An electric vehicle charging system energy efficiency dynamic self-adaptive evaluation method, system, electronic device and medium

Abstract

The application discloses an electric vehicle charging system energy efficiency dynamic self-adaptive evaluation method and system, an electronic device and a storage medium. In the method, multi-source operation data of a charging system to be evaluated is obtained, a hierarchical structure model is constructed, and key energy efficiency evaluation indexes are determined in combination with a graph neural network; objective weights of each index are calculated by using an entropy weight method to obtain an initial energy efficiency comprehensive index; the multi-source operation data and the initial energy efficiency comprehensive index are input into a pre-trained deep reinforcement learning network, index weights are dynamically adjusted according to real-time environmental parameters and equipment states, and a dynamic energy efficiency comprehensive index is obtained; and then, clustering analysis is performed on the dynamic energy efficiency comprehensive index, and a ladder performance efficiency grade evaluation result is generated according to a clustering center and a clustering boundary. The application can reduce the influence of subjective weighting, dynamically update the energy efficiency evaluation result according to the running state of the charging system, and improve the objectivity and adaptability of the energy efficiency evaluation.

Description

Technical Field

[0001] This invention belongs to the field of electrical energy efficiency assessment technology, specifically relating to a dynamic adaptive assessment method, system, electronic device, and medium for the energy efficiency of an electric vehicle charging system. Background Technology

[0002] With the continuous growth of electric vehicle ownership and charging infrastructure, the operational energy efficiency of charging stations, centralized charging sites, and multi-terminal charging systems has gradually become a crucial factor affecting the operating efficiency of power distribution networks, equipment maintenance costs, and energy conservation and emission reduction effects. Electric vehicle charging systems involve multiple stages during operation, including grid power reception, power conversion, power output, power quality, equipment standby, load dispatching, environmental changes, and operation management. Their energy consumption sources and influencing factors are multi-source, dynamic, and coupled. Therefore, how to comprehensively, objectively, and dynamically quantify the energy efficiency status of charging systems is a critical technical issue that needs to be addressed in the refined management of charging infrastructure.

[0003] Existing energy efficiency assessment methods for charging systems typically rely on charging efficiency, power factor, or single-point energy metering results as the main evaluation criteria. However, a single or limited set of indicators cannot accurately describe the true energy efficiency level under different evaluation windows, nor can they reveal the correlation between environmental parameters, equipment status, load fluctuations, and power quality.

[0004] Furthermore, when the operating status of the charging system, environmental conditions, or load structure changes, the fixed weights set for energy efficiency indicators by existing methods are difficult to continuously reflect the actual contribution of each indicator to overall energy efficiency.

[0005] Therefore, there is an urgent need for a dynamic adaptive evaluation method for the energy efficiency of electric vehicle charging systems that can dynamically adjust the evaluation results based on real-time conditions, so as to improve the comprehensiveness, objectivity and dynamic adaptability of the quantitative evaluation of the energy efficiency of charging systems. Summary of the Invention

[0006] One of the objectives of this invention is to at least solve one or more of the aforementioned problems existing in the prior art. In other words, one of the objectives of this invention is to provide a dynamic adaptive evaluation method, system, electronic device, or medium for energy efficiency of an electric vehicle charging system that meets one or more of the aforementioned requirements.

[0007] To achieve the above-mentioned objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for dynamic adaptive evaluation of the energy efficiency of an electric vehicle charging system, comprising: Acquire multi-source operational data of the electric vehicle charging system to be evaluated; Based on multi-source operational data, several key energy efficiency assessment indicators were identified, and their values ​​were obtained. The objective weights of each energy efficiency assessment index are calculated using the entropy weight method, and the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated is obtained by weighted calculation in different evaluation windows. Multi-source operating data and the initial comprehensive energy efficiency index are input into a pre-trained deep reinforcement learning network to dynamically adjust the initial comprehensive energy efficiency index and obtain the dynamic comprehensive energy efficiency index of the electric vehicle charging system to be evaluated in different evaluation windows. The deep reinforcement learning network dynamically adjusts the weights of various energy efficiency assessment indicators used to calculate the dynamic energy efficiency comprehensive index based on real-time environmental parameters and device status. Cluster analysis is performed on the dynamic energy efficiency comprehensive index, and the tiered performance efficiency level evaluation results of the electric vehicle charging system to be evaluated are generated based on the cluster centers and cluster boundaries.

