An AI regulation-based switch abnormality prompting method

Abstract

The application discloses a switch abnormality prompting method based on AI regulation, relates to the technical field of switch abnormality detection, and is used for solving the problems of high device failure rate and low maintenance efficiency. The switch internal circuit is separately marked, collection time is set, the data of the marked object is collected within the collection time, the decision tree method is used to process the data and set multiple state weights, the response time of the marked object is detected, the abnormal characteristics are generated in combination with the state weights, the marked object is introduced into different levels of abnormality detection mechanisms according to the abnormal characteristics, the monitoring signals generated by the abnormality detection mechanisms are monitored in real time, the monitoring information is acquired, the operation records of each marked object are called, the clustering algorithm is inputted and integrated with the monitoring information, the prompting index is calculated after integration, whether abnormality prompting is performed is judged according to the prompting index, the reaction efficiency of the system to abnormal fluctuations is enhanced, and the device failure rate is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of switch anomaly detection, and more particularly to a switch anomaly prompting method based on AI regulation. BACKGROUND

[0002] In existing automation systems and intelligent monitoring devices, device operation state monitoring is of great significance to the safety and stability of the system. Traditional device state monitoring methods usually rely on manual regular inspection or single parameter measurement, such as threshold judgment or simple alarm signal identification of abnormalities.

[0003] The prior art has the following disadvantages:

[0004] At present, the existing method cannot combine multi-dimensional data for comprehensive analysis, lacks deep recognition ability of abnormal state, and cannot take corresponding processing strategies for different types of switch anomalies, making it difficult to discover abnormal phenomena in time, resulting in increased system operation risk, increased device failure rate, reduced system reliability, and reduced maintenance efficiency. Therefore, a switch anomaly prompting method based on AI regulation is proposed.

[0005] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a switch anomaly prompting method based on AI regulation, which solves the problems raised in the above background technology by monitoring line state, decision tree data processing, multi-state weight analysis, and hierarchical anomaly detection and clustering mechanism.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical scheme, a switch anomaly prompting method based on AI regulation, comprising the following steps:

[0008] Step S1: marking each line in the switch, setting the collection time, and collecting the line data of the marked line within the collection time;

[0009] In step S1, the line data of the marked line is the vibration frequency of the marked line, the ground current of the marked line, and the noise intensity of the marked line;

[0010] Select a period of time as the collection time, and collect the line data of each marked line as follows within the collection time:

[0011] a plurality of time points are randomly set in the collection time, and the vibration frequency of the marker line, the ground current and the noise intensity at each time point are detected respectively to obtain the vibration frequency of the marker line, the ground current of the marker line and the noise intensity of the marker line at each time point;

[0012] Step S2: using a decision tree method to process the line data and set a plurality of state weights, detecting the line data response time of the marker line, and generating an abnormal feature in combination with the state weight;

[0013] Step S3: according to the abnormal feature, the marker line is introduced into a line abnormality detection mechanism of different levels, and the monitoring signal generated by the line abnormality detection mechanism is monitored in real time to obtain monitoring information;

[0014] Step S4: calling the closing times of each marker line in the switch, integrating the monitoring information, inputting a clustering algorithm to calculate a switch prompt index, and judging whether to generate abnormal prompt information according to the switch prompt index.

[0015] In a preferred embodiment, in step S2, the marker line vibration frequencies of adjacent time points in the collection time are subtracted, the vibration frequency of the former time point is subtracted from the vibration frequency of the latter time point as the instantaneous vibration frequency change of the latter time point, and the subtraction is performed once for every two adjacent time points to generate a plurality of instantaneous vibration frequency changes in the collection time.

[0016] When calculating the instantaneous ground current change, the ground current of the latter time point is subtracted from the ground current of the former time point as the instantaneous ground current change of the former time point.

[0017] In a preferred embodiment, in step S2, before using the decision tree method, the first time point and the last time point in the collection time are excluded and integrated into a time point data set, and the noise intensities corresponding to each time point in the time point data set are combined into a noise data set.

