Secondary AC cable insulation abnormity monitoring method based on machine learning

Abstract

The invention discloses a secondary AC cable insulation abnormity monitoring method based on machine learning, and belongs to the field of cable abnormity monitoring methods, and the method comprises the following steps: carrying out the multi-point data collection of the insulation resistance information, DC resistance information, dielectric loss information, discharge signal characteristic parameters and cable temperature information of each target cable; receiving all the collected information and parameters, performing anomaly analysis based on imported preset data, and outputting an analysis result; performing early warning grading according to the analysis result, outputting an early warning grading result, and performing visual display on the analysis result and the early warning grading result; and according to an early warning grading result, giving out an early warning notice of a corresponding level, formulating a cable maintenance priority and judging whether the staff needing to have a rest need to come to overtime. According to the invention, whether the insulation performance of the cable is abnormal or not can be comprehensively analyzed from multiple angles, the display effect of the analysis result is good, and an efficient visual display mode is provided for background workers.

Description

Technical Field

[0001] The present invention relates to the field of cable anomaly monitoring methods, and particularly to a secondary AC cable insulation anomaly monitoring method based on machine learning. Background Art

[0002] The relay protection system plays a key role in the safe and stable operation of the power grid. The ground insulation ability of the secondary AC circuit cable directly affects the reliability and accuracy of the relay protection device. If the insulation of the secondary AC circuit cable is damaged or aged, it will lead to signal distortion, increased interference, and even cause misoperation of the protection or missed judgment of faults, reducing the detection and isolation ability of the relay protection device for power system faults. However, the working environment of the relay protection secondary AC circuit cable is complex, and the insulation ability is constantly tested. Cable laying, collision, migration during construction, and chewing by small animals may damage the insulation. After long-term operation, weak defects may gradually worsen, resulting in inaccurate signal transmission and even causing misoperation and refusal to operate, threatening the safe and stable operation of the power grid.

[0003] Existing cable insulation anomaly monitoring methods usually compare the measured insulation resistance value with a specified standard value to determine whether the insulation performance of the cable is qualified. However, this monitoring method has certain defects and cannot comprehensively analyze whether the insulation performance of the cable is abnormal from multiple angles. The display effect of the analysis results is not ideal, and there is a lack of an efficient visualization display method.

[0004] Therefore, those skilled in the art have provided a secondary AC cable insulation anomaly monitoring method based on machine learning to solve the problems raised in the above background art. Summary of the Invention

[0005] Aiming at the defects in the prior art, the present invention provides a secondary AC cable insulation anomaly monitoring method based on machine learning, including the following steps:

[0006] Collect multi-point data on the insulation resistance information, DC resistance information, dielectric loss information, discharge signal characteristic parameters, and cable temperature information of each target cable;

[0007] Receive all the collected information and parameters and perform anomaly analysis based on the imported preset data, and output the analysis results;

[0008] Perform early warning grading according to the analysis results, output the early warning grading results, and visually display the analysis results and the early warning grading results;

[0009] Send out corresponding-level early warning notifications according to the early warning grading results, formulate the priority of cable maintenance, and judge whether it is necessary to arrange for employees on rest to work overtime.

[0010] As a further solution of the present invention: The specific process of the abnormality analysis is as follows:

[0011] Mark the insulation resistance of the target cable collected as A, mark the characteristic parameters of the discharge signal of the target cable collected as B, mark the temperature of the target cable collected as C, mark the DC resistance of the target cable collected as D, and mark the dielectric loss of the target cable collected as E;

[0012] Compare the insulation resistance A, the characteristic parameters of the discharge signal B, the temperature C, the DC resistance D, and the dielectric loss E of the target cable with the corresponding preset intervals respectively. If any one of the insulation resistance A, the characteristic parameters of the discharge signal B, the temperature C, the DC resistance D, and the dielectric loss E is not within the preset interval, it indicates that the insulation performance of the target cable is abnormal;

[0013] Screen out the target cables with abnormal insulation performance and mark them as Xi, i = 1···n, where n is a positive integer, and determine whether the positions of the selected cables Xi are outdoors or indoors;

[0014] Construct an abnormal cause analysis model and output a first analysis result. The first analysis result includes the predicted abnormal cause of the cable Xi and the corresponding required maintenance duration;

[0015] Construct an abnormal degree analysis model and output a second analysis result. The second analysis result includes the single - cable abnormal degree and the overall abnormal degree of the cable Xi.

