Intelligent induction type cable quick connector plugging method

Abstract

The invention relates to an intelligent induction type cable quick connector plugging method, which comprises the steps of removing noise and errors in dust particle data through a signal processing module, constructing a first data set by adopting a Kalman filtering algorithm in combination with the processed dust particle data and ambient air pressure data, and storing the first data set according to a predefined format; by acquiring an output result of the particle swarm optimization model, analyzing to obtain a motion trend and concentration change of dust, monitoring an insulation state of the cable joint in real time through an insulation state sensor of the dust removal instrument, and constructing a second data set in combination with the motion trend and concentration change of dust particles and the insulation state of the cable joint; obtaining the working intensity of the dust removal instrument through a fuzzy logic controller in the dust removal instrument; according to the insulation state of the cable joint, judging whether the insulation performance of the cable joint reaches a safety standard; and collecting real-time plugging data in the plugging process, and evaluating the accuracy and safety of the plugging operation by analyzing the real-time plugging data.

Description

Technical Field

[0001] The present invention relates to the technical field of cable connector plugging, and in particular to an intelligent inductive cable quick connector plugging method. Background Art

[0002] One of the main technical challenges when performing high-voltage cable quick connector plug-in operations is the influence of dust particles. Dust particles are easily adsorbed on the insulating surface of the cable connector. The adsorbed dust particles will seriously affect the insulation performance of the connector and may cause safety problems such as insulation breakdown. It is crucial to ensure that the insulating surface of the connector is kept clean before and after plugging. The removal of dust also requires an efficient dust removal system, which should be able to automatically adjust the frequency and intensity of dust removal based on the monitored dust movement data to ensure that the plug-in process of the cable connector is not affected by dust. Due to the complex environment and the different sizes of dust particles, the dust removal system must adapt to different situations and remove various dust particles in real time. The main technical problem is how to accurately and in real time monitor the dynamic changes of dust, and based on the monitoring data, efficiently remove dust particles to ensure the safe and reliable plug-in of high-voltage cable quick connectors. Summary of the invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the present invention aims to provide an intelligent inductive cable quick connector plugging method to ensure safe and reliable plugging of high-voltage cable quick connectors.

[0004] The present invention discloses an intelligent inductive cable quick connector plugging method, comprising the following steps:

[0005] The dust particle data in the air is detected in real time by the environmental sensor, the size and concentration of the dust particles are recorded, and the collected dust particle data is transmitted to the cloud server through the wireless network; the environmental sensor includes an optical particle counter, a laser radar and a camera device; the camera device is used to capture the movement path and speed of the dust particles; the environmental air pressure data is detected in real time by the air pressure monitoring system and transmitted to the cloud server;

[0006] Removing noise and errors from the dust particle data through a cloud server to obtain processed dust particle data, combining the processed dust particle data with ambient air pressure data, obtaining a first data set through a Kalman filter algorithm, and storing the first data set in a predefined format;

[0007] Setting initial parameters of a particle swarm optimization model through a cloud server, and inputting the first data set into the particle swarm optimization model to train the particle swarm optimization model; the initial parameters include the number, speed and position of particles; the particle swarm optimization model is used to generate a prediction result including a movement trend of dust;

[0008] The output results of the particle swarm optimization model are obtained through the built-in system of the dust removal instrument, and the movement trend and concentration change of the dust particles are analyzed; the insulation state of the cable joint is obtained through the insulation state sensor of the dust removal instrument; a second data set is constructed by combining the movement trend of the dust particles, the concentration change and the insulation state of the cable joint; the second data set is used to adjust the operation frequency and cleaning intensity of the dust removal instrument;

[0009] The working intensity of the dust removal instrument is obtained by a fuzzy logic controller in the dust removal instrument; the input parameters of the fuzzy logic controller include dust movement trend, concentration change and insulation state of cable joints, and the output parameters are the working intensity of the dust removal instrument; the working intensity of the dust removal instrument includes voltage, current, working frequency, vibration frequency and wind speed;

[0010] According to the insulation state of the cable joint, it is judged whether the insulation performance of the cable joint reaches the safety standard. If the insulation performance of the cable joint reaches the safety standard, an instruction for plugging the cable joint is issued; if the insulation performance of the cable joint does not reach the safety standard, the second data set is input into the fuzzy logic controller, the output parameters are obtained, and an instruction is issued to make the dust removal device perform the dust removal operation according to the output parameters of the fuzzy logic controller; by analyzing the insulation state of the cable joint during dust removal, the instruction for making the dust removal device perform the dust removal operation according to the output parameters of the fuzzy logic controller is continuously issued until the insulation performance of the cable joint reaches the safety standard;

[0011] By continuously monitoring the communication status between the cloud server, cable connector and dust removal instrument, the stability of data transmission is ensured and real-time plugging data during the plugging process is collected; the real-time plugging data is analyzed by the cloud server application data acquisition module to evaluate the accuracy and safety of the plugging operation; the real-time plugging data includes torque, insertion force and dust concentration.

[0012] Preferably, the real-time detection of dust particle data in the air by an environmental sensor, recording the size and concentration of dust particles, and transmitting the collected dust particle data to a cloud server via a wireless network; the environmental sensor includes an optical particle counter, a laser radar, and a camera device; the camera device is used to capture the movement path and speed of dust particles; the real-time detection of environmental air pressure data by an air pressure monitoring system and transmission to a cloud server include:

[0013] The optical particle counter detects the dust particle data in the air in real time to obtain the size and concentration information of the particles; the laser radar captures the movement path and speed of the particles; and the camera equipment records the visual image data of the particles;

[0014] The detected dust particle size, concentration, movement path and movement speed as well as visual image data are transmitted to the cloud server in real time via wireless network for summary processing;

[0015] The detected dust particle size, concentration, movement path and movement speed and visual image data are cleaned and standardized by a data preprocessing method of data cleaning and normalization, and a dust particle data model is established by a support vector machine (SVM) and a random forest algorithm;

[0016] According to the established dust particle data model, through time series analysis and regression analysis prediction methods, the concentration trend of dust particles in the air in the future period is predicted in real time to determine the change trend of air quality;

[0017] The air pressure monitoring system detects ambient air pressure data in real time and transmits the data to the cloud server;

[0018] The cloud server correlates and analyzes the air pressure data with the dust particle data to establish a prediction model between environmental meteorological conditions and air quality;

[0019] Estimate the air quality change trend in the future through meteorological forecast data.

[0020] Preferably, the method of removing noise and errors in dust particle data through a cloud server to obtain processed dust particle data, combining the processed dust particle data with ambient air pressure data, obtaining a first data set through a Kalman filter algorithm, and storing the first data set in a predefined format includes:

[0021] The high-frequency noise and random errors in the data are removed by wavelet threshold denoising method;

[0022] The signal processing module is used to extract the features of the filtered dust particle data to obtain key feature parameters, which include the concentration, particle size distribution and movement speed of the dust particles;

[0023] Acquire real-time air pressure data through an environmental air pressure monitoring system, and synchronize the air pressure data with the processed dust particle data to obtain synchronized dust particle data and air pressure data;

[0024] fusing the synchronized dust particle data and air pressure data using a Kalman filter algorithm to obtain a first data set;

[0025] The first data set includes the concentration, particle size distribution and movement speed of dust particles and ambient air pressure data;

[0026] Organizing and packaging the first data set through a predefined data format and structure, and adding metadata and timestamp information;

[0027] The first data set is stored in a document-based NoSQL database such as MongoDB.

[0028] Preferably, the initial parameters of the particle swarm optimization model are set by a cloud server, and the first data set is input into the particle swarm optimization model to train the particle swarm optimization model; the initial parameters include the number, speed and position of particles; the particle swarm optimization model is used to generate a prediction result including the movement trend of dust, including:

[0029] Selecting the number of particles according to the dimension, complexity and computing resource factors of the first data set; and generating the initial velocity and position of the particles through a random function;

[0030] The particle swarm optimization model evaluates the optimization performance of each particle through a fitness function;

[0031] Setting a convergence condition, wherein the particle swarm optimization model iterates according to the convergence condition to obtain a prediction model for generating a prediction result including a movement trend of dust;

[0032] The convergence conditions include:

[0033] Set a maximum number of iterations. When the maximum number of iterations is reached, the particle swarm optimization model is considered to have converged.

[0034] A minimum fitness threshold is set. When the optimal fitness value of multiple consecutive iterations changes less than the minimum fitness threshold, it is considered that the particle swarm optimization model has converged.

[0035] Setting a minimum particle position change threshold, when the particle position change of multiple consecutive iterations is less than the minimum particle position change threshold, it is considered that the particle swarm optimization model has converged;

[0036] If the particle swarm optimization model satisfies one or more of the above three convergence conditions, the particle swarm optimization model completes iteration to obtain a prediction model for generating a prediction result including a motion trend of dust;

[0037] The prediction model is used to perform prediction analysis on the newly collected dust particle data to generate the dust movement trend in the future.

[0038] Preferably, the output result of the particle swarm optimization model is obtained through the built-in system of the dust removal instrument, and the movement trend and concentration change of the dust particles are analyzed; the insulation state of the cable joint is obtained through the insulation state sensor of the dust removal instrument; the second data set is constructed in combination with the movement trend of the dust particles, the concentration change and the insulation state of the cable joint; the second data set is used to adjust the operation frequency and cleaning intensity of the dust removal instrument, including:

[0039] The output results of the particle swarm optimization model are obtained in real time through a wireless network; the position coordinates, movement speed and direction information of dust particles at different time points, and the numerical change curve of dust concentration are obtained;

[0040] Obtain the movement data and concentration data of dust particles through analysis; associate the movement data and concentration data of dust particles with the spatial position information of the dust removal instrument to determine the density and distribution state of dust particles encountered by the dust removal instrument in the future operation process;

[0041] Obtain the insulation resistance value threshold range and the dielectric loss factor threshold range, monitor the insulation resistance value and the dielectric loss factor at the cable joint in real time through the insulation state sensor of the dust removal instrument, and compare the collected insulation resistance value with the preset insulation resistance value threshold range, and compare the collected dielectric loss factor with the preset dielectric loss factor threshold range;

[0042] When the insulation resistance value exceeds the preset insulation resistance value threshold range, or when the dielectric loss factor exceeds the preset dielectric loss factor threshold range, it is determined that the insulation state is abnormal and a warning signal is triggered;

[0043] Correlating the abnormal insulation state data with the dust particle movement and concentration data at the corresponding time to construct a second data set;

[0044] According to the AdaBoost algorithm, a correlation model between dust pollution and abnormal insulation status is established. The probability distribution of abnormal insulation status of cable joints under different dust particle concentrations and movement trends is obtained through the correlation model.

[0045] Adjust the operating frequency and cleaning intensity of the dust removal equipment according to the predicted dust particle concentration and the probability of abnormal insulation status;

[0046] According to the abnormal probability of insulation state predicted by the model, the operation frequency and cleaning intensity of the dust removal equipment are divided into low, medium and high levels, and a probability threshold is preset for each level, namely the low probability threshold, the medium probability threshold and the high probability threshold;

[0047] When the predicted dust particle concentration or insulation state abnormal probability exceeds the low-level probability threshold, the operating frequency and cleaning intensity of the dust removal equipment are increased;

[0048] When the predicted dust particle concentration and insulation state abnormal probability are lower than the low-level probability threshold, the operating frequency and cleaning intensity of the dust removal instrument are reduced;

[0049] According to the analysis results of the correlation model, the high-risk time window for abnormal insulation status of the cable joint is predicted. Before the high-risk time window, the cable joint is dusted and cleaned according to the operating frequency and cleaning intensity of the dust removal instrument corresponding to the probability threshold level of the high-risk time window.

