Charging pile intelligent operation management method and system based on Internet of Things

Abstract

The invention relates to a charging pile intelligent operation management method and system based on the Internet of Things. The method comprises the following steps: performing Internet of Things topology construction on charging pile nodes to obtain a node communication distance matrix and a network connectivity matrix; performing deep learning dimension reduction and node screening based on the network connectivity matrix to obtain a fault feature vector; the fault feature vector and the charging pile operation state data are input into a hybrid neural network for risk calculation, a fault risk assessment value is obtained, and the hybrid neural network comprises a space perception sub-network and a time sequence analysis sub-network; performing normalization and dimension reduction mapping on the power response data of the charging pile to obtain a time domain mapping matrix; and performing particle swarm optimization calculation based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power distribution scheme, and generating a control instruction set according to the charging power distribution scheme. According to the invention, the dynamic optimization of the charging power distribution scheme is realized, and the charging efficiency and the load balance are effectively balanced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent operation of charging piles, and in particular to an intelligent operation management method and system of charging piles based on the Internet of Things. Background Art

[0002] Traditional charging pile management methods mainly rely on manual monitoring and fixed-rule scheduling, which makes it difficult to cope with increasingly complex charging demand fluctuations and equipment failures, and cannot fully utilize the overall efficiency of the charging pile cluster.

[0003] At present, the pre-fault dynamic safety assessment (DSA) method of the charging pile system has obvious shortcomings: on the one hand, the existing data-driven smart charging pile DSA method lacks generalizable fault location features, resulting in low accuracy in the assessment of unlearned fault locations; on the other hand, the traditional DSA method often takes the fault location as a fixed input, ignoring the impact of the uncertainty of the fault location on the system safety, making the reliability of the assessment result questionable. In addition, in today's IoT-connected charging pile system, the demand for flexible scheduling of smart charging piles has increased significantly, while the existing dynamic optimal energy flow technology still has a lot of room for improvement in terms of accuracy, efficiency and convergence. Especially under the influence of uncertain factors such as charging demand fluctuations and charging pile equipment failures, how to achieve coordinated control and intelligent scheduling of charging pile groups has become a technical difficulty that needs to be overcome urgently. Summary of the invention

[0004] The main purpose of the present invention is to provide a charging pile intelligent operation and management method and system based on the Internet of Things. The present invention realizes dynamic optimization of the charging power distribution scheme and effectively balances the charging efficiency and load balancing.

[0005] To achieve the above object, the present invention provides a charging pile intelligent operation and management method based on the Internet of Things, comprising the following steps:

[0006] The IoT topology is constructed for charging pile nodes to obtain the node communication distance matrix and network connectivity matrix;

[0007] Perform deep learning dimensionality reduction and node screening based on the network connectivity matrix to obtain a fault feature vector;

[0008] Inputting the fault feature vector and the charging pile operation status data into a hybrid neural network for risk calculation to obtain a fault risk assessment value, wherein the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork;

[0009] Normalize and reduce the dimension of the charging pile power response data to obtain the time domain mapping matrix;

[0010] A particle swarm optimization calculation is performed based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation scheme, and a control instruction set is generated according to the charging power allocation scheme, wherein the control instruction set includes power and time parameters of each charging pile.

[0011] The present invention also provides a charging pile intelligent operation and management system based on the Internet of Things, comprising:

[0012] The construction module is used to construct the IoT topology for charging pile nodes and obtain the node communication distance matrix and network connectivity matrix;

[0013] A screening module, used for performing deep learning dimension reduction and node screening based on the network connectivity matrix to obtain a fault feature vector;

[0014] A calculation module, used for inputting the fault feature vector and the charging pile operation status data into a hybrid neural network for risk calculation to obtain a fault risk assessment value, wherein the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork;

[0015] A mapping module is used to normalize and reduce the dimension of the charging pile power response data to obtain a time domain mapping matrix;

[0016] A generation module is used to perform particle swarm optimization calculation based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation plan, and generate a control instruction set according to the charging power allocation plan, wherein the control instruction set includes power and time parameters of each charging pile.

[0017] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0018] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0019] In summary, the technical solution provided by the present invention realizes efficient communication and state perception between charging pile nodes by constructing the Internet of Things topology and introducing the network connectivity matrix. The deep learning dimension reduction and node screening technology are used to effectively extract the fault feature vector, solve the problem that the fault location feature in the traditional method cannot be generalized, and improve the evaluation ability of the unknown fault location. A hybrid neural network architecture including a spatial perception subnetwork and a timing analysis subnetwork is designed to realize the multi-dimensional analysis of the operating state of the charging pile, and significantly improve the accuracy of the fault risk assessment. Through the construction of the time domain mapping matrix, the reduced-order mapping relationship between the charging pile operating state variables is established, which reduces the computational complexity and improves the efficiency of energy optimization. The particle swarm optimization algorithm with forgetting factor is introduced to realize the dynamic optimization of the charging power allocation scheme, and effectively balance the charging efficiency and load balancing. Based on the two-stage robust planning model and the emergency uncertainty set, a complete fault emergency response mechanism is constructed to improve the operating stability of the charging pile system under fault conditions. By designing a hierarchical control instruction structure, seamless switching between normal operating mode and emergency mode is realized, and the fault adaptability and operation reliability of the charging pile system are enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the steps of a charging pile intelligent operation and management method based on the Internet of Things in one embodiment of the present invention;

[0021] Figure 2 It is a structural block diagram of a charging pile intelligent operation and management system based on the Internet of Things in one embodiment of the present invention;

[0022] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0023] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] Reference Figure 1 This embodiment provides a charging pile intelligent operation and management method based on the Internet of Things, comprising the following steps:

[0026] S1, construct the IoT topology for the charging pile nodes to obtain the node communication distance matrix and network connectivity matrix;

[0027] Among them, the Internet of Things communication module is deployed on the charging pile device. By configuring each charging pile communication module, setting the communication parameters, including the communication frequency band, data transmission rate, working frequency, etc., and recording the device number and these communication parameters, the node deployment data is formed, which contains the basic information of each communication module. For the charging pile devices equipped with the Internet of Things communication module, high-precision positioning technology is used to determine their spatial positions. The geographical coordinate data of each charging pile is obtained by using the global positioning system or other high-precision spatial positioning means, including its longitude, latitude and altitude. Through these data, a geographical coordinate set is formed to describe the distribution of all charging pile devices in the actual geographical space. The geographical coordinate set is converted into standardized location data suitable for topological analysis. Through the conversion, the complex geographical coordinate form is simplified into a three-dimensional rectangular coordinate system that is directly applied to mathematical analysis and modeling, reflecting the relationship between charging pile devices in spatial distribution. Based on the standardized location data, the wireless communication link between the nodes is constructed. By analyzing the spatial distance between each charging pile device, a communication distance matrix is ​​generated to record the distance information between each pair of nodes in the network. In order to more accurately describe the communication capability, the signal strength is calculated by the propagation characteristics of the wireless signal. By analyzing the distance between devices and the attenuation characteristics of the signal in space, the communication distance is converted into a signal strength value to generate a signal strength matrix, which reflects the actual communication quality between each pair of nodes. The signal strength matrix is ​​normalized. The range of signal strength is limited to a standardized interval, so that different strength values ​​are more comparable, and the initial connection strength matrix is ​​obtained. By comparing the value of the initial connection strength matrix with the preset connection threshold, a binary connection matrix is ​​generated. If the connection strength of a pair of nodes exceeds the threshold, it is considered that there is a valid connection, and the value in the matrix is ​​set to 1; otherwise, it is set to 0. The binary connection matrix is ​​Laplace transformed. By comprehensively considering the number of connections and specific connection relationships of the nodes, a network connectivity matrix is ​​generated to reflect the topological structure of the entire network, including the degree of connection of each node and its importance in the overall network.

