A power distribution cabinet for environmental monitoring based on a sensor network

Abstract

The present invention relates to the technical field of distribution cabinet monitoring, especially a distribution cabinet for environmental monitoring based on a sensor network. By arranging multi-point temperature and humidity sensors on the outer side of the distribution cabinet body and using a cloud monitoring module to uniformly manage data collection and storage, it realizes all-round real-time monitoring of the environment around the distribution cabinet. The dynamic weight allocation algorithm is adopted to generate a comprehensive weight according to the historical anomaly frequency, position and environmental change rate of the sensors and assign it to the temperature and humidity feature matrix. Through the graph attention network to learn data correlation and weight changes, a global feature vector of the environmental state is generated, greatly improving the analytical ability of complex environmental data. At the same time, combined with the Gaussian process regression prediction model, it accurately predicts the future environmental trend. By comparing the deviation interval between the predicted value and the real-time value, combined with the parameter change rate, an alarm score and an alarm level are generated, and real-time alarm information is provided, effectively reducing the manual inspection cost and improving the monitoring accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution cabinet monitoring, and particularly to a power distribution cabinet that realizes environmental monitoring based on a sensor network. Background Art

[0002] With the continuous expansion of the scale and the increasing complexity of modern power distribution systems, the stability and reliability of power distribution equipment have become one of the key factors to ensure power supply safety. As an important device for power distribution, power distribution cabinets are long-term exposed to changing environments, such as high temperature, humidity, and sudden natural disasters (such as typhoons and floods), and are extremely vulnerable to environmental factors, resulting in a decline in equipment performance or even failures. This will not only affect the stability of power supply but may also cause serious safety hazards. The environmental monitoring of traditional power distribution cabinets mainly relies on manual inspections and single-point sensor measurements, and this method has the following deficiencies: Limited monitoring range: Traditional methods usually only focus on a single monitoring point of the power distribution cabinet and cannot comprehensively understand the environmental status around the power distribution cabinet; Lack of real-time performance: The frequency of manual inspections is limited and cannot quickly respond to sudden environmental changes; The traditional system lacks analysis and judgment of monitoring data, and the prediction ability has a high error rate. Summary of the Invention

[0003] To solve the above problems, the present invention provides a power distribution cabinet that realizes environmental monitoring based on a sensor network. By setting up a sensor network and performing deep learning considering environmental parameters, the orientation of environmental parameters, historical anomaly frequencies, and the current environmental change rate, automatic monitoring of the power distribution cabinet is achieved, and the accuracy of predictive analysis is effectively improved.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A power distribution cabinet that realizes environmental monitoring based on a sensor network, comprising: a power distribution cabinet body, a sensor network, and a cloud monitoring module. The cloud monitoring module includes a plurality of temperature sensors and humidity sensors arranged at different points outside the power distribution cabinet body. The sensor network is communicatively connected to the cloud monitoring module;

[0006] The cloud monitoring module includes a sensor management unit, a parameter selection unit, an environmental vector generation unit, and a predictive maintenance unit;

[0007] The sensor management unit is used to store the types of sensors in the sensor network, the orientation relative to the power distribution cabinet body, and the temperature parameters or humidity parameters obtained through the corresponding sensors;

[0008] The parameter selection unit is used to perform weighted fusion on the orientation relative to the distribution cabinet body, the historical anomaly frequency, and the current environmental change rate based on the dynamic weight allocation algorithm to obtain a comprehensive weight, and assign weights to the temperature parameter and the humidity parameter based on the comprehensive weight to generate a temperature feature matrix and a humidity feature matrix;

[0009] The environmental vector generation unit is used to use the temperature feature matrix and the humidity feature matrix as the inputs of the graph attention network at the same time, learn the correlation and weight changes of the temperature and humidity parameters between specific positions through the node attention mechanism, model the global features of the environmental state around the distribution cabinet, and generate a multi-dimensional feature vector for describing the environmental state;

[0010] The predictive maintenance unit is used to construct a Gaussian process regression prediction model based on the multi-dimensional feature vector, calculate the anomaly probability distribution by comparing the deviation interval between the predicted value generated by the Gaussian process regression prediction model and the real-time value, generate an alarm level in combination with the parameter change rate, and send maintenance information to the maintenance personnel terminal according to the alarm level.

[0011] Further, the weighted fusion of the orientation relative to the distribution cabinet body, the historical anomaly frequency, and the current environmental change rate based on the dynamic weight allocation algorithm includes the following steps:

[0012] Based on the orientation of the sensor relative to the distribution cabinet body stored in the sensor management unit, combined with the key areas of the distribution cabinet and the preset weight area influence factor, calculate the initial position weight of each sensor and store it in the position weight set;

[0013] Count the anomaly frequency and type of the sensor in the historical data, calculate the anomaly influence factor of each sensor based on the weighted time decay function, and generate a historical anomaly frequency weight set;

[0014] Call the historical temperature parameter and humidity parameter, use the sliding window difference algorithm to calculate the change rate of the continuous data, and map the change rate to the rate weight to generate the current environmental change rate weight set;

[0015] Based on the position weight set, the historical anomaly frequency weight set, and the rate weight set, fuse each weight through the dynamic weight allocation algorithm to generate the comprehensive weight of each sensor.

[0016] Further, the assigning weights to the temperature parameter and the humidity parameter based on the comprehensive weight to generate a temperature feature matrix and a humidity feature matrix includes:

[0017] Extract the orientation information of each sensor relative to the distribution cabinet body from the sensor management unit, map the position of each sensor to a coordinate point using the preset two-dimensional plane rectangular coordinate system, and generate an initial coordinate matrix containing the sensor positions;

[0018] Match the temperature parameters and humidity parameters collected in real time with the initial coordinate matrix respectively to generate an initial temperature matrix and an initial humidity matrix containing humidity data;

[0019] Call the comprehensive weight, and multiply each point at the corresponding position of the initial temperature matrix and the initial humidity matrix according to the position index to generate a temperature feature matrix and a humidity feature matrix.

