PLC-IoT street lamp electric leakage detection method and system based on edge calculation

Abstract

The invention discloses a PLC-IoT street lamp electric leakage detection method and system based on edge calculation, and relates to the technical field of smart power grids and Internet of Things. According to the invention, the self-adaptive threshold adjustment algorithm is adopted to dynamically calibrate the electric leakage detection threshold, so that the problem that the traditional fixed threshold method is easy to generate false alarm or missing alarm when the environment changes is effectively solved, and the accuracy and reliability of detection are remarkably improved; through a multi-node collaborative decision algorithm and a data fusion technology, the regional electric leakage risk can be accurately judged, misjudgment caused by local interference in single-node detection is avoided, the robustness of the system is enhanced, through a parameter optimization mechanism and a machine learning regression algorithm, an environment correction coefficient and detection algorithm parameters can be dynamically adjusted, and the reliability of the system is improved. The electric leakage risk trend is predicted, dynamic and accurate evaluation of the electric leakage risk is realized, and the intelligent level and the management efficiency of the system are improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of smart grids and Internet of Things, and particularly relates to a PLC-IoT street lamp electric leakage detection method and system based on edge computing. BACKGROUND

[0002] As an important part of smart city infrastructure, the safe operation of the city street lamp system is directly related to public safety and city management efficiency. With the continuous expansion of the city scale and the continuous increase of the street lamp density, the frequent occurrence of street lamp electric leakage accidents has become a major hidden danger of city management. Therefore, an efficient and reliable electric leakage detection and protection system needs to be established. The current street lamp electric leakage detection scheme mainly relies on centralized monitoring systems and traditional grounding resistance measurement methods. These schemes have significant defects in practical application. The centralized detection architecture needs to gather all sensor data to the center server for analysis and processing, which leads to a long detection response time and cannot meet the rapid disposal demand of electric leakage accidents. The traditional grounding resistance detection method is mature in technology, but its detection accuracy is easily affected by environmental factors, and misjudgment often occurs in complex urban electromagnetic environments, making it difficult to accurately identify real electric leakage risks.

[0003] The core challenge of street lamp electric leakage detection is caused by the technical contradiction between the dynamic change characteristics of the detection features and the environmental adaptability requirements. The electric leakage features in the street lamp power supply system will present complex dynamic change rules with factors such as seasonal changes, power load fluctuations and equipment aging degrees. This change characteristic requires that the detection system must have real-time threshold adjustment capability. However, the existing detection methods use a fixed threshold judgment mechanism and cannot dynamically calibrate according to environmental parameter changes. When the humidity increases in the plum rain season, the fixed threshold setting often causes a large number of false positives or false negatives. More complexly, the independent detection decision of a single node is easily affected by local power grid noise and equipment abnormalities, and lacks effective collaborative verification mechanism to improve detection accuracy. For example, when a street lamp produces slight current anomalies due to aging, the single-node detection system will misjudge it as an electric leakage risk and trigger unnecessary power-off protection. SUMMARY

[0004] The purpose of the present application is to provide a PLC-IoT street lamp electric leakage detection method and system based on edge computing, which significantly improves the real-time performance, accuracy and intelligent level of street lamp electric leakage detection, reduces false positives and false negatives, and enhances the robustness and management efficiency of the system.

[0005] The purpose of the present application can be achieved by the following technical solutions: The present application provides a PLC-IoT street lamp electric leakage detection method based on edge computing, comprising the following steps: The electrical parameters and environmental parameters of the street lamp nodes are acquired, a multi-dimensional feature vector containing a timestamp and a node identifier is established, and a basic data source covering the operation state of the street lamp power supply system is formed; According to the change trend of the environmental parameters in the multi-dimensional feature vector, an adaptive threshold adjustment algorithm is used to dynamically calibrate the preset leakage detection threshold, and a dynamic detection threshold suitable for the current environmental conditions is obtained. Through the dynamic detection threshold, the electrical parameters of each street lamp node are abnormally identified, the suspected leakage state is marked, and a risk identifier containing the abnormal type and the abnormal degree is generated; The electrical parameters and operation states of the adjacent nodes within a preset range around the suspected leakage node are acquired, a multi-node collaborative decision algorithm is used to analyze the parameter distribution characteristics of the adjacent nodes, and regional leakage risks or equipment aging and local interference are confirmed; According to the analysis result of the collaborative decision algorithm, the single-node detection result and the multi-node verification information are integrated through data fusion technology, the comprehensive confidence parameter is calculated, and the leakage alarm signal is generated or recorded as a monitoring event according to the confidence; A dynamic risk assessment algorithm is used to classify the confirmed leakage risks, a risk score is calculated according to the leakage current size, duration and influence range, a power-off protection instruction is triggered or maintenance reminder information is sent; Through the parameter optimization mechanism, the environmental correction coefficient in the adaptive threshold adjustment algorithm is continuously updated, the algorithm parameters are adjusted according to the verification result of the historical leakage events, and a continuously optimized detection decision mechanism is formed.

[0006] Further, a basic data source covering the operation state of the street lamp power supply system is formed, specifically including: Through the sensor array, current parameters, voltage parameters, ground resistance, temperature data, humidity data, and rainfall are collected in real time, combined with timestamps and node identifiers, and then multi-dimensional feature vectors containing timestamps and node identifiers are extracted; The principal component analysis algorithm is used to reduce the dimension of the feature vector, if any parameter in the feature vector exceeds the preset threshold range, the abnormal state is judged through the abnormal detection algorithm, the abnormal node list is obtained, the operation state and environmental parameters of the abnormal node are acquired, the K-means clustering algorithm is used to classify the abnormal nodes, and the node abnormal category is obtained; The timestamps and node identifiers are extracted from the node abnormal category, combined with the historical data set, the occurrence frequency and time distribution of each type of abnormality are calculated, the abnormal mode characteristics are obtained, the time series analysis algorithm is used to predict the abnormal probability of each node in the future time period, and the predicted abnormal distribution is obtained; According to the predicted abnormal distribution, the priority ranking of each node is generated, the node priority list for scheduling maintenance is output, and a basic data source covering the operation state of the street lamp power supply system is formed.