[0016] As a further solution of the present invention: The specific analysis process of the abnormal cause analysis model is as follows:

[0017] Mark the points with abnormal insulation performance on the cable Xi as Yi, i = 1···n, where n is a positive integer;

[0018] Extract the insulation resistance A, the characteristic parameters of the discharge signal B, the temperature C, the DC resistance D, and the dielectric loss E collected corresponding to the abnormal point Yi as variables to generate a radar chart P;

[0019] Obtain the insulation resistance, the characteristic parameters of the discharge signal, the temperature, the DC resistance, and the dielectric loss of the abnormal points in the previous secondary AC cable insulation abnormality records, and use them as variables to generate radar charts pi, i = 1···n, where n is a positive integer;

[0020] Compare the similarity between the radar chart P and the radar charts pi, extract the three radar charts pi with the highest similarity to the radar chart P, and then obtain the abnormal causes in the insulation abnormality records corresponding to the three radar charts pi;

[0021] Select the abnormal cause with the most repeated occurrences from the three abnormal causes as the predicted abnormal cause of the abnormal point Yi;

[0022] If there is no repetition among the three abnormal reasons, the abnormal reason corresponding to the radar chart pi with the highest similarity to the radar chart P is used as the predicted abnormal reason for the abnormal point Yi;

[0023] Based on the determined predicted abnormal reason, obtain the average maintenance duration T of the abnormal points with the same abnormal reason from the previous records of secondary AC cable insulation abnormalities;

[0024] Take the obtained duration T as the maintenance duration required for the abnormal point Yi.

[0025] As a further aspect of the present invention: The specific analysis process of the abnormal degree analysis model is as follows:

[0026] Obtain the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D, and dielectric loss E collected corresponding to the abnormal point Yi;

[0027] Calculate the deviations between the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D, and dielectric loss E and their corresponding preset intervals respectively, and mark them as a1, b1, c1, d1, and e1;

[0028] Calculate the single - body abnormal degree f of the abnormal point Yi = 40%*(a1 / a2)+20%*(b1 / b2)+20%*(c1 / c2)+10%*(d1 / d2)+10%*(e1 / e2), where a2, b2, c2, d2, and e2 are all preset values;

[0029] Count the number of all abnormal points Yi on the cable Xi, mark it as Z, and extract the maximum single - body abnormal degree corresponding to the abnormal point Yi on the cable Xi, mark it as f 最大 ;

[0030] Calculate the overall abnormal degree F of the cable Xi = f 最大 *(Z / z), where z is a preset value.

[0031] As a further aspect of the present invention: The specific process of visualizing the analysis results is as follows:

[0032] Display the map of the area where the cable is located on the display screen, and display the line routes of each target cable on the display screen;

[0033] Mark the line routes of the cables without abnormal insulation performance on the display screen with green lines;

[0034] Compare the overall abnormality degree F corresponding to the cable Xi with a preset value ΔF. If F > ΔF, mark the line routing of the cable Xi on the display screen with a red line. If F ≤ ΔF, mark the line routing of the cable Xi on the display screen with a yellow line;

[0035] Obtain the position of the abnormal point Yi on the cable Xi, and mark the position of the abnormal point Yi with a black dot on the line routing of the cable Xi.

[0036] As a further solution of the present invention: The specific process of the warning classification is as follows:

[0037] Extract the number of red line routings from the display screen and mark it as G1;

[0038] Extract the number of yellow line routings from the display screen and mark it as G2;

[0039] If G1 > 0, issue a level-three warning. If G1 = 0, proceed to the next step;

[0040] Compare G2 with a preset value g. If G2 > g, issue a level-two warning. If 0 < G2 ≤ g, issue a level-one warning. If G2 = 0, do not issue a warning.