[0020] Furthermore, the graph attention network is trained through the following steps:

[0021] Extract the historical temperature feature matrix and humidity feature matrix as node feature input data, combine the relative positions and environmental correlations between sensors to construct the graph structure of the sensor network, and initialize the node set and edge set;

[0022] Initialize the node features based on the parameter values of each node in the temperature feature matrix and humidity feature matrix, and initialize the edge weight matrix according to the preset edge weight calculation function;

[0023] Through the node attention mechanism of the graph attention network, calculate the attention scores based on the node features and the edge weight matrix, and update the weight values of adjacent nodes in the graph structure;

[0024] Use the attention scores to perform weighted calculation on the adjacent node features to generate multi-dimensional feature vectors;

[0025] Based on the mean square error loss function, aiming at the actual environmental parameter change trend in the training data, optimize the model output prediction value, and update the model parameters of the graph attention network through the backpropagation algorithm.

[0026] Furthermore, the historical temperature feature matrix and humidity feature matrix are obtained through the following steps:

[0027] Collect the temperature parameters and humidity parameters under different seasons, weather, natural events and human events, and generate a temperature feature matrix and a humidity feature matrix through a parameter selection unit. The natural events include typhoons, floods and southern China's damp weather, and the human events include splashing water on the main body of the power distribution cabinet from different directions.

[0028] Furthermore, constructing a Gaussian process regression prediction model based on the multi-dimensional feature vector includes the following steps:

[0029] Extract the multi-dimensional feature vector from the environmental vector generation unit and perform normalization processing on the multi-dimensional feature vector;

[0030] Construct a covariance function based on Gaussian process regression, select a radial basis kernel function combined with a bias term, use the preprocessed multi-dimensional feature vector as the training data of the Gaussian process regression prediction model, calculate the covariance matrix between any two points in the training data using the covariance function, and construct a Gaussian process distribution in combination with the mean of the sampled historical data;

[0031] Use the gradient descent method to optimize the hyperparameters in the Gaussian process regression prediction model;

[0032] Use the optimized Gaussian process regression prediction model to predict the input real-time multi-dimensional feature vector, and generate the predicted values of the temperature parameter and humidity parameter at future time points and their corresponding abnormal probability distributions.

[0033] Further, the formula of the Gaussian process regression prediction model is as follows:

[0034] ;

[0035] Among them, is the predicted value of the prediction point ; and are the mean functions of the prediction point and the training point respectively; is the covariance matrix between the prediction point and the training data; is the covariance matrix between the training data; is the prediction distribution variance; is the identity matrix; is the observed value of the training data.

[0036] Further, the generation of the alarm level includes the following steps:

[0037] Obtain the predicted value from the Gaussian process regression prediction model, compare it with the real-time monitoring value, and calculate the deviation value between the two;

[0038] According to the calculated deviation value, combined with the preset deviation interval, divide the deviation value into normal, slight deviation, moderate deviation and severe deviation;

[0039] Use the prediction distribution variance provided by the Gaussian process regression model to calculate the cumulative probability within the deviation interval;

[0040] Weight the cumulative probability within the deviation interval and the parameter change rate to obtain the alarm score;

[0041] Match the alarm level according to the alarm score.

[0042] Further, the sending of maintenance information to the maintenance personnel terminal according to the alarm level includes the following steps:

[0043] Obtain the generated alarm level from the predictive maintenance unit, where the alarm level includes normal, minor anomaly, moderate anomaly, and severe anomaly;

[0044] If the alarm level is normal, no maintenance information is sent to the maintenance personnel terminal;

[0045] If the alarm level is minor anomaly, moderate anomaly, or severe anomaly, maintenance information is sent to the maintenance personnel terminal.

[0046] The beneficial effects of the present invention are as follows: By arranging temperature sensors and humidity sensors at different points outside the power distribution cabinet body, and through the cloud monitoring module to uniformly manage the data acquisition and storage of these sensors, the limitations of the traditional method that only focuses on a single monitoring point are overcome, and all-round real-time monitoring of the environmental state around the power distribution cabinet is achieved. By adopting the dynamic weight assignment algorithm, the historical anomaly frequency, position, and environmental change rate of the sensors are used as key parameters for generating comprehensive weights. The weighted temperature feature matrix and humidity feature matrix are input through the graph attention network to fully learn the correlation and weight changes of the sensor data, and a global feature vector of the environmental state is generated. This method significantly improves the ability to analyze complex environmental data, and combines the Gaussian process regression prediction model to accurately predict future trends. By comparing the deviation interval between the Gaussian process regression prediction value and the real-time value, and combining the parameter change rate to generate an alarm score, the alarm level is further generated based on the alarm score and an alarm is issued, significantly reducing the manual inspection cost and false alarm rate. Description of the Drawings

[0047] Figure 1 is a schematic structural diagram of a power distribution cabinet for environmental monitoring based on a sensor network in the present invention.