[0008] As a preferred implementation method, several key energy efficiency assessment indicators are determined based on multi-source operational data, including: Taking the comprehensive energy efficiency of the charging system as the target layer, a criterion layer is constructed according to energy input, energy conversion, energy output, operation management and grid friendliness, and multiple candidate indicators are set under each criterion layer to form a hierarchical structure model; Multiple key energy efficiency assessment indicators were selected from a number of candidate indicators.

[0009] As a preferred embodiment, the method further includes: Based on graph neural networks, multiple factors affecting the energy loss of the charging system are treated as nodes in the graph structure, and the relationships between these factors are treated as edges in the graph structure. Local structural features are aggregated through graph convolutional network layers in graph neural networks, and attention coefficients between nodes are calculated through graph attention network layers. Based on the attention coefficient after model convergence, the implicit relationships between different influencing factors are quantified to determine key energy efficiency assessment indicators.

[0010] As a preferred implementation method, the objective weights of each energy efficiency evaluation index are calculated using the entropy weight method, and the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated is obtained by weighted calculation in different evaluation windows, specifically including: Calculate the information entropy of each energy efficiency assessment indicator to identify the data volatility of each energy efficiency assessment indicator; Assign higher objective weights to energy efficiency assessment indicators with higher information entropy; The initial comprehensive energy efficiency index is calculated based on objective weights.

[0011] As a preferred implementation method, the objective weights of each energy efficiency evaluation index are calculated using the entropy weight method, and the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated is obtained by weighted calculation in different evaluation windows. This also includes: The various energy efficiency assessment indicators are normalized to eliminate dimensions.

[0012] As a preferred implementation, the deep reinforcement learning network adopts the Actor-Critic architecture, where the state space includes a dynamic energy efficiency comprehensive index, real-time environmental parameters, and device status; and the action space is the weight adjustment strategy for each energy efficiency evaluation index. As a further preferred implementation, the reward function of the deep reinforcement learning network is determined based on the degree of improvement in the overall energy efficiency index, system stability, and adjustment costs.

[0013] On the other hand, the present invention also provides a dynamic adaptive evaluation system for energy efficiency of an electric vehicle charging system, comprising: The terminal subsystem is used to collect multi-source operational data of the electric vehicle charging system to be evaluated. The edge subsystem is used for cleaning and storing multi-source operational data; The cloud-based subsystem includes an indicator determination module, an initial comprehensive energy efficiency index calculation module, a dynamic comprehensive energy efficiency index calculation module, and an evaluation result generation module. The indicator determination module is used to determine multiple key energy efficiency assessment indicators based on multi-source operating data and obtain indicator values. The initial energy efficiency comprehensive index calculation module is used to calculate the objective weights of each energy efficiency assessment indicator using the entropy weight method, and to calculate the initial energy efficiency comprehensive index of the electric vehicle charging system to be evaluated in different evaluation windows using weighted calculation. The dynamic energy efficiency comprehensive index calculation module is used to input multi-source operating data and initial energy efficiency comprehensive index into a pre-trained deep reinforcement learning network to dynamically adjust the initial energy efficiency comprehensive index and obtain the dynamic energy efficiency comprehensive index of the electric vehicle charging system to be evaluated in different evaluation windows. The deep reinforcement learning network dynamically adjusts the weights of various energy efficiency assessment indicators used to calculate the dynamic energy efficiency comprehensive index based on real-time environmental parameters and device status. The evaluation result generation module is used to perform cluster analysis on the dynamic energy efficiency comprehensive index. Based on the cluster centers and cluster boundaries, it generates the tiered performance efficiency level evaluation results for the electric vehicle charging system to be evaluated. On the other hand, the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the dynamic adaptive evaluation method for energy efficiency of an electric vehicle charging system as described above.

[0014] On the other hand, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the dynamic adaptive evaluation method for energy efficiency of an electric vehicle charging system as described above.

[0015] Compared with existing technologies, the electric vehicle charging system energy efficiency dynamic adaptive evaluation method, system, electronic device, and medium provided by this invention have the following beneficial effects: This invention utilizes the entropy weight method to determine objective weights based on the data distribution characteristics of various energy efficiency assessment indicators, and calculates the initial comprehensive energy efficiency index based on these objective weights. Compared to subjective weighting methods that rely on human experience or expert scoring, this method can correlate indicator weights with the volatility, variability, and discriminative power of actual operating data, reducing the impact of human factors on the evaluation results and improving the objectivity of the initial comprehensive energy efficiency index.