[0041] As a further solution of the present invention: The process of formulating the cable maintenance priority is specifically as follows:

[0042] Plan the cable Xi corresponding to the red line routing in the display screen into the first sequence;

[0043] Select any point in the map of the display screen as the center of the circle, draw a circle according to the center of the circle and a preset radius, search for the abnormal points Yi within the circle, count the number of abnormal points Yi within the circle and mark it as W;

[0044] Compare W with a preset value w. If W > w, screen out the area where the corresponding circle is located as the area to be maintained, and then plan the area to be maintained into the second sequence;

[0045] Maintain the entire cable Xi in the first sequence as the first priority. Among them, if the number of cables Xi in the first sequence is multiple, maintain them in order from largest to smallest according to the overall abnormality degree F corresponding to the cable Xi;

[0046] Maintain the abnormal point Yi within the area to be maintained in the second sequence as the second priority. Among them, if the number of areas to be maintained in the second sequence is multiple, calculate the abnormal weight value of each area to be maintained. The abnormal weight value refers to the sum of the individual abnormality degrees f corresponding to all the abnormal points Yi within the area to be maintained, and then maintain each area to be maintained in order from largest to smallest according to the abnormal weight value;

[0047] Maintain the abnormal points Yi that are not within the area to be maintained and on the first-sequence cables Xi as the third priority.

[0048] As a further solution of the present invention: The specific process of determining whether to arrange for the employees on rest to work overtime is as follows:

[0049] Calculate the total duration T required for maintenance corresponding to all abnormal points Yi, and mark it as t_total;

[0050] Count the number of areas to be maintained in the second sequence and mark it as M, and then obtain the number G1 of the red-line routes;

[0051] Count the number of all on-duty maintenance personnel currently, and mark it as W;

[0052] Calculate the current abnormal maintenance workload weight value S = (t_total / W) * (M + 0.5G1 + m) / m, where m is a preset value;

[0053] Compare the abnormal maintenance workload weight value S with the preset value s. If S ≤ s, it is not necessary to arrange for the employees on rest to work overtime; if S > s, it is necessary to arrange for the employees on rest to work overtime.

[0054] As a further solution of the present invention: The specific arrangement for the employees on rest to work overtime is as follows:

[0055] If s < S ≤ 2s, arrange three employees on rest to work overtime;

[0056] If 2s < S ≤ 3s, arrange five employees on rest to work overtime;

[0057] If 3s < S, arrange ten employees on rest to work overtime;

[0058] The employees arranged to work overtime are preferentially arranged in ascending order of their on-duty time this month.

[0059] As a further solution of the present invention: The specific process of sending out warning notifications at corresponding levels according to the warning classification results is as follows:

[0060] If the warning level is a first-level warning, the frame of the display screen flashes yellow light;

[0061] If the warning level is a second-level warning, the frame of the display screen flashes red light;

[0062] If the warning level is a third-level warning, the frame of the display screen is constantly lit with red light.

[0063] The beneficial effects of the present invention are reflected in:

[0064] This application comprehensively analyzes the insulation performance of a cable from multiple perspectives, including the insulation resistance, discharge signal characteristic parameters, temperature, DC resistance, and dielectric loss of the target cable. Then, it quickly and accurately determines whether there are any abnormalities in the cable's insulation performance, and based on this analysis, predicts the causes of cable abnormalities. At the same time, this application adopts a more efficient visual display method for the analysis results of the cable, with a better display effect for the analysis results. It not only facilitates the staff to understand the degree of abnormality of each cable but also helps to determine the areas with severe cable abnormalities. In addition, this application formulates corresponding maintenance priorities according to the degree of cable abnormality and determines whether to arrange for off-duty employees to work overtime. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0066] Figure 1 It is a flowchart of a method for monitoring insulation abnormalities of secondary AC cables based on machine learning. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0068] As mentioned in the background art of this application, through research, it is found that the existing methods for monitoring insulation abnormalities of cables usually compare the measured insulation resistance value with the specified standard value to determine whether the insulation performance of the cable is qualified. However, this monitoring method has certain defects and cannot comprehensively analyze whether the insulation performance of the cable is abnormal from multiple perspectives. The display effect of the analysis results is not ideal, and there is a lack of an efficient visual display method.

[0069] To solve the above defects, this application discloses a method for monitoring insulation abnormalities of secondary AC cables based on machine learning, which can comprehensively analyze whether the insulation performance of the cable is abnormal from multiple perspectives, has a good display effect for the analysis results, and provides an efficient visual display method for the back-end staff.

[0070] The following will introduce in detail how the solution of this application solves the above technical problems with reference to the drawings.