[0048] Figure 2 is a flow chart of the training steps of the graph attention network in the present invention. Detailed Embodiments

[0049] Please refer to Figure 1-2 As shown, the present invention relates to a power distribution cabinet for environmental monitoring based on a sensor network, including: a power distribution cabinet body, a sensor network, and a cloud monitoring module. The cloud monitoring module includes a number of temperature sensors and humidity sensors arranged at different points outside the power distribution cabinet body, and the sensor network is communicatively connected to the cloud monitoring module;

[0050] The cloud monitoring module includes a sensor management unit, a parameter selection unit, an environmental vector generation unit, and a predictive maintenance unit;

[0051] The sensor management unit is used to store the type of each sensor in the sensor network, the orientation relative to the power distribution cabinet body, and the temperature parameter or humidity parameter obtained through the corresponding sensor;

[0052] The parameter selection unit is used to perform weighted fusion on the orientation relative to the main body of the power distribution cabinet, the historical abnormal frequency, and the current environmental change rate based on the dynamic weight allocation algorithm to obtain a comprehensive weight, and assign weights to the temperature parameter and the humidity parameter based on the comprehensive weight to generate a temperature feature matrix and a humidity feature matrix;

[0053] The environmental vector generation unit is used to use the temperature feature matrix and the humidity feature matrix as the inputs of the graph attention network at the same time, learn the correlation and weight changes of the temperature and humidity parameters between specific positions through the node attention mechanism, model the global features of the environmental state around the power distribution cabinet, and generate a multi-dimensional feature vector for describing the environmental state;

[0054] The predictive maintenance unit is used to construct a Gaussian process regression prediction model based on the multi-dimensional feature vector, calculate the abnormal probability distribution by comparing the deviation interval between the predicted value generated by the Gaussian process regression prediction model and the real-time value, generate an alarm level in combination with the parameter change rate, and send maintenance information to the maintenance personnel terminal according to the alarm level.

[0055] Specifically, this power distribution cabinet system consists of a power distribution cabinet body, a sensor network, and a cloud monitoring module. The power distribution cabinet body is made of high-strength galvanized steel plates and is coated with an anti-corrosion coating to meet the long-term use requirements in harsh environments. The cabinet complies with the IP54 protection level, has dust and waterproof performance, and can protect the internal equipment from external interference under changing environmental conditions. Multiple temperature sensors and humidity sensors are arranged outside the cabinet, and these sensors are distributed according to accurately calculated monitoring points to ensure coverage of key areas, such as high-temperature areas (the top of the power distribution cabinet) and moisture-prone areas (the bottom and sides of the power distribution cabinet). The sensor network adopts a hybrid communication architecture, mainly composed of a wired RS485 bus and a wireless LoRa module. Among them, the wired network is responsible for the real-time data transmission of sensors with high data throughput and is suitable for areas with dense sensors, while the wireless module is applied to areas far from the cabinet or with limited wiring to achieve low-power long-distance data transmission. Each sensor node is equipped with a high-performance microcontroller (MCU), such as an STM32 series chip, which is used to collect environmental data in real time and perform preliminary filtering processing to remove high-frequency noise in the collected signals. The cloud monitoring module, as the core hardware unit of the entire system, includes a sensor management unit, a parameter selection unit, an environmental vector generation unit, and a predictive maintenance unit. The sensor management unit is implemented by an embedded server, which uses a high-performance ARM architecture processor to run the Linux operating system and is configured with a large-capacity solid-state storage device for storing sensor types, installation locations, and real-time temperature and humidity parameter data. This unit communicates with the sensor network through the Modbus TCP protocol to achieve low-latency data transmission and storage. The parameter selection unit is implemented by FPGA hardware, and its parallel computing ability is used to perform weighted calculations on a large amount of sensor data. Based on the dynamic weight assignment algorithm, this unit comprehensively considers the sensor position, historical anomaly frequency, and environmental change rate to generate a comprehensive weight and transmits the result to the subsequent module at high speed. The internal of the FPGA is designed with a hierarchical pipeline structure, significantly improving the real-time performance of weight calculation. The environmental vector generation unit is composed of high-performance GPU (such as NVIDIA Jetson series) hardware, which is mainly used to perform deep learning modeling on the weighted temperature and humidity feature matrix. Through the parallel computing ability optimized by the CUDA architecture, this unit runs the graph attention network model, learns the correlation and weight distribution between sensor data, and generates a multi-dimensional feature vector for describing the global state of the environment. The predictive maintenance unit takes industrial-grade server hardware as the core and deploys a Gaussian process regression model. The server uses a multi-core x86 architecture processor. By calling the feature vector output by the environmental vector generation unit, the predictive maintenance unit makes a high-precision prediction of the future environmental state. Its main tasks include calculating the deviation value in real time, evaluating the anomaly probability, and generating an alarm level in combination with the parameter change rate. The alarm information is sent to the maintenance terminal through the Ethernet module, including the alarm level, possible anomaly reasons, and recommended maintenance measures.

[0056] Further, the weighted fusion of the orientation relative to the distribution cabinet body, the historical anomaly frequency, and the current environmental change rate based on the dynamic weight assignment algorithm includes the following steps:

[0057] Based on the orientation of the sensor relative to the distribution cabinet body stored in the sensor management unit, combined with the key areas of the distribution cabinet and the preset weight area influence factor, calculate the initial position weight of each sensor and store it in the position weight set;

[0058] Count the anomaly frequency and type of the sensor in the historical data, calculate the anomaly influence factor of each sensor based on the weighted time decay function, and generate the historical anomaly frequency weight set;

[0059] Call the historical temperature parameters and humidity parameters, use the sliding window difference algorithm to calculate the change rate of continuous data, and map the change rate to the rate weight to generate the current environmental change rate weight set;

[0060] Based on the position weight set, the historical anomaly frequency weight set, and the rate weight set, fuse each weight through the dynamic weight assignment algorithm to generate the comprehensive weight of each sensor.