[0018] In the collection time, the instantaneous vibration frequency change of the last time point is excluded in the order of time points, and the instantaneous vibration frequency changes are combined into a vibration frequency data set;

[0019] In the collection time, the instantaneous ground current change of the first time point is excluded in the order of time points, and the instantaneous ground current changes are combined into a ground current data set;

[0020] The time point data set, the noise data set, the vibration frequency data set and the ground current data set are used as inputs to construct a decision tree model to screen target line data.

[0021] In a preferred embodiment, in step S2, the step of constructing a decision tree model to screen target line data is as follows:

[0022] Data preparation: the time point data set, the noise data set, the vibration frequency data set, and the ground current data set are taken as inputs, and the target line data is taken as output;

[0023] Setting the root node: two decision trees are constructed using the vibration frequency data set and the ground current data set, respectively, and the maximum values in the vibration frequency data set and the ground current data set are selected as the root nodes of the two decision trees;

[0024] Constructing a decision tree: taking the root node as a parent node, connecting a left child node and a right child node to each parent node, and repeating the operation with the child nodes as new parent nodes until the number of nodes is consistent with the number of data in the vibration frequency data set or the ground current data set, filling the data in the vibration frequency data set or the ground current data set into the nodes in descending order, and recording the time points corresponding to the data in each node and the noise intensity at the corresponding time points from the time point data set and the noise data set;

[0025] Screening time points: the number of layers of the two decision trees is counted, all nodes in the middle layers of the two decision trees are marked, the time points corresponding to the maximum noise intensity are selected in the recording information of the marked nodes in the two decision trees, and the selected time points in the two decision trees are compared;

[0026] Feature selection: if the selected time points in the two decision trees are the same time point, the instantaneous vibration frequency change and the instantaneous ground current change at the corresponding time point are taken as the target line data; if the selected time points in the two decision trees are not the same time point, all time points between the two time points are selected in the time point data set, and the instantaneous vibration frequency change and the instantaneous ground current change at the middle time point are taken as the target line data;

[0027] Target output: the screened target line data is output as output data.

[0028] In a preferred embodiment, in step S2, a state weight is set according to the target line data, and the specific steps are as follows:

[0029] The instantaneous vibration frequency change and the instantaneous ground current change in the target line data are set as input variables, and the input variables are taken as logistic regression intermediate variables to generate state weights: where i is the serial number of the three input variables, is the state weight generated for the i-th input variable, e is the natural base, is the i-th logistic regression intermediate variable;

[0030] When the input variable is the instantaneous vibration frequency change amount, the generated state weight is set as the vibration frequency weight; when the input variable is the instantaneous geoelectric current change amount, the generated state weight is set as the geoelectric current weight.

[0031] In a preferred embodiment, in step S2, the line data response time of the marker line includes the response time of the marker line vibration frequency and the response time of the marker line geoelectric current; the time difference between the time point corresponding to the maximum value of the marker line vibration frequency and the time point corresponding to the minimum value of the marker line vibration frequency in the collection time is calculated as the response time of the marker line vibration frequency, and the response time of the marker line geoelectric current is calculated in the same way;

[0032] The vibration frequency weight, the geoelectric current weight, the response time of the marker line vibration frequency and the response time of the marker line geoelectric current are weighted and summed to generate the abnormal feature of the marker line.

[0033] In a preferred embodiment, in step S3, when the abnormal feature of the marker line exceeds the preset abnormal threshold, the marker line is matched with the preset high-level detection mechanism; otherwise, the marker line is matched with the preset low-level detection mechanism;

[0034] In each layer of detection, it is judged whether the line information of the marker line meets the detection condition, and if it meets the condition, it enters the next layer, and if it does not meet the condition, a monitoring signal is generated;

[0035] The monitoring information is the number of monitoring signals generated after the marker line passes through the line abnormal detection mechanism.

[0036] In a preferred embodiment, in step S4, the closing times of each marker line are standardized to generate the results of the marker line after standardization of the closing times;

[0037] The results of the marker line after standardization of the closing times are multiplied by the number of monitoring signals corresponding to the marker line to generate a marker line prompt index, and the marker line prompt index is merged into a prompt index data set and input into a clustering algorithm for screening:

[0038] The data in the prompt index data set is arranged from small to large, the median of the data is selected as the core point, a screening difference value is set, the previous data or the next data of the core point is marked, and when the absolute value of the difference between the marker point and the core point is lower than the screening difference value, the marker point and the core point are screened out. The marker point is used as a new core point for repeated operation, and the operation is stopped when the calculation result of the absolute value of the difference between the marker point and the core point exceeds the screening difference value, and the average value of the data in the screened prompt index data set is calculated as the switch prompt index.