[0071] Please refer to Figure 1, in the embodiment of the present invention, a method for monitoring abnormal insulation of secondary AC cables based on machine learning includes the following steps: collecting multi-point data on the insulation resistance information, DC resistance information, dielectric loss information, discharge signal characteristic parameters, and cable temperature information of each target cable; receiving all the collected information and parameters and performing abnormal analysis based on the imported preset data, and outputting the analysis result; performing early warning grading according to the analysis result, outputting the early warning grading result, and visually displaying the analysis result and the early warning grading result; sending out corresponding-level early warning notifications according to the early warning grading result, formulating the priority of cable maintenance, and judging whether it is necessary to arrange for employees on rest to work overtime. This application can comprehensively analyze whether the insulation performance of the cable is abnormal from multiple angles, has a good display effect for the analysis result, and provides an efficient visual display method for the back-end staff.

[0072] In this embodiment, the specific process of abnormal analysis is as follows: Mark the insulation resistance of the target cable collected as A, mark the discharge signal characteristic parameters of the target cable collected as B, mark the temperature of the target cable collected as C, mark the DC resistance of the target cable collected as D, and mark the dielectric loss of the target cable collected as E; Compare the insulation resistance A, discharge signal characteristic parameters B, temperature C, DC resistance D, and dielectric loss E of the target cable with the corresponding preset intervals respectively. If any one of the insulation resistance A, discharge signal characteristic parameters B, temperature C, DC resistance D, and dielectric loss E is not within the preset interval, it indicates that the insulation performance of the target cable is abnormal; Screen out the target cables with abnormal insulation performance and mark them as Xi, where i = 1···n, and n is a positive integer, and determine whether the positions of the selected cables Xi are outdoors or indoors; Construct an abnormal cause analysis model and output a first analysis result, where the first analysis result includes the predicted abnormal cause of the cable Xi and the corresponding maintenance duration required; Construct an abnormal degree analysis model and output a second analysis result, where the second analysis result includes the individual abnormal degree and the overall abnormal degree of the cable Xi. This application comprehensively analyzes the insulation performance of the cable from the insulation resistance, discharge signal characteristic parameters, temperature, DC resistance, and dielectric loss of the target cable, and then quickly and accurately determines whether there is an abnormality in the insulation performance of the cable, and predicts the cause of the cable abnormality based on this analysis.

[0073] In this embodiment, the specific analysis process of the abnormal cause analysis model is as follows: Mark the points with abnormal insulation performance on the cable Xi as Yi, where i = 1···n, and n is a positive integer; Extract the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D, and dielectric loss E collected corresponding to the abnormal point Yi as variables to generate a radar chart P; Obtain the insulation resistance, discharge signal characteristic parameters, temperature, DC resistance, and dielectric loss of the abnormal points in the previous secondary AC cable insulation abnormality records, and use them as variables to generate radar charts pi, where i = 1···n, and n is a positive integer; Compare the similarity between the radar chart P and the radar charts pi, extract the three radar charts pi with the highest similarity to the radar chart P, and then obtain the abnormal causes in the abnormal records corresponding to the three radar charts pi; Screen out the abnormal cause with the most repeated times from the three abnormal causes as the predicted abnormal cause of the abnormal point Yi; If there is no repetition among the three abnormal causes, use the abnormal cause corresponding to the radar chart pi with the highest similarity to the radar chart P as the predicted abnormal cause of the abnormal point Yi; Based on the determined predicted abnormal cause, obtain the average maintenance duration T of the abnormal points with the same abnormal cause from the previous secondary AC cable insulation abnormality records; Use the obtained duration T as the maintenance duration required for the abnormal point Yi. This setting can quickly and accurately analyze the abnormal causes of cables with abnormal insulation performance.

[0074] In this embodiment, the specific analysis process of the abnormal degree analysis model is as follows: Obtain the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D, and dielectric loss E collected corresponding to the abnormal point Yi; Calculate the deviations between the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D, and dielectric loss E and the corresponding preset intervals respectively, and mark them as a1, b1, c1, d1, and e1; Calculate the single abnormal degree f of the abnormal point Yi = 40%*(a1 / a2) + 20%*(b1 / b2) + 20%*(c1 / c2) + 10%*(d1 / d2) + 10%*(e1 / e2), where a2, b2, c2, d2, and e2 are all preset values; Count the number of all abnormal points Yi on the cable Xi, mark it as Z, and extract the maximum single abnormal degree corresponding to the abnormal point Yi on the cable Xi, mark it as f 最大 ; Calculate the overall abnormal degree F of the cable Xi = f 最大 *(Z / z), where z is a preset value. This setting can not only determine the abnormal degree of a single abnormal point of the cable, but also determine the overall abnormal degree of the cable.

