A CVT abnormal channel detection method based on an M-LOF algorithm, a medium and a terminal

Abstract

The application belongs to the technical field of mutual inductor detection, and relates to a CVT abnormal channel detection method based on an M-LOF algorithm, a medium and a terminal. (N+L)×M ; S20, differences between channels are found to form a new matrix, a channel difference matrix is extracted from the first N+step rows, and a data sequence X ij is formed; ij S30, distances between a stitching domain point and other points in the data sequence X ij , local reachable density of the stitching domain point, average value of local reachable density of all points in the k domain of the stitching domain point and a LOF score are calculated; S40, whether the stitching domain has an abnormal point is judged, if yes, a probability is calculated, an abnormal channel is output, and if not, the first step rows in the training domain are deleted, the stitching domain is spliced to the training domain, and a new training domain is formed. The application has simple process and accurate result, the M-LOF model is used to detect the CVT abnormal channel, and the accuracy and efficiency of the CVT abnormal channel detection are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of mutual inductor detection, and particularly relates to a CVT abnormal channel detection method based on an M-LOF algorithm, a medium and a terminal. BACKGROUND

[0002] Capacitive voltage transformers (CVT) are widely used in high-voltage voltage measurement occasions of 110 kV and above due to good insulation performance and low cost. In the past, the periodic verification method was generally used to evaluate the measurement performance of the CVT, but this method needs to be implemented under the power-off state of the substation, which has a greater impact on the operation of the power system. In recent years, evaluating the measurement state of the CVT and monitoring the measurement error in real time under the non-power-off condition has become an important research direction in the field of CVT operation and maintenance.

[0003] In the prior art, the PCA method is used to qualitatively judge the error channel of the capacitive voltage transformer, but the PCA method is seriously dependent on the strong correlation of the signal and the distribution of the noise. In practice, due to the influence of the aging degree of the equipment, the load, the temperature, the humidity and the like, the data signals among the three phases of the capacitive voltage transformer are in a weak correlation and a strong noise state, which seriously affects the accuracy of the error channel judgment. In addition, the traditional LOF algorithm can also be used to detect the abnormal channel of the CVT. Although the traditional LOF algorithm does not need to assume that the data conforms to a specific probability distribution and can quantify the abnormality of each data point, it cannot consider the change characteristics of the time dimension, thereby leading to the accuracy of the abnormal channel monitoring. In addition, when there are repeated data points in the traditional LOF algorithm, the average reachable distance of these points is zero, and the local reachable density becomes infinite, which brings some troubles to the calculation. Moreover, the LOF algorithm needs to calculate the distance between all data points, which causes the time complexity of the entire algorithm to be too complex, so the traditional LOF algorithm also has defects. Patent No. CN116027249A provides a kind of metering system mutual inductor fault diagnosis method based on improved ELM, including: obtaining test sample data by using data acquisition system;The mutual inductor fault data is normalized and PCA principal component analysis processing, respectively, to avoid the data attributes of a large initial domain from covering up those small data, while reducing the dimensionality of the required data, removing a large amount of redundant information contained in the data;ELM extreme learning machine classification is used as a classification algorithm to diagnose and classify data samples. This patent also uses the PCA method to detect the mutual inductor, and still has the problem of low detection accuracy.

[0004] Therefore, how to improve the accuracy of the capacitive voltage transformer abnormal channel detection is a problem that needs to be solved by the personnel in this technical field. SUMMARY

[0005] In view of the deficiencies of the prior art, the purpose of the present application is to provide a CVT abnormal channel detection method based on M-LOF algorithm to solve the problem of low accuracy of abnormal channel detection of capacitive voltage transformer in the prior art. In addition, the present application also provides a CVT abnormal channel detection medium and terminal based on M-LOF algorithm.