[0050] Preferably, the working intensity of the dust removal instrument is obtained by a fuzzy logic controller in the dust removal instrument; the input parameters of the fuzzy logic controller include dust movement trend, concentration change and insulation state of the cable joint, and the output parameter is the working intensity of the dust removal instrument; the working intensity of the dust removal instrument includes voltage, current, working frequency, vibration frequency and wind speed, including:

[0051] Obtain dust particle movement trends and concentration changes as well as cable joint insulation status;

[0052] Classify dust concentration into: low dust concentration, medium dust concentration and high dust concentration;

[0053] The dust movement trend is divided into: stable dust movement trend, slowly rising dust movement trend and rapidly rising dust movement trend;

[0054] The insulation status of the cable joint is divided into: normal insulation status of the cable joint, slightly abnormal insulation status of the cable joint and seriously abnormal insulation status of the cable joint;

[0055] According to Mamdan i reasoning method, a fuzzy rule base is constructed;

[0056] Through the centroid defuzzification method, the result of fuzzy reasoning is converted into the specific working parameters of the dust removal instrument; and an instruction is issued to make the dust removal instrument work according to the specific working parameters.

[0057] Preferably, the method of judging whether the insulation performance of the cable joint meets the safety standard according to the insulation state of the cable joint, and issuing an instruction to perform a plugging operation on the cable joint if the insulation performance of the cable joint meets the safety standard; inputting the second data set into the fuzzy logic controller to obtain output parameters and issuing an instruction to make the dust removal device perform a dust removal operation according to the output parameters of the fuzzy logic controller if the insulation performance of the cable joint does not meet the safety standard; and continuously issuing an instruction to make the dust removal device perform a dust removal operation according to the output parameters of the fuzzy logic controller by analyzing the insulation state of the cable joint during dust removal until the insulation performance of the cable joint meets the safety standard, comprises:

[0058] Preset the insulation status index threshold, collect the insulation resistance value and dielectric loss factor of the cable joint, and calculate the insulation status index IS:

[0059] IS=0.7×(R / R0)+0.3×(tanδ0 / tanδ),

[0060] Where R is the measured insulation resistance value, R0 is the insulation resistance threshold, tanδ is the measured dielectric loss factor, and tanδ0 is the dielectric loss factor threshold;

[0061] When IS is greater than or equal to the insulation state index threshold, it is determined that the insulation performance of the cable joint meets the safety standard; when IS is less than the insulation state index threshold, it is determined that the insulation performance of the cable joint does not meet the safety standard;

[0062] If the insulation performance of the cable connector reaches the safety standard, an instruction to perform the plugging operation of the cable connector is issued;

[0063] If the insulation performance of the cable joint does not meet the safety standard, the second data set is input into the fuzzy logic controller;

[0064] According to Mamdan i reasoning method, a fuzzy rule base is constructed;

[0065] Through the centroid defuzzification method, the result of fuzzy reasoning is converted into the specific working parameters of the dust removal instrument; and an instruction is issued to make the dust removal instrument work according to the specific working parameters;

[0066] When the insulation state index IS is greater than the insulation state index threshold for multiple consecutive times, it is determined that the insulation performance of the cable joint has reached the safety standard, and an instruction to perform the plugging operation of the cable joint is issued;

[0067] Collect the insertion depth of the connector, the displacement and speed of the plugging mechanism, calculate the deviation between the actual insertion depth of the connector and the target depth, and adjust the movement of the plugging mechanism according to the deviation between the actual insertion depth and the target depth until the plugging operation of the cable connector is completed;

[0068] After completing the plugging operation of the cable connector, collect the axial compression force and judge whether the required tightening degree is achieved based on the axial compression force.

[0069] Preferably, the communication status between the cloud server, the cable connector and the dust removal instrument is continuously monitored to ensure the stability of data transmission and collect real-time plugging data during the plugging process; the real-time plugging data is analyzed by the cloud server application data acquisition module to evaluate the accuracy and safety of the plugging operation; the real-time plugging data includes torque, insertion force and dust concentration, including:

[0070] The communication status between the cloud server, cable connector and dust removal instrument is monitored through the heartbeat mechanism, and the round-trip delay is recorded;

[0071] The data transmission traffic between the cloud server, cable connector and dust removal instrument is shaped and restricted through the token bucket algorithm; the token generation rate is dynamically adjusted according to the water level of the token passing through;

[0072] When the water level of the token bucket is higher than a certain percentage for a period of time, a traffic anomaly alarm is triggered;

[0073] Collect the torque, insertion force and dust concentration in the plugging environment during the plugging process as real-time plugging data;

[0074] Preprocess the real-time plug-in data and compress the data through the run-length encoding algorithm;

[0075] The real-time plug-in data is filtered and smoothed by a Kalman filter algorithm to obtain the plug-in data;

[0076] Establishing a plugging operation evaluation model for judging the accuracy and safety of the current plugging operation through a decision tree algorithm, inputting the plugging data into the plugging operation evaluation model; judging the accuracy and safety of the current plugging operation;

[0077] Collect the plug-in data, perform cluster analysis on the plug-in data, use the elbow method to evaluate different clustering number K values, obtain the optimal clustering number, classify the plug-in data through the K-means algorithm, and count the number, proportion and parameter distribution status of various plug-in operations to generate a plug-in operation quality report.

[0078] The advantage of the intelligent induction cable quick connector plug-in method disclosed in the present invention is that the present invention solves the problem of the influence of dust particles on the insulation performance during the plug-in operation of high-voltage cable quick connectors in high-altitude areas. Due to the thin air at high altitudes, dust particles are easily suspended and adsorbed on the insulating surface of the cable connector, resulting in a decrease in insulation performance and even causing insulation breakdown. The present invention accurately monitors the dynamic changes of suspended dust through real-time monitoring and intelligent control systems, and automatically adjusts the dust removal frequency and intensity according to the monitoring data to ensure that the cable connector remains clean before and after plugging. The system can adapt to complex environments and different dust particle sizes, remove various dust particles in real time, and ensure the safe and reliable plug-in of high-voltage cable connectors in extreme environments, solving the technical problem that traditional methods cannot effectively deal with dust problems in high altitude areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 The present invention is a flow chart of an intelligent inductive cable quick connector plugging method.

[0080] Figure 2The present invention is a schematic diagram of an intelligent inductive cable quick connector plugging method.

[0081] Figure 3 This is another schematic diagram of an intelligent inductive cable quick connector plugging method disclosed in the present invention. DETAILED DESCRIPTION

[0082] like Figure 1 to Figure 3 As shown, the intelligent inductive cable quick connector plugging method described in the present disclosure may specifically include:

[0083] Step S101, using an environmental sensor to detect dust particle data in the air in real time, record the size and concentration of the dust particles, and transmit the collected dust particle data to a cloud server via a wireless network; the environmental sensor includes an optical particle counter, a laser radar, and a camera device; the camera device is used to capture the movement path and speed of dust particles; the air pressure monitoring system is used to detect the ambient air pressure data in real time and transmit it to the cloud server;

[0084] Specifically, the optical particle counter detects the dust particle data in the air in real time to obtain the size and concentration information of the particles; the laser radar captures the movement path and speed of the particles; and the camera equipment records the visual image data of the particles;

[0085] The detected dust particle size, concentration, movement path and movement speed as well as visual image data are transmitted to the cloud server in real time via wireless network for summary processing;

[0086] The detected dust particle size, concentration, movement path and movement speed and visual image data are cleaned and standardized by a data preprocessing method of data cleaning and normalization, and a dust particle data model is established by a support vector machine (SVM) and a random forest algorithm;

[0087] According to the established dust particle data model, through time series analysis and regression analysis prediction methods, the concentration trend of dust particles in the air in the future period is predicted in real time to determine the change trend of air quality;

[0088] The air pressure monitoring system detects ambient air pressure data in real time and transmits the data to the cloud server;

[0089] The cloud server correlates and analyzes the air pressure data with the dust particle data to establish a prediction model between environmental meteorological conditions and air quality;

[0090] Estimate the air quality change trend in the future through meteorological forecast data.

[0091] More specifically, the environmental sensors include optical particle counters, lidars and cameras; the environmental sensors detect dust particle data in the air in real time, record the size and concentration of dust particles, and capture the movement path and speed of dust particles through cameras; the environmental sensors transmit the collected dust particle data to the cloud server via a wireless network; the air pressure monitoring system detects environmental air pressure data in real time and transmits it to the cloud server.

[0092] The environmental sensor detects the dust particle data in the air in real time through an optical particle counter to obtain the size and concentration information of the particles. At the same time, the laser radar captures the movement path and speed of the particles, and the camera records the visual image data of the particles. The environmental sensor transmits a series of multi-dimensional data such as the size, concentration, movement path, movement speed and visual image of the detected dust particles to the cloud server through a wireless network in real time for summary processing. After receiving the multi-dimensional data of dust particles from the environmental sensor, the cloud server uses data cleaning and normalization data preprocessing methods to clean and standardize the data, and establishes a dust particle data model through support vector machine SVM and random forest algorithm to mine the correlation and regularity between data. The SVM algorithm can effectively process high-dimensional data and is suitable for classification under small sample data; the random forest algorithm can handle nonlinear relationships and is not easy to overfit. According to the established dust particle data model, the cloud server uses time series analysis and regression analysis prediction methods to predict the change trend of dust particle concentration in the air in the future period in real time, judge the change trend of air quality, and provide data support for environmental management decisions. The air pressure monitoring system detects the ambient air pressure data in real time and transmits the data to the cloud server. Air pressure data is closely related to dust particle data, and changes in air pressure will affect the movement and diffusion of dust particles. Incorporating air pressure data into the analysis and prediction model can comprehensively characterize the changing patterns of air quality. The cloud server correlates the air pressure data with the dust particle data, studies the influence of air pressure changes on the movement of dust particles, and establishes a prediction model between environmental meteorological conditions and air quality. The trend of air quality changes in the future is estimated through meteorological forecast data. Environmental sensors and air pressure monitoring systems continuously collect data and transmit it to the cloud server. The server uses the gradient descent method and cross-validation method to self-optimize and iteratively update the prediction model, continuously improving the accuracy and reliability of air quality prediction. During the optimization process, the gradient descent method is used to find the optimal value of the model parameters, and the cross-validation method is used to select the best combination of model parameters to avoid the problem of overfitting or underfitting of the model.