[0016] Meanwhile, this invention inputs multi-source operating data and the initial comprehensive energy efficiency index into a pre-trained deep reinforcement learning network to dynamically adjust the weights of the indicators used to calculate the dynamic comprehensive energy efficiency index in combination with real-time environmental parameters and equipment status. This enables a comprehensive balance between the degree of energy efficiency improvement, system stability, and adjustment costs, allowing the comprehensive energy efficiency index to be dynamically updated as the charging system's operating status changes, thereby improving the evaluation results' adaptability to environmental changes, equipment status changes, and load fluctuations. Attached Figure Description

[0017] Figure 1 This is a flowchart of the dynamic adaptive evaluation method for energy efficiency of electric vehicle charging systems according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the hierarchical structure model according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the GNN network according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the architecture of a deep reinforcement learning network according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of the electric vehicle charging system energy efficiency dynamic adaptive evaluation system according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0019] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the described elements without departing from the scope of the invention. Various processes or components may be appropriately omitted, substituted, or added to the various examples. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Furthermore, features described with respect to some examples may be combined into other examples.

[0020] Reference Figure 1 This invention provides a dynamic adaptive evaluation method for the energy efficiency of an electric vehicle charging system, which mainly includes the following steps S100-S500.

[0021] S100: Acquire multi-source operational data of the electric vehicle charging system to be evaluated.

[0022] Specifically, step S100 acquires full-dimensional data of the charging system operation in real time through a multi-source sensor network. The multi-source operation data includes electrical parameters, environmental parameters, and operating status data.

[0023] As a feasible example, electrical parameters include data such as RMS voltage, RMS current, active power, and harmonic content; environmental parameters include data related to the charging system's operating environment, such as temperature and humidity; and operating status data includes data such as charging start / stop time, equipment operating status, and fault codes.

[0024] In some implementations, the evaluation window can be set to a period of 7 days, 1 month, or half a year, depending on the needs of energy efficiency management or assessment, and the data can be segmented according to peak, flat, and valley electricity consumption periods to ensure the representativeness and comprehensiveness of the data.

[0025] In some implementations, the multi-source operational data is also cleaned and filtered to eliminate transient interference errors, providing a high-quality data foundation for subsequent analysis.

[0026] S200 determines several key energy efficiency assessment indicators based on multi-source operating data and obtains the values ​​of each indicator.

[0027] In some implementations, step S200 determines key energy efficiency assessment indicators by constructing a hierarchical structure model.

[0028] Specifically, the hierarchical structure model is as follows: Figure 2 As shown, it includes the target layer, the criteria layer, and the indicator layer.

[0029] The target layer, located at the top of the model, clearly states the ultimate purpose of the evaluation, namely, the comprehensive energy efficiency evaluation of the charging system.

[0030] The criteria layer acts as a bridge connecting the overall goal and specific indicators, breaking down the overall goal into several independent yet interconnected core dimensions. As a feasible example, the criteria layer includes power supply reliability, charging equipment efficiency, power quality, operational status, and ancillary service capabilities. The indicator layer is the most specific layer in the model. It refines each criterion layer dimension into multiple quantifiable and measurable specific indicators, which serve as key energy efficiency assessment indicators.

[0031] Figure 2 The intermediate indicator layer includes fifteen quantifiable candidate indicators: voltage qualification rate, system availability, fault repair time, overall system efficiency, loss rate classification, standby power consumption, harmonic distortion rate, power factor, voltage fluctuation rate, load factor, peak-valley electricity price ratio, revenue per unit of electricity, V2G response time, frequency regulation capacity ratio, and renewable energy consumption rate. These indicators can be obtained directly from the monitoring system or calculated based on multi-source operational data.

[0032] To further screen key energy efficiency evaluation indicators from candidate indicators, in some embodiments, step S200 also includes a key influencing factor screening mechanism based on graph neural networks, so as to realize the automatic mining and dynamic optimization of complex relationships between different energy loss factors.