[0075] In this embodiment, the specific process of visualizing the analysis results is as follows: The map of the area where the cables are located is displayed on a display screen, and the line routes of each target cable are also displayed on the display screen; The line routes of the cables with no abnormal insulation performance on the display screen are marked with green lines; The overall abnormal degree F corresponding to the cable Xi is compared with a preset value ΔF. If F > ΔF, the line route of the cable Xi on the display screen is marked with a red line. If F ≤ ΔF, the line route of the cable Xi on the display screen is marked with a yellow line; The position of the abnormal point Yi on the cable Xi is obtained, and the position of the abnormal point Yi is marked with a black dot on the line route of the cable Xi. This application adopts a more efficient visual display method for the analysis results of the cables, which not only facilitates the staff to understand the abnormal degree of each cable, but also helps to determine the areas with severe cable abnormalities.

[0076] In this embodiment, the specific process of early warning classification is as follows: The number of red line routes is extracted from the display screen and marked as G1; The number of yellow line routes is extracted from the display screen and marked as G2; If G1 > 0, a third-level early warning is issued. If G1 = 0, the next step is entered; G2 is compared with a preset value g. If G2 > g, a second-level early warning is issued. If 0 < G2 ≤ g, a first-level early warning is issued. If G2 = 0, no early warning is issued. This setting facilitates the staff to determine the abnormal degree of the current cables according to the early warning level.

[0077] In this embodiment, the process of formulating the cable maintenance priority is as follows: The cable Xi corresponding to the red line routing in the display screen is planned into the first sequence; Select any point in the map of the display screen as the center of the circle, draw a circle according to the center of the circle and the preset radius, search for the abnormal points Yi within the circle, count the number of abnormal points Yi within the circle and mark it as W; Compare W with the preset value w. If W > w, then screen out the area where the corresponding circle is located as the area to be maintained, and then plan the area to be maintained into the second sequence; The cables Xi in the first sequence are maintained as the first priority as a whole. Among them, if the number of cables Xi in the first sequence is multiple, they are maintained in turn according to the overall abnormal degree F of the corresponding cables Xi from large to small; The abnormal points Yi within the area to be maintained in the second sequence are maintained as the second priority. Among them, if the number of areas to be maintained in the second sequence is multiple, calculate the abnormal weight values of each area to be maintained. The abnormal weight value refers to the sum of the individual abnormal degrees f corresponding to all abnormal points Yi within the area to be maintained, and then maintain each area to be maintained in turn according to the abnormal weight value from large to small; The abnormal points Yi that are not within the area to be maintained and on the cable Xi in the first sequence are maintained as the third priority. This setting can quickly and accurately determine which cables need to be maintained preferentially. Among them, the more abnormal points there are within the area to be maintained, the more serious the cable abnormal disaster area this area is, and there are great influencing factors in this area, which need to be solved in time. And those with multiple abnormal points on a single cable indicate that there are serious abnormalities in the quality of the cable itself or the environment it is in. If not solved in time, it is very likely that this cable will overheat and catch fire, thereby affecting other cables.

[0078] In this embodiment, the specific process of determining whether to arrange the employees on rest to work overtime is as follows: Calculate the total sum of the maintenance required time T corresponding to all abnormal points Yi, and mark it as t_total; Count the number of areas to be maintained in the second sequence and mark it as M, and then obtain the number G1 of the red line routings; Count the number of all on-duty maintenance personnel currently, and mark it as W; Calculate the current abnormal maintenance workload weight value S = (t_total / W) * (M + 0.5G1 + m) / m, where m is a preset value; Compare the abnormal maintenance workload weight value S with the preset value s. If S ≤ s, then there is no need to arrange the employees on rest to work overtime; If S > s, then it is necessary to arrange the employees on rest to work overtime. This setting can quickly determine whether to arrange the employees on rest to work overtime.

[0079] In this embodiment, the employees arranged to work overtime while on rest are as follows: if s < S ≤ 2s, three employees on rest are arranged to work overtime; if 2s < S ≤ 3s, five employees on rest are arranged to work overtime; if 3s < S, ten employees on rest are arranged to work overtime; the employees arranged to work overtime are preferentially arranged in ascending order of their on-duty time this month. This setting can quickly determine how many personnel are needed to work overtime.