[0093] For example, the environmental sensor collects 100 air samples per second through an optical particle counter, and counts the number of particles with diameters of 0.5μm, 1.0μm, 2.5μm and 5.0μm in each sample. At the same time, the lidar scans the dust particles in the air at a frequency of 10Hz to obtain the movement trajectory and instantaneous speed of the particles. The camera device records video images of dust particles at a speed of 30 frames per second, with a resolution of 1920×1080 pixels. These multi-dimensional data are transmitted to the cloud server in real time through the Wi-Fi6 network, with a transmission delay of less than 10ms and a data throughput of 1Gbps. The cloud server uses the ApacheSpark big data processing framework to clean and normalize the received data, remove outliers and invalid values, and standardize the data to the [0,1] interval. Then, the SVM algorithm and random forest algorithm in the scikit-learn machine learning library are used to establish a dust particle data model with features such as particle concentration, movement trajectory, and speed as input. Among them, the SVM algorithm uses the RBF kernel function, the penalty coefficient C = 1.0, and the kernel function parameter gamma = 0.1; the random forest algorithm uses 100 decision trees, and the maximum depth of each decision tree is 10. The performance of the model was evaluated through 5-fold cross validation, and the accuracy rate reached more than 95%. Using the established data model, the Prophet time series prediction algorithm and the multivariate linear regression algorithm were used to predict the particle concentration change trend in the next 24 hours, and the mean absolute percentage error MAPE was less than 10%. At the same time, the air pressure monitoring system collects air pressure data once a minute with a resolution of 0.1hPa. The Pearson correlation analysis of the air pressure data and the particle concentration data showed that the correlation coefficient between the two reached more than 0.8, indicating that the change in air pressure has a significant effect on the particle concentration. Therefore, the air pressure data is used as an additional feature, together with the particle concentration, motion trajectory, speed and other features, and input into the multi-layer perceptron (MLP) neural network model to predict the air quality index AQI in the next 24 hours. The MLP model uses 3 hidden layers, each containing 64, 32, and 16 neurons respectively. The activation function is ReLU, the optimization algorithm is Adam, and the learning rate is 0.01. The weather forecast data is used to estimate the changes in air pressure in the next 24 hours, which are input into the trained MLP model to obtain the corresponding AQI prediction value. Finally, the TensorBoard visualization tool is used to monitor the training process and performance indicators of the model in real time, and the early stopping method (EarlyStopping) and learning rate decay strategy are used to avoid model overfitting. Through continuous iterative optimization, the mean absolute error MAE of AQI prediction is controlled within 5.

[0094] In other embodiments of the present disclosure, corresponding models can also be established according to the above ideas. For example, in coastal areas, due to the proximity to sea water, the salinity and humidity of the air will also affect the accuracy and safety of cable plugging. The above method can be used to establish a humidity model and / or salinity model to predict the air quality.

[0095] Step S102, removing noise and errors in the dust particle data through the cloud server to obtain processed dust particle data, combining the processed dust particle data with the ambient air pressure data, obtaining a first data set through a Kalman filter algorithm, and storing the first data set in a predefined format.

[0096] Specifically, the high-frequency noise and random errors in the data are removed by wavelet threshold denoising method;

[0097] The signal processing module is used to extract the features of the filtered dust particle data to obtain key feature parameters, which include the concentration, particle size distribution and movement speed of the dust particles;

[0098] Acquire real-time air pressure data through an environmental air pressure monitoring system, and synchronize the air pressure data with the processed dust particle data to obtain synchronized dust particle data and air pressure data;

[0099] fusing the synchronized dust particle data and air pressure data using a Kalman filter algorithm to obtain a first data set;

[0100] The first data set includes the concentration, particle size distribution and movement speed of dust particles and ambient air pressure data;

[0101] Organizing and packaging the first data set through a predefined data format and structure, and adding metadata and timestamp information;

[0102] The first data set is stored in a document-based NoSQL database such as MongoDB.

[0103] More specifically, the cloud server uses a signal processing module to remove noise and errors in the dust particle data, combines the processed dust particle data with the ambient air pressure data, uses a Kalman filter algorithm to construct a first data set, and stores it in a predefined format.

[0104] The cloud server preprocesses the received dust particle data through the signal processing module, and uses the wavelet threshold denoising method to remove high-frequency noise and random errors in the data. The signal processing module extracts features from the filtered dust particle data to obtain a series of key characteristic parameters such as particle concentration, particle size distribution and movement speed. The cloud server obtains real-time air pressure data from the ambient air pressure monitoring system and synchronizes the air pressure data with the processed dust particle data. The Kalman filter algorithm is used to fuse the synchronized dust particle data and air pressure data. By establishing a state space model and an observation model, the system state is recursively estimated to obtain the optimal data fusion result. The state vector includes the position and velocity information of the dust particles, and the observation vector includes the concentration and particle size distribution data measured by the sensor. The Kalman filter algorithm initializes the state estimate and error covariance matrix, and sets the initial value based on prior knowledge. In the prediction process, the state transfer matrix and control input are used to predict the current state and calculate the prediction error covariance matrix. In the update process, the observation value and observation matrix are used to calculate the Kalman gain, update the state estimate and error covariance matrix. The formula for the prediction and update process is as follows:

[0105] During the prediction process, the state prediction is x_pred = A*x_est + B*u;

[0106] The error covariance prediction is P_pred = A*P_est*A^T+Q;

[0107] During the update process, the Kalman gain K=P_pred*H^T*(H*P_pred*H^T+R)^(-1) is calculated;

[0108] The state estimate is updated to x_est = x_pred + K*(zH*x_pred);

[0109] The error covariance is updated to P_est = (IK*H)*P_pred;

[0110] Among them, A is the state transition matrix, B is the control input matrix, Q is the process noise covariance matrix, H is the observation matrix, R is the observation noise covariance matrix, and I is the identity matrix. The Kalman filter algorithm continuously iterates the prediction and update process until the convergence condition is reached or all data are processed to minimize the variance of the estimation error. The fused data includes the concentration, particle size distribution, movement speed and ambient air pressure data of dust particles to form the first data set. The cloud server organizes and encapsulates the first data set according to the predefined data format and structure, and adds metadata and timestamp information. The first data set is stored in a document-based NoSQL database such as MongoDB, and its flexible data model and high scalability are used to store and manage the heterogeneous first data set. The database adopts a distributed architecture and uses data sharding and replication mechanisms. The cloud server indexes the stored first data set and accelerates the retrieval and analysis of data through indexing.

[0111] Exemplarily, the signal processing module uses a wavelet threshold denoising method to preprocess the dust particle data, selects the Daubechies4 wavelet basis function to decompose the data into 5 layers, and removes the noise in the high-frequency detail coefficient according to the threshold parameter λ=σsqrt(2log(N)), where σ is the noise standard deviation and N is the data length. In the feature extraction process, the particle image velocimetry method is used to calculate the particle movement speed, and the image sequence is cross-correlated to obtain the particle displacement vector. Combined with the image sampling interval time, the speed resolution is calculated to reach 0.1m / s. When the Kalman filter algorithm fuses the dust particle data and the air pressure data, the state vector includes the particle position (x, y, z) and velocity (vx, vy, vz), and the observation vector includes the particle concentration, diameter and air pressure value. The initial state estimation is calculated based on the first frame image, the process noise covariance matrix Q is set according to the acceleration standard deviation of the particle, and the observation noise covariance matrix R is set according to the measurement error of the sensor. The state transfer matrix A in the prediction step considers the uniform motion of the particle, and the observation matrix H in the update step associates the state variable with the observation. Iterate the prediction and update steps 50 times, processing 100 data points each time, that is, 5 seconds of data. The first fused data set is encapsulated in JSON format, and each data point contains timestamp, particle concentration, diameter, speed, and air pressure fields. The MongoDB database uses the WiredTiger storage engine and distributes data to 4 shard nodes through hash sharding. Each shard node has 2 copies to ensure data reliability. Indexing the timestamp field increases the query speed by more than 10 times.

[0112] Step S103, setting initial parameters of the particle swarm optimization model through the cloud server, and inputting the first data set into the particle swarm optimization model to train the particle swarm optimization model; the initial parameters include the number, speed and position of particles; the particle swarm optimization model is used to generate a prediction result including the movement trend of dust.

[0113] Specifically, the number of particles is selected according to the dimension, complexity and computing resource factors of the first data set; and the initial speed and position of the particles are generated by a random function;

[0114] The particle swarm optimization model evaluates the optimization performance of each particle through a fitness function;

[0115] Setting a convergence condition, wherein the particle swarm optimization model iterates according to the convergence condition to obtain a prediction model for generating a prediction result including a movement trend of dust;

[0116] The convergence conditions include:

[0117] Set a maximum number of iterations. When the maximum number of iterations is reached, the particle swarm optimization model is considered to have converged.

[0118] A minimum fitness threshold is set. When the optimal fitness value of multiple consecutive iterations changes less than the minimum fitness threshold, it is considered that the particle swarm optimization model has converged.

[0119] Setting a minimum particle position change threshold, when the particle position change of multiple consecutive iterations is less than the minimum particle position change threshold, it is considered that the particle swarm optimization model has converged;

[0120] If the particle swarm optimization model satisfies one or more of the above three convergence conditions, the particle swarm optimization model completes iteration to obtain a prediction model for generating a prediction result including a motion trend of dust;

[0121] The prediction model is used to perform prediction analysis on the newly collected dust particle data to generate the dust movement trend in the future.

[0122] More specifically, the cloud server sets the initial parameters of the particle swarm optimization model, including the number, speed and position of particles, and inputs the first data set into the particle swarm optimization model; trains the particle swarm optimization model, including the particle swarm continuously adjusting its position in the solution space to determine the optimal solution for the movement law and adsorption pattern of dust particles, so that the particle swarm optimization model generates a prediction result including the movement trend of dust.

[0123] The cloud server sets the initial parameters of the particle swarm optimization model according to the characteristic dimension and data volume of the first data set. Combined with the dimension, complexity and computing resource factors of the data set, the number of particles that can obtain good optimization effects within a reasonable time is selected through experimental comparison of different numbers of particles. If the dimension of the data set is high or the complexity is large, the number of particles is increased; if the computing resources are limited, the number of particles is reduced. The initial velocity and position of the particles are generated by a random function. The cloud server inputs the first data set into the particle swarm optimization model, and the data set includes the concentration, particle size distribution, movement speed and ambient air pressure information of the dust particles; preferably, it can also include the adsorption amount data of the dust on the surfaces of different materials. The particle swarm optimization model evaluates the optimization performance of each particle through the fitness function. The fitness function combines the accuracy of the dust movement trend prediction and the accuracy of the adsorption amount prediction. For the dust movement trend prediction, the mean square error MSE and the mean absolute error MAE are used to measure the deviation between the predicted value and the actual value; for the adsorption amount prediction, the relative error RE and the determination coefficient R-square are used to measure the degree of fit between the predicted value and the actual value. The results of each evaluation index are weighted averaged to obtain a comprehensive prediction accuracy score as part of the fitness function. During the model training process, the particle swarm continuously updates its speed and position in the solution space, and adjusts the movement direction and step size of the particles through the two indicators of individual optimal and global optimal, so that the particles continue to converge to the optimal solution. Individual optimal refers to the optimal solution found by a single particle in the historical search, and global optimal refers to the optimal solution found by the entire particle swarm in the current search. The speed and position of the particle are updated by comparing the current fitness of the particle with the fitness corresponding to the individual optimal and global optimal. The number of iterations is set according to the convergence speed and performance requirements of the particle swarm optimization algorithm. Through experimental comparison of different numbers of iterations, the number of iterations that can achieve good convergence effect within a reasonable time is selected. During the iteration process, if multiple consecutive iterations fail to significantly improve the optimal solution, the iteration is terminated in advance. At the same time, an upper limit on the number of iterations is set to prevent the algorithm from falling into an infinite loop. In each iteration, excellent particles are selected according to the fitness function to update the speed and position, and the remaining particles are randomly generated again. Through continuous iterative optimization, the particle swarm eventually converges to the optimal solution of the movement law and adsorption mode of dust particles. The convergence conditions are based on the following three aspects: setting the maximum number of iterations. When the maximum number of iterations is reached, the algorithm is considered to have converged; setting the minimum fitness threshold. When the change in the optimal fitness value of multiple consecutive iterations is less than the minimum fitness threshold, the algorithm is considered to have converged; setting the minimum particle position change threshold. When the change in the particle position of multiple consecutive iterations is less than the minimum particle position change threshold, the algorithm is considered to have converged. Combining the above three convergence conditions, avoid the algorithm from stopping iteration too early or too late.Based on the trained prediction model, the cloud server performs prediction analysis on the newly collected dust particle data to generate prediction results of dust movement trend and adsorption amount in the future. The prediction time span can be set according to demand. The prediction results are presented in a visual way, showing the changes in the spatiotemporal distribution of dust concentration and adsorption amount through curves and heat maps.