[0033] Specifically, define the graph structure: ; in, V This represents a set of nodes, where each node represents an influencing factor (such as overall system efficiency, temperature, harmonic distortion rate, etc.), and the node feature vector is... h v ; E A boundary set, representing the relationship between factors (such as causal relationship, statistical association); A It is an adjacency matrix. A ij Represents a node i and j The strength of the connection between them.

[0034] This GNN network consists of an input layer, two Graph Conventional Network (GCN) layers, one Graph Attention Network (GAT) layer, and an output layer, as shown in the structure below. Figure 3 As shown.

[0035] The GCN layer is used to aggregate the neighbor information of each node in the graph in order to capture local structural features.

[0036] The GCN layer is based on spectral graph convolution theory and achieves information propagation through Laplacian matrix normalization. The convolution calculation formula is as follows: .

[0037] In the formula It adds a self-connected adjacency matrix to ensure that the node's own information is not ignored; yes The degree matrix is ​​used for normalization; H (l) It is the first l The node feature matrix of the layer; W (l) It is a trainable weight matrix.

[0038] The GAT layer introduces an attention mechanism on top of GCN, dynamically learning the relationship weights between nodes to highlight the role of key factors. The GAT layer achieves selective information aggregation by calculating attention coefficients between nodes. Its core formulas include attention coefficient calculation and feature updating.

[0039] The formula for calculating the attention coefficient is: ; The feature update formula is: .

[0040] in a It is an attention vector. W It is a weight matrix. h i It is a node feature. This represents vector concatenation. N ( i ) is a node i The GAT layer allows each node to assign different weights to its neighbors, thereby focusing on important connections.

[0041] By inputting the key energy efficiency evaluation indicators selected by the hierarchical model into the aforementioned GNN network, and continuously training through backpropagation to obtain the convergence value of the attention coefficient, the implicit relationships between various factors in the factor library can be quantitatively expressed. Based on the output of the graph network, the key energy efficiency evaluation indicators are further screened, removing indicators with minimal impact on energy efficiency and those heavily dependent on other factors. This further overcomes the static limitations of traditional factor libraries and significantly improves the comprehensiveness and accuracy of key influencing factor discovery.

[0042] S300 uses the entropy weight method to calculate the objective weights of the key energy efficiency evaluation indicators obtained from S200, and then calculates the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated in different evaluation windows.

[0043] Since different energy efficiency assessment indicators have different dimensions and different numerical ranges, the indicators are normalized before calculating the objective weights.

[0044] The normalization calculation formula is: .

[0045] Normalization is used to convert indicators with different dimensions to comparable numerical scales, so that indicators such as overall efficiency, harmonic distortion rate, load rate, and V2G response time can participate in unified weight calculation and comprehensive evaluation.

[0046] After normalization, an indicator matrix is ​​constructed based on n indicator data from m charging stations. X Next, the information entropy of each indicator is calculated. e j : For the j For each indicator, you can first calculate the proportion. : .

[0047] in, This represents the sum of the normalized values ​​of the j-th indicator across all evaluation objects or evaluation windows.

[0048] Then, the information entropy of the j-th indicator is calculated. : .

[0049] Information entropy Used to characterize the j The uncertainty of the distribution of indicators in different evaluation objects or evaluation windows is used to identify the data volatility of various energy efficiency assessment indicators.

[0050] Objective weight According to information entropy Sure: .

[0051] The greater the data volatility, the stronger the distinguishing ability of the indicator in the evaluation sample and the higher its contribution to the comprehensive evaluation. Step S300 uses the entropy weight method to effectively identify indicators with high data volatility and assign them higher weights.

[0052] After calculating the objective weights of each indicator, a weighted average is performed based on the objective weights and normalized indicator values ​​to obtain the initial Energy Efficiency Index (EEI), which serves as the core parameter for determining energy efficiency levels and achieving quantitative energy efficiency assessment. .

[0053] inw j For the first j The weight of each indicator x j For the first j The normalized value of the indicator.

[0054] S400 inputs multi-source operating data and the initial comprehensive energy efficiency index into a pre-trained deep reinforcement learning network to dynamically adjust the initial comprehensive energy efficiency index and obtain the dynamic comprehensive energy efficiency index (D-EEI) of the electric vehicle charging system to be evaluated in different evaluation windows.