[0039] When the switch prompt index exceeds the preset prompt threshold, an abnormal prompt information is generated.

[0040] Technical effects and advantages of the present application:

[0041] The present application marks each line in the switch, sets the collection time, collects the data of the marked object within the collection time, processes the data by using the decision tree method, sets multiple state weights, detects the response time of the marked object, generates the abnormal features in combination with the state weights, imports the marked object into different levels of abnormal detection mechanisms according to the abnormal features, monitors the monitoring signals generated by the abnormal detection mechanisms in real time and obtains the monitoring information, calls the operation records of each marked object, calculates the prompt index after the clustering algorithm is integrated with the monitoring information, judges whether to perform the abnormal prompt according to the prompt index, can realize the automatic identification and hierarchical detection of the abnormal state, effectively improves the fault diagnosis accuracy and response speed of the system, enhances the reaction efficiency of the system to abnormal fluctuations, and reduces the equipment failure rate. BRIEF DESCRIPTION OF DRAWINGS

[0042] Fig. 1 The implementation flowchart of the switch abnormal prompt method based on AI regulation of the present application.

[0043] Fig. 2 The step schematic diagram of the switch abnormal prompt method based on AI regulation of the present application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0045] Embodiment 1, a switch abnormal prompt method based on AI regulation, as shown in the figure, includes the following steps: Figs. 1-2

[0046] Step S1: marking each line in the switch, setting the collection time, collecting the line data of the marked line within the collection time;

[0047] Step S2: processing the line data by using the decision tree method and setting multiple state weights, detecting the line data response time of the marked line, and generating the abnormal features in combination with the state weights;

[0048] Step S3: according to the abnormal features, importing the marked line into different levels of line abnormal detection mechanisms, monitoring the monitoring signals generated by the line abnormal detection mechanisms in real time and obtaining the monitoring information;

[0049] ​Step S4: calling the closing times of each marked line in the switch, integrating with the monitoring information, and inputting into the clustering algorithm to calculate the switch prompt index. According to the switch prompt index, it is judged whether to generate abnormal prompt information.

[0050] The specific implementation is as follows:

[0051] In step S1, each line in the switch is marked, and the line data of each marked line is collected.

[0052] The line data of the marked line is the vibration frequency of the marked line, the ground current of the marked line, and the noise intensity of the marked line.

[0053] It should be noted that the information database is a structured collection for storing, managing and retrieving data. In this example, the information database is an information database for storing switch information, which contains the stored current interval of each line under normal operation.

[0054] By collecting the state data of the marked line, multiple state weights are set to judge the abnormality of the marked line, thereby improving the accuracy of line judgment.

[0055] A period of time is selected as the collection time, and the line data of each marked line is collected in the collection time as follows:

[0056] When collecting the vibration frequency of the marked line and the ground current of the marked line, multiple time points are randomly set in the collection time, and the vibration frequency and the ground current of the marked line are detected at each time point by using a vibration sensor and a ground current monitor to obtain the vibration frequency and the ground current of the marked line at each time point.

[0057] When collecting the noise intensity of the marked line, the noise of the marked line is detected at each time point by using a noise meter in the collection time, and the detection result is taken as the noise intensity of the corresponding time point.

[0058] It should be noted that the time length of the collection time is not unique, for example, the time length of the collection time is set to half an hour, etc., which is not described here.

[0059] In step S2, the vibration frequency of the marked line at adjacent time points in the collection time is subtracted, specifically, the vibration frequency of the marked line at the former time point is subtracted from the vibration frequency of the marked line at the latter time point to obtain the instantaneous vibration frequency change at the latter time point. The subtraction is performed once for every two adjacent time points, and multiple instantaneous vibration frequency changes are generated in the collection time.