[0006] In order to solve the above technical problems, the present application adopts the following technical solutions:

[0007] In the first aspect, the present application provides a CVT abnormal channel detection method based on M-LOF algorithm, comprising the following steps:

[0008] S10, collecting training data and test data at the secondary side of the M-channel capacitive voltage transformer, and splicing the data to form a matrix T (N+L)×M ;

[0009] S20, difference between channels to form a new matrix Extract the first N+step rows of the difference matrix between channels, and form a data sequence X ij ;

[0010] S30, calculate the data sequence X ij The midpoint distance between points, the local reachable density of the point, the average value of the local reachable density of all points in the k domain of the point, and the LOF score;

[0011] S40, judge whether there is an abnormal point in the splicing domain, if yes, calculate the probability, output the abnormal channel, if not, delete the first step rows in the training domain, splice the splicing domain to the training domain, and form a new training domain.

[0012] Further, the specific steps of step S10 are as follows:

[0013] S101, collect training data and test data to form a training data matrix Tr N×M and a test data matrix Te L×M , wherein N is the length of the training data, M is the number of signal channels, and L is the length of the test data, considering the ratio difference and angle difference;

[0014] S102, splice the test data matrix Te L×M to the training data matrix Tr N×M , and form a new matrix T (N+L)×M .

[0015] Further, the specific steps of step S20 are as follows:

[0016] S201, difference between channels to form a matrix wherein This represents the difference between the i-th channel and the j-th channel, where i ≠ j, and has N+L rows and 1 column.

[0017] S202, the step size is represented as step, extract the matrix. The first N+step rows, where step < N / 4, are used to form the training field, and the (N+1)th to (N+step)th rows are used to form the concatenation field, thus creating the data sequence X. ij ={x P}, where P = 1, 2, ..., N+step.

[0018] Furthermore, the specific steps of step S30 are as follows:

[0019] S301. For each point P in the splicing domain, calculate the distance between point P and other points O, where O ≠ P, and sort them from nearest to farthest.

[0020] S302. For each point P in the splicing domain, calculate the local reachability density of point P;

[0021] First, calculate the reachability distances of the k nearest neighbors of point P, where k = (N + step) / 2. The expression for the k-th reachability distance from point P to point O is as follows:

[0022]

[0023] Where, d k (O) is the k-th furthest distance from point O in the sorting, and d(P, O) is the straight-line distance from point P to point O.

[0024] Then, the local reachability density of point P is calculated. The local reachability density of point P is the reciprocal of the average reachability distance of point P's k nearest neighbors, as expressed below:

[0025]

[0026] Where, N k (P) represents all points within the k-th distance of point P, including points at the k-th distance but excluding point P, |N k (P)| represents its number;

[0027] S303. Calculate the average local reachability density of all points in the k-neighborhood of point P in the splicing domain;

[0028] S304. The LOF score is obtained by comparing the result calculated in step S303 with the result calculated in step S302, as shown in the following expression:

[0029]

[0030] Furthermore, the specific steps of step S40 are as follows:

[0031] S401, Let m be the number of window moves, where m∈N * Points with abnormal LOF scores significantly greater than the threshold thr = 1.2 and whose regions are within the splicing domain are selected. If the difference matrix between all channels in the splicing domain... If no outliers are identified, then the first step row of the training domain is deleted, and the difference matrix between all channels is removed. After the N+(m-1)*step+1th row to the N+m*stepth row is concatenated into the training domain, a new training domain is formed. At the same time, the N+m*step+1th row to the N+(m+1)*stepth row becomes a new concatenation domain, forming the data sequence X. ij ={x P};

[0032] S402, The number of window moves is denoted as m, where m∈N * Points with abnormal LOF scores significantly greater than the threshold thr = 1.2 and whose regions are within the splicing domain are selected. If the difference matrix between all channels in the splicing domain... If anomalies are identified, the frequency of their occurrence within the stitching domain is counted and mapped to their respective single channels. The probability of an anomaly in a single channel within the stitching region is considered an abnormal channel; if the probability is less than a threshold, it is considered a normal channel. Simultaneously, the training domain remains unchanged, and the difference matrix between all channels is maintained. The concatenation field is a matrix From row N+m*step+1 to row N+(m+1)*step+1, form the data sequence X. ij ={x P Repeat steps S30 to S40.