[0055] For the same charging system, its energy efficiency level will be affected by real-time environmental parameters, equipment status, load fluctuations and changes in operating status. In order to make up for the shortcomings of the initial comprehensive energy efficiency index, which is only based on static objective weights and linear weighting and cannot adapt to changes in environment or equipment status, step S400 inputs the multi-source operating data in the current evaluation window, the initial comprehensive energy efficiency index and features related to environment and equipment status into the deep reinforcement learning network, so that the network outputs a dynamic adjustment result that matches the current operating status.

[0056] In some implementations, deep reinforcement learning networks employ an Actor-Critic architecture. Figure 4 The diagram shown is a schematic of its architecture.

[0057] In the Actor-Critic architecture, the state space includes the dynamic energy efficiency comprehensive index, real-time environmental parameters, and equipment status; the action space is the weight adjustment strategy for each energy efficiency evaluation indicator.

[0058] The reward function is designed as follows: .

[0059] in α , β , γ Δ is the weighting coefficient. EEI To the degree of improvement in energy efficiency index, S As a system stability indicator, C To optimize costs.

[0060] During training, the training samples consist of key energy efficiency assessment indicators within the evaluation window, an initial comprehensive energy efficiency index, environmental parameters, and equipment status. The deep reinforcement learning network adjusts the indicator weights used to calculate the initial comprehensive energy efficiency index according to a weight adjustment strategy, and calculates the corresponding dynamic comprehensive energy efficiency index. Subsequently, reward feedback is calculated based on the degree of improvement in the comprehensive energy efficiency index, system stability, and adjustment costs. The weight adjustment strategy is evaluated based on the reward feedback. The network parameters are updated in the above cycle of status, action, reward, and evaluation until the preset training termination condition is met.

[0061] As an example of a reward function feedback evaluation weight adjustment strategy, it is assumed that the model prioritizes energy efficiency. α , β , γ Let's take values ​​of 0.6, 0.2, and 0.2 respectively. Assume the action... a j (Initial weight 0.1) Corresponding indicator x j Increased x j = 0.5, this action leads to a decrease in stability ( S = -0.05), no optimization cost ( C = 0), calculated to R ( s , a j = 0.02. Similarly, the reward function value can be calculated when other indicators change, and the weights can be adjusted according to the reward function value corresponding to different indicator changes. w j The energy is then redistributed to obtain the final dynamic comprehensive energy efficiency index.

[0062] During the inference phase, the pre-trained deep reinforcement learning network receives the state input of the current evaluation window, outputs the weight adjustment strategy corresponding to the current evaluation window, and then generates the corrected index weights based on the strategy, and generates a dynamic comprehensive energy efficiency index.

[0063] Through the processing in step S400, the generated dynamic energy efficiency comprehensive index not only reflects the objective weighting results contained in the initial energy efficiency comprehensive index, but also reflects the real-time environmental parameters and equipment status changes in the current evaluation window, so that the energy efficiency evaluation results can be dynamically updated with the operating status of the charging system.

[0064] S500 performs cluster analysis on the dynamic energy efficiency comprehensive index and generates a tiered performance efficiency level assessment result for the electric vehicle charging system to be evaluated based on the cluster centers and cluster boundaries.

[0065] In some implementations, K-means clustering can be used for cluster analysis.

[0066] Specifically, the dynamic energy efficiency comprehensive index within multiple evaluation windows is used as the input for clustering, and the cluster centers and cluster boundaries are calculated.

[0067] The formula for calculating cluster centers is: .

[0068] in, C i Indicates the first iOne cluster, μ i express C i Cluster centers.

[0069] Based on the clustering results and the dynamic energy efficiency comprehensive index value, the system energy efficiency level can be classified. Table 1 shows a specific hierarchical classification method.

[0070]

[0071] Table 1 By using the D-EEI sample distribution to form cluster centers and cluster boundaries, the system can reduce the subjectivity caused by relying entirely on manually fixed thresholds, and enable the energy efficiency level to be adaptively divided according to the distribution of evaluation samples.

[0072] Embodiments of the present invention also provide a dynamic adaptive evaluation system for the energy efficiency of an electric vehicle charging system, the structural schematic diagram of which is shown below. Figure 5 As shown, it includes a terminal subsystem, an edge subsystem, and a cloud subsystem.

[0073] The terminal subsystem is used to collect multi-source operating data of the electric vehicle charging system to be evaluated. Specifically, it includes sensors deployed at key nodes such as charging piles and power distribution networks to non-intrusively collect electrical parameters such as voltage, current, power, and harmonics, as well as environmental parameters such as temperature and humidity.