[0124] Exemplarily, the cloud server selects the number of particles (such as 100) that can achieve better optimization effect within a reasonable time (such as 10 minutes) by comparing different numbers of particles (such as 50, 100, and 200) according to the feature dimensions (such as 20 features) and data volume (such as 10,000 samples) of the comprehensive data set. The initial velocity and position of the particle are generated by a random function, the velocity range is [-1, 1], and the position range covers the entire solution space. The cloud server inputs the first data set into the particle swarm optimization model, which includes the concentration, particle size distribution, movement speed and ambient air pressure information of dust particles, as well as the adsorption amount data of dust on the surfaces of different materials. The particle swarm optimization model evaluates the optimization performance of each particle through the fitness function. For the prediction of dust movement trend, the mean square error MSE and mean absolute error MAE evaluation indicators are used, and the weights of MSE and MAE are 0.6 and 0.4 respectively; for the prediction of adsorption amount, the relative error RE and determination coefficient R-square evaluation indicators are used, and the weights of RE and R-square are 0.7 and 0.3 respectively. The results of each evaluation index are weighted averaged to obtain a comprehensive prediction accuracy score as part of the fitness function. During the model training process, the particle swarm continuously updates its speed and position in the solution space, and adjusts the movement direction and step size of the particles through the two indicators of individual optimal and global optimal, so that the particles continue to converge to the optimal solution. Through experimental comparison of different iteration numbers (such as 500, 1000, and 2000), the number of iterations (such as 1000) that can achieve a good convergence effect within a reasonable time (such as 30 minutes) is selected. During the iteration process, if the fitness value of the optimal solution fails to increase by 1% after 50 consecutive iterations, the iteration is terminated in advance. At the same time, the upper limit of the number of iterations is set to 5000 to prevent the algorithm from falling into an infinite loop. In each iteration, the top 20% of excellent particles are selected according to the fitness function for speed and position update, and the remaining particles are randomly generated again. The following three convergence conditions are considered comprehensively: reaching the maximum number of iterations of 1000; the optimal fitness value changes less than 0.01 for 30 consecutive iterations; the particle position changes less than 0.001 for 30 consecutive iterations. When any convergence condition is met, the algorithm is considered to have converged to the optimal solution. Based on the trained prediction model, the cloud server performs predictive analysis on the newly collected dust particle data to generate dust movement trends and adsorption prediction results for the next 24 hours, 48 ​​hours, and 72 hours. The prediction results are presented in the form of curves and heat maps. The curves show the dust concentration and adsorption change trends at different time granularities (such as every hour, every 6 hours), and the heat maps show the dust concentration and adsorption distribution at different spatial locations (such as different areas, different heights).

[0125] Step S104, obtaining the output results of the particle swarm optimization model through the built-in system of the dust removal instrument, and analyzing the movement trend and concentration change of the dust particles; obtaining the insulation state of the cable joint through the insulation state sensor of the dust removal instrument; constructing a second data set based on the movement trend of the dust particles, the concentration change and the insulation state of the cable joint; the second data set is used to adjust the operating frequency and cleaning intensity of the dust removal instrument.

[0126] Specifically, the output results of the particle swarm optimization model are obtained in real time through a wireless network; the position coordinates, movement speed and direction information of dust particles at different time points, and the numerical change curve of dust concentration are obtained;

[0127] Obtain the movement data and concentration data of dust particles through analysis; associate the movement data and concentration data of dust particles with the spatial position information of the dust removal instrument to determine the density and distribution state of dust particles encountered by the dust removal instrument in the future operation process;

[0128] Obtain the insulation resistance value threshold range and the dielectric loss factor threshold range, monitor the insulation resistance value and the dielectric loss factor at the cable joint in real time through the insulation state sensor of the dust removal instrument, and compare the collected insulation resistance value with the preset insulation resistance value threshold range, and compare the collected dielectric loss factor with the preset dielectric loss factor threshold range;

[0129] When the insulation resistance value exceeds the preset insulation resistance value threshold range, or when the dielectric loss factor exceeds the preset dielectric loss factor threshold range, it is determined that the insulation state is abnormal and a warning signal is triggered;

[0130] Correlating the abnormal insulation state data with the dust particle movement and concentration data at the corresponding time to construct a second data set;

[0131] According to the AdaBoost algorithm, a correlation model between dust pollution and abnormal insulation status is established. The probability distribution of abnormal insulation status of cable joints under different dust particle concentrations and movement trends is obtained through the correlation model.

[0132] Adjust the operating frequency and cleaning intensity of the dust removal equipment according to the predicted dust particle concentration and the probability of abnormal insulation status;

[0133] According to the abnormal probability of insulation state predicted by the model, the operation frequency and cleaning intensity of the dust removal equipment are divided into low, medium and high levels, and a probability threshold is preset for each level, namely the low probability threshold, the medium probability threshold and the high probability threshold;

[0134] When the predicted dust particle concentration or insulation state abnormal probability exceeds the low-level probability threshold, the operating frequency and cleaning intensity of the dust removal equipment are increased;

[0135] When the predicted dust particle concentration and insulation state abnormal probability are lower than the low-level probability threshold, the operating frequency and cleaning intensity of the dust removal instrument are reduced;

[0136] According to the analysis results of the correlation model, the high-risk time window for abnormal insulation status of the cable joint is predicted. Before the high-risk time window, the cable joint is dusted and cleaned according to the operating frequency and cleaning intensity of the dust removal instrument corresponding to the probability threshold level of the high-risk time window.

[0137] More specifically, the built-in system of the dust removal instrument obtains the output results of the particle swarm optimization model, and analyzes the movement trend and concentration changes of the dust; the insulation status sensor of the dust removal instrument monitors the insulation status of the cable joint in real time; and the second data set is constructed by combining the movement trend of dust particles, concentration changes and insulation status of the cable joint.

[0138] The built-in system of the dust removal instrument obtains the output results of the particle swarm optimization model in the cloud server in real time through the wireless network, including the movement trend and concentration change prediction data of dust particles in the future. The built-in system analyzes and extracts the obtained prediction data to obtain the position coordinates, movement speed and direction information of dust particles at different time points, as well as the numerical change curve of dust concentration. The built-in system associates the analyzed dust movement and concentration data with the spatial position information of the dust removal instrument itself to determine the density and distribution state of dust particles encountered by the dust removal instrument in the future operation process, providing a basis for the subsequent cleaning strategy formulation. The insulation state sensor of the dust removal instrument adopts the capacitive or resistive measurement principle to monitor the insulation resistance value and dielectric loss factor at the cable joint in real time. The sensor collects data at regular intervals and compares the collected insulation resistance value and dielectric loss factor with the preset threshold. The setting of the threshold needs to be optimized and adjusted according to the actual engineering experience and equipment characteristics. By collecting a large amount of historical operation data, analyzing the distribution state of insulation resistance value and dielectric loss factor, combining expert experience and equipment characteristics, multiple threshold levels are set, and different warning and disposal measures are taken according to different threshold levels. When the insulation resistance value or dielectric loss factor exceeds the threshold range, the insulation state is judged to be abnormal and an early warning signal is triggered. The built-in system associates the abnormal insulation state data with the dust particle movement and concentration data at the corresponding time, uses the timestamp as the association field, aligns the two types of data according to time, forms an associated data set containing dust particle concentration, movement trend and insulation state fields, and constructs the second data set. The built-in system analyzes and mines the second data set, judges dust pollution and insulation state abnormality by manual annotation or setting certain threshold conditions, and associates the annotation results with other fields in the second data set to form a complete training data set. Using the AdaBoost algorithm, an association model between dust pollution and insulation state abnormality is established to obtain the probability distribution of cable joint insulation state abnormality under different dust particle concentrations and movement trends. Based on the analysis results of the association model, the built-in system dynamically adjusts the operating parameters and cleaning strategy of the dust removal instrument. A fuzzy control algorithm is designed to dynamically adjust the operating frequency, cleaning intensity and dust removal cycle parameters of the dust removal instrument according to the predicted dust particle concentration and insulation state abnormality probability. The control algorithm needs to balance the dust removal effect and energy consumption to achieve the best overall performance. An adaptive learning mechanism is introduced to adjust and optimize the parameters of the control algorithm online based on actual operation feedback and effect evaluation. According to the abnormal probability of insulation state predicted by the model, the operation frequency and cleaning intensity of the dust removal instrument are divided into three levels: low, medium and high, and a probability threshold is preset for each level.When the predicted dust particle concentration or insulation state abnormality probability exceeds the low-level probability threshold, the control algorithm increases the operating frequency and cleaning intensity of the dust removal instrument, and increases the number and time of dust removal of the cable joint; when the predicted dust particle concentration and insulation state abnormality probability are lower than the low-level probability threshold, the control algorithm reduces the operating frequency and cleaning intensity of the dust removal instrument, extends the dust removal cycle of the cable joint, and saves energy and resource consumption. The built-in system can predict the high-risk time window of abnormal insulation state of the cable joint in the future based on the analysis results of the correlation model, and perform preventive dust removal and cleaning on the cable joint in advance before the high-risk time window to reduce the risk of abnormal insulation state.