[0033] In a second aspect, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0034] Thirdly, the present invention also provides an electronic terminal, comprising: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to cause the terminal to perform the method described above.

[0035] The CVT abnormal channel detection method, medium, and terminal based on the M-LOF algorithm provided by this invention have at least the following advantages compared with existing technologies:

[0036] The prior art qualitatively judges the error channel of the capacitive voltage transformer by the PCA mode, but the PCA mode seriously depends on the strong correlation of the signal and the distribution of the noise, in the actual application, the data signals among the three phases of the capacitive voltage transformer are in a weak correlation and a strong noise state due to the influences of the aging degree of the equipment, the load, the temperature, the humidity and the like, and seriously affect the accuracy of the error channel judgment; or the traditional LOF algorithm also has the disadvantage of low accuracy. The method has the advantages of simple flow and accurate result, detects the abnormal channel of the CVT through the M-LOF model, that is, the training set is continuously updated along with the data collection, and when the repeated points appear, a very small value is added to the reachable distance, so that the situation that the reachable distance is equal to zero is avoided, and the detection accuracy is improved; the data set is grouped, and the grouped data set is further divided into a training domain and a splicing domain, then the LOF value of each point in the splicing domain in each subset is calculated, so that the algorithm efficiency is improved, and the accuracy and efficiency of the whole detection method are greatly improved compared with the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the scheme of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0038] Figure 1 A flow chart of the CVT abnormal channel detection method based on the M-LOF algorithm provided by the embodiment of the present application is shown in the figure.

[0039] Figure 2 The original waveform graph of the matrix T in the CVT abnormal channel detection method based on the M-LOF algorithm provided by the embodiment of the present application is shown in the figure. (55+10)×3

[0040] Figure 3 The LOF value size diagram of the A-B phase splicing domain when the window moving number m=1 in the CVT abnormal channel detection method based on the M-LOF algorithm provided by the embodiment of the present application is shown in the figure.

[0041] Figure 4 The LOF value size diagram of the B-C phase splicing domain when the window moving number m=1 in the CVT abnormal channel detection method based on the M-LOF algorithm provided by the embodiment of the present application is shown in the figure.

[0042] Figure 5 The LOF value size diagram of the C-A phase splicing domain when the window moving number m=1 in the CVT abnormal channel detection method based on the M-LOF algorithm provided by the embodiment of the present application is shown in the figure.

[0043] ​Figure 6 A-B phase splicing domain LOF value size schematic diagram when the window moving number m=2 in the CVT abnormal channel detection method based on the M-LOF algorithm provided for the embodiment of the application is shown in FIG. 6;

[0044] Figure 7 B-C phase splicing domain LOF value size schematic diagram when the window moving number m=2 in the CVT abnormal channel detection method based on the M-LOF algorithm provided for the embodiment of the application is shown in FIG. 7;

[0045] Figure 8 C-A phase splicing domain LOF value size schematic diagram when the window moving number m=2 in the CVT abnormal channel detection method based on the M-LOF algorithm provided for the embodiment of the application is shown in FIG. 8. DETAILED DESCRIPTION

[0046] In order to facilitate the understanding of the present application, the present application will be described more fully below with reference to the accompanying drawings. The preferred embodiments of the present application are shown in the drawings. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terms used in the specification of the present application herein are only for the purpose of describing the specific embodiments and are not intended to limit the present application.

[0048] The present application provides a CVT abnormal channel detection method based on the M-LOF algorithm, which is applied to the error detection process of a capacitive voltage transformer. The CVT abnormal channel detection method based on the M-LOF algorithm comprises the following steps:

[0049] S10, collecting training data and test data at the secondary side of the M-channel capacitive voltage transformer, and forming a matrix T by splicing the data (N+L)×M ;

[0050] S20, forming a new matrix by differencing between channels extracting the first N+step rows of the difference value matrix between channels, and forming a data sequence X ij ;

[0051] S30, calculating the data sequence X ij the distance between the splicing domain points and other points, the local reachable density of the splicing domain points, the average value of the local reachable density of all points in the k domain of the splicing domain points, and the LOF score;

[0052] S40, judging whether the splicing domain has an abnormal point, if yes, calculating a probability, outputting an abnormal channel, if not, deleting the first step rows in the training domain, splicing the splicing domain to the training domain, and forming a new training domain.