[0007] Further, according to the environmental parameter change trend in the multi-dimensional feature vector, an adaptive threshold adjustment algorithm is used to dynamically calibrate the preset leakage detection threshold, specifically including: The humidity parameter and temperature parameter are collected from the sensor to determine the real-time environmental condition of the environmental parameter change, and the change trend is analyzed according to the environmental condition. The time series analysis algorithm is used to obtain the environmental parameter change trend; If the humidity parameter exceeds the preset humidity reference value, the resistance threshold is reduced by the linear interpolation algorithm to obtain the adjusted resistance threshold, and the adjusted current threshold is obtained; An adaptive threshold adjustment algorithm is used to fuse the environmental parameter change trend to determine the dynamic detection threshold that adapts to the current environmental condition. By comparing the dynamic detection threshold with the real-time leakage data, it is determined whether there is a leakage anomaly, and the leakage detection result is obtained and used to update the environmental condition database to record the dynamic detection threshold and the leakage state, and obtain the environmental adaptive feedback data.

[0008] Further, the electrical parameters of each street lamp node are identified for abnormalities, specifically including: The electrical parameter data, including node current and grounding resistance, are obtained from each street lamp node through the sensor and stored as a time series data set. The time series data set is processed using a sliding window method to calculate the dynamic change trend of each node current and grounding resistance to obtain dynamic change characteristics; According to the dynamic change characteristics, a k-means algorithm is used to perform cluster analysis on the electrical parameters of each street lamp node to determine the dynamic current threshold and dynamic resistance threshold. If the node current value exceeds the dynamic current threshold, it is marked as a current anomaly; If the grounding resistance value is lower than the dynamic resistance threshold, it is marked as a resistance anomaly, and a suspected leakage state is generated. The suspected leakage state is classified by a decision tree algorithm, and the abnormal type and abnormal degree are determined in combination with the dynamic change characteristics; According to the abnormal type and abnormal degree, a risk identifier containing a timestamp, node identifier, abnormal type and abnormal degree is generated and stored as structured data. A time series analysis method is used to detect the periodic change trend of the risk identifier to obtain the risk identifier containing the abnormal type and abnormal degree.

[0009] Further, a multi-node collaborative decision algorithm is used to analyze the parameter distribution characteristics of adjacent nodes, specifically including: The electrical parameters and operating states of the suspected leakage node and its adjacent nodes within the preset range are obtained. Voltage, current and power data are collected through a sensor network, and then a multi-node collaborative decision algorithm is used for analysis to calculate the distribution characteristics of the electrical parameters of adjacent nodes and determine the parameter distribution pattern; If more than a preset proportion of the electrical parameters of the nodes in the adjacent nodes deviate from the normal distribution, the abnormal node group is identified through a clustering algorithm, and the regional electric leakage risk is determined. The abnormal persistence is determined. According to the abnormal persistence result, if the single node abnormal duration exceeds the preset threshold, the device aging is determined through the device aging model analysis. If the single node abnormal duration does not exceed the preset threshold, the local interference is determined through the interference detection algorithm analysis of the environmental data.

[0010] Further, according to the confidence degree, an electric leakage alarm signal is generated or recorded as a monitoring event, specifically including: Real-time acquisition of electrical parameters by sensors deployed at each street lamp node and abnormality detection to obtain single node detection results, acquisition of verification data from multiple node sensors, alignment of node data through time synchronization technology, and obtaining of multi-node verification information; Using a weighted average algorithm to fuse single node detection results and multi-node verification information, calculating a comprehensive confidence parameter, and if the comprehensive confidence parameter exceeds a preset confidence threshold, generating an electric leakage alarm signal and storing it in an alarm database; If the comprehensive confidence parameter is lower than the preset confidence threshold, the data is recorded as a monitoring event and stored in an event database. Through time series analysis algorithm, the data in the monitoring event database is analyzed for trend, the abnormal mode is determined, and the subsequent observation strategy is updated.

[0011] Further, a dynamic risk assessment algorithm is used to classify the confirmed electric leakage risk, specifically including: Real-time acquisition of electric leakage current, duration and influence range electrical system parameters by sensors to form an original monitoring data set, pre-processing of the collected data to remove noise and outliers, using a weighted calculation method to assign different weights according to the electric leakage current size, duration and influence range, calculating the risk score through a weighted formula to obtain a quantitative risk value; According to the comparison of the quantitative risk value and the preset threshold, the risk level is divided. If the quantitative risk value exceeds the preset emergency threshold, the power-off protection instruction is triggered. If it is within the preset warning range, the maintenance reminder information is sent; According to the power-off execution state or the reminder sending state, the risk score and the processing result are recorded to the log database. Through analysis of the log data, a machine learning regression algorithm is used to predict the electric leakage risk trend, and the weight parameters of the risk assessment algorithm are adjusted according to the trend, which is used to optimize the evaluation model.

[0012] Further, the environmental correction coefficient in the self-adaptive threshold adjustment algorithm is continuously updated through a parameter optimization mechanism, specifically including: The initial value of the environmental correction coefficient is obtained through historical electric leakage event data, the false alarm rate under each environmental condition is calculated by statistical analysis method, the baseline value of the environmental correction coefficient is determined, if the false alarm rate is higher than the preset threshold, the sensitivity coefficient of the corresponding environmental condition is reduced by gradient descent algorithm, and the adjusted sensitivity coefficient is obtained; The adaptive threshold adjustment algorithm is updated according to the adjusted sensitivity coefficient, a new environmental correction coefficient is obtained, an updated detection decision model is generated, the updated detection decision model is verified through real-time collected electric leakage data, the model false alarm rate is evaluated by cross-validation method, and the model performance is judged; If the model false alarm rate is lower than the preset threshold, the environmental condition characteristics are extracted through historical data analysis, the correlation coefficient of the environmental condition and the false alarm rate is obtained, the weight of the parameter optimization mechanism is adjusted according to the correlation coefficient, the optimized environmental correction coefficient is obtained, and the detection decision mechanism is updated to generate a continuously optimized adaptive threshold adjustment algorithm.

[0013] Further, after updating the environmental correction coefficient in the adaptive threshold adjustment algorithm, it further includes: establishing a state monitoring database based on the collaborative verification results of multiple nodes, storing the abnormal patterns and disposal records of each node, and identifying repetitive abnormal characteristics through pattern matching technology.