[0074] The edge subsystem includes edge computing nodes or regional controllers, which are responsible for the initial cleaning, filtering and local storage of the collected raw data, and can perform simple logical judgments and real-time control to reduce the transmission and computing pressure on the cloud subsystem and improve the system's robustness when the network is unstable.

[0075] The cloud-based subsystem includes an indicator determination module, an initial comprehensive energy efficiency index calculation module, a dynamic comprehensive energy efficiency index calculation module, and an evaluation result generation module.

[0076] The input to the indicator determination module is the processed multi-source operational data.

[0077] The initial energy efficiency comprehensive index calculation module is used to calculate the objective weights of each energy efficiency assessment indicator using the entropy weight method, and to calculate the initial energy efficiency comprehensive index of the electric vehicle charging system to be evaluated in different evaluation windows using weighted calculation. The dynamic energy efficiency comprehensive index calculation module is used to input multi-source operating data and initial energy efficiency comprehensive index into a pre-trained deep reinforcement learning network to dynamically adjust the initial energy efficiency comprehensive index and obtain the dynamic energy efficiency comprehensive index of the electric vehicle charging system to be evaluated in different evaluation windows. The deep reinforcement learning network dynamically adjusts the weights of various energy efficiency assessment indicators used to calculate the dynamic energy efficiency comprehensive index based on real-time environmental parameters and device status. The evaluation result generation module is used to perform cluster analysis on the dynamic energy efficiency comprehensive index. Based on the cluster centers and cluster boundaries, it generates the tiered performance efficiency level evaluation results of the electric vehicle charging system to be evaluated.

[0078] This application also provides an electronic device, including: a processor, a memory, and optional communication interfaces and input / output interfaces.

[0079] The processor can be a central processing unit (CPU), a digital signal processor (DSP), a programmable logic device (FPGA), an application-specific integrated circuit (ASIC), or other processing cores capable of executing instructions. In this embodiment, the processor is the control center of the electronic device, responsible for running the computer program stored in the memory.

[0080] Memory can be any type of volatile or non-volatile storage medium, such as random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive (HDD), solid-state drive (SSD), etc. Memory is used to store operating systems, various data, and computer programs.

[0081] When the instructions of a computer program stored in memory are executed by a processor, the electronic device performs the energy efficiency dynamic adaptive evaluation method for the electric vehicle charging system described in the above embodiments.

[0082] This application also provides a computer-readable storage medium. A computer program is stored on this computer-readable storage medium.

[0083] The computer program contains a series of instructions that, when loaded and executed by the processor of an electronic device, enable the electronic device to implement the dynamic adaptive energy efficiency evaluation method for the electric vehicle charging system described in the above embodiments.

[0084] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0085] The above description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of embodiments of this disclosure upon considering the specification and practicing the disclosure herein. This invention is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A dynamic adaptive evaluation method for the energy efficiency of an electric vehicle charging system, characterized in that, include: Acquire multi-source operational data of the electric vehicle charging system to be evaluated; Based on the multi-source operation data, several key energy efficiency assessment indicators were determined, and the indicator values ​​were obtained. The objective weights of each energy efficiency evaluation index are calculated using the entropy weight method, and the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated is obtained by weighted calculation in different evaluation windows. The multi-source operating data and the initial comprehensive energy efficiency index are input into a pre-trained deep reinforcement learning network to dynamically adjust the initial comprehensive energy efficiency index and obtain the dynamic comprehensive energy efficiency index of the electric vehicle charging system to be evaluated in different evaluation windows. The deep reinforcement learning network dynamically adjusts the weights of each energy efficiency evaluation indicator used to calculate the dynamic energy efficiency comprehensive index based on real-time environmental parameters and device status. Cluster analysis is performed on the dynamic energy efficiency comprehensive index, and the tiered performance efficiency level evaluation results of the electric vehicle charging system to be evaluated are generated based on the cluster centers and cluster boundaries.

2. The method for dynamic adaptive evaluation of energy efficiency of electric vehicle charging systems according to claim 1, characterized in that, Based on the aforementioned multi-source operational data, several key energy efficiency assessment indicators were identified, including: Taking the comprehensive energy efficiency of the charging system as the target layer, a criterion layer is constructed according to energy input, energy conversion, energy output, operation management and grid friendliness, and multiple candidate indicators are set under each criterion layer to form a hierarchical structure model; Multiple key energy efficiency assessment indicators were obtained by screening from the multiple candidate indicators.