[0060] Differences between the marker line ground currents of adjacent time points in the collection time are calculated, specifically, the difference between the marker line ground current of the later time point and the marker line ground current of the former time point is taken as the instantaneous ground current variation of the former time point, and similarly, a plurality of instantaneous ground current variations are generated within the collection time;

[0061] It should be explained that the greater the instantaneous vibration frequency variation or the greater the instantaneous ground current variation, the more likely it is to cause the line to be abnormal.

[0062] After the first time point and the last time point in the collection time are removed, the time point data set is integrated, and the noise intensities corresponding to each time point in the time point data set are combined into a noise data set;

[0063] The instantaneous vibration frequency variations are combined into a vibration frequency data set after the instantaneous vibration frequency variation of the last time point is removed in the collection time according to the time point sequence;

[0064] The instantaneous ground current variations are combined into a ground current data set after the instantaneous ground current variation of the first time point is removed in the collection time according to the time point sequence;

[0065] The time point data set, the noise data set, the vibration frequency data set, and the ground current data set are used as inputs to construct a decision tree model, and the specific steps are as follows:

[0066] Data preparation: The time point data set, the noise data set, the vibration frequency data set, and the ground current data set are used as inputs, and the target line data is used as output.

[0067] Set the root node: two decision trees are constructed using the vibration frequency data set and the ground current data set, respectively, the instantaneous vibration frequency variations and the instantaneous ground current variations in the two data sets are split downward in descending order, the largest instantaneous vibration frequency variation and the largest instantaneous ground current variation are taken as the root nodes of the two decision trees, respectively, the time point corresponding to the instantaneous vibration frequency variation is selected from the time point data set and recorded, the noise intensity of the recorded time point is selected from the noise data set and recorded, and the instantaneous ground current variation is operated in the same way;

[0068] Building decision tree: take the root node as the first parent node, connect two child nodes through each parent node, select the two largest instantaneous vibration frequency changes in the vibration frequency dataset that are not nodes when building a decision tree using the vibration frequency dataset, and record the time points of the corresponding node instantaneous vibration frequency changes and the noise intensity of the corresponding time points from the time point dataset and the noise dataset, respectively. The left and right child nodes are respectively taken as two new parent nodes to derive down to each data in the dataset and integrate into the decision tree. The same processing is performed when building a decision tree using the geoelectric current dataset;

[0069] Screening time points: count the number of layers of the two decision trees, mark all nodes in the middle layer of the two decision trees, call the recorded information of the marked nodes, and select the time points corresponding to the maximum noise intensity in the recorded information of the marked nodes in the two decision trees. Compare the selected time points of the two decision trees;

[0070] Feature selection: if the selected time points of the two decision trees are the same time point, the instantaneous vibration frequency change and the instantaneous geoelectric current change corresponding to the time point are taken as the target line data; if the selected time points of the two decision trees are not the same time point, all time points between the two time points are selected in the time point dataset, and the instantaneous vibration frequency change and the instantaneous geoelectric current change of the middle time point are taken as the target line data;

[0071] Target output: output the selected target line data as output data.

[0072] It should be noted that in the decision tree, the parent node and the left and right child nodes connected are in different layers, and the left and right child nodes are in the same layer. When counting the number of layers of the two decision trees, if the number of layers of the decision tree is even, all nodes in the middle two layers of the decision tree are marked. If the selected time points of the two decision trees are not the same time point and the number of time points between the two time points in the time point dataset is even, the instantaneous vibration frequency change and the instantaneous geoelectric current change corresponding to the time point with the maximum noise intensity among the middle two time points are taken as the target line data.