[0014] The application provides a PLC-IoT street lamp electric leakage detection system based on edge computing, which is used to realize a PLC-IoT street lamp electric leakage detection method based on edge computing, comprising: A basic data source construction module acquires electrical parameters and environmental parameters of street lamp nodes in real time through a sensor array, combines time stamps and node identifiers to form a multi-dimensional feature vector, and constructs a basic data source covering the running state of the street lamp power supply system; A dynamic threshold calibration module dynamically calibrates the preset electric leakage detection threshold value by using an adaptive threshold adjustment algorithm according to the environmental parameter change trend in the multi-dimensional feature vector, adjusts the resistance and current threshold values respectively by linear interpolation and weighted average algorithm according to the change of humidity and temperature parameters, and determines the dynamic detection threshold value suitable for the current environmental condition; An abnormality identification and risk identification module obtains electrical parameter data from each street lamp node, stores it as a time series data set, processes and calculates the dynamic change trend by using a sliding window method, determines the dynamic threshold value by using k-means algorithm clustering analysis, marks the current and resistance abnormalities, generates a suspected electric leakage state, classifies and determines the abnormal type and degree by using a decision tree algorithm, and generates a risk identification containing time stamp, node identifier and other information; A multi-node collaborative decision-making module obtains electrical parameters and operating states of the suspected electric leakage node and its adjacent nodes, analyzes parameter distribution characteristics of the adjacent nodes through a multi-node collaborative decision-making algorithm, and if electrical parameters of more than a preset proportion of the adjacent nodes deviate from normal distribution, it is determined that there is a regional electric leakage risk, and it is further determined that the equipment is aging or there is local interference; A confidence assessment and decision-making module fuses single-node detection results and multi-node verification information, calculates a comprehensive confidence parameter using a weighted average algorithm, generates an electric leakage alarm signal and stores it in an alarm database if the comprehensive confidence parameter exceeds a preset confidence threshold, and records it as a monitoring event and stores it in an event database if it is below the threshold, and updates subsequent observation strategies through trend analysis; A risk level division module calculates a risk score according to each parameter using a weighted calculation method, divides the risk level, triggers a power-off protection instruction if the risk score exceeds an emergency threshold, sends a maintenance reminder if it is in a warning range, predicts the risk trend using a machine learning regression algorithm by analyzing log data, and optimizes the evaluation model; A parameter optimization and continuous updating module obtains an initial value of an environmental correction coefficient based on historical electric leakage event data, calculates a false alarm rate under each environmental condition to determine a baseline value, adjusts a sensitivity coefficient through a gradient descent algorithm, updates an adaptive threshold adjustment algorithm, obtains a new environmental correction coefficient, verifies and evaluates the model performance, extracts environmental condition features if the model false alarm rate is below a threshold, adjusts the parameter optimization mechanism weight, and continuously optimizes the detection decision-making mechanism.

[0015] The present application has the following advantages: The present application directly completes data processing and analysis on edge devices near street light nodes through edge computing technology, greatly shortens data transmission and processing time, thereby realizing rapid response to electric leakage accidents and meeting the demand for rapid disposal, and at the same time, using an adaptive threshold adjustment algorithm, dynamically calibrating the electric leakage detection threshold according to real-time environmental parameters (such as humidity and temperature), effectively solving the problem of false positives or false negatives of traditional fixed threshold methods when the environment changes, significantly improving the accuracy and reliability of detection; Through a multi-node collaborative decision-making algorithm, the electrical parameters and operating states of the suspected electric leakage node and its adjacent nodes are analyzed, and a clustering algorithm is used to identify abnormal node groups, so as to accurately determine the regional electric leakage risk and avoid false positives caused by local interference in single-node detection, thereby enhancing the robustness of the system, in addition, through data fusion technology, single-node detection results and multi-node verification information are integrated, a comprehensive confidence parameter is calculated, and the electric leakage risk is further confirmed, realizing the transition from single-point detection to multi-point collaborative verification, and significantly improving the credibility of the detection results; Through the parameter optimization mechanism, the environmental correction coefficient and the detection algorithm parameters are dynamically adjusted according to historical electric leakage event data and false alarm rate analysis, so that the system can continuously adapt to new operating environments and conditions, reduce false positives and false negatives, at the same time, the machine learning regression algorithm is used to predict the electric leakage risk trend, and the weight parameters of the risk assessment model are adjusted according to the trend, realizing the dynamic and accurate assessment of the electric leakage risk, and improving the intelligent level and management efficiency of the system. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to better understand and implement, the technical solutions of the present application are described in detail below in conjunction with the drawings.

[0017] Figure 1 A flowchart of a PLC-IoT street lamp electric leakage detection method based on edge computing provided for Embodiment 1 of the present application; Figure 2 A flowchart of a PLC-IoT street lamp electric leakage detection method based on edge computing provided for Embodiment 1 of the present application; Figure 3 A flowchart of a PLC-IoT street lamp electric leakage detection method based on edge computing provided for Embodiment 1 of the present application; Figure 4 A structural diagram of a PLC-IoT street lamp electric leakage detection system based on edge computing provided for Embodiment 2 of the present application. DETAILED DESCRIPTION

[0018] In order to better understand and implement, the technical solutions of the present application are described in detail below in conjunction with the drawings.

[0019] The terms used in the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein means and includes any or all possible combinations of one or more associated listed items.

[0020] The specific embodiments, features and effects of the present application according to the preferred embodiments are described in detail below in conjunction with the drawings.

[0021] Embodiment 1 Please refer to Figures 1-3 The embodiment provides a PLC-IoT street lamp electric leakage detection method based on edge computing, comprising the following steps: Among them, PLC-IoT street lamp is a kind of intelligent street lamp system combined with power line carrier communication (PLC) and Internet of Things (IoT) technology, realizes communication and control between devices through power line transmission data, at the same time, uses Internet of Things technology to realize remote monitoring, fault diagnosis and intelligent management, so as to improve the operation efficiency and maintenance convenience of street lamp system.

[0022] S1, the current, voltage, grounding resistance and other electrical parameters of street lamp node and temperature, humidity, rainfall and other environmental parameters are obtained, multi-dimensional monitoring data are collected in real time through sensor array, multi-dimensional feature vector containing time stamp and node identification is established, and basic data source covering the running state of street lamp power supply system is formed.