[0028] S2, deep learning dimension reduction and node screening are performed based on the network connectivity matrix to obtain the fault feature vector;

[0029] Specifically, the network connectivity matrix is ​​input into the deep autoencoder network for feature extraction. The deep autoencoder network is an unsupervised learning model that compresses and reconstructs data through an encoder and decoder structure. The deep autoencoder contains three layers of encoders and three layers of decoders. The encoder part extracts the deep features of the network connectivity matrix layer by layer, maps the high-dimensional data to a low-dimensional latent space representation, while the decoder part reconstructs the input matrix from the low-dimensional representation, and ensures that the extracted features are representative by minimizing the reconstruction error. Through this step, the network connectivity matrix is ​​converted into an initial feature set containing key features. The initial feature set is subjected to principal component analysis, the data is reduced in dimension, the principal component features are extracted, and a reduced dimension feature matrix is ​​formed to effectively describe the main structural relationship between nodes in the network. A node importance evaluation function is constructed based on the reduced dimension feature matrix. The node importance evaluation function measures the importance of each node in the network by calculating the degree centrality and betweenness centrality of the node, where the degree centrality describes the direct connection ability of the node, and the betweenness centrality reflects the role of the node as an intermediate bridge in the global network. By combining these indicators, the importance of each node is quantified to obtain a node score vector. Perform a Top-K screening operation on the node score vector, select the most important nodes, and generate a target node index set. Top-K screening sorts the node score vector and selects the top K nodes with the highest scores as target nodes. These nodes occupy a key position in the network and have high priority and representativeness. According to the target node index set, extract the relevant subgraph structure from the network connectivity matrix to form a target node connection submatrix. Calculate the shortest path distance for the target node connection submatrix. The shortest path distance is an important indicator for measuring the network distance between nodes, which can reflect the degree of mutual proximity of the target nodes in the topological structure. Generate the node distance matrix by calculating the shortest path distance for the connection submatrix. Normalize the node distance matrix so that all distance values ​​are converted to a unified standard range to obtain the normalized distance feature. Perform a tensor splicing operation on the reduced dimension feature matrix and the normalized distance feature to generate a fault feature vector.

[0030] S3, inputting the fault feature vector and the charging pile operation status data into the hybrid neural network for risk calculation to obtain the fault risk assessment value, the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork;

[0031] It should be noted that the fault feature vector is input into the spatial perception subnetwork of the hybrid neural network. This subnetwork captures the spatial relationship between nodes in the network. The structure includes three layers of graph attention layers, each of which contains a multi-head attention mechanism and a residual connection. The multi-head attention mechanism can capture the interaction between nodes from multiple angles, while the residual connection ensures that information is not lost excessively during feature transmission. In the first layer of graph attention layer, the network performs preliminary processing on the input fault feature vector and extracts a feature representation with 128 dimensions; the second layer of graph attention layer further compresses the features to 64 dimensions, and extracts the key patterns of node spatial association through deeper learning; the third layer of graph attention layer further compresses the features to 32 dimensions to obtain the final spatial perception output features. At the same time, the operating status data of the charging pile is processed, and the data is segmented using the sliding window method to generate a state sequence matrix. The sliding window segmentation process can retain the time series characteristics of the operating status, and at the same time, segmentation makes the data more representative in time series analysis. The generated state sequence matrix is ​​input into the time series analysis subnetwork of the hybrid neural network. The subnetwork contains two layers of bidirectional long short-term memory (BiLSTM) networks. The first layer of the bidirectional long short-term memory layer consists of 64 hidden units and the output dimension is 128. The main function of this layer is to capture the dynamic characteristics of the forward and reverse directions in the time series and generate the initial time series feature matrix. The initial time series feature matrix is ​​input into the second layer of the bidirectional long short-term memory layer for feature extraction. The second layer of the network contains 32 hidden units and the output dimension is 64. Through more refined time modeling, the output features of time series analysis are generated to effectively summarize the dynamic change law of the charging pile operation status data. The spatial perception output features and the time series analysis output features are fused. The spatial perception output features and the time series analysis output features represent the spatial distribution characteristics and dynamic time series characteristics of the charging pile network respectively. The combination of the two more comprehensively reflects the fault risk of the charging pile system. Through the feature fusion operation, the two features are spliced ​​into a unified fusion feature vector. The fused feature vector is input into the multi-classification layer of the network to calculate the fault risk probability. The multi-classification layer classifies according to different risk levels in the training data and outputs the fault risk probability distribution vector. The fault risk probability distribution vector contains probability values ​​corresponding to multiple risk levels, and each value represents the risk possibility of that level. Multi-classification calculations normalize the output into a probability form by introducing an activation function (such as softmax) to ensure that the sum of the probabilities of all risk levels is 1. The weighted risk score is calculated based on the fault risk probability distribution vector to obtain the final fault risk assessment value. The probability value of each risk level is multiplied by its corresponding risk weight coefficient and then summed. The risk weight coefficient is set according to the severity of different risk levels in actual applications. For example, higher risk levels are assigned higher weights, thereby paying more attention to the impact of serious faults. After weighted summation, a comprehensive risk assessment value is obtained to quantify the overall fault risk level of the current system.

[0032] S4, normalizing and reducing the dimension of the charging pile power response data to obtain a time domain mapping matrix;

[0033] Specifically, the power response data of the charging pile is sampled, and the power response matrix is ​​constructed by collecting the power data of different charging piles at specific time points. In the matrix, the row vector represents the identification of each charging pile, and the column vector corresponds to the sampling time series point. Each element represents the power value of a charging pile at a specific time point. In order to capture the characteristics of the charging pile power in the dynamic change process, the power response matrix is ​​calculated by step response. Step response is one of the core methods for analyzing the dynamic characteristics of the system. By calculating the transient response of power change, a step response data set is generated. This data set records the dynamic response process of the charging pile when the power changes, such as the transition behavior when the power switches from a lower state to a higher state or vice versa. The step response data set is normalized to map the numerical range of the data to a standard interval, such as [0,1], to eliminate the influence of the power amplitude difference between different charging piles, and to form a standardized step response matrix. The standardized step response matrix is ​​decomposed in state space, and the matrix is ​​decomposed into a state variable matrix and an output matrix. The former is used to describe the internal state of the system, and the latter is used to represent the external output of the system. State space decomposition can divide a complex dynamic system into several independent subsystems, thereby significantly reducing the complexity of calculations while retaining key information in the dynamic characteristics of the system. The state variable matrix is ​​transformed into a dimensionality reduction matrix, and the high-dimensional state space is mapped into a lower-dimensional subspace, while retaining the main dynamic characteristics contained in the original data to the greatest extent, and the reduced-order state matrix is ​​obtained. At the same time, combined with the output matrix, a reduced-order mapping function is constructed by designing a state observer to describe the dynamic relationship between the reduced-order state and the output of the system. The reduced-order mapping function is discretized on the time axis, and the continuous time dynamics are converted into a series of discrete time nodes to generate a time domain discrete sequence, which reflects the change process of the reduced-order state at discrete time points. The time domain discrete sequence is reconstructed to construct a time domain mapping matrix. In the time domain mapping matrix, each element represents the state transition probability between adjacent time nodes, which describes the state change trend of the charging pile between two time points. In the process of matrix reconstruction, the statistical characteristics between adjacent time nodes are extracted and converted into a quantized probability form through the analysis and operation of the discrete sequence.