[0053] The application has simple process and accurate result, detects the CVT abnormal channel through the M-LOF model, and improves the accuracy and efficiency of the CVT abnormal channel detection.

[0054] In order to enable the personnel in the technical field to better understand the application scheme, the technical scheme in the embodiment of the application will be clearly and completely described below in combination with the drawings.

[0055] The application provides a CVT abnormal channel detection method based on an M-LOF algorithm, which is applied to an error detection process of a capacitive voltage transformer. Figure 1 As shown in the figure, the CVT abnormal channel detection method based on the M-LOF algorithm comprises the following steps:

[0056] S10, collecting training data and test data at the secondary side of an M-channel capacitive voltage transformer, and splicing the data to form a matrix T (N+L)×M ;

[0057] Specifically, the specific steps of step S10 are as follows:

[0058] S101, collecting the training data and the test data to form a training data matrix Tr N×M and a test data matrix Te L×M , wherein N is the length of the training data, M is the number of signal channels, and L is the length of the test data, and the ratio difference and the angle difference are considered;

[0059] S102, splicing the test data matrix Te L×M to the training data matrix Tr N×M in time sequence, and forming a new matrix T (N+L)×M .

[0060] S20, forming a new matrix by differencing between channels ij extracting the first N+step rows of the difference value matrix between channels, and forming a data sequence X

[0061] Specifically, the specific steps of step S20 are as follows:

[0062] S201, forming a matrix by differencing between channels , wherein indicates the difference between the i-th channel and the j-th channel, i≠j, the number of rows is N+L, and the number of columns is 1;

[0063] S202, moving the step length, and extracting the matrix the first N+step rows of the front N+step column, step < N / 4, wherein the first N rows are training domains, and the N+1 row to the N+step row are splicing domains, and form a data sequence X ij P , wherein P = 1, 2,..., N+step.

[0064] S30, calculating the data sequence X ij the distance between the splicing domain point and other points, the local reachable density of the splicing domain point, the average value of the local reachable density of all points in the k-neighborhood of the splicing domain point, and the LOF score;

[0065] Specifically, the specific steps of step S30 are as follows:

[0066] S301, for each point P in the splicing domain, calculating the distance between the point P and other points O, O≠P, and sorting from near to far;

[0067] Further, the distance calculation formula is obtained according to the improved Manhattan distance formula, and the formula is as follows:

[0068]

[0069] , wherein x P , x O are the amplitudes (phase angles) of the point P and the point O respectively, and d(P, O) is the distance between the point P and the point O;

[0070] S302, for each point P in the splicing domain, calculating the local reachable density of the point P;

[0071] First, the k-nearest neighbor reachable distance of the point P is calculated, wherein k can be taken as (N+step) / 2, and the k-th reachable distance expression of the point P to the point O is as follows:

[0072]

[0073] , wherein d k (k) is the k-th distance of the point O, and d(P, O) is the straight-line distance between the point P and the point O;

[0074] Then, the local reachable density of the point P is calculated, and the local reachable density of the point P is the reciprocal of the average reachable distance of the k-nearest neighbors of the point P, and the expression is as follows:

[0075]

[0076] , wherein N k (P) is all points within the k-th distance of the point P, including the point on the k-th distance, and excluding the point P, and |N k (P)| is the number thereof; ​

[0077] S303, calculate the average value of the local reachable density of all points in the k-neighborhood of the point P in the stitching domain;

[0078] S304, obtain the LOF score by the ratio of the result of step S303 and the result of step S302, and the expression is as follows:

[0079]

[0080] S40, judge whether the stitching domain has an abnormal point, if yes, calculate the probability, output the abnormal channel, and if not, delete the first step rows in the training domain, and form a new training domain after the stitching domain is spliced to the training domain.