[0023] Further, the basic data source covering the running state of street lamp power supply system is formed, specifically comprising: Through the sensor array, current parameters, voltage parameters, grounding resistance, temperature data, humidity data and rainfall are collected in real time, combined with time stamp and node identification, and then the multi-dimensional feature vector containing time stamp and node identification is extracted; The principal component analysis algorithm is adopted to reduce the dimension of the feature vector, if any parameter in the feature vector exceeds the preset threshold range, the abnormal state is judged through the abnormal detection algorithm, the abnormal node list is obtained, the running state and environmental parameters of the abnormal node are obtained, the K-means clustering algorithm is adopted to classify the abnormal nodes, and the node abnormal category is obtained; The time stamp and node identification are extracted from the node abnormal category, combined with the historical data set, the occurrence frequency and time distribution of each category of abnormality are calculated, the abnormal mode characteristics are obtained, the time series analysis algorithm is adopted to predict the abnormal probability of each node in the future time period, and the predicted abnormal distribution is obtained; According to the predicted abnormal distribution, the priority of each node is generated, the node priority list for scheduling maintenance is output, and the basic data source covering the running state of street lamp power supply system is formed.

[0024] Specifically, by collecting the electrical parameters and environmental parameters of the street lamp nodes in real time, a multi-dimensional feature vector is constructed and a basic data source is formed, the principal component analysis, anomaly detection and clustering algorithm are used to accurately identify and classify the abnormal nodes, the historical data are used to predict the abnormal probability and generate a maintenance priority list, so as to realize the comprehensive monitoring and accurate maintenance of the street lamp power supply system. This method effectively solves the problems of incomplete data collection, inaccurate anomaly detection and lack of scientific basis for maintenance decision in traditional detection methods, significantly improves the operation efficiency and maintenance pertinence of the street lamp system, and enhances the public safety and urban management efficiency.

[0025] S2, according to the environmental parameter change trend in the multi-dimensional feature vector, an adaptive threshold adjustment algorithm is used to dynamically calibrate the preset leakage detection threshold, if the humidity parameter exceeds the preset humidity reference value, the resistance threshold is reduced, if the temperature parameter is lower than the preset temperature reference value, the current threshold is increased, and the dynamic detection threshold adapting to the current environmental condition is obtained.

[0026] Further, according to the environmental parameter change trend in the multi-dimensional feature vector, an adaptive threshold adjustment algorithm is used to dynamically calibrate the preset leakage detection threshold, specifically including: S21, the humidity parameter and the temperature parameter are collected from the sensor, the real-time environmental condition of the environmental parameter change is determined, the change trend is analyzed according to the environmental condition, and the time series analysis algorithm is used to obtain the environmental parameter change trend; S22, if the humidity parameter exceeds the preset humidity reference value, the resistance threshold is reduced by a linear interpolation algorithm to obtain the adjusted resistance threshold, if the temperature parameter is lower than the preset temperature reference value, the current threshold is increased by a weighted average algorithm to obtain the adjusted current threshold; S23, an adaptive threshold adjustment algorithm is used to determine the dynamic detection threshold adapting to the current environmental condition by fusing the environmental parameter change trend, the leakage detection result is obtained by comparing the dynamic detection threshold with the real-time leakage data to judge whether there is leakage anomaly, and is used to update the environmental condition database to record the dynamic detection threshold and the leakage state, and obtain the environmental adaptive feedback data.

[0027] Specifically, by using the adaptive threshold adjustment algorithm, the leakage detection threshold is dynamically calibrated according to the environmental parameter change trend in the multi-dimensional feature vector, which solves the problem that the traditional fixed threshold detection method is prone to false alarm or missed alarm when the environment changes, not only improves the accuracy and reliability of the leakage detection, but also realizes the accurate judgment of the leakage anomaly by comparing the dynamic detection threshold with the real-time leakage data, and continuously optimizes the detection performance through the environmental adaptive feedback data, enhances the intelligent level and environmental adaptability of the system.

[0028] S3, the electrical parameters of each street lamp node are abnormally identified by dynamic detection threshold, if the node current value exceeds the corresponding dynamic current threshold or the grounding resistance value is lower than the corresponding dynamic resistance threshold, the node is marked as a suspected electric leakage state, and a risk identification containing the abnormal type and the abnormal degree is generated.

[0029] Further, the electrical parameters of each street lamp node are abnormally identified, specifically including: The electrical parameter data including node current and grounding resistance are obtained from each street lamp node by a sensor, stored as a time series data set, and the time series data set is processed by a sliding window method to calculate the dynamic change trend of each node current and grounding resistance to obtain dynamic change characteristics; According to the dynamic change characteristics, the electrical parameters of each street lamp node are analyzed by k-means algorithm, the dynamic current threshold and the dynamic resistance threshold are determined, if the node current value exceeds the dynamic current threshold, it is marked as current abnormality; If the grounding resistance value is lower than the dynamic resistance threshold, it is marked as resistance abnormality, a suspected electric leakage state is generated, the suspected electric leakage state is classified by a decision tree algorithm, and the abnormal type and the abnormal degree are determined according to the dynamic change characteristics; According to the abnormal type and the abnormal degree, a risk identification containing a time stamp, a node identification, an abnormal type and an abnormal degree is generated, stored as structured data, and a time series analysis method is used to detect the periodic change trend of the risk identification to obtain the risk identification containing the abnormal type and the abnormal degree.

[0030] Specifically, the electrical parameters of the street lamp node are abnormally identified by dynamic detection threshold, the time series data is processed by a sliding window method, the dynamic threshold is determined by k-means clustering analysis, the abnormal node is accurately marked and the risk identification is generated, the suspected electric leakage state is further classified by a decision tree algorithm, and the periodic change of the risk identification is detected by time series analysis to provide a basis for maintenance. This method effectively improves the accuracy of electric leakage detection, solves the problem of inaccurate detection of traditional fixed threshold, and enhances the intelligent level and maintenance efficiency of the system.

[0031] S4, the electrical parameters and operating states of the adjacent nodes within a preset range around the suspected electric leakage node are obtained, the parameter distribution characteristics of the adjacent nodes are analyzed by a multi-node collaborative decision algorithm, if more than a preset proportion of nodes in the adjacent nodes appear similar abnormalities, the regional electric leakage risk is confirmed, if only a single node is abnormal, it is judged as equipment aging or local interference.