[0061] In some embodiments, taking the temperature sensor arranged at the top of the power distribution cabinet as an example, since this position is close to the heat source area (such as switches and cables), the temperature parameters collected by it have important reference value for environmental monitoring. First, according to the physical position of the sensor, its initial position weight is calculated. Assuming that the importance of the top area is higher than that of other areas, the physical position of the sensor is weighted by the regional influence factor, so that its weight accounts for a larger proportion in the position set. This process is achieved through a preset regional weight matrix, and the matrix values are determined by combining historical experience data and can be further adjusted according to the actual usage environment. Subsequently, the historical data of this temperature sensor is analyzed to count the frequency and type of its abnormal events. For example, if this sensor has detected abnormal values exceeding the normal temperature range multiple times, then based on the weighted time decay model, a higher influence weight is given to recent abnormalities, so as to generate the historical abnormal frequency weight. In this way, the system can dynamically identify the severity of abnormal events and their importance to the overall environmental monitoring. In addition, the system also calls the temperature change data of this sensor in real time, and uses the sliding window difference algorithm to calculate the change rate of the current environment. Within a set time interval, the difference in temperature change is calculated point by point and the average change rate and the maximum change rate are counted. Taking a typhoon weather as an example, the system detects that the temperature parameters fluctuate greatly in a short time, and generates a change rate weight by mapping the relationship between the rate and the weight. This weight can reflect the dynamic characteristics of the current environment and provide early warning support for sudden environmental changes. After calculating the initial position weight, the historical abnormal frequency weight and the change rate weight, the system fuses the three through a dynamic weight allocation algorithm to generate a comprehensive weight. The fusion algorithm adjusts the weight allocation ratio according to the specific requirements of the environment where the sensor is located. For example, in a stable environment, the position weight and the historical abnormal weight account for the main proportion, while in a sudden environment, the weight of the change rate weight will be dynamically increased to ensure that the comprehensive weight can timely reflect the current monitoring requirements. The generated comprehensive weight is used to weight the temperature and humidity parameters to form a feature matrix. Taking the feature matrix as the input, the system further uses a deep learning model (such as a graph attention network) to learn and model the correlation and weight distribution between the data. Combining with the Gaussian process regression prediction model, the system can achieve high-precision prediction of future environmental parameters based on these features, so as to provide strong support for anomaly detection and maintenance planning.

[0062] Further, the weighting the temperature parameter and the humidity parameter based on the comprehensive weight to generate a temperature feature matrix and a humidity feature matrix includes:

[0063] Extract the orientation information of each sensor relative to the power distribution cabinet body from the sensor management unit, map the position of each sensor to a coordinate point using a preset two-dimensional plane rectangular coordinate system, and generate an initial coordinate matrix containing the sensor positions;

[0064] Match the temperature parameters and humidity parameters collected in real time with the initial coordinate matrix respectively to generate an initial temperature matrix and an initial humidity matrix containing humidity data;

[0065] Call the comprehensive weight, and multiply each point at the corresponding position of the initial temperature matrix and the initial humidity matrix according to the position index to generate a temperature feature matrix and a humidity feature matrix.

[0066] In some embodiments, first, extract the stored sensor position data from the sensor management unit. The relative position of each sensor is mapped through a preset two-dimensional plane rectangular coordinate system based on the spatial layout of the power distribution cabinet body. Assuming that the bottom left corner of the power distribution cabinet body is the origin and the top right corner is the diagonal point of the coordinate system, the relative position of each sensor (such as top, bottom, side) is converted into specific coordinate points. For example, the position of the temperature sensor in the center of the top is mapped to the coordinate point (0.5, 1.0), and the position of the humidity sensor on the right side of the bottom is mapped to the coordinate point (1.0, 0.0). The position data of all sensors are stored in matrix form to form an initial coordinate matrix. Secondly, the system collects temperature parameters and humidity parameters from the sensor network in real time to generate two independent data sets. These parameters are matched point by point with the initial coordinate matrix according to the unique identifier of the sensor to associate the environmental data with the physical space position. The matched temperature data and humidity data are respectively stored as an initial temperature matrix and an initial humidity matrix, where the rows correspond to the spatial positions of the sensors and the columns record the corresponding environmental parameter values. For example, the data at the i-th row and j-th column of the temperature matrix collected at a certain moment represents the temperature reading of the sensor located at the coordinate point (0.5, 1.0). Next, the system calls the comprehensive weight to assign the calculated weight to each sensor to characterize the importance of the sensor. The comprehensive weight combines the regional influence factor of the sensor position, the historical anomaly frequency, and the environmental change rate, and these weights have been dynamically calculated in the previous steps. The comprehensive weight is stored in the form of a one-dimensional vector, where each element corresponds to a sensor. The system performs a weighting operation on each element of the initial temperature matrix and the initial humidity matrix in turn. The weighting operation is completed in the form of matrix point-by-point multiplication: through the position index, multiply the comprehensive weight by the data point at the corresponding position of the initial matrix to generate a weighted temperature feature matrix and a humidity feature matrix. For example, if the comprehensive weight of the coordinate point (0.5, 1.0) is 1.5 and the corresponding initial temperature value is 30°C, the final generated eigenvalue is 30×1.5 = 45. The calculation process of the humidity feature matrix is similar.

[0067] Furthermore, the graph attention network is trained through the following steps:

[0068] Extract the historical temperature feature matrix and humidity feature matrix as the node feature input data. Combine the relative positions and environmental correlations between sensors to construct the graph structure of the sensor network, and initialize the node set and edge set;

[0069] Initialize the node features based on the parameter values of each node in the temperature feature matrix and humidity feature matrix, and initialize the edge weight matrix according to the preset edge weight calculation function;

[0070] Through the node attention mechanism of the graph attention network, calculate the attention scores based on the node features and the edge weight matrix, and update the weight values of adjacent nodes in the graph structure;

[0071] Use the attention scores to perform weighted calculations on the adjacent node features to generate multi-dimensional feature vectors;

[0072] Based on the mean square error loss function, with the actual environmental parameter change trend in the training data as the target, optimize the model output prediction value, and update the model parameters of the graph attention network through the backpropagation algorithm.