[0139] For example, the built-in system of the dust removal instrument obtains the output results of the particle swarm optimization model in real time at a rate of 150Mbps through the wireless network of the IEEE 802.11ac standard, including the movement trend and concentration change prediction data of dust particles in the next 24 hours. The built-in system uses JSON format to parse the prediction data, extract the position coordinates (accuracy of 0.1 meters), movement speed (accuracy of 0.1 meters / second) and direction (accuracy of 1 degree) of dust particles, and the numerical change curve of dust concentration (time granularity of 1 hour). The built-in system performs spatial correlation analysis on the parsed data and the GPS location information of the dust removal instrument (accuracy of 1 meter) to identify the high-pollution areas (dust concentration exceeds 1000 / cubic centimeter) on the future operation path of the dust removal instrument. The insulation state sensor adopts the capacitive measurement principle, with a range of 0.1pF to 10pF, an accuracy of 0.01pF, and collects data every 1 second. The sensor is calibrated at the factory to obtain the safety threshold range of insulation resistance value and dielectric loss factor (insulation resistance value greater than 100MΩ, dielectric loss factor less than 0.01). The built-in system aligns the abnormal insulation status data (data beyond the threshold range) with the dust particle data at the corresponding time, and constructs an associated data set containing fields such as timestamp, dust particle concentration, movement speed, movement direction, insulation resistance value and dielectric loss factor. The built-in system cleans the associated data set, removes missing values ​​and outliers, and then scales the data using the Z-score normalization method. Using the AdaBoost algorithm, with dust particle concentration, movement speed and movement direction as features, and whether the insulation status is abnormal as a label, a binary classification association model of dust pollution and insulation status abnormality is established. The model uses 10-fold cross validation, and the average accuracy rate reaches more than 95%. Based on the association model, a fuzzy control algorithm is designed for the built-in system. According to the abnormal insulation state probability predicted by the model, the operation frequency and cleaning intensity of the dust removal instrument are divided into three levels: low, medium and high, with corresponding probability thresholds of 0.2, 0.5 and 0.8 respectively. When the predicted probability is lower than 0.2, the dust removal instrument runs once a day with a cleaning intensity of level 1; when the predicted probability is between 0.2 and 0.5, the dust removal instrument runs twice a day with a cleaning intensity of level 2; when the predicted probability is higher than 0.5, the dust removal instrument runs three times a day with a cleaning intensity of level 3. The operation time of the dust removal instrument is determined by the high pollution time window predicted by the particle swarm optimization model. It starts running one hour before the high pollution time window and stops running one hour after the end of the high pollution time window. Through the fuzzy control algorithm, the dust removal instrument can dynamically adjust the operation strategy according to the predicted dust pollution situation and the risk of abnormal insulation state, while ensuring the dust removal effect, saving energy and cost to the maximum extent.

[0140] Step S105, obtaining the working intensity of the dust removal instrument through the fuzzy logic controller in the dust removal instrument; the input parameters of the fuzzy logic controller include dust movement trend, concentration change and insulation status of the cable connector, and the output parameters are the working intensity of the dust removal instrument; the working intensity of the dust removal instrument includes voltage, current, working frequency, vibration frequency and wind speed.

[0141] Specifically, the movement trend and concentration changes of dust particles and the insulation status of cable joints are obtained;

[0142] Classify dust concentration into: low dust concentration, medium dust concentration and high dust concentration;

[0143] The dust movement trend is divided into: stable dust movement trend, slowly rising dust movement trend and rapidly rising dust movement trend;

[0144] The insulation status of the cable joint is divided into: normal insulation status of the cable joint, slightly abnormal insulation status of the cable joint and seriously abnormal insulation status of the cable joint;

[0145] According to Mamdan i reasoning method, a fuzzy rule base is constructed;

[0146] Through the centroid defuzzification method, the result of fuzzy reasoning is converted into the specific working parameters of the dust removal instrument; and an instruction is issued to make the dust removal instrument work according to the specific working parameters.

[0147] More specifically, the dust removal instrument includes a fuzzy logic controller, the input parameters of the fuzzy logic controller include dust movement trends, concentration changes and insulation status of cable connectors, and the output parameters include the working intensity of the dust removal instrument, including voltage, current, working frequency, vibration frequency and wind speed.

[0148] The fuzzy logic controller in the dust removal instrument obtains the dust particle movement trend, concentration change and cable joint insulation status as input parameters in real time through the RS-485 bus. Among them, the dust particle movement trend and concentration change data come from the dust sensor of the dust removal instrument, which is collected once at a certain time through the RS-485 bus; the cable joint insulation status data comes from the insulation status monitoring device, which is collected once at a certain time through the RS-485 bus. The controller uses a triangular membership function to fuzzify the input parameters and divide the dust concentration into three fuzzy subsets, including low dust concentration, medium dust concentration and high dust concentration. The dust movement trend is divided into stable dust movement trend, slowly rising dust movement trend and rapidly rising dust movement trend. The cable joint insulation status is divided into normal cable joint insulation status, slightly abnormal cable joint insulation status and seriously abnormal cable joint insulation status. The fuzzy logic controller builds a fuzzy rule base based on the Mamdan i reasoning method. The premise part of the rule is composed of fuzzy subsets of input parameters, and the conclusion part is composed of fuzzy subsets of output parameters. If the dust concentration is high, the dust movement trend is rising rapidly, and the insulation state of the cable joint is seriously abnormal, then the working intensity of the dust removal instrument is high. The fuzzy logic controller uses the centroid defuzzification method to convert the results of fuzzy reasoning into specific working parameters of the dust removal instrument. Assume that the membership function of the working intensity of the dust removal instrument obtained by fuzzy reasoning is μ(x), where x is the value of the working intensity. Discretize μ(x) into n points, the horizontal coordinate of each point is xi, the vertical coordinate is μ(xi), i=1, 2,..., n. Calculate the horizontal coordinate of the center of gravity as:

[0149] x*=(x1μ(x1)+x2μ(x2)+…+xnμ(xn)) / (μ(x1)+μ(x2)+…+μ(xn)).

[0150] x is the specific value of the working intensity of the dust removal instrument after defuzzification. The controller divides the working voltage of the dust removal instrument into three fuzzy subsets: low, medium and high. The working current is divided into three fuzzy subsets: low, medium and high. The working frequency is divided into three fuzzy subsets: low, medium and high. The vibration frequency is divided into three fuzzy subsets: low, medium and high. The working wind speed is divided into three fuzzy subsets: low, medium and high. According to the position of x* in each fuzzy subset, the specific values ​​of the corresponding working voltage, current, frequency, vibration frequency and wind speed are determined and sent to the actuator of the dust removal instrument through the RS-485 bus. The actuator of the dust removal instrument adjusts the output voltage and current of the power module according to the received working parameters, controls the vibration frequency of the vibration motor, changes the speed and air volume of the dust suction fan, and realizes precise control of the dust removal intensity. The fuzzy logic controller also has adaptive learning ability. By online monitoring of the actual operation effect of the dust removal instrument, the neural network algorithm is used to dynamically optimize and adjust the fuzzy rule base. A feedforward neural network is used. The input parameters include dust concentration, dust movement trend and insulation status of cable joints. The output parameters include the working voltage, current, frequency, vibration frequency and wind speed of the dust removal instrument. The activation function of the hidden layer and the output layer is the sigmoid function. The neural network is trained using the back propagation algorithm. At regular intervals, a batch of new data is collected through the dust sensor and insulation status monitoring device of the dust removal instrument as training samples, and the neural network is incrementally trained to achieve dynamic optimization of the fuzzy rule base. The optimization goal is to minimize the energy consumption and dust concentration of the dust removal instrument and maximize the insulation resistance value of the cable joint. By setting appropriate weight coefficients, the multi-objective optimization problem is converted into a single-objective optimization problem.

[0151] For example, the fuzzy logic controller obtains the dust particle movement trend and concentration change data from the dust sensor of the dust removal instrument every 1 second at a baud rate of 9600bps through the RS-485 bus, and obtains the cable joint insulation status data from the insulation status monitoring device every 5 seconds. The controller uses a triangular membership function to fuzzy the input parameters, dividing the domain of dust concentration into 0-2000 / cubic centimeter, the domain of dust movement trend into 0-30 / cubic centimeter / hour, and the domain of cable joint insulation status into 0-2000MΩ. The Mamdan i reasoning method uses the minimum membership and maximum membership synthesis rules to match the fuzzy subsets of the input parameters through 27 I F-THEN rules to calculate the membership of the output parameters. For example, the matching degree of the rule "IF dust concentration is high AND dust movement trend is rising rapidly AND cable joint insulation state is seriously abnormal THEN dust removal instrument working intensity is high" is min(μhigh(x concentration),μrapidly rising(x trend),μseverely abnormal(x insulation))=0.8. The centroid defuzzification method discretizes the membership distribution curve μ(x) of the output parameter in its domain into 100 points, and the formula is as follows:

[0152] x*=(x1μ(x1)+x2μ(x2)+……+x100μ(x100)) / (μ(x1)+μ(x2)+……+μ(x100))

[0153] According to the above formula, the horizontal coordinate x of the center of gravity is calculated as the specific work intensity value after defuzzification. If x*=0.75, the fuzzy subset partitioning table of the working parameters is queried to obtain the working voltage of 250V, the working current of 2.25A, the working frequency of 26.25Hz, the vibration frequency of 32.5Hz, and the working wind speed of 17.25m / s, which are sent to the actuator of the dust removal instrument through the RS-485 bus at a baud rate of 19200bps. The three-layer feedforward neural network is incrementally trained through 100 sets of new data collected by the dust removal instrument every hour, using the mean square error as the loss function, and the batch gradient descent method to update the weight matrix. The learning rate is 0.01, the momentum factor is 0.9, and the maximum number of iterations is 1000. The energy consumption weight coefficient is set to 0.5, the dust concentration weight coefficient is set to 0.3, and the insulation resistance weight coefficient is set to 0.2. The original objective function f(x) = 0.5 energy consumption + 0.3 dust concentration - 0.2 * insulation resistance value is transformed into minimizing f(x). The genetic algorithm is used to search for the optimal weight coefficient combination. The population size is 50, the crossover probability is 0.8, the mutation probability is 0.1, and the evolution is 100 generations. The optimal weight coefficient combination [0.6, 0.2, 0.2] is obtained as the training target of the neural network model, so that the fuzzy rule base can be dynamically optimized according to changes in environmental conditions and equipment status.

[0154] Step S106, judging whether the insulation performance of the cable connector meets the safety standard according to the insulation state of the cable connector, if the insulation performance of the cable connector meets the safety standard, issuing an instruction to perform the plugging operation of the cable connector; if the insulation performance of the cable connector does not meet the safety standard, inputting the second data set into the fuzzy logic controller, obtaining the output parameters and issuing an instruction to make the dust removal device perform the dust removal operation according to the output parameters of the fuzzy logic controller; by analyzing the insulation state of the cable connector during dust removal, continuously issuing instructions to make the dust removal device perform the dust removal operation according to the output parameters of the fuzzy logic controller until the insulation performance of the cable connector meets the safety standard.