[0032] Further, the parameter distribution characteristics of the adjacent nodes are analyzed by a multi-node collaborative decision algorithm, specifically including: The electrical parameters and operating states of the suspected electric leakage node and its adjacent nodes within a preset range are acquired, voltage, current and power data are collected through a sensor network, and a multi-node collaborative decision algorithm is used for analysis to calculate the distribution characteristics of the electrical parameters of the adjacent nodes and determine the parameter distribution mode; If the electrical parameters of more than a preset proportion of nodes in the adjacent nodes deviate from the normal distribution, an abnormal node group is identified through a clustering algorithm, and it is determined to be a regional electric leakage risk. If only a single node electrical parameter deviates from the normal distribution, the historical data trend of the node is detected through a time series analysis algorithm to determine the abnormal persistence. According to the abnormal persistence result, if the single node abnormal duration exceeds a preset threshold, a device aging model analysis is performed to determine device aging. If the single node abnormal duration does not exceed the preset threshold, environmental data is analyzed through an interference detection algorithm to determine local interference.

[0033] Specifically, the electrical parameters and operating states of the suspected electric leakage node and its surrounding adjacent nodes are accurately analyzed through a multi-node collaborative decision algorithm, thereby effectively distinguishing regional electric leakage risks from single node abnormalities. This method significantly improves the accuracy and reliability of electric leakage detection, reduces false positives and false negatives caused by local interference or device aging, enhances the robustness and adaptability of the system, and provides a more scientific and efficient electric leakage detection solution for city management.

[0034] S5, according to the analysis result of the collaborative decision algorithm, the single node detection result and the multi-node verification information are integrated through data fusion technology, the comprehensive confidence parameter is calculated, and the leakage alarm signal or the monitoring event is recorded according to the confidence.

[0035] The comprehensive confidence parameter is represented as: , represents the abnormal confidence detected by a single node. The closer the value is to 1, the more confident the node is that there is an abnormality. The closer the value is to 0, the less abnormal or the lower the abnormal confidence. represents the comprehensive confidence after verification of multiple adjacent nodes. The closer the value is to 1, the more nodes consistently support the abnormality. The closer the value is to 0, the less support for the abnormality. represents the weight of the single node detection result. represents the weight of the multi-node verification information.

[0036] Further, the leakage alarm signal or the monitoring event is generated according to the confidence, specifically including: S51, real-time acquisition of electrical parameters (such as current, voltage, grounding resistance, etc.) through sensors deployed at each street lamp node and abnormality detection to obtain single node detection results, acquisition of verification data from multi-node sensors, alignment of node data through time synchronization technology to obtain multi-node verification information, S52, the single-node detection result and the multi-node verification information are fused by using a weighted average algorithm, a comprehensive confidence parameter is calculated, and if the comprehensive confidence parameter exceeds a preset confidence threshold, an electric leakage alarm signal is generated and stored in an alarm database; S53, if the comprehensive confidence parameter is lower than the preset confidence threshold, the data is recorded as a monitoring event and stored in an event database, trend analysis is performed on the data in the monitoring event database by using a time series analysis algorithm, an abnormal mode is determined, and a subsequent observation strategy is updated.

[0037] Specifically, the single-node detection result and the multi-node verification information are integrated by using a data fusion technology, the comprehensive confidence parameter is calculated, the electric leakage alarm signal is accurately generated or the monitoring event is recorded, the accuracy and reliability of the electric leakage detection are significantly improved, the false positives and false negatives in the traditional single-node detection are effectively reduced, and the credibility of the detection result is enhanced through the multi-node collaborative verification. Meanwhile, the trend analysis is performed on the monitoring event by using the time series analysis, the potential electric leakage risk can be warned in advance and the subsequent observation strategy can be optimized, and the intelligent level and the dynamic adaptability of the system are further improved, thereby providing a powerful guarantee for the safe operation of the street lamp system.

[0038] S6, a dynamic risk assessment algorithm is used to classify the confirmed electric leakage risk, a risk score is calculated according to the electric leakage current size, duration and influence range, a power-off protection instruction is triggered if the risk score exceeds an emergency disposal threshold, and a maintenance reminder information is sent if it is in a warning range.

[0039] Further, the dynamic risk assessment algorithm is used to classify the confirmed electric leakage risk, specifically including: The electric leakage current, duration and influence range electrical system parameters are collected in real time by using a sensor to form an original monitoring data set, the collected data is preprocessed to remove noise and outliers, a weighted calculation method is used, different weights are given according to the electric leakage current size, duration and influence range, the risk score is calculated by using a weighted formula, and a quantitative risk value is obtained; According to the comparison of the quantitative risk value and the preset threshold, the risk level is divided, if the quantitative risk value exceeds the preset emergency threshold, the power-off protection instruction is triggered, and if it is in the preset warning range, the maintenance reminder information is sent; According to the power-off execution state or the reminder sending state, the risk score and the processing result are recorded to a log database, the leakage risk trend is predicted by using a machine learning regression algorithm through analyzing the log data, and the weight parameters of the risk assessment algorithm are adjusted according to the trend, which is used to optimize the evaluation model, so as to realize the dynamic and accurate evaluation and effective management of the electric leakage risk.

[0040] The risk score is expressed as: , Indicates the magnitude of leakage current. Indicates duration, Indicates the scope of influence. The weighting of leakage current. Weights representing duration, The weight represents the scope of influence.

[0041] Specifically, by employing a dynamic risk assessment algorithm, this invention can calculate a risk score based on the magnitude, duration, and impact range of the leakage current, and classify the leakage risk accordingly. This method effectively solves the problems of inaccurate risk assessment and inability to dynamically adjust in traditional leakage detection. Through weighted calculation methods and machine learning regression algorithms, the system can dynamically adjust weight parameters, accurately predict leakage risk trends, and trigger power outage protection commands or send maintenance reminders based on the risk level. This process not only improves the accuracy and timeliness of leakage risk assessment but also optimizes the assessment model, enabling dynamic management and effective control of leakage risks, and providing strong protection for the safe operation of street lighting systems.

[0042] S7. The environmental correction coefficient in the adaptive threshold adjustment algorithm is continuously updated through the parameter optimization mechanism. The algorithm parameters are adjusted according to the verification results of historical leakage events. If historical data shows that the false alarm rate is high under certain environmental conditions, the corresponding sensitivity coefficient is reduced, thus forming a continuously optimized detection decision mechanism.