[0080] In this embodiment, issuing a warning notice at the corresponding level according to the warning classification result is specifically as follows: if the warning level is a first-level warning, the border of the display screen flashes yellow light; if the warning level is a second-level warning, the border of the display screen flashes red light; if the warning level is a third-level warning, the border of the display screen is constantly lit with red light. This setting facilitates the backstage staff to quickly determine the current warning level.

[0081] The present invention comprehensively analyzes the insulation performance of the cable from multiple angles including the insulation resistance, discharge signal characteristic parameters, temperature, DC resistance, and dielectric loss of the target cable, and then quickly and accurately determines whether there is an abnormality in the insulation performance of the cable. Based on this analysis, the reason for the cable abnormality is predicted. At the same time, the present application adopts a more efficient visual display method for the analysis results of the cable, and the display effect of the analysis results is better. It is not only convenient for the staff to understand the abnormality degree of each cable, but also convenient to determine the area with serious cable abnormalities. In addition, according to the abnormality degree of the cable, the present application formulates the corresponding maintenance priority level and determines whether it is necessary to arrange employees on rest to work overtime.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or equivalently replace some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered by the scope of the claims and the description of the present invention.

Claims

1. A secondary AC cable insulation abnormality monitoring method based on machine learning, characterized in that: The following steps are involved: Perform multi-point data collection on the insulation resistance information, DC resistance information, dielectric loss information, discharge signal characteristic parameters and cable temperature information of each target cable; Receive all collected information and parameters and perform abnormal analysis based on the imported preset data, and output the analysis results; Perform early warning classification based on the analysis results, output the early warning classification results, and visualize the analysis results and the early warning classification results; According to the warning classification results, a warning notification of the corresponding level is issued, the cable maintenance priority is determined, and it is determined whether it is necessary to arrange for employees on rest to work overtime.

2. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 1, characterized in that: The specific process of the abnormal analysis is as follows: The collected insulation resistance of the target cable is marked as A, the collected discharge signal characteristic parameter of the target cable is marked as B, the collected temperature of the target cable is marked as C, the collected DC resistance of the target cable is marked as D, and the collected dielectric loss of the target cable is marked as E; The insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D and dielectric loss E of the target cable are respectively compared with the corresponding preset intervals. If any one of the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D and dielectric loss E is not within the preset interval, it means that the insulation performance of the target cable is abnormal; Filter out target cables with abnormal insulation performance and mark them as Xi, where i=1···n, where n is a positive integer, and determine whether the location of the filtered cable Xi is outdoor or indoor; Construct an abnormality cause analysis model and output a first analysis result, where the first analysis result includes a predicted abnormality cause of the cable Xi and a corresponding maintenance time required; An abnormality degree analysis model is constructed to output a second analysis result, which includes the individual abnormality degree and the overall abnormality degree of the cable Xi.

3. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 2 is characterized in that: The specific analysis process of the abnormal cause analysis model is as follows: The points on the cable Xi where the insulation performance is abnormal are marked as Yi, i=1···n, where n is a positive integer; Extract the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D and dielectric loss E corresponding to the abnormal point Yi as variables to generate a radar chart P; Obtain the insulation resistance, discharge signal characteristic parameters, temperature, DC resistance and dielectric loss of the abnormal point in the previous secondary AC cable insulation abnormality records, and use them as variables to generate a radar chart pi, i = 1···n, where n is a positive integer; Compare the radar chart P with the radar chart pi for similarity, extract the three radar charts pi with the highest similarity to the radar chart P, and then obtain the abnormal causes in the insulation abnormality records corresponding to the three radar charts pi; The abnormal cause with the most repetitions is selected from the three abnormal causes as the predicted abnormal cause of the abnormal point Yi; If there is no duplication among the three abnormal causes, the abnormal cause corresponding to the radar chart pi with the highest similarity to the radar chart P is taken as the predicted abnormal cause of the abnormal point Yi; Based on the determined predicted abnormal cause, the average maintenance time T of abnormal points with the same abnormal cause is obtained from the previous secondary AC cable insulation abnormality records; The acquired duration T is used as the required maintenance duration corresponding to the abnormal point Yi.

4. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 3 is characterized in that: The specific analysis process of the abnormality degree analysis model is as follows: Obtain the insulation resistance A, discharge signal characteristic parameter B, temperature C, DC resistance D and dielectric loss E corresponding to the abnormal point Yi; Calculate the deviations between the insulation resistance A, the discharge signal characteristic parameter B, the temperature C, the DC resistance D and the dielectric loss E and the corresponding preset intervals, and mark them as a1, b1, c1, d1 and e1 respectively; Calculate the monomer abnormality level f corresponding to the abnormal point Yi = 40%*(a1 / a2)+20%*(b1 / b2)+20%*(c1 / c2)+10%*(d1 / d2)+10%*(e1 / e2), where a2, b2, c2, d2 and e2 are all preset values; Count the number of all abnormal points Yi on the cable Xi, marked as Z, and extract the maximum single abnormal degree corresponding to the abnormal point Yi on the cable Xi, marked as f 最大 ; Calculate the overall abnormality level F corresponding to the cable Xi = f 最大 *(Z / z), where z is a preset value.

5. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 4 is characterized in that: The specific process of visualization of the analysis results is as follows: Display a map of the area where the cables are located on a display screen, and show the route of each target cable on the display screen; The cables with no insulation abnormalities are marked with green lines on the display screen; The overall abnormality level F corresponding to the cable Xi is compared with the preset value ΔF. If F>ΔF, the line alignment of the cable Xi on the display screen is marked with a red line. If F≤ΔF, the line alignment of the cable Xi on the display screen is marked with a yellow line. Get the position of the abnormal point Yi on the cable Xi, and mark the position of the abnormal point Yi with a black dot on the line of the cable Xi.

6. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 5, characterized in that: The specific process of the warning classification is as follows: Extract the number of red line traces from the display, marked as G1; Extract the number of yellow line traces from the display, marked as G2; If G1>0, a third-level warning is issued, and if G1=0, the next step is entered; Compare G2 with the preset value g. If G2>g, a second-level warning is issued. If 0<G2≤g, a first-level warning is issued. If G2=0, no warning is issued.

7. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 6 is characterized in that: The cable maintenance priority setting process is as follows: Plan the cable Xi corresponding to the red line in the display screen to the first sequence; Select any point on the map on the display screen as the center of the circle, draw a circle based on the center of the circle and the preset radius, search for abnormal points Yi in the circle, count the number of abnormal points Yi in the circle and mark them as W; Compare W with the preset value w. If W>w, select the area where the corresponding circle is located as the area to be maintained, and then plan the area to be maintained into the second sequence; The cables Xi in the first sequence are maintained as a whole as the first priority, wherein, if there are multiple cables Xi in the first sequence, they are maintained in descending order according to the overall abnormality levels F corresponding to the cables Xi; The abnormal point Yi in the area to be maintained in the second sequence is maintained as the second priority. If there are multiple areas to be maintained in the second sequence, the abnormal weight value of each area to be maintained is calculated. The abnormal weight value refers to the sum of the monomer abnormality degrees f corresponding to all abnormal points Yi in the area to be maintained. Then, each area to be maintained is maintained in turn according to the abnormal weight value from large to small. The abnormal point Yi that is not in the area to be maintained and on the first sequence cable Xi is maintained as the third priority.

8. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 7 is characterized in that: The specific process of determining whether it is necessary to arrange for employees on rest to work overtime is as follows: Calculate the sum of the maintenance time T corresponding to all abnormal points Yi, marked as ttotal; Count the number of areas to be maintained in the second sequence and mark them as M, and then obtain the number of red line routings G1; Count the number of all maintenance personnel currently on duty, marked as W; Calculate the current abnormal maintenance workload weight value S = (ttotal / W)*(M+0.5G1+m) / m, where m is a preset value; The abnormal maintenance workload weight value S is compared with the preset value s. If S≤s, there is no need to arrange employees on rest to work overtime; if S>s, it is necessary to arrange employees on rest to work overtime.

9. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 8, characterized in that: The employees who are arranged to take a break to work overtime are as follows: If s<S≤2s, three employees who are on break will be arranged to work overtime; If 2s<S≤3s, five employees who are on break will be arranged to work overtime; If 3s < S, arrange ten employees who are on break to work overtime; Employees who are scheduled to work overtime will be given priority in the order of their working hours this month from the shortest to the longest.

10. The method for monitoring secondary AC cable insulation abnormality based on machine learning according to claim 9, characterized in that: The specific warning notifications of corresponding levels issued according to the warning classification results are: If the warning level is level one, the border of the display screen will flash yellow; If the warning level is level 2, the border of the display screen will flash red; If the warning level is level three, the border of the display screen will be always red.

Images