[0155] Specifically, the insulation status index threshold is preset, the insulation resistance value and dielectric loss factor of the cable joint are collected, and the insulation status index IS is calculated:

[0156] IS=0.7×(R / R0)+0.3×(tanδ0 / tanδ),

[0157] Where R is the measured insulation resistance value, R0 is the insulation resistance threshold, tanδ is the measured dielectric loss factor, and tanδ0 is the dielectric loss factor threshold;

[0158] When IS is greater than or equal to the insulation state index threshold, it is determined that the insulation performance of the cable joint meets the safety standard; when IS is less than the insulation state index threshold, it is determined that the insulation performance of the cable joint does not meet the safety standard;

[0159] If the insulation performance of the cable connector reaches the safety standard, an instruction to perform the plugging operation of the cable connector is issued;

[0160] If the insulation performance of the cable joint does not meet the safety standard, the second data set is input into the fuzzy logic controller;

[0161] According to Mamdan i reasoning method, a fuzzy rule base is constructed;

[0162] Through the centroid defuzzification method, the result of fuzzy reasoning is converted into the specific working parameters of the dust removal instrument; and an instruction is issued to make the dust removal instrument work according to the specific working parameters;

[0163] When the insulation state index IS is greater than the insulation state index threshold for multiple consecutive times, it is determined that the insulation performance of the cable joint has reached the safety standard, and an instruction to perform the plugging operation of the cable joint is issued;

[0164] Collect the insertion depth of the connector, the displacement and speed of the plugging mechanism, calculate the deviation between the actual insertion depth of the connector and the target depth, and adjust the movement of the plugging mechanism according to the deviation between the actual insertion depth and the target depth until the plugging operation of the cable connector is completed;

[0165] After completing the plugging operation of the cable connector, collect the axial compression force and judge whether the required tightening degree is achieved based on the axial compression force.

[0166] More specifically, based on the insulation state of the cable connector, it is determined whether the insulation performance of the cable connector meets the safety standard; if the insulation performance does not meet the standard, a second data set is input into the fuzzy logic controller so that the dust removal instrument performs dust removal operations according to the output parameters of the fuzzy logic controller; the insulation state of the cable connector during dust removal is analyzed, and the dust removal operation is continued until the safety standard is met; if the insulation state meets the safety standard, an instruction is issued to perform the plugging operation of the cable connector.

[0167] The insulation state monitoring device collects the insulation resistance value and dielectric loss factor of the cable joint in real time, and transmits the collected data to the insulation performance evaluation module through industrial Ethernet. The insulation performance evaluation module sets the insulation resistance threshold and the dielectric loss factor tanδ threshold according to the voltage level of the cable joint. The evaluation module divides the collected insulation resistance value and dielectric loss factor by their thresholds respectively to obtain relative values, and calculates the insulation state index IS=0.7×(R / R0)+0.3×(tanδ0 / tanδ), where R is the measured insulation resistance value, R0 is the insulation resistance threshold, tanδ is the measured dielectric loss factor, and tanδ0 is the dielectric loss factor threshold. When IS is greater than or equal to the insulation state index threshold, it is determined that the insulation performance meets the standard, and when IS is less than the insulation state index threshold, it is determined that the insulation performance does not meet the standard. If the insulation performance evaluation result does not meet the standard, the insulation performance evaluation module immediately sends a warning signal to the data fusion module, and the data fusion module transmits the insulation state monitoring data, dust concentration and movement trend data in the second data set to the fuzzy logic controller through the ModbusTCP protocol. The fuzzy logic controller adopts Mamdan i-type fuzzy reasoning system and sets several fuzzy rules in the form of I F-THEN. If the insulation state index IS is small and the dust concentration is high, the dust removal intensity should be increased. The antecedent of the fuzzy rule is composed of the fuzzy subsets of the input variables, and the consequent is composed of the fuzzy subsets of the output variables. The minimum membership method is used for rule matching and activation. In the fuzzy reasoning process, the reasoning results of each rule are aggregated to obtain the fuzzy subsets of the output variables. The centroid method is used for defuzzification to obtain the clear value of the dust removal intensity, and then the optimal working parameter combination of the dust removal instrument is calculated, including the working voltage, current, frequency, vibration frequency and wind speed, and the control command is sent to the actuator of the dust removal instrument through the ModbusTCP protocol. According to the received control command, the actuator of the dust removal instrument starts the dust removal operation, adjusts the output voltage and current of the power module, controls the vibration frequency of the vibration motor, changes the speed and air volume of the dust suction fan, and performs centralized dust removal on the cable joint and its surrounding environment. During the dust removal operation, the insulation status monitoring device continuously collects the insulation parameters of the cable joint, and the insulation performance evaluation module calculates the insulation status index IS in real time. When the evaluation results are greater than the insulation status index threshold for multiple consecutive times, the dust removal operation is judged to be effective, and a dust removal completion signal is sent to the data fusion module. After receiving the dust removal completion signal, the data fusion module sends a stop work instruction to the dust removal instrument and writes the control bit in the Modbus holding register of the dust removal instrument. After receiving the stop work instruction, the dust removal instrument compares the value of the control bit with the preset stop flag bit. If they are consistent, the dust removal operation is stopped immediately, and a stop confirmation signal is returned to the data fusion module. At the same time, the insulation status monitoring data of the cable joint and the dust removal operation parameter information are recorded in the database to form a cable joint dust removal operation report, which provides data support for subsequent preventive tests and status maintenance.After the dust removal instrument stops working, the data fusion module sends a connector ready signal to the cable connector operation control module. After receiving the ready signal, the cable connector operation control module performs the plug-in operation of the cable connector. An inductive displacement sensor is installed on the outer shell of the cable connector to measure the insertion depth of the connector in real time. A photoelectric encoder is installed on the plug-in mechanism to measure the displacement and speed of the plug-in mechanism in real time. The deviation between the actual insertion depth of the connector and the target depth is calculated through the feedback signal of the displacement sensor and the encoder, and the movement of the plug-in mechanism is adjusted through the PID control algorithm to achieve precise alignment and plug-in, and the male and female connectors are inserted into place. After the plug-in is completed, the axial compression force of the cable connector is measured by the pressure sensor to determine whether the required tightening degree is achieved. The cable path conductivity is tested by the contact conduction tester to ensure reliable plug-in. After the plug-in operation is completed, the cable connector operation control module reads the latest data of the insulation status monitoring device and conducts a second confirmation of the insulation performance of the cable connector. If the test results meet the standards for multiple consecutive times, the insulation maintenance and plug-in operation of the entire cable connector are completed, and the system returns to the normal monitoring state. During the entire process, the fuzzy logic controller dynamically optimizes fuzzy rules and membership functions based on historical data and expert experience using an adaptive adjustment mechanism, and uses an objective function that integrates dust removal effect and energy consumption to continuously improve dust removal performance.

[0168] Exemplarily, the insulation state monitoring device collects the insulation resistance value and dielectric loss factor of the cable joint every 1 minute, with a resolution of 0.1MΩ and 1×10-6 respectively, and transmits it to the insulation performance evaluation module through the RS-485 bus at a baud rate of 9600bps. The evaluation module adopts a segmented threshold judgment method. For cable joints with voltage levels of 10kV and below, the insulation resistance threshold R0 is 100MΩ, and the dielectric loss factor threshold tanδ0 is 40×10-4. After calculating the relative value, the weighted average formula IS=0.7×(R / R0)+0.3×(tanδ0 / tanδ) is used to obtain the insulation state index, and the weight coefficient is optimized by the decision tree algorithm. When IS is less than 0.9 for three consecutive times, the early warning interrupt of the data fusion module is triggered through the Modbus TCP protocol, and the collected 512 bytes of data are encapsulated in the Modbus data frame and sent to the fuzzy logic controller. Based on the preset 125 fuzzy rules, the fuzzy logic controller uses the maximum-minimum synthetic reasoning method, obtains the dust removal intensity clarity value through the central average defuzzification method, and then determines the five working parameters of the dust removal instrument through the table lookup method, among which the working voltage range is 100V-300V, the step length is 10V, and the frequency range is 10Hz-30Hz, the step length is 1Hz. After receiving the Modbus control command, the dust removal instrument writes the dust removal intensity value into the Modbus coil, starts the dust removal operation, and returns the current working status (such as normal, fault, alarm, etc.) to the data fusion module through the Modbus coil every 30 seconds. The insulation performance evaluation module calculates IS every 2 minutes during the dust removal operation. When the IS is greater than or equal to 0.9 for three consecutive times, the Modbus predefined abnormal response code 0x0A is sent to the data fusion module to trigger the dust removal completion event. After receiving the 0x0A response code, the data fusion module writes the stop flag 0x55AA to the Modbus holding register of the dust removal instrument. After the dust removal instrument detects the flag, it stops immediately and writes back 0xAA55 for response confirmation. At the same time, the latest equipment operation data (such as cumulative operation time, number of faults, etc.) is stored in the equipment log table of the database. The range of the inductive displacement sensor on the cable connector housing is 0mm-100mm, the accuracy is 0.1mm, and the data is transmitted every 5 milliseconds through the 4mA-20mA current loop. The resolution of the photoelectric encoder on the plug-in mechanism is 1000 pulses / turn, and the displacement and speed data are transmitted every 1 millisecond through the RS-422 serial interface. The plug-in process control uses a three-closed-loop PID algorithm, with position loop P=2.5, I=0.05, D=0.01, speed loop P=10, I=0.1, D=0.05, and current loop P=20, I=1, D=0.1.After the plug-in is completed, the pressure sensor measures the axial clamping force at a sampling frequency of 0.5Hz, with a range of 0N-1000N and an accuracy of 1N. When the clamping force is stable within the range of 800N±10N, the plug-in is determined to be in place. The continuity tester outputs a 5V, 50mA DC signal. When the on-resistance is less than 1Ω, the conduction is determined to be successful. Finally, the cable joint operation control module reads the data of the insulation status monitoring device and performs a t test with a confidence level of 95%. When the t values ​​are within the critical value for three consecutive times and the on-resistance is less than 1Ω for three consecutive times, the "qualified" mark is written to the joint status database to complete the plug-in operation. Based on the dust removal effect and energy consumption data of the past week, the fuzzy logic controller performs self-learning optimization once a day, uses a genetic algorithm to find the best, with a crossover probability of 0.8, a mutation probability of 0.1, and 100 iterations, and updates the fuzzy rule base of the optimal individual to the controller's EPROM. The objective function FIT = 0.8 × (IS1-IS0) + 0.2 × (P0-P1), where IS0 and IS1 represent the insulation state index before and after dust removal, respectively, and P0 and P1 represent the energy consumption per unit time before and after dust removal, respectively. The fuzzy rule base is optimized by maximizing the objective function.

[0169] Step S107, by continuously monitoring the communication status between the cloud server, the cable connector and the dust removal instrument, ensure the stability of data transmission and collect real-time plugging data during the plugging process; analyze the real-time plugging data through the cloud server application data acquisition module to evaluate the accuracy and safety of the plugging operation; the real-time plugging data includes torque, insertion force and dust concentration.

[0170] Specifically, the communication status between the cloud server, cable connector, and dust removal instrument is monitored through the heartbeat mechanism, and the round-trip delay is recorded;

[0171] The data transmission traffic between the cloud server, cable connector and dust removal instrument is shaped and restricted through the token bucket algorithm; the token generation rate is dynamically adjusted according to the water level of the token passing through;

[0172] When the water level of the token bucket is higher than a certain percentage for a period of time, a traffic anomaly alarm is triggered;

[0173] Collect the torque, insertion force and dust concentration in the plugging environment during the plugging process as real-time plugging data;

[0174] Preprocess the real-time plug-in data and compress the data through the run-length encoding algorithm;

[0175] The real-time plug-in data is filtered and smoothed by a Kalman filter algorithm to obtain the plug-in data;

[0176] Establishing a plugging operation evaluation model for judging the accuracy and safety of the current plugging operation through a decision tree algorithm, inputting the plugging data into the plugging operation evaluation model; judging the accuracy and safety of the current plugging operation;

[0177] Collect the plug-in data, perform cluster analysis on the plug-in data, use the elbow method to evaluate different clustering number K values, obtain the optimal clustering number, classify the plug-in data through the K-means algorithm, and count the number, proportion and parameter distribution status of various plug-in operations to generate a plug-in operation quality report.