[0034] S5, performing particle swarm optimization calculation based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation plan, and generating a control instruction set according to the charging power allocation plan, the control instruction set including the power and time parameters of each charging pile.

[0035] Among them, based on the time domain mapping matrix and the fault risk assessment value, a multi-objective optimization function is constructed. The design of this function aims to optimize the charging efficiency, load balance and system safety of the charging pile at the same time. The charging efficiency sub-goal is to increase the charging amount of the charging pile per unit time to meet the user's demand for fast charging; the load balancing sub-goal ensures a more uniform power distribution between different charging piles by minimizing the load difference of the charging piles in the system to avoid overloading or idleness of individual charging piles; the safety constraint sub-goal is used to limit the power and time distribution of the charging pile within a safe range to prevent equipment damage or safety accidents caused by power overload or abnormal charging time. Through the multi-objective optimization function, the dynamic characteristics of the charging pile reflected in the time domain mapping matrix are combined with the potential risks of the nodes revealed in the fault risk assessment value. After constructing the multi-objective optimization function, weights are assigned to each sub-goal to form a comprehensive objective function. The weight allocation is based on the importance of different sub-goals and the needs of actual application scenarios. For example, more attention is paid to charging efficiency during peak charging hours, while load balance and safety need to be improved in scenarios where system stability is prioritized. The comprehensive objective function unifies different optimization objectives into a computable form by weighted summation of each sub-goal. Based on the comprehensive objective function, a particle swarm optimizer with a forgetting factor is constructed, and the particle swarm P0 is initialized. The initial particle swarm contains M particles, each particle corresponds to a charging power allocation scheme, and its dimension is 2N, where N is the number of charging piles, the first N dimensions represent the power allocation value of each charging pile, and the last N dimensions represent the time allocation value of each charging pile. This structure enables particles to simultaneously represent the power and time allocation schemes of charging piles, providing flexibility for optimization calculations. After the initial particle swarm is generated, the fitness value set F is calculated for each particle, and the fitness value reflects the pros and cons of the particle's corresponding scheme. By comparing the fitness values, the particle's historical optimal solution ph and the global optimal solution pg are extracted. In order to prevent particles from falling into local optimal solutions and better adapt to dynamic optimization needs, the forgetting factor λ is introduced to dynamically update the historical optimal solution. The forgetting factor will impose a certain attenuation weight on the past optimal solution, so that particles can adjust their direction more quickly when exploring new solutions, thereby improving the effect of global optimization. The speed update calculation is performed on the initial particle swarm. The speed update process combines the historical optimal solution, the global optimal solution and the current particle position to ensure that the particles can move towards the historical experience and the global optimal direction at the same time. The calculated particle velocity matrix is ​​superimposed with the current position to generate an updated particle position matrix. In order to ensure the effectiveness of the optimization results, the updated particle position matrix is ​​subjected to boundary constraints to ensure that the power allocation value and time allocation value of each charging pile in the particle are within a reasonable range and will not exceed the safe operation parameters of the charging pile. The global optimal solution is extracted from the updated particle position matrix to obtain the charging power allocation plan, which includes the optimal power value and charging time period for each charging pile.The global optimal solution is extracted based on the comparison of fitness values, and the particle that makes the comprehensive objective function value optimal is selected as the final solution. Based on the charging power allocation plan, a set of control instructions is generated. Each control instruction contains the power and time parameters of the corresponding charging pile. These instructions will be sent to the charging pile execution system to adjust the actual charging behavior.

[0036] A hierarchical clustering analysis is performed on the historical fault data of the charging pile. The historical fault data records various fault conditions that occur in the system at different operating stages, including equipment abnormalities, power fluctuations, and communication interruptions. Through hierarchical clustering analysis, these data are grouped according to fault characteristics to form a high-dimensional fault feature matrix, revealing the potential correlation between different faults. Time series probability prediction is performed based on the fault feature matrix. By introducing a time series analysis model (such as LSTM or ARIMA), the time dependency and trend characteristics in the historical data are captured to generate a 24-hour fault prediction sequence. The sequence shows the distribution of possible faults in the system within the next day and their probability of occurrence. The 24-hour fault prediction sequence and the control instruction set are input into a two-stage decision model generator to construct a two-stage robust planning model. The two-stage robust planning model divides the system operation into two stages: the first stage is responsible for normal operation power allocation to ensure efficient and balanced operation of the charging pile under normal conditions; the second stage is used for emergency dispatch and designs a dynamic adjustment plan when a fault occurs. The objective function combines the operating cost and the emergency cost. The operating cost is used to measure the system resource consumption under normal conditions, while the emergency cost focuses on the adjustment cost after the fault occurs. By jointly optimizing the variables of the two stages, the model can reserve a certain adjustment space for emergency scenarios while meeting the normal operation requirements, thereby improving the robustness of the system. In order to deal with the uncertainty of emergency scenarios, the two-stage robust planning model is decomposed into emergency scenarios to generate an emergency uncertainty set U. A basic scenario set S0 is constructed, which contains all single-device failure scenarios and reflects the basic situation when a single charging pile fails; a combined scenario set S1 is generated based on the correlation between devices, and the combined scenario considers the possibility and impact of multiple devices failing at the same time; the combined scenario is screened by setting a probability threshold, and scenarios with extremely low probability of occurrence are removed to obtain a more streamlined and practical emergency uncertainty set U. Based on the emergency uncertainty set U, a constraint generation subproblem is constructed to extract a new emergency constraint condition set C. These constraints are used to limit the scheduling behavior of the system when a fault occurs, such as the power adjustment range and time switching window of the charging pile, to ensure the feasibility of the scheduling scheme in the emergency scenario. The constraints are applied to each instruction in the control instruction set, and the feasibility is verified to obtain a robust instruction subset R. The robust instruction subset contains control instructions that can still maintain reliability in emergency scenarios, reflecting the robustness and flexibility of the system in dealing with uncertain scenarios. The emergency dispatch strategy is designed based on the robust instruction subset R and the emergency uncertainty set U. The emergency dispatch strategy defines the automatic switching behavior of the system when a fault is detected, such as dynamically adjusting the power distribution of the faulty equipment, reallocating the charging tasks of the surrounding charging piles, or shortening the execution time of certain tasks to reduce the impact. These emergency dispatch strategies are encapsulated into the control instruction set to form the final target instruction set.Each instruction in the target instruction set is divided into a normal execution part and an emergency switching part. The normal execution part is used to ensure that the system runs as planned in the absence of faults, while the emergency switching part is automatically triggered when a fault occurs to dynamically adjust the system's operating status.

