is not obvious, and the value of m is difficult to determine. The Bayesian Information Criterion (BIC) is an effective method for adaptively estimating the dimension of the source signal. It can determine the model complexity by penalizing the number of parameters, thereby avoiding overfitting; at the same time, the likelihood function is introduced to ensure the goodness of model fit through maximum likelihood estimation. The BIC criterion function is as follows:

[0045]

[0046]

[0047] Where: N is the number of sampling points; M is the number of rows in the matrix; L is the number of columns in the matrix, where L+M-1=N; y is the number of main signals; λ j (j=1,2,....y) are the singular values ​​in the matrix H arranged in descending order; is the variance corresponding to each y;

[0048] Since the exponential function has a rapid growth, when the number of sampling points is large, it will lead to gross errors and a huge amount of calculation. Therefore, this paper adopts the logarithmic form of the BIC criterion function, defined as the improved Bayesian Information Criterion (IBIC). The IBIC criterion function is

[0049]

[0050] Where: N is the number of sampling points; L is the number of columns of the matrix, where L+M-1=N; M is the number of rows of the matrix; y is the number of main signals; λ j (j=1,2,....y) are the singular values ​​in the matrix H arranged in descending order; is the variance corresponding to each y;

[0051] Assuming that the number of main signals is y, the IBIC value corresponding to each y is calculated by the above formula, and the value of m is the y value when IBIC(y) reaches the maximum.

[0052] The present invention can effectively study the moisture process and identify the moisture state, and can be applied to actual projects to evaluate the moisture problem of power cables in humid environments, providing more powerful guarantees for the safe operation of cables. It is of great significance for power cable status inspection, ensuring the stable operation of power systems, improving equipment asset utilization, and reducing production costs.

Claims

1. A method for evaluating the moisture level of a 66kV AC XLPE cable, characterized by: The following steps are included: Step 1: According to the cable distributed parameter model, establish the reflection coefficient spectrum of the cable head end containing the noise signal, perform data processing through the Hankle matrix, and obtain the cable positioning component and the corresponding amplitude; Step 2: The number of cable distribution model parameters with low singular value mutation values ​​is penalized by the improved Bayesian information criterion IBIC to prevent overfitting, and the likelihood function is used to ensure the goodness of model fitting.

2. A method for evaluating the moisture degree of a 66kV AC XLPE cable according to claim 1, characterized in that: Step 1 The Hankle matrix processes the head-end reflection coefficient spectrum data x(k) (k=1, 2, ..., N), and the obtained M×L order Hankle matrix H is Where: N is the number of sampling points; M is the number of rows of the matrix; L is the number of columns of the matrix, where L+M-1=N; Perform singular value decomposition on the matrix H and obtain the eigenvalue γ i Construct the N×m-order Vandermonde matrix G as The column vector X is constructed using the collected head-end reflection coefficient spectrum data: X=[x(1),x(2),…,x(N)] T According to the least squares method, the signal amplitude K corresponding to each frequency component can be obtained i ,have K=[K1,K2,…,K m ] K=(G T G) -1 G T X。 3. A method for evaluating the moisture degree of a 66kV AC XLPE cable according to claim 1, characterized in that: The improved Bayesian Information Criterion IBIC described in step 2 is: Where: N is the number of sampling points; M is the number of rows in the matrix; L is the number of columns in the matrix, where L+M-1=N; y is the number of main signals; λ j (j=1,2,....y) are the singular values ​​in the matrix H arranged in descending order; is the variance corresponding to each y.

Images