[0178] More specifically, during the plugging operation, the communication status between the cloud server, cable connector and dust removal instrument is continuously monitored to ensure the stability of data transmission, and real-time plugging data during the plugging process is collected, including torque, insertion force and dust concentration; the cloud server applies a data acquisition module to analyze the real-time plugging data and evaluate the accuracy and safety of the plugging operation.

[0179] The communication status monitoring module adopts a heartbeat mechanism, sending a heartbeat packet to the cloud server, cable connector and dust removal instrument at regular intervals, and recording the round-trip delay. If the round-trip delay of multiple consecutive heartbeat packets exceeds a certain time, or there is no response to multiple consecutive heartbeat packets, a communication abnormality alarm is triggered, and the timestamp, node ID and abnormality type information of the abnormal moment are written into the communication log database. The communication status monitoring module also monitors the data transmission traffic between the cloud server, cable connector and dust removal instrument, and uses the token bucket algorithm to shape and limit the traffic. The token generation rate is adjusted according to the statistical distribution of historical data transmission traffic and the volatility of traffic. The token bucket capacity is set according to the network delay and jitter characteristics to ensure that the impact of burst traffic on the network can be effectively alleviated. When the water level of the token bucket is higher than a certain percentage for a period of time, a traffic abnormality alarm is triggered, and the token generation rate is dynamically adjusted according to the water level to achieve adaptive control of traffic. The cable connector is equipped with a torque sensor, a pressure sensor and a dust concentration sensor. The torque sensor and the pressure sensor collect the torque and insertion force data during the plugging process at a set frequency, and the dust concentration sensor collects the dust concentration data of the plugging environment at a set frequency. The sensor uses the industrial Ethernet protocol EtherCAT to communicate with the data acquisition module. The data acquisition module preprocesses the real-time plugging data, including timestamp alignment, data format conversion and outlier filtering, and compresses the data using the run-length encoding algorithm. The processed real-time plugging data is packaged into JSON format and sent to the cloud server through the MQTT protocol in a reliable transmission mode of QoS level 1. After receiving the real-time plugging data, the data analysis engine of the cloud server uses the Kalman filter algorithm to filter and smooth the data, obtain the plugging data, and input the plugging data into the plugging operation evaluation model. The model adopts decision tree algorithm, based on the statistical distribution of historical plugging operation data, extracts 11 characteristic indicators, including peak value, slope, fluctuation range, root mean square value, waveform factor of torque signal, rise time, peak time, steady-state value of insertion force signal, average value, variance, and exceedance rate of dust concentration signal, and uses ReliefF algorithm to automatically select the key feature subset that has the greatest impact on plugging quality from the candidate feature set. A decision tree is generated using CHAID algorithm, and the optimal parameter combination of decision tree is selected through grid search and 10-fold cross validation method, including the maximum depth of split nodes, minimum number of samples and impurity measurement. The plugging data flows through the decision tree classifier to judge the accuracy and safety of the current plugging operation. When the instantaneous value of torque or insertion force exceeds the safety threshold for multiple times in a row, it is judged that there is a safety hazard in the plugging operation, and the safety alarm is immediately triggered to notify the cable connector operation control module to suspend plugging. According to the standard deviation of torque, slope of insertion force and dust concentration, it is judged whether the plugging operation is correct.If the plugging operation is correct, the current plugging parameters and environmental parameters are recorded in the plugging operation log database as an excellent plugging case. The data analysis engine performs cluster analysis on the recent plugging data at regular intervals, uses the elbow method to evaluate different clustering number K values, selects the K value corresponding to the inflection point where the SSE decrease rate slows down as the optimal clustering number, and uses the K-means algorithm to cluster the plugging data into three categories: excellent plugging, qualified plugging, and unqualified plugging. The number, proportion, and parameter distribution of each type of plugging operation are counted, and a plugging operation quality report is generated and pushed to on-site operators and managers through Web services. The data analysis engine uses an incremental learning algorithm. Whenever the cumulative newly added plugging operation data reaches a certain threshold, a certain amount of data is randomly selected as a training set, and the remaining data is used as a test set. The decision tree classifier and K-means clusterer are retrained and evaluated, and grid search and cross-validation are used to continuously optimize the parameters of the decision tree and clusterer to continuously improve the accuracy and reliability of the plugging operation quality evaluation.

[0180] Exemplarily, the communication status monitoring module adopts a heartbeat mechanism based on the UDP protocol. The heartbeat packet size is 64 bytes, including node ID, timestamp, sequence number and other information. It is sent through the MQTT protocol with a QoS level of 0. The round-trip delay is calculated by the Ping-Pong algorithm. If there are three consecutive timeouts (RTT>500ms) or two packet losses, the node status is marked as abnormal, and the timestamp, duration, abnormal type and other information of the event are recorded and written into the abnormal log table of the MySQL database. The data transmission traffic is collected in real time through the network traffic probe once per second, and the token bucket algorithm is used for traffic shaping. The token generation rate is set to 100Mbps and the bucket depth is set to 8MB. When the token bucket water level is higher than 80% for 5 consecutive seconds, the traffic abnormality alarm is triggered, and the alarm time, traffic value, token bucket parameters and other information are written into the traffic abnormality log table, and the cloud management system is notified through Webhooks to dynamically adjust the token generation rate, with an adjustment step of 10Mbps. The cable connector uses an integrated spoke-type six-dimensional force sensor, which can simultaneously measure forces and torques in three orthogonal directions. The ranges are Fx / Fy±1000N, Fz±2000N, Mx / My / Mz±50Nm, and the resolutions are 0.1N and 0.01Nm, respectively. The data is uploaded every 1ms through the EtherCAT bus with a bandwidth of 100MHz. The dust concentration sensor adopts the principle of laser scattering, with a range of 0-10000 / cubic centimeter, a resolution of 1 / cubic centimeter, and a sampling frequency of 10Hz. After pre-processing such as linear interpolation, outlier removal, and digital filtering, the sensor data is compressed using Huffman coding, with a compression ratio of 3:1, and then packaged into JSON format and sent to the cloud data receiving module through the MQTT protocol (QoS level 1), with a transmission delay of less than 100ms. The plug-in operation evaluation model builds a decision tree based on the C4.5 algorithm, selects the optimal features through the information gain ratio, and dynamically prunes through the heuristic method. The feature vector includes 8 indicators such as the peak value, slope, and fluctuation range of the force signal, and 3 indicators such as the mean, standard deviation, and exceedance rate of the dust concentration signal. The training set contains 10,000 historical plug-in cases, 10-fold cross validation, and the minimum number of leaf node samples is set to 20, and the accuracy rate reaches more than 95%. When the torque exceeds 45Nm or the insertion force exceeds 850N, the decision tree outputs a safety alarm and triggers the plug-in operation to be suspended through the OPCUA protocol; when the feature vector falls into the normal plug-in range, the plug-in parameters are written into the excellent plug-in case table of the SQLite database. The K-means clustering algorithm uses the maximum expectation algorithm for parameter estimation, and the elbow method determines the optimal cluster number K=3. Parallel computing is achieved through MapReduce, processing 10 minutes of data per minute, and generating a clustering report that is pushed to the Grafana dashboard.Incremental learning is triggered every time 1,000 new data are added. The Mini-BatchK-Means algorithm updates the cluster center, and the AdaBoost algorithm updates the decision tree parameters. The training time is less than 10 seconds.

[0181] In the description of the present disclosure, it is necessary to understand that the directions or positional relationships indicated by directional words such as "front, back, up, down, left, right", "lateral, vertical, horizontal" and "top, bottom" are usually based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present disclosure and simplifying the description. Unless otherwise specified, these directional words do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the scope of protection of the present disclosure.

[0182] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all of these changes and deformations should fall within the protection scope of the claims of the present disclosure.

Claims

1. An intelligent inductive cable quick connector plugging method, characterized in that: The following steps are involved: The dust particle data in the air is detected in real time by the environmental sensor, the size and concentration of the dust particles are recorded, and the collected dust particle data is transmitted to the cloud server through the wireless network; the environmental sensor includes an optical particle counter, a laser radar and a camera device; the camera device is used to capture the movement path and speed of the dust particles; the environmental air pressure data is detected in real time by the air pressure monitoring system and transmitted to the cloud server; Removing noise and errors from the dust particle data through a cloud server to obtain processed dust particle data, combining the processed dust particle data with ambient air pressure data, obtaining a first data set through a Kalman filter algorithm, and storing the first data set in a predefined format; Setting initial parameters of a particle swarm optimization model through a cloud server, and inputting the first data set into the particle swarm optimization model to train the particle swarm optimization model; the initial parameters include the number, speed and position of particles; the particle swarm optimization model is used to generate a prediction result including a movement trend of dust; The output results of the particle swarm optimization model are obtained through the built-in system of the dust removal instrument, and the movement trend and concentration change of the dust particles are analyzed; the insulation state of the cable joint is obtained through the insulation state sensor of the dust removal instrument; a second data set is constructed by combining the movement trend of the dust particles, the concentration change and the insulation state of the cable joint; the second data set is used to adjust the operation frequency and cleaning intensity of the dust removal instrument; The working intensity of the dust removal instrument is obtained by a fuzzy logic controller in the dust removal instrument; the input parameters of the fuzzy logic controller include dust movement trend, concentration change and insulation state of cable joints, and the output parameters are the working intensity of the dust removal instrument; the working intensity of the dust removal instrument includes voltage, current, working frequency, vibration frequency and wind speed; According to the insulation state of the cable joint, it is judged whether the insulation performance of the cable joint reaches the safety standard. If the insulation performance of the cable joint reaches the safety standard, an instruction for plugging the cable joint is issued; if the insulation performance of the cable joint does not reach the safety standard, the second data set is input into the fuzzy logic controller, the output parameters are obtained, and an instruction is issued to make the dust removal device perform the dust removal operation according to the output parameters of the fuzzy logic controller; by analyzing the insulation state of the cable joint during dust removal, the instruction for making the dust removal device perform the dust removal operation according to the output parameters of the fuzzy logic controller is continuously issued until the insulation performance of the cable joint reaches the safety standard; By continuously monitoring the communication status between the cloud server, cable connector and dust removal instrument, the stability of data transmission is ensured and real-time plugging data during the plugging process is collected; the real-time plugging data is analyzed by the cloud server application data acquisition module to evaluate the accuracy and safety of the plugging operation; the real-time plugging data includes torque, insertion force and dust concentration.