[0037] In one example, the IoT topology is constructed for the charging pile nodes to obtain the node communication distance matrix and the network connectivity matrix, including:

[0038] Deploy the IoT communication module to each charging pile device to configure the communication parameters and obtain node deployment data, which includes the device number and communication parameters of the IoT communication module;

[0039] Perform spatial positioning on the charging pile device equipped with the Internet of Things communication module to obtain a set of geographic coordinates, wherein the set of geographic coordinates includes the longitude, latitude and altitude values ​​of each charging pile;

[0040] The geographic coordinate set is converted into a network topology coordinate system to obtain standardized location data. The standardized location data represents the spatial distribution of nodes through three-dimensional rectangular coordinates (x, y, z);

[0041] Wireless communication links are constructed for the nodes in the standardized location data to obtain a node communication distance matrix, and a signal attenuation function is calculated based on the node communication distance matrix to obtain a signal strength matrix;

[0042] Performing interval normalization operation on the signal strength matrix to obtain an initial connection strength matrix, and comparing the initial connection strength matrix with a preset connection threshold to obtain a binary connection matrix;

[0043] The binary connection matrix is ​​transformed by Laplace matrix to obtain the network connectivity matrix.

[0044] In this example, the IoT communication module is deployed to each charging pile device, and the communication parameters are configured for each module. These communication parameters include communication frequency, transmission rate, signal strength, etc., which are used to ensure that the module can operate normally and communicate reliably with other nodes. After deployment, each module is assigned a unique device number (e.g. Indicates The number of the charging pile) and record its communication parameters Node deployment data is represented as The collection of charging piles reflects the identification and communication capabilities of each charging pile. The charging pile equipment equipped with the IoT communication module is spatially positioned. The geographic coordinate data of each charging pile is obtained through high-precision positioning technology (such as GPS or differential positioning), including longitude ( The label can accurately describe the physical location of the charging pile in three-dimensional space, forming a complete set of geographic coordinates: in is the total number of charging piles. In order to facilitate network topology analysis, the geographic coordinate set is converted into standardized location data in a three-dimensional rectangular coordinate system. The spherical coordinates (longitude and latitude) are converted into rectangular coordinates using the following formula:

[0045] ;

[0046] ;

[0047] ;

[0048] in, is the mean radius of the Earth, It is The altitude of the node, and are its longitude and latitude respectively. Through these formulas, the spatial position of each node is expressed as Represents, generating a standardized location data set:

[0049] ;

[0050] After obtaining the standardized location data, the wireless communication link between the nodes is constructed. The Euclidean distance between each pair of nodes is calculated, and the distance formula is:

[0051] ;

[0052] in, Representation Node and nodes The spatial distance between them. By calculating all node pairs, the node communication distance matrix is ​​generated. , where each element Indicates Nodes and The distance between nodes. Based on the node communication distance matrix , calculate the signal attenuation function and generate the signal strength matrix. Signal attenuation is expressed by the following formula:

[0053] ;

[0054] in, Is a node The signal emitted at the node The intensity received at Is a node The transmission power, is the path loss exponent (usually between 2 and 4, depending on the propagation environment), Is a node and Through this calculation, we get the signal strength matrix , where each element Indicates the communication signal strength between two nodes. In order to make the data of the signal strength matrix more intuitive and easier to process, it is normalized. The normalization formula is:

[0055] ;

[0056] Normalized initial connection strength matrix All signal strength values ​​are compressed to the range of [0, 1] to eliminate the influence of signal power differences between different nodes on the results. With the preset connection threshold A comparison operation is performed to determine if there is a valid connection between the nodes. , then the node and nodes There is a communication connection between them, otherwise it is considered that there is no connection. This process generates a binary connection matrix , whose elements are defined as:

[0057] ;

[0058] Binarized connection matrix Perform Laplace matrix transformation to generate the network connectivity matrix. The Laplace matrix is ​​defined as:

[0059] ;

[0060] in, is the degree matrix, which is a diagonal matrix with diagonal elements Representation Node The degree of a node is the number of all its connections:

[0061] ;

[0062] The network connectivity matrix generated by Laplace matrix transformation It can effectively reflect the connection characteristics of nodes in the network and their importance in the entire topology.

[0063] In one example, deep learning dimensionality reduction and node screening are performed based on the network connectivity matrix to obtain a fault feature vector, including:

[0064] The network connectivity matrix is ​​input into the deep autoencoder network for feature extraction to obtain an initial feature set. The deep autoencoder network includes a three-layer encoder and a three-layer decoder structure.

[0065] Perform principal component analysis on the initial feature set to obtain a reduced-dimensional feature matrix, and construct a node importance evaluation function based on the reduced-dimensional feature matrix to obtain a node score vector. The node score vector is calculated using node degree centrality and betweenness centrality.

[0066] Perform a Top-K screening operation on the node score vector to obtain the target node index set, and extract the subgraph structure from the network connectivity matrix based on the target node index set to obtain the target node connection submatrix;

[0067] Calculate the shortest path distance for the target node connection submatrix to obtain the node distance matrix, and normalize the node distance matrix to obtain the normalized distance feature;

[0068] The fault feature vector is obtained by performing tensor concatenation operation on the reduced dimension feature matrix and the normalized distance feature.

[0069] In this example, the network connectivity matrix The input is fed into a deep autoencoder network for feature extraction. The deep autoencoder network is an unsupervised learning model that consists of three layers of encoders and three layers of decoders. The encoder gradually maps high-dimensional data to a low-dimensional latent space to compress information and extract key features, while the decoder remaps the compressed features back to the original space to minimize reconstruction errors. Assume that the input network connectivity matrix ,in is the number of nodes, indicating the size of the network. After the first layer of encoder, the data is mapped to a hidden layer with an output dimension of :

[0070] ;

[0071] in, is the weight matrix, is the bias vector, is an activation function, such as ReLU. The second encoder layer will Mapped again to a lower dimensional hidden layer with dimension :

[0072] ;

[0073] Similarly, after the third encoder layer, the low-dimensional feature representation of the latent space is obtained. The decoder part works in reverse, converting Gradually decode back to the original space, by minimizing the reconstruction error To optimize the network weights. Among them, is the reconstructed connectivity matrix. The extracted latent features is defined as the initial feature set, which represents the key characteristics of the network. The principal component analysis is performed on the initial feature set to reduce the dimension and extract the main features. Principal component analysis can find the principal component directions of the data, which can explain the maximum variance in the data. Assume that the initial feature set is the matrix ,right Calculate the covariance matrix:

[0074] ;

[0075] Among them, express The characteristic correlation of the covariance matrix is ​​then calculated, and the eigenvalues ​​and eigenvectors are selected. The largest eigenvalues ​​and their corresponding eigenvectors form a reduced-dimensional feature matrix The reduced-dimensional feature matrix retains the main information of the network structure while significantly reducing the data complexity. , construct the node importance evaluation function to obtain the node score vector The importance of a node is measured by centrality metrics in graph theory, including degree centrality and betweenness centrality. Representation Node The number of direct connections is defined as:

[0076] ;

[0077] in is the binary adjacency matrix element, indicating the node and Whether they are directly connected. Betweenness centrality It reflects the role of the node in all shortest paths and is defined as:

[0078] ;

[0079] in, Is a node and nodes The number of shortest paths between These paths pass through the nodes Through normalization, the two centrality indicators are combined to obtain the comprehensive node score vector . Perform Top-K screening on the node score vector and select the one with the highest score nodes as the target nodes and generate the target node index set According to the target node index set, from the network connectivity matrix Extract the subgraph structure of the target node and obtain the target node connection submatrix . Connect the submatrix to the target node Perform shortest path distance calculation to generate node distance matrix . Shortest path distance Is a node and The path weight and the minimum path length between Perform normalization processing, map the distance value to the interval [0,1], and generate a normalized distance feature matrix . Reduce the dimension of the feature matrix And the normalized distance feature matrix Perform tensor concatenation operations. The concatenated tensor integrates the structural characteristics and spatial distance characteristics of the nodes to form the final fault feature vector .