[0073] Since the vibration frequency and the geoelectric current change in a single direction when the line fails, the data detected by the noise intensity is imported into the decision tree for processing and output of the target line data, which improves the accuracy of subsequent data analysis;

[0074] According to the target line data, the state weight is set, and the specific steps are as follows:

[0075] Set input: set the instantaneous vibration frequency change and the instantaneous geoelectric current change in the target line data as input variables, respectively;

[0076] Constructing a logistic regression model: generate state weight for each input variable as a logistic regression intermediate variable: where i is the serial number of the three input variables, generate state weight for the i-th input variable, e is the natural base, is the i-th logistic regression intermediate variable;

[0077] Weight setting: when the input variable is the instantaneous vibration frequency change, set the generated state weight as the vibration frequency weight; when the input variable is the instantaneous ground current change, set the generated state weight as the ground current weight;

[0078] The line data response time of the marker line includes the response time of the marker line vibration frequency and the response time of the marker line ground current; calculate the time difference between the time point corresponding to the maximum value of the marker line vibration frequency and the time point corresponding to the minimum value of the marker line vibration frequency in the collection time as the response time of the marker line vibration frequency, and calculate the response time of the marker line ground current in the same way.

[0079] Weighted sum of vibration frequency weight, ground current weight, response time of marker line vibration frequency and response time of marker line ground current to generate abnormal features of marker line.

[0080] It should be noted that the larger the abnormal features of the marker line, the more likely the marker line will be abnormal, and the more it needs to be detected in depth.

[0081] In step S3, the line anomaly detection mechanism is divided into high-level detection mechanism and low-level detection mechanism. When the abnormal features of the marker line exceed the preset abnormal threshold, the marker line is matched with the high-level detection mechanism; otherwise, the marker line is matched with the low-level detection mechanism.

[0082] It should be noted that the high level, low level and abnormal threshold can be set according to the actual situation, for example, setting double-layer detection as low level, setting five-layer detection as high level, setting abnormal threshold as 0.6, etc., which will not be analyzed here.

[0083] After the marker line is introduced into the line anomaly detection mechanism of different levels according to the abnormal features of the marker line, it is judged in each detection whether the line information of the marker line meets the detection condition. If it meets the condition, it enters the next layer, otherwise a monitoring signal is generated;

[0084] The monitoring information is the number of monitoring signals generated after the marker line passes through the line anomaly detection mechanism. The more the number of monitoring signals, the more likely the marker line is abnormal.

[0085] It should be noted that the detection conditions in the line anomaly detection mechanism are set by experimenters in the art according to actual conditions, and the line information acquisition channels of the marker lines are not unique, and will not be analyzed here.

[0086] In step S4, the more times the line is closed, the faster the corresponding line insulation layer is corroded, the more likely the corresponding line is to have an anomaly, and the more likely the switch as a whole is to have an anomaly.

[0087] After calling the closing times of each marker line in the switch, the Max-min normalization method is used for normalization processing: , is the closing time of the marker line, is the minimum value of the closing times of each marker line in the switch, is the maximum value of the closing times of each marker line in the switch, is the normalized result of the closing time of the corresponding marker line.

[0088] After multiplying the normalized result of the closing time of the marker line and the number of monitoring signals corresponding to the marker line, a marker line prompt index is generated, and all the marker line prompt indexes in the switch are combined into a prompt index data set;

[0089] The prompt index data set is input into a clustering algorithm for screening, and the specific steps are as follows:

[0090] Set the core point: arrange the data in the prompt index data set from small to large, and select the median of the data as the core point;

[0091] Core clustering: set a screening difference, start with the core point, expand to both sides according to the screening difference, mark the data before or after the core point, compare the absolute value of the difference between the marked point and the core point with the screening difference, if the absolute value of the difference is lower than the screening difference, then the marked point and the core point are merged into the screened prompt index data set, and the marked point is used as a new core point for repeated operation, until the calculation result of the absolute value of the difference between the marked point and the core point exceeds the screening difference, then stop operation.

[0092] Calculate the switch prompt index: calculate the average value of the data in the screened prompt index data set as the switch prompt index.

[0093] Compare the switch prompt index with the preset prompt threshold value, when the switch prompt index exceeds the preset prompt threshold value, generate an abnormal prompt information and send it to the user end.

[0094] It should be noted that the abnormal prompt information is a text information prompting that there is a risk, and the setting of the screening difference and the prompt threshold value is not unique, and can be set according to the specific situation such as the service life of the switch, which will not be described here.

[0095] Finally, it should be noted that the terminology used herein, such as first and second, is merely used for the convenience of the reader and is not intended to, and should not, be taken to imply or provide any actual relationship between, or order of, such entities or actions.