[0081] Specifically, the specific steps of step S40 are as follows:

[0082] S401, mark the window moving number as m, m belongs to N * , screen out the points with the abnormal score LOF far greater than the threshold value thr=1.2 and the region being the stitching domain, if all the channel difference matrices of the stitching domain do not identify the abnormal points, delete the first step rows in the training domain, and form a new training domain after all the channel difference matrices and the N+(m-1)*step+1th row to the N+m*stepth row are spliced to the training domain. ij P ;

[0083] S402, mark the window moving number as m, m belongs to N * , screen out the points with the abnormal score LOF far greater than the threshold value thr=1.2 and the region being the stitching domain, if all the channel difference matrices of the stitching domain identify the abnormal points, count the frequency of the abnormal points in the stitching domain, and map to the single channel, the probability of the single channel abnormal point in the stitching region is the abnormal channel, and less than the threshold value is the normal channel, at the same time, the training domain remains unchanged, all the channel difference matrices of the stitching domain are the N+m*step+1th row to the N+(m+1)*step+1th row of the matrix , and form a data sequence X ij ={x P}, and repeat the steps S30 to S40.

[0084] The embodiment of the application further provides a computer readable storage medium, which has a computer program stored thereon, and the computer program is executed by a processor to realize any one of the methods in the embodiment.

[0085] ​​The embodiment of the present application also provides an electronic terminal, comprising a processor and a memory; the memory is used for storing a computer program, and the processor is used for executing the computer program stored in the memory, so that the terminal executes any method in the embodiment.

[0086] The computer readable storage medium in the embodiment can be understood by those skilled in the art that all or part of the steps of the above-mentioned method embodiments can be completed by the hardware of the computer program. The computer program mentioned above can be stored in a computer readable storage medium. The program executes the steps of the above-mentioned method embodiments when executed; and the storage medium mentioned above includes ROM, RAM, magnetic disc or optical disc and various storage program codes.

[0087] The electronic terminal provided by the embodiment comprises a processor, a memory, a transceiver and a communication interface, the memory and the communication interface are connected with the processor and the transceiver and complete communication between each other, the memory is used for storing a computer program, the communication interface is used for communication, and the processor and the transceiver are used for running the computer program, so that the electronic terminal executes the steps of the above method.

[0088] The following takes N=55, M=3 and L=10 as an example to illustrate the detection process of the CVT abnormal channel detection method based on the M-LOF algorithm provided by the embodiment of the present application.

[0089] S10, training data and test data are established;

[0090] Specifically, S101, a training data matrix Tr 55×3 and a test data matrix Te 10×3 are formed.

[0091] S102, the test data matrix Te 10×3 is spliced into the training data matrix Tr 55×3 in time sequence, a new matrix T (55+10)×3 is formed, and the image formed by the matrix is as shown in Figure 2 .

[0092] S20, data preprocessing;

[0093] Specifically, S201, each channel is subtracted to form a matrix wherein i={1, 2, 3}, j={1, 2, 3}, i≠j;

[0094] S202, step=5 is set, the first N+step rows (N=55, step=5) of the matrix are extracted, and a data sequence X ij ={x P} (wherein P=1, 2,..., 55+5) is formed.

[0095] S30. For each point P in the splicing domain, find the k-nearest-neighbor and calculate the LOF score;

[0096] Specifically, S301, for each point P in the splicing domain, calculate the distance between point P and other points O, where O ≠ P, and sort them from nearest to farthest;

[0097] S302 to S304, let k = (55+5) / 2 = 25, and calculate the data sequence X according to steps S302 to S304. ij The LOF score of the midpoint of the concatenated domain is used to obtain the score matrix. (where i = {A, B, C}, j = {A, B, C}, i ≠ j, and p is a point in the concatenation domain).