[0080] In one example, the fault feature vector and the charging pile operation status data are input into a hybrid neural network for risk calculation to obtain a fault risk assessment value. The hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork, including:

[0081] The fault feature vector is input into the spatial perception subnetwork in the hybrid neural network for feature extraction. The spatial perception subnetwork contains three graph attention layers. Each graph attention layer contains a multi-head attention mechanism and a residual connection. The output dimension of the first graph attention layer is 128, the output dimension of the second graph attention layer is 64, and the output dimension of the third graph attention layer is 32, and the spatial perception output feature is obtained;

[0082] The charging pile operation status data is segmented by sliding windows to obtain a state sequence matrix, and the state sequence matrix is ​​input into the first bidirectional long short-term memory layer of the time series analysis subnetwork in the hybrid neural network. The first bidirectional long short-term memory layer contains 64 hidden units and the output dimension is 128, and the initial time series feature matrix is ​​obtained.

[0083] The initial time series feature matrix is ​​input into the second bidirectional long short-term memory layer of the time series analysis subnetwork for feature extraction. The second bidirectional long short-term memory layer contains 32 hidden units and the output dimension is 64, so as to obtain the output features of the time series analysis.

[0084] Perform feature fusion on the spatial perception output features and the time series analysis output features to obtain a fused feature vector, and perform multi-classification calculation on the fused feature vector to obtain a fault risk probability distribution vector;

[0085] The weighted risk score is calculated based on the failure risk probability distribution vector to obtain the failure risk assessment value. The weighted risk score is calculated by summing the product of the probability value of each risk level and the corresponding risk weight coefficient.

[0086] In this example, the fault feature vector Input to the spatial perception subnetwork of the hybrid neural network. represents the number of nodes, Represents the feature dimension of each node. The spatial perception subnetwork gradually extracts features through three layers of graph attention layers, each of which contains a multi-head attention mechanism and residual connection. The core of the graph attention layer is to calculate the correlation between nodes and perform weighted aggregation of neighbor node features through the attention mechanism. Graph attention of layers, nodes The output features of are defined as:

[0087] ;

[0088] in, Indicates The node in The characteristics of the layer, Is a node The neighbor set of is the weight matrix, Is a node and nodes The attention weight between is calculated by the following formula:

[0089] ;

[0090] in, is a learnable attention vector, Represents the feature concatenation operation. The output dimension of the first graph attention layer is 128, and the generated feature matrix is Through similar calculations, the output dimensions of the second and third layers are 64 and 32 respectively, and the spatial perception output features are obtained . Preprocess the charging pile operation status data and generate the state sequence matrix through sliding window segmentation ,in is the number of time windows, It is the feature dimension within a single window (such as power, voltage and other parameters). The sliding window method can capture the time series characteristics of the data while retaining the local characteristics of each time segment. Input to the first bidirectional long short-term memory layer (BilSTM) of the time series analysis subnetwork of the hybrid neural network. The first layer of BiLSTM contains 64 hidden units and the output dimension is 128. For each time step , the output of the bidirectional LSTM is calculated as follows:

[0091] ;

[0092] in, and are the output features of the forward and reverse LSTMs respectively. The output of the entire sequence forms the initial time series feature matrix The initial time series feature matrix is ​​input to the second bidirectional long short-term memory layer for feature extraction. The second layer of BiLSTM contains 32 hidden units and the output dimension is 64, which finally generates the time series analysis output feature After completing spatial perception and time series analysis, the two features are fused. and timing analysis output characteristics Generate a fused feature vector by fusion through feature splicing The fused feature vector is input into the multi-classification layer to calculate the fault risk probability. The multi-classification layer uses the softtmax activation function to convert the output into the probability distribution of each risk level:

[0093] ;

[0094] in, Is the risk level The probability of is the classifier for the level The unnormalized score of is the total number of risk levels. The resulting failure risk probability distribution vector is . Calculate the weighted risk score based on the failure risk probability distribution vector to obtain the final failure risk assessment value The weighted risk score is calculated using the following formula:

[0095] ;

[0096] in, Is the risk level The weight of the level reflects the impact of the level on the overall risk of the system. is the probability of the corresponding level. Through weighted calculation, the evaluation value Quantify the failure risk level of the charging pile system in its current state.

[0097] In one example, the charging pile power response data is normalized and dimensionally reduced to obtain a time domain mapping matrix, including:

[0098] The power response data of the charging pile is sampled to obtain a power response matrix, the row vector of the power response matrix represents the charging pile identifier, and the column vector represents the time series sampling point;

[0099] Perform step response calculation on the power data in the power response matrix to obtain a step response data set, which records the dynamic response process of the charging pile at the moment of power change;

[0100] Performing normalization operation on the step response data set to obtain a standardized step response matrix, and performing state space decomposition on the standardized step response matrix to obtain a state variable matrix and an output matrix;

[0101] Performing dimension reduction transformation on the state variable matrix to obtain a reduced-order state matrix, and designing a state observer for the reduced-order state matrix and the output matrix to obtain a reduced-order mapping function;

[0102] The reduced-order mapping function is discretized on the time axis to obtain a time-domain discrete sequence, and a matrix reconstruction operation is performed on the time-domain discrete sequence to obtain a time-domain mapping matrix. Each element of the time-domain mapping matrix represents the state transition probability between adjacent time nodes.