[0096] Also, the terms "comprise", "comprising", or any other variation thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0097] In this document, the terms "a" or "an" are used, as is common in patent documents, to include one or more than one, independent of any other instances or usages of "at least one." The term "or" is used in the inclusive sense (and not the exclusive sense), so that when used, for example, in a list of two or more items, the term "or" means "any of the items can be present, or any combination of the items can be present." The term "set" is used in the set-theoretic sense, and not the mechanical sense, unless specifically stated otherwise. The term "substantially" is used to account for variations in, for example, manufacturing tolerances, inter- and intra- individual differences, measurements of a parameter in different instances of the same thing, etc. Such terms are also used when describing various conditions or circumstances. Thus, "substantially" can be used to describe a situation where an aspect or parameter is not exactly as desired, but where the variation is not significant in terms of achieving the overall functionality of the aspect or parameter. Such terms of degree should be interpreted accordingly.

[0098] The various embodiments described in this specification can be implemented in hardware, software, or a combination thereof. The various embodiments can also employ various computer-implemented methods which can include receiving, sending, accessing, processing, or storing information, or any combination thereof. In one embodiment, software or programs that embody the methods can be stored in a memory and executed by a processor.

[0099] The above description of disclosed embodiments is not intended to be exhaustive or to be unduly limited by the details given herein. The various embodiments can be modified in various ways, and the application can be implemented in other forms without departing from the spirit or scope of the application. Therefore, the application is not limited to the disclosed embodiments, but rather the intent is to cover all modifications falling within the spirit and scope of the application.

Claims

1. An AI regulation-based switch abnormality prompting method, characterized by, Comprise the following steps: Step S1: mark each line in the switch, set the collection time, collect the line data of the marked line within the collection time; In step S1, the line data of the marked line is the vibration frequency of the marked line, the ground current of the marked line and the noise intensity of the marked line; Select a period of time as the collection time, and collect the line data of each marked line within the collection time as follows: Randomly set multiple time points within the collection time, and detect the vibration frequency, ground current and noise intensity of the marked line at each time point to obtain the vibration frequency, ground current and noise intensity of the marked line at each time point; Step S2: using decision tree method to process the line data and set multiple state weights, detecting the line data response time of the marked line, combining the state weights to generate abnormal characteristics; Step S3: according to the abnormal characteristics, the marked line is introduced into different levels of line anomaly detection mechanism, the monitoring signal generated by the line anomaly detection mechanism is monitored in real time, and the monitoring information is obtained; Step S4: call the closing times of each marked line in the switch, integrate the monitoring information, input the clustering algorithm to calculate the switch prompt index, and judge whether to generate abnormal prompt information according to the switch prompt index.

2. The switch abnormal prompt method based on AI regulation according to claim 1, wherein: In step S2, the vibration frequency of the marked line at adjacent time points within the collection time is subtracted, the vibration frequency at the former time point is subtracted from the vibration frequency at the latter time point as the instantaneous vibration frequency change at the former time point, and the subtraction is performed once for every two adjacent time points to generate multiple instantaneous vibration frequency changes within the collection time; When calculating the instantaneous ground current change, the ground current at the latter time point is subtracted from the ground current at the former time point as the instantaneous ground current change at the former time point.

3. The switch abnormal prompt method based on AI regulation according to claim 2, wherein: In step S2, before using the decision tree method, the first time point and the last time point within the collection time are excluded and integrated into a time point data set, and the noise intensity corresponding to each time point in the time point data set is combined into a noise data set; Within the collection time, the instantaneous vibration frequency change of the last time point is excluded according to the time point order, and the instantaneous vibration frequency changes are combined into a vibration frequency data set; Within the collection time, the instantaneous ground current change of the first time point is excluded according to the time point order, and the instantaneous ground current changes are combined into a ground current data set; The time point data set, the noise data set, the vibration frequency data set and the ground current data set are used as input to construct a decision tree model to screen target line data.