[0098] S40, Window movement and outlier probability calculation;

[0099] Specifically, S401, with the window moving times m=1, filters out points in the splicing domain whose outlier scores (LOF) are much greater than the threshold thr=1.2, i.e., points where 55≤p≤60, such as... Figures 3 to 5 As shown, Figures 3 to 5 The horizontal axis represents the point number (in the splicing domain), the vertical axis represents the LOF value, and the dashed line represents the threshold thr = 1.2. No outliers (LOF values ​​less than the threshold) were identified in the inter-channel difference matrix of the splicing domain. Therefore, the first step = 5 rows in the training domain are deleted, and the matrix... Lines 56 to 60 are concatenated with the training domain to form a new training domain. Simultaneously, lines 61 to 65 form a new concatenation domain, which together form the data sequence X. ij ={x P Repeat steps S30 and S40.

[0100] S402, with the window moving m = 2 times, filter out points with anomaly scores (LOF) significantly greater than the threshold (thr = 1.2) and whose regions are within the concatenation domain, i.e., points where 61 ≤ p ≤ 65. Figures 6 to 8 As shown, Figures 6 to 8 The horizontal axis represents the point number (spoofing domain), the vertical axis represents the LOF value, and the dashed line represents the threshold thr = 1.2. The inter-channel difference matrix in the splicing domain identified outliers, with 5 outliers in phase AB and 5 outliers in phase CA. Phase BC had no outliers. The mapping to the statistical results of their respective single channels is shown in Table 1.

[0101] Table 1

[0102] Channel A B C Outlier statistical value 10 5 5

[0103] There are a total of 10 points for each phase (A, B, and C). The probability values ​​for each phase are shown in Table 2.

[0104] Table 2

[0105] Channel A B C Outlier probability value 10 / 10 5 / 10 5 / 10

[0106] The probability threshold is set to 90%, and it is known that the A-phase channel is abnormal. The training domain remains unchanged, and the splicing domain is the data collected at the subsequent time point of 5 time points, and a data sequence X is formed ij = {x P}Repeat steps S30 and S40.

[0107] Compared with the prior art, the CVT abnormal channel detection method, medium and terminal based on the M-LOF algorithm described in the above embodiment, in the prior art, the error channel of the capacitive voltage transformer is qualitatively judged by the PCA method, but the PCA method is seriously dependent on the strong correlation of the signal and the distribution of the noise. In practice, due to the influence of the aging degree of the equipment, the load, the temperature, the humidity and the like, the data signals among the three phases of the capacitive voltage transformer are in a weak correlation and a strong noise state, which seriously affects the accuracy of the error channel judgment. Or the traditional LOF algorithm also has the disadvantage of low accuracy. The process of the present application is simple, and the result is accurate. The M-LOF model is used to detect the CVT abnormal channel. That is, the training set is continuously updated with the data collection. At the same time, when the repeated points appear, a very small value is added to the reachable distance to avoid the situation that the reachable distance is equal to zero, thereby improving the detection accuracy. The data set is grouped, and the grouped data set is further divided into a training domain and a splicing domain. Then, the LOF value of each point in the splicing domain in each subset is calculated, thereby improving the algorithm efficiency. The accuracy and efficiency of the whole detection method are greatly improved compared with the prior art.

[0108] Obviously, the above-described embodiments are only preferred embodiments of the present application, but not all the embodiments of the present application. The preferred embodiments of the present application are shown in the drawings, but do not limit the patent scope of the present application. The present application can be realized in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or equivalently replace some technical features. Any equivalent structure made by using the content of the present application specification and drawings, directly or indirectly applied to other related technical fields, is also within the patent protection scope of the present application.