[0073] In some embodiments, first, the system extracts the historical temperature feature matrix and humidity feature matrix as input data. These feature matrices are generated from the historical temperature and humidity parameters collected by sensors, and have been weighted through a pre - processing step, integrating information on the spatial location of the sensors, the historical anomaly frequency, and the environmental change rate. These feature matrices are mapped to the node feature space, where each node represents a sensor, and the feature vector of each node contains the temperature and humidity parameters of the sensor and its comprehensive weight value. Based on the relative positions between sensors and their environmental correlations, the system constructs the graph structure of the sensor network, including a node set and an edge set. Nodes represent sensors, and edges represent the associations between sensors. In the initialization stage, the system assigns feature vectors to each node according to the parameter values in the temperature and humidity feature matrices, and at the same time generates an edge weight matrix based on the relative positions and environmental correlations of the sensors. The edge weight calculation uses a preset function, such as the Gaussian kernel function, to calculate the weight value based on the Euclidean distance between sensors. A higher weight value indicates a stronger correlation between two sensors. For example, sensors located in the same area have a higher weight, while sensors at a long distance have a lower weight. The initialized graph structure can intuitively reflect the relationship between the spatial location and environmental parameters in the sensor network. During the training process, the system calculates the attention scores between each node and its adjacent nodes through the node attention mechanism of the graph attention network. The node attention mechanism dynamically adjusts the weight values of adjacent nodes through the joint calculation of node features and the edge weight matrix. For example, for areas with large environmental fluctuations, the attention scores of relevant nodes will increase to enhance the sensitivity to environmental changes in key areas. Through the calculation of attention scores, the system updates the weight values of adjacent nodes in the graph structure to ensure that the representation ability of the graph is closer to the actual environmental changes. After the weight update, the system uses the attention scores to perform weighted calculations on the features of adjacent nodes to generate multi - dimensional feature vectors. These feature vectors integrate the environmental information of the node itself and its adjacent nodes, and are a high - level representation of the overall environmental state of the sensor network. For example, for areas where the temperature and humidity parameters between adjacent nodes have a strong correlation, the feature vectors will more prominently reflect the environmental state of the area. During the training process, the output of the model is compared with the actual environmental parameter change trend, and the mean squared error (MSE) is used as the loss function to optimize the prediction results. Through the backpropagation algorithm, the system dynamically adjusts the model parameters of the graph attention network to gradually reduce the prediction error. For example, for continuous model training iterations, the system will identify the difference between the current prediction result and the target value, and further optimize by adjusting the weights of nodes and edges.

[0074] Further, the historical temperature feature matrix and humidity feature matrix are obtained through the following steps:

[0075] Collect temperature parameters and humidity parameters under different seasons, weather conditions, natural events, and human events, and generate a temperature feature matrix and a humidity feature matrix through a parameter selection unit. The natural events include typhoons, floods, and spring tides, and the human events include splashing water on the main body of the power distribution cabinet from different directions.

[0076] In some embodiments, first, to construct complete historical data, the system deploys multiple temperature sensors and humidity sensors. These sensors are arranged at key points on the outer side of the main body of the power distribution cabinet, including positions such as the top, sides, and bottom that are vulnerable to environmental impacts. The sensors collect data in real time, and their sampling frequency is dynamically adjusted according to the event type. For example, during stable seasons, the sampling frequency can be set to once per hour, while under natural events or human interventions, the sampling frequency is increased to once per minute to capture rapidly changing environmental conditions. The data collection covers diverse weather conditions under the four seasons, including sunny days, rainy days, and high-humidity environments. At the same time, the impacts of special natural events are emphasized. For example, the strong winds and heavy rains brought by typhoons on the external environment of the power distribution cabinet; the possible long-term high-humidity state under flood conditions; the condensate water problems caused by spring tides (high-humidity weather in spring). In these cases, the system can capture the sharp increase in external humidity in real time through high-precision humidity sensors and synchronously record the temperature fluctuations caused by the evaporation of condensate water through temperature sensors. In addition, to simulate the impact of emergencies on the environment of the power distribution cabinet, various human intervention experiments are carried out. For example, splashing water on the main body of the power distribution cabinet from different directions to simulate possible external liquid leakage or rainwater intrusion; at the same time, by controlling the amount and frequency of water splashing, the impacts on sensors at different points are recorded. These experimental data effectively supplement special situations that may not be covered in natural environment data. All the collected temperature and humidity data are transmitted through the sensor network to the parameter selection unit for processing. In the parameter selection unit, the system first performs preliminary calibration and denoising processing on the collected data to filter out outliers caused by sensor hardware drift or sudden interference signals. Subsequently, according to the environmental characteristics under different collection conditions, the system assigns weights to the data using a dynamic weight allocation algorithm. For example, under typhoon conditions, a higher weight is assigned to the sensors in the top area, while under flood conditions, more emphasis is placed on the data of the sensors in the bottom area. This dynamic weight assignment method ensures that the feature matrix can accurately reflect the comprehensive importance of sensors under different conditions. After the weight assignment process, the system stores the processed temperature and humidity data in a matrix form, generating a temperature feature matrix and a humidity feature matrix respectively. Each row of the matrix corresponds to a sensor, and each column corresponds to a data point under a specific condition. For example, the temperature data of the top sensor under typhoon conditions will occupy a higher weight in the temperature feature matrix, and the relevant row of the humidity feature matrix highlights the fluctuation range of its humidity value.