[0043] Furthermore, the environmental correction coefficients in the adaptive threshold adjustment algorithm are continuously updated through a parameter optimization mechanism, specifically including: The initial value of the environmental correction coefficient is obtained by using historical leakage event data. The false alarm rate under each environmental condition is calculated by using statistical analysis methods to determine the baseline value of the environmental correction coefficient. If the false alarm rate is higher than the preset threshold, the sensitivity coefficient of the corresponding environmental condition is reduced by using the gradient descent algorithm to obtain the adjusted sensitivity coefficient. The adaptive threshold adjustment algorithm is updated based on the adjusted sensitivity coefficient to obtain a new environmental correction coefficient, and an updated detection decision model is generated. The updated detection decision model is verified by real-time collected leakage current data, and the model false alarm rate is evaluated by cross-validation method to determine the model performance. If the model's false alarm rate is lower than a preset threshold, environmental condition features are extracted through historical data analysis to obtain the correlation coefficient between environmental conditions and the false alarm rate. The weights of the parameter optimization mechanism are adjusted based on the correlation coefficient to obtain the optimized environmental correction coefficient, which is then used to update the detection decision mechanism and generate a continuously optimized adaptive threshold adjustment algorithm.

[0044] Specifically, by dynamically adjusting the environmental correction coefficient in the adaptive threshold algorithm through the parameter optimization mechanism, the sensitivity of the high false alarm rate environment condition is reduced based on historical data, the detection decision mechanism is continuously optimized, the problem of high false alarm rate of traditional fixed threshold detection method under different environmental conditions is solved, the accuracy and reliability of the leakage detection are improved, and the system intelligence and environmental adaptability are enhanced.

[0045] Further, after updating the environmental correction coefficient in the adaptive threshold adjustment algorithm, it further includes: establishing a state monitoring database based on the collaborative verification results of multiple nodes, storing the abnormal patterns and disposal records of each node, identifying repetitive abnormal features through pattern matching technology, and providing data support for subsequent preventive maintenance and equipment replacement decisions.

[0046] Further, the repetitive abnormal features are identified through the pattern matching technology, specifically including: Obtain the node verification results of the multi-node collaborative verification, store them to the state monitoring database, generate a data set containing abnormal patterns and disposal records, analyze the abnormal patterns in the initial data set using pattern matching technology, identify repetitive abnormal features, obtain an abnormal feature set, and if the repetitive abnormal features in the abnormal feature set exceed a preset threshold, group the abnormal feature set through a clustering algorithm to determine the abnormal pattern category. According to the abnormal pattern category, extract the corresponding disposal records from the state monitoring database to generate an association mapping of abnormal patterns and disposal records, analyze the occurrence frequency of repetitive abnormal features and the time sequence of disposal records through the association mapping, and obtain the trend features of the abnormal patterns. Classify the trend features using a decision tree algorithm to determine whether to trigger a preventive maintenance or equipment replacement decision, generate maintenance decision support data, store the maintenance decision support data in the state monitoring database, update the abnormal patterns and disposal records, and form a closed-loop data processing flow.

[0047] Specifically, by establishing a state monitoring database based on the collaborative verification results of multiple nodes and using pattern matching technology to identify repetitive abnormal features, strong data support is provided for preventive maintenance and equipment replacement decisions, solving the problem of incomplete identification of abnormal features and lack of effective preventive maintenance decision support in traditional detection methods. It can accurately determine whether to trigger preventive maintenance or equipment replacement, so as to take measures in advance, avoid potential leakage risks and equipment failures, improve the operation safety and maintenance efficiency of the street lamp system, and reduce the operation and maintenance cost.

[0048] Embodiment 2 Please refer to Figure 4 The embodiment provides a PLC-IoT street lamp leakage detection system based on edge computing, which is used for realizing a PLC-IoT street lamp leakage detection method based on edge computing. The basic data source construction module collects electrical parameters (current, voltage, ground resistance) and environmental parameters (temperature, humidity, rainfall) of the street lamp node in real time through the sensor array, combines the time stamp and node identifier to form a multi-dimensional feature vector, and constructs a basic data source covering the operation state of the street lamp power supply system to provide comprehensive data support for subsequent analysis; The dynamic threshold calibration module dynamically califies the preset leakage detection threshold according to the environmental parameter change trend in the multi-dimensional feature vector using an adaptive threshold adjustment algorithm, adjusts the resistance and current thresholds according to the changes in humidity and temperature parameters through linear interpolation and weighted average algorithm, determines the dynamic detection threshold that adapts to the current environmental conditions, and improves the accuracy of leakage detection; The abnormality identification and risk identification module obtains electrical parameter data from each street lamp node, stores it as a time series data set, processes and calculates the dynamic change trend using a sliding window method, determines the dynamic threshold using k-means algorithm clustering analysis, marks current and resistance abnormalities, generates suspected leakage states, classifies abnormal types and degrees through decision tree algorithm, and generates risk identification containing time stamp, node identifier and other information; The multi-node collaborative decision-making module obtains electrical parameters and operating states of suspected leakage nodes and their adjacent nodes, analyzes the parameter distribution characteristics of adjacent nodes through a multi-node collaborative decision-making algorithm, and if the electrical parameters of more than a preset proportion of adjacent nodes deviate from the normal distribution, it is judged as a regional leakage risk; if only a single node is abnormal, further analyze its abnormal persistence, and judge it as equipment aging or local interference; The confidence evaluation and decision-making module fuses single-node detection results and multi-node verification information, calculates the comprehensive confidence parameter using a weighted average algorithm, generates a leakage alarm signal if the comprehensive confidence parameter exceeds a preset confidence threshold, and stores it in the alarm database; if it is below the threshold, it is recorded as a monitoring event and stored in the event database, and the subsequent observation strategy is updated through trend analysis; The risk level division module collects leakage current, duration and influence range parameters in real time, forms an original monitoring data set and pre-processes it, calculates the risk score according to each parameter using a weighted calculation method, and divides the risk level. If the risk score exceeds the emergency threshold, the power-off protection instruction is triggered; if it is in the warning range, the maintenance reminder information is sent, the risk trend is predicted using machine learning regression algorithm by analyzing log data, and the evaluation model is optimized; The parameter optimization and continuous updating module obtains initial values of environment correction coefficients based on historical electric leakage event data, calculates false alarm rates under each environment condition to determine baseline values, adjusts sensitivity coefficients through a gradient descent algorithm if the false alarm rate is higher than a preset threshold, updates the adaptive threshold adjustment algorithm, obtains new environment correction coefficients, verifies and evaluates the model performance, extracts environment condition features if the model false alarm rate is lower than the threshold, adjusts the parameter optimization mechanism weight, and continuously optimizes the detection decision mechanism.