Claims

1. A method for detecting abnormal channels of CVT based on M-LOF algorithm, characterized in that, The method comprises the following steps: S10, the M-channel capacitor voltage transformer secondary side collects training data and test data, and the data is spliced to form a matrix T (N+L)×M ; The specific steps of the step S10 are as follows: S101、Collecting training data and test data, forming a training data matrix Tr N×M and a test data matrix Te L×M Wherein N is the length of the training data, M is the number of signal channels, L is the length of the test data, considering the ratio difference and angle difference; S102、per time order will test data matrix Te L×M Spliced to training data matrix Tr N×M After, form new matrix T (N+L)×M ; S20, difference between each channel is made to form a new matrix extract the first N+step rows of the difference matrix between channels and form a data sequence X ij ; The specific steps of the step S20 are as follows: S201、Each channel difference is formed into a matrix wherein represents the difference between the ith channel and the jth channel, i≠j, the number of rows is N+L, and the number of columns is 1. S202, the moving step is represented as step, extracting the first N+step rows of the matrix , step ij P} where P=1, 2, …, N+step;​ S30, calculating the data sequence X ij The distance between the domain point and other points, the local reachable density of the domain point, the average value of the local reachable density of all points in the k-neighborhood of the domain point, and the LOF score. S40, judging whether the splicing domain has an abnormal point, if yes, calculating a probability, outputting an abnormal channel, if not, deleting the first step rows in the training domain, splicing the splicing domain to the training domain, and forming a new training domain.

2. The CVT anomaly path detection method based on the M-LOF algorithm according to claim 1, characterized in that, The specific steps of the step S30 are as follows: S301, for each point P in the splicing domain, calculating the distance between the point P and other points O, O≠P, and sorting from near to far; S302, for each point P in the splicing domain, calculating the local reachable density of the point P; First, calculating the reachable distance of the k nearest neighbors of the point P, wherein k is (N+step) / 2, and the k reachable distance from the point P to the point O is expressed as follows: where d k (O) is the kth order distance of point O, d(P, O) is the straight-line distance from point P to point O; Then, calculating the local reachable density of the point P, which is the reciprocal of the average reachable distance of the k nearest neighbors of the point P, and the expression is as follows: where N k (P) is the number of points within the kth distance of point P, including points on the kth distance, not including point P, |N k (P)| is the number of points. S303, calculating the average value of the local reachable densities of all points in the k neighborhood of the point P in the splicing domain; S304, obtaining the LOF score through the ratio of the calculation result of the step S303 to the calculation result of the step S302, and the expression is as follows:

3. The CVT anomaly path detection method based on M-LOF algorithm according to claim 1, characterized in that, The specific steps of the step S40 are as follows: S401, Let m be the number of window moves, where m∈N * Points with abnormal LOF scores significantly greater than the threshold thr = 1.2 and whose regions are within the splicing domain are selected. If the difference matrix between all channels in the splicing domain... If no outliers are identified, then the first step row of the training domain is deleted, and the difference matrix between all channels is removed. After the N+(m-1)*step+1th row to the N+m*stepth row is concatenated into the training domain, a new training domain is formed. At the same time, the N+m*step+1th row to the N+(m+1)*stepth row becomes a new concatenation domain, forming the data sequence X. ij ={x P }; S402, the window moving number of times is marked as m, m∈N * , the points with the abnormal score LOF far greater than the threshold value thr=1.2 and the region being a splicing domain are screened out, if all the inter-channel difference value matrices of the splicing domain identify abnormal points, the frequency of the abnormal points appearing in the splicing domain is counted and mapped to the single channel belonging to the abnormal points, the probability of the single channel abnormal points in the splicing region is an abnormal channel, less than the threshold value is a normal channel, at the same time, the training domain remains unchanged, all the inter-channel difference value matrices The splicing domain is the N+m*step+1th row to the N+(m+1)*step+1th row of the matrix , and a data sequence X ij ={x P} is formed, and the steps S30 to S40 are repeated.

4. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is executed by the processor to realize the method in any one of claims 1 to 3.

5. An electronic terminal, characterized in that It comprises: A processor and a memory; The memory is used for storing a computer program, and the processor is used for executing the computer program stored in the memory to enable the terminal to execute the method in any one of claims 1 to 3.

Images