[0077] Further, the construction of the Gaussian process regression prediction model based on the multi-dimensional feature vector includes the following steps:

[0078] Extract the multi-dimensional feature vector from the environmental vector generation unit and perform normalization processing on the multi-dimensional feature vector;

[0079] Construct a covariance function based on Gaussian process regression, select a radial basis kernel function combined with a bias term, use the preprocessed multi-dimensional feature vector as the training data of the Gaussian process regression prediction model, calculate the covariance matrix between any two points in the training data using the covariance function, and construct a Gaussian process distribution in combination with the mean of the sampled historical data;

[0080] Use the gradient descent method to optimize the hyperparameters in the Gaussian process regression prediction model;

[0081] Utilize the optimized Gaussian process regression prediction model to predict the input real-time multi-dimensional feature vector, and generate the predicted values of the temperature parameter and humidity parameter at future time points and their corresponding abnormal probability distributions.

[0082] First, extract the multi-dimensional feature vectors generated by the graph attention network from the environmental vector generation unit. These vectors synthesize the temperature and humidity parameters of the sensors, the spatial position relationship, and the dynamic weight characteristics, and can comprehensively reflect the global state of the environment around the power distribution cabinet. Since different feature components may have different numerical ranges, it is necessary to normalize the multi-dimensional feature vectors to eliminate the influence of different dimensions on the model calculation. The normalization adopts the maximum-minimum normalization method, linearly mapping the value of each component to a preset interval. The normalized feature vectors are input into the Gaussian process regression model. After the feature vectors are normalized, the system models the data through the Gaussian process regression model. The core of the model is the design of the covariance function. The system adopts the form of combining the radial basis kernel function (RBF kernel) with a bias term. The radial basis kernel function measures the similarity between data points by calculating the Euclidean distance between multi-dimensional feature vectors. The bias term is used to capture the noise in the data and enhance the model's tolerance to uncertainty. The parameters of the covariance function, such as the signal variance, noise variance, and length scale, can be flexibly adjusted to adapt to different environmental monitoring scenarios. After constructing the covariance function, the system calculates the covariance matrix using historical temperature and humidity data. Each element of the covariance matrix represents the correlation between historical data points. At the same time, combining the means of these data, a Gaussian distribution of multi-dimensional feature vectors is established to describe the probability model of future trends. In the training stage, the system optimizes the model parameters by minimizing the negative log marginal likelihood function. The optimization process uses the gradient descent algorithm to dynamically adjust the hyperparameters of the covariance function, enabling the model to more accurately capture the distribution law of the data. After completing the model training, the system uses the real-time obtained multi-dimensional feature vectors for prediction. Calculate the correlation between the real-time data and the historical data through the covariance function, and combine the trained covariance matrix and mean to generate the predicted values of the temperature parameter and humidity parameter at future time points. At the same time, the Gaussian process regression also outputs the probability distribution of the predicted values. This probability distribution not only provides the predicted values but also quantifies the uncertainty of the prediction results through the variance of the distribution. Taking a certain sudden weather event as an example, the system may predict that the humidity value of a certain sensor within the next hour is 80%, but at the same time indicates that the variance of the prediction is large, reflecting the potential risk of significant humidity fluctuations.

[0083] Further, the formula of the Gaussian process regression prediction model is as follows:

[0084] ;

[0085] Where, is the predicted value of the prediction point ; and are the mean functions of the prediction point and the training point respectively; is the covariance matrix between the prediction point and the training data; is the covariance matrix between training data; is the variance of the prediction distribution; is the identity matrix; is the observation value of the training data.

[0086] Specifically, first, the value of the prediction point is calculated through the following process. The model assumes that the change of environmental parameters follows a Gaussian distribution, and the value of the prediction point consists of the prior mean and the contribution of historical observation data to its correction. The prior mean is a preliminary estimate of the value of the prediction point, usually assumed to be zero mean to simplify the calculation. Subsequently, the system corrects the prior mean of the prediction point through the covariance matrix. These covariance matrices reflect the correlation between the prediction point and the historical data. Specifically, the covariance matrix is calculated through a radial basis kernel function, and the kernel function measures the similarity according to the distance between the prediction point and the training data. The closer the distance, the stronger the correlation. The magnitude of the correction is determined by the deviation between the observation value and the mean of the historical data. The covariance matrix distributes the influence of this deviation to the prediction point according to the weight, and finally obtains the corrected prediction value. This calculation method based on statistical characteristics ensures the reliability of the prediction result and reduces the influence of single-point data error. In addition, the uncertainty of the prediction result is quantified by the variance of the prediction distribution. The system calculates the propagation effect of the correlation between the prediction point and the historical data in the covariance matrix. The magnitude of the prediction variance reflects the confidence of the model in the value of the prediction point. For example, when there is less data or greater variation around the prediction point, the variance will increase, indicating a higher uncertainty in the prediction result. In applications, the model generates a covariance matrix through real-time collected multi-dimensional feature vectors and combines the observation values of historical data to calculate the predicted value of future environmental parameters. For example, under typhoon conditions, the model may predict the humidity value of the top sensor within the next hour, and attach a prediction variance to remind the maintenance personnel that the humidity may fluctuate violently.

[0087] Furthermore, the generation of the alarm level includes the following steps:

[0088] Obtain the predicted value from the Gaussian process regression prediction model, compare it with the real-time monitoring value, and calculate the deviation value between the two;

[0089] According to the calculated deviation value and in combination with the preset deviation interval, divide the deviation value into normal, slight deviation, moderate deviation, and severe deviation;

[0090] Utilize the prediction distribution variance provided by the Gaussian process regression model to calculate the cumulative probability within the deviation interval;

[0091] Weight the cumulative probability within the deviation interval and the parameter change rate to obtain the alarm score;

[0092] Match the alarm level according to the alarm score.