[0049] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any modification, change, and modification of the above embodiments, which does not depart from the technical solution of the present application, is still within the scope of the present application.

Claims

1. An edge computing-based PLC-IoT electric leakage detection method for street lamps, characterized in that: The method comprises the following steps: Obtain the electrical parameters and environmental parameters of the street lamp nodes, establish a multi-dimensional feature vector containing a timestamp and a node identifier, and form a basic data source covering the operating state of the street lamp power supply system; According to the change trend of the environmental parameters in the multi-dimensional feature vector, a self-adaptive threshold adjustment algorithm is used to dynamically calibrate the preset leakage detection threshold, and a dynamic detection threshold that adapts to the current environmental conditions is obtained; Through the dynamic detection threshold, the electrical parameters of each street lamp node are abnormally identified, the suspected leakage state is marked, and a risk identifier containing the abnormal type and the abnormal degree is generated; Obtain the electrical parameters and operating state of the adjacent nodes within a preset range around the suspected leakage node, analyze the parameter distribution characteristics of the adjacent nodes using a multi-node collaborative decision algorithm, confirm the regional leakage risk, or judge that the device is aging or there is local interference; According to the analysis result of the collaborative decision algorithm, the single-node detection result and the multi-node verification information are integrated through a data fusion technology, a comprehensive confidence parameter is calculated, and a leakage alarm signal is generated or a monitoring event is recorded according to the confidence; A dynamic risk assessment algorithm is used to classify the confirmed leakage risk, a risk score is calculated according to the leakage current size, duration and influence range, and a power-off protection instruction or maintenance reminder information is sent; Through a parameter optimization mechanism, the environmental correction coefficient in the self-adaptive threshold adjustment algorithm is continuously updated, the algorithm parameters are adjusted according to the verification result of the historical leakage event, and a continuously optimized detection decision mechanism is formed.

2. The PLC-IoT based edge computing based electric leakage detection method for street light as claimed in claim 1 wherein: The basic data source covering the operating state of the street lamp power supply system is formed, specifically including: Real-time acquisition of current parameters, voltage parameters, grounding resistance, temperature data, humidity data, and rainfall through a sensor array, combination of a timestamp and a node identifier, and extraction of a multi-dimensional feature vector containing a timestamp and a node identifier; Principal component analysis algorithm is used to reduce the dimension of the feature vector to obtain an abnormal node list, obtain the operating state and environmental parameters of the abnormal nodes, and use K-means clustering algorithm to classify the abnormal nodes to obtain node abnormal categories; From the node abnormal categories, extract the timestamp and node identifier, combine the historical data set, calculate the occurrence frequency and time distribution of each type of abnormality, obtain the abnormal mode characteristics, and use time series analysis algorithm to predict the abnormal probability of each node in the future time period to obtain the predicted abnormal distribution; According to the predicted abnormal distribution, the priority of each node is generated, the node priority list for scheduling maintenance is output, and the basic data source covering the operating state of the street lamp power supply system is formed. 3.The PLC-IoT based edge computing based electric leakage detection method for street light according to claim 1, wherein: According to the change trend of the environmental parameters in the multi-dimensional feature vector, a self-adaptive threshold adjustment algorithm is used to dynamically calibrate the preset leakage detection threshold, specifically including: From the sensor, humidity parameters and temperature parameters are collected to determine the real-time environmental conditions of the change in environmental parameters, the change trend is analyzed according to the environmental conditions, and a time series analysis algorithm is used to obtain the change trend of the environmental parameters; If the humidity parameter exceeds the preset humidity reference value, the resistance threshold is reduced through a linear interpolation algorithm to obtain the adjusted resistance threshold, and if the temperature parameter is lower than the preset temperature reference value, the current threshold is increased through a weighted average algorithm to obtain the adjusted current threshold; Adopting the adaptive threshold adjustment algorithm, the dynamic detection threshold adaptive to the current environmental condition is determined by fusing the environmental parameter change trend, and the leakage detection result is obtained by comparing the dynamic detection threshold with the real-time leakage data, and is used to update the environmental condition database to record the dynamic detection threshold and the leakage state, and obtain the environmental adaptive feedback data.

4. The edge computing based PLC-IoT electric leakage detection method for street light according to claim 1, wherein: The electrical parameters of each street lamp node are abnormally identified, specifically including: The electrical parameter data of each street lamp node is obtained through the sensor, including the node current and the grounding resistance, and is stored as a time series data set, and the time series data set is processed by using the sliding window method to calculate the dynamic change trend of the node current and the grounding resistance, and the dynamic change characteristic is obtained; According to the dynamic change characteristic, the k-means algorithm is used for cluster analysis of the electrical parameters of each street lamp node to determine the dynamic current threshold and the dynamic resistance threshold, and if the node current value exceeds the dynamic current threshold, it is marked as current abnormality; If the grounding resistance value is lower than the dynamic resistance threshold, it is marked as resistance abnormality, and a suspected leakage state is generated, and the suspected leakage state is classified by using the decision tree algorithm, and the abnormal type and the abnormal degree are determined in combination with the dynamic change characteristic; According to the abnormal type and the abnormal degree, the risk identification containing the time stamp, the node identification, the abnormal type and the abnormal degree is generated and stored as structured data, and the periodic change trend of the risk identification is detected by using the time series analysis method to obtain the risk identification containing the abnormal type and the abnormal degree.