[0103] In this example, the power response data of the charging pile is sampled to form a power response matrix ,in Indicates the number of charging piles. Indicates the number of time series sampling points. In , each row represents the power response record of a charging pile over the entire time series, and each column represents the power value of all charging piles at a specific time point. In order to analyze the dynamic response characteristics of the charging pile at the moment of power change, the power response matrix is ​​calculated for step response. Step response is one of the core characteristics of system dynamics, which is used to describe the transient behavior of the system when it jumps from one steady state to another. By calculating the response curve of power change, a step response data set is generated. , where each set of data describes the step response process of the charging pile at a specific moment. For example, the response process of the recording system , satisfying the following form:

[0104] ;

[0105] in, and denote the initial and final steady-state power values, respectively, is the power change, is the time constant of the response speed, is the time when the power change starts. The generated step response data set is normalized to map the power response data of different charging piles to a standard range, such as [0,1], for unified processing. The normalized matrix It is expressed as:

[0106] ;

[0107] in, is the raw step response data, and are the minimum and maximum values ​​in the data set respectively. Perform state space decomposition on the standardized step response matrix to obtain the state variable matrix And the output matrix The state-space model is of the form:

[0108] ;

[0109] ;

[0110] in, is a state variable, is the state transition matrix, is the input matrix, is the system input, is the output variable, and The system is modeled numerically and extracted from the standardized matrix In order to simplify the system, the state variable matrix Perform dimensionality reduction transformation, extract the main features through singular value decomposition or principal component analysis, and obtain the reduced-order state matrix ,in After the dimensionality reduction transformation, the reduced-order mapping function is constructed by designing the state observer:

[0111] ;

[0112] ;

[0113] in, is the state space parameter after reduction. The reduced-order mapping function is discretized on the time axis to generate a discrete sequence in the time domain. The discretization process converts the continuous time dynamics into the state value at the discrete time point. Let the discrete time step be , and obtain the discretized state equation:

[0114] ;

[0115] in, is the discrete state transfer matrix, Is a discrete input matrix. Perform matrix reconstruction on the time domain discrete sequence to generate the time domain mapping matrix The elements of the time domain mapping matrix are Indicates at time node and The state transition probability between is calculated by the following formula:

[0116] ;

[0117] This matrix provides the dynamic behavior of the charging pile at each time node, which can effectively support the analysis and optimization of power response.

[0118] In one example, a particle swarm optimization calculation is performed based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation scheme, and a control instruction set is generated according to the charging power allocation scheme. The control instruction set includes power and time parameters of each charging pile, including:

[0119] A multi-objective optimization function is constructed based on the time domain mapping matrix and the fault risk assessment value. The multi-objective optimization function includes a charging efficiency sub-objective, a load balancing sub-objective, and a safety constraint sub-objective.

[0120] Perform weight distribution calculation on the multi-objective optimization function to obtain a weighted comprehensive objective function;

[0121] Based on the comprehensive objective function, a particle swarm optimizer with a forgetting factor is constructed to obtain the initial particle swarm P0. The initial particle swarm P0 contains M particles, and the dimension of each particle is 2N, where N is the number of charging piles, the first N dimensions represent power allocation, and the last N dimensions represent time allocation;

[0122] Calculate the historical optimal solution ph and the global optimal solution pg for each particle in the initial particle swarm P0, obtain the particle fitness value set F, and introduce the forgetting factor λ to dynamically update the historical optimal solution;

[0123] Based on the forgetting factor λ, the velocity of the initial particle group is updated and calculated to obtain the particle velocity matrix;

[0124] The particle velocity matrix is ​​superimposed with the current position to obtain the updated particle position matrix, and the particle position matrix is ​​subjected to boundary constraint processing to ensure that the power and time parameters are within the valid range;

[0125] The global optimal solution is extracted from the updated particle position matrix to obtain the charging power allocation scheme, which includes the optimal charging power value and charging time period of each charging pile;

[0126] A control instruction set is generated based on the charging power allocation plan, and the control instruction set includes power and time parameters of each charging pile.

[0127] In this example, according to the time domain mapping matrix and failure risk assessment value , construct a multi-objective optimization function. The core of multi-objective optimization is to optimize multiple objective functions at the same time. The multi-objective optimization function includes three sub-objectives: charging efficiency sub-objective, load balancing sub-objective and safety constraint sub-objective. Charging efficiency sub-objective It is used to maximize the total charging power of all charging piles, which is defined as:

[0128] ;

[0129] in, It is The power value of the charging pile, the negative sign indicates that it is transformed into a minimization problem. Load balancing sub-goal It is used to minimize the imbalance of charging pile load in the system and is defined as:

[0130] ;

[0131] in, is the average power of all charging piles. Safety constraint sub-goal It is used to limit the allocation range of power and time to ensure that the parameters meet the system safety requirements. It is defined in the form of a penalty function:

[0132] ;

[0133] in, and are the maximum allowable values ​​of power and time respectively. These three sub-goals are combined into a comprehensive objective function To balance the importance of different goals, a weight is assigned to each sub-goal The comprehensive objective function is in the form of:

[0134] ;

[0135] in, . By adjusting the weights, some goals are prioritized, such as increasing the weight of charging efficiency during peak hours and focusing more on load balancing during off-peak hours. Based on the comprehensive objective function, a particle swarm optimizer with a forgetting factor is constructed. The particle swarm optimizer gradually approaches the global optimal solution by simulating the movement of particles in the solution space. Initialize a particle swarm ,Include particles, each with a dimension of , among which the former The dimension represents the power distribution of the charging pile. The dimension represents the time distribution. Let particle The location is , the speed is For each particle, calculate its fitness value , which is the value of the comprehensive objective function. Record the historical optimal solution of each particle and the global optimal solution In order to enhance the adaptability to dynamic environments, the forgetting factor is introduced Dynamically update the historical optimal solution:

[0136] ;

[0137] Based on the forgetting factor, update the speed and position of each particle. The speed update formula is:

[0138] ;

[0139] in, is the inertia weight, is the learning factor, is a random number. The position update formula is:

[0140] ;

[0141] To ensure that the power and time allocation of particles are within the valid range, the updated particle positions are bounded:

[0142] ;

[0143] ;

[0144] Through multiple iterations, the global optimal solution is extracted , which is the optimal power allocation and time allocation scheme for all charging piles. Assume that the final charging power allocation scheme is The time allocation plan is Based on the charging power allocation plan, a set of control instructions is generated. Each instruction contains the charging pile number, power parameters and time parameters, for example:

[0145] ;

[0146] All instruction sets constitute the system's charging scheduling plan for the charging piles to execute.

[0147] In one example, the intelligent operation and management method of charging piles based on the Internet of Things also includes:

[0148] Perform hierarchical clustering analysis on the historical fault data of charging piles to obtain the fault feature matrix, and perform time series probability prediction based on the fault feature matrix to obtain a 24-hour fault prediction sequence;

[0149] The 24-hour fault prediction sequence and control instruction set are input into the two-stage decision model generator to obtain a two-stage robust planning model. The first-stage variable of the two-stage robust planning model is the normal operation power allocation, the second-stage variable is the emergency dispatch plan, and the objective function includes the operation cost term and the emergency cost term.

[0150] The emergency scenario of the two-stage robust planning model is decomposed to obtain the emergency uncertainty set U. The emergency uncertainty set U is generated by the following steps: first, a basic scenario set S0 is constructed, which includes all single-device failure scenarios, then a combined scenario set S1 is generated based on the correlation between devices, and finally, the emergency uncertainty set U is obtained by screening through probability thresholds;

[0151] Based on the emergency uncertainty set U, a constraint generation sub-problem is constructed to obtain a new emergency constraint condition set C. The emergency constraint condition set C is applied to each instruction in the control instruction set for feasibility verification to obtain a robust instruction subset.

[0152] An emergency dispatch strategy is designed based on the robust instruction subset R and the emergency uncertainty set U, and the emergency dispatch strategy is encapsulated into the control instruction set to obtain the target instruction set. Each instruction in the target instruction set contains a normal execution part and an emergency switching part. When a fault is detected, the corresponding emergency dispatch strategy is automatically triggered.