[0093] In some embodiments, first, the system extracts predicted values of environmental parameters from the Gaussian process regression prediction model, such as the future change trends of temperature and humidity. These predicted values are generated through the model's learning of historical data and the input of real-time multi-dimensional feature vectors, and are estimates of the future environmental state. At the same time, the real-time monitoring values are collected in real time by sensors deployed at different points of the power distribution cabinet, reflecting the actual environmental state. Subsequently, the system compares the predicted values with the real-time monitoring values and calculates the deviation value between the two. The system further uses the prediction distribution variance provided by the Gaussian process regression model to calculate the cumulative probability of the predicted value and the actual value within the deviation interval. The prediction distribution variance reflects the uncertainty of the model for the predicted value, and the cumulative probability of the deviation value within a specific interval can be obtained through the integration of the probability density function. This probability measures the confidence level of the deviation, that is, the likelihood of the deviation value within the current interval. Next, the system combines the cumulative probability with the change rate of the environmental parameters. The change rate is calculated by the continuous data of the sensor through the sliding window difference algorithm, reflecting the dynamics of the environmental state change. For example, when the change rate of humidity is high, even a small deviation may mean a potential risk. By performing a weighted calculation on the cumulative probability and the change rate, the system generates an alarm score. The weighting coefficient is dynamically adjusted according to the actual application scenario to adapt to different monitoring requirements. Finally, the corresponding alarm level is matched according to the alarm score. The alarm level division includes: Normal: The score is lower than the preset threshold, and no measures need to be taken; Slight anomaly: The score is close to the threshold, indicating that attention is needed; Moderate anomaly: The score significantly exceeds the threshold, and it is recommended to check; Severe anomaly: The score significantly exceeds the threshold, triggering an emergency response. For example, in a humid weather environment, when the predicted value of the humidity sensor is 95%, the real-time value is 98%, the deviation value is 3%, and the cumulative probability is high, the system may generate an alarm at the slight anomaly level. If the change rate of humidity is fast and the deviation value further increases, the alarm level may be upgraded to moderate anomaly or even severe anomaly.

[0094] Further, the sending of maintenance information to the maintenance personnel terminal according to the alarm level includes the following steps:

[0095] Obtain the generated alarm level from the predictive maintenance unit, and the alarm level includes normal, slight anomaly, moderate anomaly, and severe anomaly;

[0096] If the alarm level is normal, no maintenance information is sent to the maintenance personnel terminal;

[0097] If the alarm level is slight anomaly, moderate anomaly, or severe anomaly, maintenance information is sent to the maintenance personnel terminal.

[0098] Specifically, the system first obtains the generated alarm level from the predictive maintenance unit, which is comprehensively analyzed based on the results of the Gaussian process regression prediction model, real-time monitoring values, deviation values, and parameter change rates. The alarm levels are classified as normal, minor anomaly, moderate anomaly, and severe anomaly, corresponding to different environmental states and response requirements respectively.

[0099] After receiving the alarm level, the system performs hierarchical processing according to the preset logic. For the case where the alarm level is "normal", the system determines that the current environmental state is within the safe range and no intervention measures are required, so no maintenance information will be sent to the maintenance personnel side. This design can avoid unnecessary notifications, reduce information interference and labor costs. If the alarm level is "minor anomaly", the system will send a reminder-type maintenance information to the maintenance personnel side. This information mainly includes: the specific location of the sensor, the monitored environmental parameters, the predicted deviation value, and the alarm level. For example, "Sensor A (location: top of the distribution cabinet) detects a minor humidity anomaly, the current humidity is 85%, and the predicted deviation value is 3%." Such information is intended to prompt the maintenance personnel to pay attention to the environmental changes at this point, but no immediate action is required. For the alarm level of "moderate anomaly", the generated maintenance information by the system will be more detailed and include recommended measures. For example, the system may prompt the maintenance personnel to check some equipment components vulnerable to the environment, or provide short-term monitoring suggestions to further track the changes of environmental parameters. For example, "Sensor B (location: north side of the distribution cabinet) detects a moderate humidity anomaly, the current humidity is 90%, and the deviation value is 7%. It is recommended to check the sealing state on the north side of the distribution cabinet to avoid further deterioration." When the alarm level reaches "severe anomaly", the system will generate an emergency maintenance information and notify the maintenance personnel in real time through multiple channels (such as text messages, emails, or dedicated maintenance applications). Such information not only includes the anomaly details and recommended measures, but may also be attached with a brief analysis of the estimated risks. For example, "Sensor C (location: inside the distribution cabinet) detects a severe temperature anomaly, the current temperature is 60°C, and the deviation value is 15%. There is a risk of equipment overheating. Please check the internal cooling device of the distribution cabinet and repair potential problems."

[0100] The above embodiments are only descriptions of the preferred embodiments of the present invention and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A power distribution cabinet for realizing environmental monitoring based on a sensor network, characterized in that: include: A power distribution cabinet body, a sensor network, and a cloud monitoring module, wherein the cloud monitoring module includes a plurality of temperature sensors and humidity sensors disposed at different points outside the power distribution cabinet body, and the sensor network is communicatively connected with the cloud monitoring module; The cloud monitoring module includes a sensor management unit, a parameter selection unit, an environmental vector generation unit and a predictive maintenance unit; The sensor management unit is used to store the type of each sensor in the sensor network, the position relative to the power distribution cabinet body, and the temperature parameter or humidity parameter obtained by the corresponding sensor; The parameter selection unit is used to perform weighted fusion of the relative position of the power distribution cabinet body, the historical abnormal frequency and the current environmental change rate based on a dynamic weight allocation algorithm to obtain a comprehensive weight, and weight the temperature parameter and the humidity parameter based on the comprehensive weight to generate a temperature feature matrix and a humidity feature matrix; The environmental vector generation unit is used to use the temperature feature matrix and the humidity feature matrix as inputs of the graph attention network at the same time, learn the correlation and weight changes between temperature and humidity parameters at specific positions through the node attention mechanism, model the global characteristics of the environmental state around the distribution cabinet, and generate a multi-dimensional feature vector for describing the environmental state; The predictive maintenance unit is used to construct a Gaussian process regression prediction model based on a multidimensional feature vector, calculate the abnormal probability distribution by comparing the deviation interval between the predicted value generated by the Gaussian process regression prediction model and the real-time value, and generate an alarm level in combination with the parameter change rate, and send maintenance information to the maintenance personnel according to the alarm level.

2. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 1, characterized in that: The weighted fusion of the relative position of the power distribution cabinet body, the historical abnormal frequency and the current environmental change rate based on the dynamic weight allocation algorithm includes the following steps: Based on the orientation of the sensor relative to the power distribution cabinet body stored in the sensor management unit, combined with the key area of ​​the power distribution cabinet and the preset weight area influence factor, the initial position weight of each sensor is calculated and stored in the position weight set; Count the abnormal frequency and type of sensors in historical data, calculate the abnormal impact factor of each sensor based on the weighted time decay function, and generate a historical abnormal frequency weight set; Call the historical temperature and humidity parameters, use the sliding window difference algorithm to calculate the change rate of continuous data, and map the change rate to a rate weight to generate a current environment change rate weight set; Based on the location weight set, historical abnormal frequency weight set and rate weight set, the weights are fused through a dynamic weight allocation algorithm to generate a comprehensive weight for each sensor.

3. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 1, characterized in that: The step of weighting the temperature parameters and the humidity parameters based on the comprehensive weights to generate the temperature characteristic matrix and the humidity characteristic matrix comprises: Extract the orientation information of each sensor relative to the power distribution cabinet body from the sensor management unit, map each sensor position into a coordinate point using a preset two-dimensional plane rectangular coordinate system, and generate an initial coordinate matrix containing the sensor position; Match the real-time collected temperature parameters and humidity parameters with the initial coordinate matrix respectively, and generate an initial temperature matrix and an initial humidity matrix containing humidity data; The comprehensive weights are called and multiplied point by point with the corresponding positions of the initial temperature matrix and the initial humidity matrix according to the position index to generate the temperature feature matrix and the humidity feature matrix.

4. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 1, characterized in that: The graph attention network is trained by the following steps: Extract the historical temperature feature matrix and humidity feature matrix as node feature input data, combine the relative position and environmental correlation between sensors, build the graph structure of the sensor network, and initialize the node set and edge set; Initialize the node features based on the parameter values ​​of each node in the temperature feature matrix and the humidity feature matrix, and initialize the edge weight matrix according to the preset edge weight calculation function; Through the node attention mechanism of the graph attention network, the attention score is calculated based on the node features and the edge weight matrix, and the weight values ​​of the adjacent nodes in the graph structure are updated; Use attention scores to perform weighted calculations on adjacent node features to generate a multi-dimensional feature vector; Based on the mean square error loss function, the actual environmental parameter change trend in the training data is taken as the target to optimize the model output prediction value, and the model parameters of the graph attention network are updated through the back propagation algorithm.

5. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 4, characterized in that: The historical temperature feature matrix and humidity feature matrix are obtained by the following steps: The temperature parameters and humidity parameters under different seasons, weather, natural events and man-made events are collected and a temperature characteristic matrix and a humidity characteristic matrix are generated through a parameter selection unit. The natural events include typhoons, floods and return of the south wind, and the man-made events include pouring water on the distribution cabinet body from different directions.

6. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 1, characterized in that: The Gaussian process regression prediction model based on the multidimensional feature vector is constructed, comprising the following steps: extracting a multi-dimensional feature vector from an environment vector generating unit, and performing normalization processing on the multi-dimensional feature vector; Based on Gaussian process regression, the covariance function is constructed. The radial basis kernel function is selected in combination with the bias term. The preprocessed multidimensional feature vector is used as the training data of the Gaussian process regression prediction model. The covariance matrix between any two points in the training data is calculated using the covariance function, and the Gaussian process distribution is constructed in combination with the mean of the sampled historical data. Optimize hyperparameters in Gaussian process regression prediction model using gradient descent method; The optimized Gaussian process regression prediction model is used to predict the input real-time multidimensional feature vector to generate the predicted values ​​of temperature parameters and humidity parameters at future time points and their corresponding abnormal probability distributions.

7. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 6, characterized in that: The formula of the Gaussian process regression prediction model is as follows: ； in, Forecast point The predicted value of and are the mean functions of the prediction points and training points respectively; is the covariance matrix between the prediction point and the training data; is the covariance matrix between training data; is the prediction distribution variance; is the identity matrix; is the observed value of the training data.

8. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 7, characterized in that: The generation of the alarm level comprises the following steps: Obtain the predicted value from the Gaussian process regression prediction model, compare it with the real-time monitoring value, and calculate the deviation between the two; According to the calculated deviation value and the preset deviation range, the deviation value is divided into normal, slight deviation, moderate deviation and severe deviation; Using the variance of the prediction distribution provided by the Gaussian process regression model, the cumulative probability within the deviation interval is calculated; The cumulative probability and parameter change rate within the deviation interval are weighted to obtain the alarm score; Match the alarm level according to the alarm score.

9. A power distribution cabinet for realizing environmental monitoring based on a sensor network according to claim 1, characterized in that: The sending of maintenance information to the maintenance personnel according to the alarm level comprises the following steps: Acquire the generated alarm level from the predictive maintenance unit, where the alarm level includes normal, slightly abnormal, moderately abnormal, and severely abnormal; If the alarm level is normal, no maintenance information will be sent to the maintenance personnel; If the alarm level is a minor abnormality, a moderate abnormality, or a serious abnormality, a maintenance message is sent to the maintenance personnel.

Images