4. The switch abnormal prompt method based on AI regulation according to claim 3, wherein: In step S2, the steps of constructing the decision tree model to screen the target line data are as follows: Data preparation: the time point dataset, noise dataset, vibration frequency dataset, and ground current dataset are taken as inputs, and the target line data are taken as outputs; Setting root nodes: two decision trees are constructed using the vibration frequency dataset and the ground current dataset, respectively, and the maximum values in the vibration frequency dataset and the ground current dataset are selected as the root nodes of the two decision trees, respectively; Building decision trees: the root nodes are taken as parent nodes, and each parent node is connected to two child nodes. After the connection is completed, the child nodes are taken as new parent nodes for repeated operations until the number of nodes is consistent with the number of data in the vibration frequency dataset or the ground current dataset. The data in the vibration frequency dataset or the ground current dataset are filled into the nodes in descending order, and the time points corresponding to the data in each node and the noise intensity at the corresponding time points are retrieved from the time point dataset and the noise dataset and recorded; Screening time points: the number of layers of the two decision trees is counted, all nodes in the middle layer of the two decision trees are marked, the time points corresponding to the maximum noise intensity are selected in the recording information of the nodes in the two decision trees, and the selected time points in the two decision trees are compared; Feature selection: if the selected time points in the two decision trees are the same time point, the instantaneous vibration frequency change and the instantaneous ground current change at the corresponding time point are taken as the target line data; if the selected time points in the two decision trees are not the same time point, all time points between the two time points are selected in the time point dataset, and the instantaneous vibration frequency change and the instantaneous ground current change at the middle time point are taken as the target line data; Target output: the selected target line data are output as output data.

5. The AI-based switch abnormality prompting method of claim 4, wherein: in step S2, a state weight is set according to the target line data, and the specific steps are as follows: when the input variable is the instantaneous vibration frequency change, the generated state weight is set as the vibration frequency weight; and when the input variable is the instantaneous ground current change, the generated state weight is set as the ground current weight. The instantaneous vibration frequency change amount and the instantaneous ground current change amount in the target line data are set as input variables, and the input variables are taken as state weights of the logistic regression intermediate variables: where i is the serial number of the three input variables, is the state weight generated for the i th input variable, e is the natural base, is the i th logistic regression intermediate variable; 6. The AI-based switch abnormality prompting method of claim 4, wherein: in step S2, the line data response time of the marked line includes the response time of the vibration frequency of the marked line and the response time of the ground current of the marked line; the time difference between the time point corresponding to the maximum value of the vibration frequency of the marked line and the time point corresponding to the minimum value of the vibration frequency of the marked line in the collection time is calculated as the response time of the vibration frequency of the marked line, and the response time of the ground current of the marked line is calculated in the same way; the vibration frequency weight, the ground current weight, the response time of the vibration frequency of the marked line, and the response time of the ground current of the marked line are weighted and summed to generate the abnormality feature of the marked line.

7. The AI-based switch abnormality prompting method of claim 6, wherein: in step S3, when the abnormality feature of the marked line exceeds the preset abnormality threshold, the marked line is matched with the preset high-level detection mechanism; otherwise, the marked line is matched with the preset low-level detection mechanism. ​ In each layer detection, whether the line information of the marked line meets the detection condition is judged, if yes, the next layer is entered, if not, a monitoring signal is generated; The monitoring information is the number of monitoring signals generated after the marked line passes through the line anomaly detection mechanism. 8.The switch anomaly prompt method based on AI regulation according to claim 7, characterized in that: In step S4, the closing times of each marked line are standardized to generate the results of the marked line after standardization of the closing times; The results of the marked line after standardization of the closing times are multiplied with the number of monitoring signals corresponding to the marked line to generate a marked line prompt index, the marked line prompt index is combined into a prompt index data set and input into a clustering algorithm for screening: The data in the prompt index data set is arranged from small to large, the median of the data is selected as a core point, a screening difference value is set, the previous data or the next data of the core point is marked, if the absolute value of the difference between the marked point and the core point is lower than the screening difference value, the marked point and the core point are screened out, the marked point is taken as a new core point for repeated operation, and the operation is stopped until the calculation result of the absolute value of the difference between the marked point and the core point exceeds the screening difference value, and the average value of the data in the screened prompt index data set is taken as a switch prompt index; If the switch prompt index exceeds a preset prompt threshold, an anomaly prompt information is generated.

Images