5. The edge computing based PLC-IoT street light electric leakage detection method according to claim 1, characterized in that: The parameter distribution characteristics of adjacent nodes are analyzed by using the multi-node collaborative decision algorithm, specifically including: The electrical parameters and operating states of the suspected leakage nodes and the adjacent nodes within the preset range thereof are obtained, the voltage, current and power data are collected through the sensor network, and then the multi-node collaborative decision algorithm is used for analysis to calculate the distribution characteristics of the electrical parameters of the adjacent nodes and determine the parameter distribution mode; If the electrical parameters of more than a preset proportion of nodes in the adjacent nodes deviate from the normal distribution, the abnormal node group is identified by using the clustering algorithm, and it is judged as regional leakage risk, and the abnormal persistence is determined; According to the abnormal persistence result, if the single node abnormal duration exceeds the preset threshold, the device aging model is analyzed to determine the device aging, and if the single node abnormal duration does not exceed the preset threshold, the environmental data is analyzed by using the interference detection algorithm to determine the local interference.

6. The edge computing based PLC-IoT electric leakage detection method for street light according to claim 1, wherein: The leakage alarm signal is generated or recorded as a monitoring event according to the confidence, specifically including: The single node detection result is obtained by real-time collection of electrical parameters through the sensor deployed at each street lamp node and abnormal detection, and the verification data is obtained from the multi-node sensor, and the multi-node verification information is obtained by aligning the node data through time synchronization technology; The comprehensive confidence parameter is calculated by using the weighted average algorithm to fuse the single node detection result and the multi-node verification information, and if the comprehensive confidence parameter exceeds the preset confidence threshold, the leakage alarm signal is generated and stored in the alarm database. If the comprehensive confidence parameter is lower than the preset confidence threshold, the data is recorded as a monitoring event, stored in an event database, and the data in the monitoring event database is analyzed by a time series analysis algorithm to determine an abnormal pattern and update a subsequent observation strategy.

7. The edge computing based PLC-IoT leakage detection method for street light as claimed in claim 1, wherein: The dynamic risk assessment algorithm is used to classify the confirmed electric leakage risk, specifically including: The sensor is used to collect the electric leakage current, duration and influence range of the electrical system parameters in real time, and the collected data is preprocessed to remove noise and outliers. A weighted calculation method is used to assign different weights to the electric leakage current, duration and influence range, and the risk score is calculated by a weighted formula to obtain a quantitative risk value. According to the comparison of the quantitative risk value and the preset threshold, if the quantitative risk value exceeds the preset emergency threshold, the power-off protection instruction is triggered; if it is within the preset warning range, the maintenance reminder information is sent; According to the power-off execution state or the reminder sending state, the risk score and the processing result are recorded to the log database, the leakage risk trend is predicted by analyzing the log data and using a machine learning regression algorithm, and the weight parameters of the risk assessment algorithm are adjusted according to the trend to optimize the evaluation model. 8.The PLC-IoT based edge computing based electric leakage detection method for street light according to claim 1, wherein: The parameter optimization mechanism is used to continuously update the environmental correction coefficient in the adaptive threshold adjustment algorithm, specifically including: The initial value of the environmental correction coefficient is obtained from the historical electric leakage event data, the false positive rate under each environmental condition is calculated by using a statistical analysis method, the baseline value of the environmental correction coefficient is determined, and the adjusted sensitivity coefficient is obtained. The adaptive threshold adjustment algorithm is updated according to the adjusted sensitivity coefficient, a new environmental correction coefficient is obtained, an updated detection decision model is generated, the updated detection decision model is verified by real-time collection of electric leakage data, the model false positive rate is evaluated by using a cross-validation method, and the model performance is determined. The correlation coefficient between the environmental condition and the false positive rate is obtained, the weight of the parameter optimization mechanism is adjusted according to the correlation coefficient, the optimized environmental correction coefficient is obtained, and the detection decision mechanism is updated to generate a continuously optimized adaptive threshold adjustment algorithm. 9.The PLC-IoT based edge computing based electric leakage detection method of street light according to claim 8, wherein: After updating the environmental correction coefficient in the adaptive threshold adjustment algorithm, a state monitoring database based on multi-node collaborative verification results is established, the abnormal patterns and disposal records of each node are stored, and repetitive abnormal features are identified by using a pattern matching technology.

10. An edge computing based PLC-IoT street light leakage detection system for implementing the edge computing based PLC-IoT street light leakage detection method according to any one of claims 1-9, characterized in that: Including: The basic data source construction module collects the electrical parameters and environmental parameters of the street lamp nodes in real time through a sensor array, combines the time stamp and node identifier, and constructs a basic data source covering the running state of the street lamp power supply system; The dynamic threshold calibration module uses the adaptive threshold adjustment algorithm to dynamically calibrate the preset electric leakage detection threshold according to the environmental parameter change trend in the multi-dimensional feature vector, adjusts the resistance and current thresholds respectively by using linear interpolation and weighted average algorithm according to the change of humidity and temperature parameters, and determines the dynamic detection threshold that adapts to the current environmental conditions; The abnormality recognition and risk identification module obtains electrical parameter data from each street lamp node, processes and calculates dynamic change trends using a sliding window method, determines dynamic thresholds using k-means algorithm clustering analysis, generates suspected electric leakage states, classifies and determines abnormal types and degrees through a decision tree algorithm, and generates risk identification; The multi-node collaborative decision-making module obtains electrical parameters and operating states of suspected electric leakage nodes and their adjacent nodes, analyzes parameter distribution characteristics of adjacent nodes through a multi-node collaborative decision-making algorithm, and if electrical parameters of more than a preset proportion of adjacent nodes deviate from normal distribution, determines that there is a regional electric leakage risk, and further determines that the equipment is aging or there is local interference; The confidence assessment and decision-making module fuses single-node detection results and multi-node verification information, calculates a comprehensive confidence parameter using a weighted average algorithm, and if the comprehensive confidence parameter exceeds a preset confidence threshold, generates an electric leakage alarm signal and stores it in an alarm database; If it is below the threshold, it is recorded as a monitoring event and stored in an event database; The risk level division module calculates a risk score according to each parameter using a weighted calculation method, divides the risk level, and if the risk score exceeds an emergency threshold, triggers a power-off protection instruction; If it is in the warning range, a maintenance reminder information is sent, the risk trend is predicted using a machine learning regression algorithm by analyzing log data, and the evaluation model is optimized; The parameter optimization and continuous updating module obtains initial values of environmental correction coefficients based on historical electric leakage event data, calculates a baseline value by determining a false alarm rate under each environmental condition, adjusts sensitivity coefficients through a gradient descent algorithm, updates an adaptive threshold adjustment algorithm, obtains new environmental correction coefficients, and verifies and evaluates the performance of the model.

Images