[0153] In this example, hierarchical clustering analysis is performed on the historical fault data of charging piles. Assume that the historical fault data contains Charging stations in the past Fault record matrix at each time point , where each element Indicates Charging piles at time Based on this data, a hierarchical clustering algorithm is used to classify charging piles with similar fault characteristics by calculating the similarity between charging piles (such as Euclidean distance or cosine similarity). The similarity measure of clustering is defined as:

[0154] Similarity ;

[0155] By constructing a clustering tree and setting an appropriate cutting threshold, we can obtain Clusters are formed, each cluster represents a group of charging piles with similar fault characteristics. Based on the clustering results, a fault feature matrix is ​​constructed. , where each row represents the change of the average fault intensity of a cluster over time. Based on the fault feature matrix Perform time series probability forecasting to predict failures in the next 24 hours. Use a time series forecasting model (such as LSTM or ARIMA) and input the matrix , the training model captures the dynamic characteristics of the time series and outputs a 24-hour fault prediction sequence Each element Indicates The cluster is in The 24-hour fault prediction sequence and control instruction set Input into the two-stage decision model generator to build a two-stage robust planning model. The first-stage variable of the two-stage robust planning model is the normal operation power allocation, denoted as ,in It is The second-stage variable is the emergency dispatch plan, denoted as ,in It is The power adjustment of each charging pile in an emergency scenario. The objective function of the model consists of an operating cost term and an emergency cost term:

[0156] ;

[0157] in, is the unit cost of operating power, is the unit cost of emergency adjustment. The goal is to optimize the cost of normal operation and emergency response simultaneously. In order to improve the robustness, the emergency scenario decomposition of the two-stage robust planning model is performed to generate the emergency uncertainty set . Build a basic scene set , including all single device failure scenarios, each scenario Simulate the fault behavior of only one charging station. Generate a set of combined scenarios based on the correlation between devices . The correlation is calculated using historical fault data, such as charging piles and The correlation is expressed as:

[0158] Relevance ;

[0159] Select a combination of devices with high correlation to generate a joint failure scenario. , the probability of screening occurrence is higher than The emergency uncertainty set Based on the emergency uncertainty set , construct constraint generation subproblems to generate new sets of emergency constraints The constraints include bounds on power and time allocation:

[0160] ;

[0161] At the same time, combined with the power demand changes in uncertain scenarios, specific power constraints are generated for each scenario:

[0162] ;

[0163] in, Representation scene Next The minimum demand for charging piles. Apply the emergency constraint set to each instruction in the control instruction set Perform feasibility verification, eliminate instructions that cannot meet the constraints, and obtain a robust instruction subset . Based on a robust instruction subset and the emergency uncertainty set , design emergency dispatch strategies. Emergency dispatch strategies define power adjustment plans according to different scenarios to ensure that the system can respond quickly when a fault occurs. For example, when a charging pile is detected In case of a fault, reduce its power allocation , while dynamically increasing the power of surrounding related charging piles , meeting the overall needs. Encapsulate these emergency strategies into a control instruction set to generate a target instruction set. Each instruction in the target instruction set includes a normal execution part and an emergency switching part, in the form of:

[0164] normal: , Emergency: ;

[0165] When a fault is detected, the emergency dispatch strategy is automatically triggered to adjust power and time.

[0166] Reference Figure 2 This embodiment provides a charging pile intelligent operation and management system based on the Internet of Things, including:

[0167] Construction module 1 is used to construct the Internet of Things topology for charging pile nodes and obtain the node communication distance matrix and network connectivity matrix;

[0168] Screening module 2 is used to perform deep learning dimension reduction and node screening based on the network connectivity matrix to obtain a fault feature vector;

[0169] Calculation module 3, used for inputting the fault feature vector and the charging pile operation status data into the hybrid neural network for risk calculation to obtain the fault risk assessment value, the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork;

[0170] Mapping module 4, used for normalizing and reducing the dimension of the charging pile power response data to obtain a time domain mapping matrix;

[0171] The generation module 5 is used to perform particle swarm optimization calculation based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation plan, and generate a control instruction set according to the charging power allocation plan, wherein the control instruction set includes power and time parameters of each charging pile.

[0172] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the description in the above method embodiment, which will not be repeated here.

[0173] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0174] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0175] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0176] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0177] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0178] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A charging pile intelligent operation and management method based on the Internet of Things, characterized in that: The following steps are involved: The IoT topology is constructed for charging pile nodes to obtain the node communication distance matrix and network connectivity matrix; Perform deep learning dimensionality reduction and node screening based on the network connectivity matrix to obtain a fault feature vector; Inputting the fault feature vector and the charging pile operation status data into a hybrid neural network for risk calculation to obtain a fault risk assessment value, wherein the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork; Normalize and reduce the dimension of the charging pile power response data to obtain the time domain mapping matrix; A particle swarm optimization calculation is performed based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation scheme, and a control instruction set is generated according to the charging power allocation scheme, wherein the control instruction set includes power and time parameters of each charging pile.

2. The method for intelligent operation and management of charging piles based on the Internet of Things according to claim 1 is characterized in that: The IoT topology is constructed for the charging pile nodes to obtain the node communication distance matrix and the network connectivity matrix, including: Deploy the IoT communication module to each charging pile device to configure communication parameters and obtain node deployment data, wherein the node deployment data includes the device number and communication parameters of the IoT communication module; Performing spatial positioning on the charging pile device equipped with the Internet of Things communication module to obtain a set of geographic coordinates, wherein the set of geographic coordinates includes the longitude, latitude and altitude values ​​of each charging pile; Converting the geographic coordinate set into a network topology coordinate system to obtain standardized location data, wherein the standardized location data represents the spatial distribution of nodes through three-dimensional rectangular coordinates (x, y, z); Establishing wireless communication links for the nodes in the standardized location data to obtain a node communication distance matrix, and calculating a signal attenuation function based on the node communication distance matrix to obtain a signal strength matrix; Performing interval normalization operation on the signal strength matrix to obtain an initial connection strength matrix, and performing a comparison operation on the initial connection strength matrix and a preset connection threshold to obtain a binary connection matrix; The binary connection matrix is ​​subjected to Laplace matrix transformation to obtain a network connectivity matrix.

3. The method for intelligent operation and management of charging piles based on the Internet of Things according to claim 2 is characterized in that: The deep learning dimension reduction and node screening based on the network connectivity matrix to obtain the fault feature vector includes: Inputting the network connectivity matrix into a deep autoencoder network for feature extraction to obtain an initial feature set, wherein the deep autoencoder network comprises a three-layer encoder and a three-layer decoder structure; Performing principal component analysis on the initial feature set to obtain a reduced dimension feature matrix, and constructing a node importance evaluation function based on the reduced dimension feature matrix to obtain a node score vector, wherein the node score vector is calculated by node degree centrality and betweenness centrality; Performing a Top-K screening operation on the node score vector to obtain a target node index set, and extracting a subgraph structure from the network connectivity matrix according to the target node index set to obtain a target node connection submatrix; Calculating the shortest path distance for the target node connection submatrix to obtain an inter-node distance matrix, and normalizing the inter-node distance matrix to obtain a normalized distance feature; A tensor concatenation operation is performed on the dimension reduction feature matrix and the normalized distance feature to obtain a fault feature vector.