2. The intelligent inductive cable quick connector plugging method according to claim 1, characterized in that: The environmental sensor detects the dust particle data in the air in real time, records the size and concentration of the dust particles, and transmits the collected dust particle data to the cloud server through a wireless network; the environmental sensor includes an optical particle counter, a laser radar, and a camera device; the camera device is used to capture the movement path and speed of the dust particles; The air pressure monitoring system detects ambient air pressure data in real time and transmits it to the cloud server, including: The optical particle counter detects the dust particle data in the air in real time to obtain the size and concentration information of the particles; the laser radar captures the movement path and speed of the particles; the camera equipment records the visual image data of the particles; The detected dust particle size, concentration, movement path and movement speed as well as visual image data are transmitted to the cloud server in real time via wireless network for summary processing; The detected dust particle size, concentration, movement path and movement speed and visual image data are cleaned and standardized by a data preprocessing method of data cleaning and normalization, and a dust particle data model is established by a support vector machine (SVM) and a random forest algorithm; According to the established dust particle data model, through time series analysis and regression analysis prediction methods, the concentration trend of dust particles in the air in the future period is predicted in real time to determine the change trend of air quality; The air pressure monitoring system detects ambient air pressure data in real time and transmits the data to the cloud server; The cloud server correlates and analyzes the air pressure data with the dust particle data to establish a prediction model between environmental meteorological conditions and air quality; Estimate the air quality change trend in the future through meteorological forecast data.

3. The intelligent inductive cable quick connector plugging method according to claim 1 is characterized in that: The method of removing noise and error in the dust particle data by using a cloud server to obtain processed dust particle data, combining the processed dust particle data with ambient air pressure data, obtaining a first data set by using a Kalman filter algorithm, and storing the first data set in a predefined format includes: The high-frequency noise and random errors in the data are removed by wavelet threshold denoising method; The signal processing module is used to extract the features of the filtered dust particle data to obtain key feature parameters, which include the concentration, particle size distribution and movement speed of the dust particles; Acquire real-time air pressure data through an environmental air pressure monitoring system, and synchronize the air pressure data with the processed dust particle data to obtain synchronized dust particle data and air pressure data; fusing the synchronized dust particle data and air pressure data using a Kalman filter algorithm to obtain a first data set; The first data set includes the concentration, particle size distribution and movement speed of dust particles and ambient air pressure data; Organizing and packaging the first data set through a predefined data format and structure, and adding metadata and timestamp information; The first data set is stored in a document-based NoSQL database such as MongoDB.

4. The intelligent inductive cable quick connector plugging method according to claim 1, characterized in that: The initial parameters of the particle swarm optimization model are set through the cloud server, and the first data set is input into the particle swarm optimization model to train the particle swarm optimization model; the initial parameters include the number, speed and position of particles; The particle swarm optimization model is used to generate prediction results including the movement trend of dust, including: Selecting the number of particles according to the dimension, complexity and computing resource factors of the first data set; and generating the initial velocity and position of the particles through a random function; The particle swarm optimization model evaluates the optimization performance of each particle through a fitness function; Setting a convergence condition, wherein the particle swarm optimization model iterates according to the convergence condition to obtain a prediction model for generating a prediction result including a movement trend of dust; The convergence conditions include: Set a maximum number of iterations. When the maximum number of iterations is reached, the particle swarm optimization model is considered to have converged. A minimum fitness threshold is set. When the optimal fitness value of multiple consecutive iterations changes less than the minimum fitness threshold, it is considered that the particle swarm optimization model has converged. Setting a minimum particle position change threshold, when the particle position change of multiple consecutive iterations is less than the minimum particle position change threshold, it is considered that the particle swarm optimization model has converged; If the particle swarm optimization model satisfies one or more of the above three convergence conditions, the particle swarm optimization model completes iteration to obtain a prediction model for generating a prediction result including a motion trend of dust; The prediction model is used to perform prediction analysis on the newly collected dust particle data to generate the dust movement trend in the future.

5. The intelligent inductive cable quick connector plugging method according to claim 1, characterized in that: The output result of the particle swarm optimization model is obtained through the built-in system of the dust removal instrument, and the movement trend and concentration change of the dust particles are analyzed; the insulation state of the cable joint is obtained through the insulation state sensor of the dust removal instrument; and a second data set is constructed by combining the movement trend of the dust particles, the concentration change and the insulation state of the cable joint; The second data set is used to adjust the operating frequency and cleaning intensity of the dust removal instrument, including: The output results of the particle swarm optimization model are obtained in real time through a wireless network; the position coordinates, movement speed and direction information of dust particles at different time points, and the numerical change curve of dust concentration are obtained; Obtain the movement data and concentration data of dust particles through analysis; associate the movement data and concentration data of dust particles with the spatial position information of the dust removal instrument to determine the density and distribution state of dust particles encountered by the dust removal instrument in the future operation process; Obtain the insulation resistance value threshold range and the dielectric loss factor threshold range, monitor the insulation resistance value and the dielectric loss factor at the cable joint in real time through the insulation state sensor of the dust removal instrument, and compare the collected insulation resistance value with the preset insulation resistance value threshold range, and compare the collected dielectric loss factor with the preset dielectric loss factor threshold range; When the insulation resistance value exceeds the preset insulation resistance value threshold range, or when the dielectric loss factor exceeds the preset dielectric loss factor threshold range, it is determined that the insulation state is abnormal and a warning signal is triggered; Correlating the abnormal insulation state data with the dust particle movement and concentration data at the corresponding time to construct a second data set; According to the AdaBoost algorithm, a correlation model between dust pollution and abnormal insulation status is established. The probability distribution of abnormal insulation status of cable joints under different dust particle concentrations and movement trends is obtained through the correlation model. Adjust the operating frequency and cleaning intensity of the dust removal equipment according to the predicted dust particle concentration and the probability of abnormal insulation status; According to the abnormal probability of insulation state predicted by the model, the operation frequency and cleaning intensity of the dust removal equipment are divided into low, medium and high levels, and a probability threshold is preset for each level, namely the low probability threshold, the medium probability threshold and the high probability threshold; When the predicted dust particle concentration or insulation state abnormal probability exceeds the low-level probability threshold, the operating frequency and cleaning intensity of the dust removal equipment are increased; When the predicted dust particle concentration and insulation state abnormal probability are lower than the low-level probability threshold, the operating frequency and cleaning intensity of the dust removal instrument are reduced; According to the analysis results of the correlation model, the high-risk time window for abnormal insulation status of the cable joint is predicted. Before the high-risk time window, the cable joint is dusted and cleaned according to the operating frequency and cleaning intensity of the dust removal instrument corresponding to the probability threshold level of the high-risk time window.

6. The intelligent inductive cable quick connector plugging method according to claim 1, characterized in that: The working intensity of the dust removal instrument is obtained by the fuzzy logic controller in the dust removal instrument; the input parameters of the fuzzy logic controller include dust movement trend, concentration change and insulation state of the cable joint, and the output parameter is the working intensity of the dust removal instrument; the working intensity of the dust removal instrument includes voltage, current, working frequency, vibration frequency and wind speed, including: Obtain dust particle movement trends and concentration changes as well as cable joint insulation status; Classify dust concentration into: low dust concentration, medium dust concentration and high dust concentration; The dust movement trend is divided into: stable dust movement trend, slowly rising dust movement trend and rapidly rising dust movement trend; The insulation status of the cable joint is divided into: normal insulation status of the cable joint, slightly abnormal insulation status of the cable joint and seriously abnormal insulation status of the cable joint; According to Mamdani reasoning method, a fuzzy rule base is constructed; Through the centroid defuzzification method, the result of fuzzy reasoning is converted into the specific working parameters of the dust removal instrument; and an instruction is issued to make the dust removal instrument work according to the specific working parameters.

7. The intelligent inductive cable quick connector plugging method according to claim 1, characterized in that: The method comprises: judging whether the insulation performance of the cable joint meets the safety standard according to the insulation state of the cable joint, and issuing an instruction to perform a plugging operation on the cable joint if the insulation performance of the cable joint meets the safety standard; inputting a second data set into a fuzzy logic controller to obtain output parameters and issuing an instruction to make a dust removal device perform a dust removal operation according to the output parameters of the fuzzy logic controller if the insulation performance of the cable joint does not meet the safety standard; and continuously issuing an instruction to make a dust removal device perform a dust removal operation according to the output parameters of the fuzzy logic controller by analyzing the insulation state of the cable joint during dust removal until the insulation performance of the cable joint meets the safety standard, including: Preset the insulation status index threshold, collect the insulation resistance value and dielectric loss factor of the cable joint, and calculate the insulation status index IS: IS=0.7×(R / R0)+0.3×(tanδ0 / tanδ), Where R is the measured insulation resistance value, R0 is the insulation resistance threshold, tanδ is the measured dielectric loss factor, and tanδ0 is the dielectric loss factor threshold; When IS is greater than or equal to the insulation state index threshold, it is determined that the insulation performance of the cable joint meets the safety standard; when IS is less than the insulation state index threshold, it is determined that the insulation performance of the cable joint does not meet the safety standard; If the insulation performance of the cable connector reaches the safety standard, an instruction to perform the plugging operation of the cable connector is issued; If the insulation performance of the cable joint does not meet the safety standard, the second data set is input into the fuzzy logic controller; According to Mamdani reasoning method, a fuzzy rule base is constructed; Through the centroid defuzzification method, the result of fuzzy reasoning is converted into the specific working parameters of the dust removal instrument; and an instruction is issued to make the dust removal instrument work according to the specific working parameters; When the insulation state index IS is greater than the insulation state index threshold for multiple consecutive times, it is determined that the insulation performance of the cable joint has reached the safety standard, and an instruction to perform the plugging operation of the cable joint is issued; Collect the insertion depth of the connector, the displacement and speed of the plugging mechanism, calculate the deviation between the actual insertion depth of the connector and the target depth, and adjust the movement of the plugging mechanism according to the deviation between the actual insertion depth and the target depth until the plugging operation of the cable connector is completed; After completing the plugging operation of the cable connector, collect the axial compression force and judge whether the required tightening degree is achieved based on the axial compression force.

8. The intelligent inductive cable quick connector plugging method according to claim 1, characterized in that: The communication status between the cloud server, the cable connector and the dust removal instrument is continuously monitored to ensure the stability of data transmission and collect real-time plugging data during the plugging process; the real-time plugging data is analyzed by the cloud server application data acquisition module to evaluate the accuracy and safety of the plugging operation; the real-time plugging data includes torque, insertion force and dust concentration, including: The communication status between the cloud server, cable connector and dust removal instrument is monitored through the heartbeat mechanism, and the round-trip delay is recorded; The data transmission traffic between the cloud server, cable connector and dust removal instrument is shaped and restricted through the token bucket algorithm; the token generation rate is dynamically adjusted according to the water level of the token passing through; When the water level of the token bucket is higher than a certain percentage for a period of time, a traffic anomaly alarm is triggered; Collect the torque, insertion force and dust concentration in the plugging environment during the plugging process as real-time plugging data; Preprocess the real-time plug-in data and compress the data through the run-length encoding algorithm; The real-time plug-in data is filtered and smoothed by a Kalman filter algorithm to obtain the plug-in data; Establishing a plugging operation evaluation model for judging the accuracy and safety of the current plugging operation through a decision tree algorithm, inputting the plugging data into the plugging operation evaluation model; judging the accuracy and safety of the current plugging operation; Collect the plug-in data, perform cluster analysis on the plug-in data, use the elbow method to evaluate different clustering number K values, obtain the optimal clustering number, classify the plug-in data through the K-means algorithm, and count the number, proportion and parameter distribution status of various plug-in operations to generate a plug-in operation quality report.

Images