4. The method for intelligent operation and management of charging piles based on the Internet of Things according to claim 3 is characterized in that: The fault feature vector and the charging pile operation status data are input into a hybrid neural network for risk calculation to obtain a fault risk assessment value, wherein the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork, including: The fault feature vector is input into the spatial perception subnetwork in the hybrid neural network for feature extraction, wherein the spatial perception subnetwork includes three graph attention layers, each graph attention layer includes a multi-head attention mechanism and a residual connection, the output dimension of the first graph attention layer is 128, the output dimension of the second graph attention layer is 64, and the output dimension of the third graph attention layer is 32, to obtain spatial perception output features; Sliding window segmentation is performed on the charging pile operation status data to obtain a state sequence matrix, and the state sequence matrix is ​​input into the first bidirectional long short-term memory layer of the time series analysis subnetwork in the hybrid neural network, the first bidirectional long short-term memory layer includes 64 hidden units, and the output dimension is 128, to obtain an initial time series feature matrix; Input the initial time series feature matrix into the second bidirectional long short-term memory layer of the time series analysis subnetwork for feature extraction, wherein the second bidirectional long short-term memory layer includes 32 hidden units and the output dimension is 64, to obtain the time series analysis output feature; Performing feature fusion on the spatial perception output feature and the time series analysis output feature to obtain a fused feature vector, and performing multi-classification calculation on the fused feature vector to obtain a fault risk probability distribution vector; A weighted risk score is calculated based on the failure risk probability distribution vector to obtain a failure risk assessment value, wherein the weighted risk score is calculated by summing the product of the probability value of each risk level and the corresponding risk weight coefficient.

5. The method for intelligent operation and management of charging piles based on the Internet of Things according to claim 4 is characterized in that: The charging pile power response data is normalized and dimensionally reduced to obtain a time domain mapping matrix, including: Sampling the charging pile power response data to obtain a power response matrix, wherein the row vector of the power response matrix represents the charging pile identifier, and the column vector represents the time series sampling point; Performing a step response calculation on the power data in the power response matrix to obtain a step response data set, wherein the step response data set records the dynamic response process of the charging pile at the moment of power change; Performing a normalization operation on the step response data set to obtain a standardized step response matrix, and performing state space decomposition on the standardized step response matrix to obtain a state variable matrix and an output matrix; Performing a dimension reduction transformation on the state variable matrix to obtain a reduced-order state matrix, and performing a state observer design on the reduced-order state matrix and the output matrix to obtain a reduced-order mapping function; The reduced-order mapping function is discretized on the time axis to obtain a time-domain discrete sequence, and a matrix reconstruction operation is performed on the time-domain discrete sequence to obtain a time-domain mapping matrix, wherein each element of the time-domain mapping matrix represents a state transition probability between adjacent time nodes.

6. The method for intelligent operation and management of charging piles based on the Internet of Things according to claim 5 is characterized in that: The particle swarm optimization calculation is performed based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation scheme, and a control instruction set is generated according to the charging power allocation scheme, wherein the control instruction set includes power and time parameters of each charging pile, including: Constructing a multi-objective optimization function based on the time domain mapping matrix and the fault risk assessment value, wherein the multi-objective optimization function includes a charging efficiency sub-objective, a load balancing sub-objective, and a safety constraint sub-objective; Performing weight distribution calculation on the multi-objective optimization function to obtain a weighted comprehensive objective function; Based on the comprehensive objective function, a particle swarm optimizer with a forgetting factor is constructed to obtain an initial particle swarm P0, wherein the initial particle swarm P0 includes M particles, and the dimension of each particle is 2N, where N is the number of charging piles, the first N dimensions represent power allocation, and the last N dimensions represent time allocation; Calculate the historical optimal solution ph and the global optimal solution pg for each particle in the initial particle swarm P0 to obtain a particle fitness value set F, and introduce a forgetting factor λ to dynamically update the historical optimal solution; Performing velocity update calculation on the initial particle group based on the forgetting factor λ to obtain a particle velocity matrix; The particle velocity matrix is ​​superimposed on the current position to obtain an updated particle position matrix, and the particle position matrix is ​​subjected to boundary constraint processing to ensure that the power and time parameters are within a valid range; Extracting a global optimal solution from the updated particle position matrix to obtain a charging power allocation scheme, wherein the charging power allocation scheme includes an optimal charging power value and a charging time period for each charging pile; A control instruction set is generated based on the charging power allocation scheme, and the control instruction set includes power and time parameters of each charging pile.

7. The method for intelligent operation and management of charging piles based on the Internet of Things according to claim 6 is characterized in that: The IoT-based charging pile intelligent operation and management method also includes: Perform hierarchical clustering analysis on the historical fault data of charging piles to obtain a fault feature matrix, and perform time series probability prediction based on the fault feature matrix to obtain a 24-hour fault prediction sequence; Inputting the 24-hour fault prediction sequence and the control instruction set into a two-stage decision model generator to obtain a two-stage robust planning model, wherein the first-stage variable of the two-stage robust planning model is the normal operating power allocation, the second-stage variable is the emergency dispatch plan, and the objective function includes an operating cost term and an emergency cost term; The two-stage robust programming model is decomposed into emergency scenarios to obtain an emergency uncertainty set U, which is generated by the following steps: first, a basic scenario set S0 is constructed, including all single-device failure scenarios, then a combined scenario set S1 is generated based on the correlation between devices, and finally, the emergency uncertainty set U is obtained by screening through a probability threshold; Based on the emergency uncertainty set U, a constraint generation sub-problem is constructed to obtain a new emergency constraint condition set C, and the emergency constraint condition set C is applied to each instruction in the control instruction set to perform feasibility verification to obtain a robust instruction subset; An emergency dispatch strategy is designed based on the robust instruction subset R and the emergency uncertainty set U, and the emergency dispatch strategy is encapsulated into the control instruction set to obtain a target instruction set. Each instruction in the target instruction set includes a normal execution part and an emergency switching part. When a fault is detected, the corresponding emergency dispatch strategy is automatically triggered.

8. A charging pile intelligent operation and management system based on the Internet of Things, characterized in that: For implementing the steps of the method according to any one of claims 1 to 7, the system comprises: The construction module is used to construct the IoT topology for charging pile nodes and obtain the node communication distance matrix and network connectivity matrix; A screening module, used for performing deep learning dimension reduction and node screening based on the network connectivity matrix to obtain a fault feature vector; A calculation module, used for inputting the fault feature vector and the charging pile operation status data into a hybrid neural network for risk calculation to obtain a fault risk assessment value, wherein the hybrid neural network includes a spatial perception subnetwork and a timing analysis subnetwork; A mapping module is used to normalize and reduce the dimension of the charging pile power response data to obtain a time domain mapping matrix; A generation module is used to perform particle swarm optimization calculation based on the time domain mapping matrix and the fault risk assessment value to obtain a charging power allocation plan, and generate a control instruction set according to the charging power allocation plan, wherein the control instruction set includes power and time parameters of each charging pile.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images