3. The method for dynamic adaptive evaluation of energy efficiency of electric vehicle charging systems according to claim 1, characterized in that, The method further includes: Based on graph neural networks, multiple factors affecting the energy loss of the charging system are treated as nodes in the graph structure, and the relationships between these factors are treated as edges in the graph structure. Local structural features are aggregated through the graph convolutional network layer in the graph neural network, and the attention coefficients between nodes are calculated through the graph attention network layer. Based on the attention coefficient after model convergence, the implicit relationships between different influencing factors are quantified to determine the key energy efficiency evaluation indicators.

4. The method for dynamic adaptive evaluation of energy efficiency of electric vehicle charging systems according to claim 1, characterized in that, The objective weights of each energy efficiency evaluation index are calculated using the entropy weight method, and the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated is obtained by weighted calculation in different evaluation windows, specifically including: Calculate the information entropy of each of the energy efficiency assessment indicators to identify the data volatility of each of the energy efficiency assessment indicators; The higher the information entropy of the energy efficiency evaluation index, the higher its objective weight. The initial comprehensive energy efficiency index is calculated based on the objective weights.

5. The method for dynamic adaptive evaluation of energy efficiency of an electric vehicle charging system according to claim 1, characterized in that, The objective weights of each energy efficiency evaluation index are calculated using the entropy weight method, and the initial comprehensive energy efficiency index of the electric vehicle charging system to be evaluated is obtained by weighted calculation in different evaluation windows. The method also includes: The energy efficiency assessment indicators described above are normalized to eliminate dimensions.

6. The method for dynamic adaptive evaluation of energy efficiency of an electric vehicle charging system according to claim 1, characterized in that, The deep reinforcement learning network adopts an Actor-Critic architecture, wherein the state space includes the dynamic energy efficiency comprehensive index, real-time environmental parameters, and device status; and the action space is the weight adjustment strategy for each of the energy efficiency evaluation indicators.

7. The method for dynamic adaptive evaluation of energy efficiency of an electric vehicle charging system according to claim 6, characterized in that, The reward function of the deep reinforcement learning network is determined based on the degree of improvement in the comprehensive energy efficiency index, system stability, and adjustment costs.

8. A dynamic adaptive evaluation system for the energy efficiency of an electric vehicle charging system, characterized in that, include: The terminal subsystem is used to collect multi-source operational data of the electric vehicle charging system to be evaluated. An edge subsystem is used to clean and store the multi-source operational data; The cloud-based subsystem includes an indicator determination module, an initial energy efficiency comprehensive index calculation module, a dynamic energy efficiency comprehensive index calculation module, and an evaluation result generation module. The indicator determination module is used to determine multiple key energy efficiency evaluation indicators based on the multi-source operating data and obtain indicator values. The initial energy efficiency comprehensive index calculation module is used to calculate the objective weights of each energy efficiency evaluation index using the entropy weight method, and to calculate the initial energy efficiency comprehensive index of the electric vehicle charging system to be evaluated in different evaluation windows using weighted calculation. The dynamic energy efficiency comprehensive index calculation module is used to input the multi-source operating data and the initial energy efficiency comprehensive index into a pre-trained deep reinforcement learning network to dynamically adjust the initial energy efficiency comprehensive index and obtain the dynamic energy efficiency comprehensive index of the electric vehicle charging system to be evaluated in different evaluation windows.

9. The deep reinforcement learning network dynamically adjusts the weights of each energy efficiency evaluation indicator used to calculate the dynamic energy efficiency comprehensive index based on real-time environmental parameters and device status; The evaluation result generation module is used to perform cluster analysis on the dynamic energy efficiency comprehensive index, and generate the tiered performance efficiency level evaluation results of the electric vehicle charging system to be evaluated based on the cluster centers and cluster boundaries.

10. An electronic device, characterized in that, The system includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the dynamic adaptive evaluation method for energy efficiency of an electric vehicle charging system as described in any one of claims 1 to 7.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the dynamic adaptive evaluation method for energy efficiency of an electric vehicle charging system as described in any one of claims 1 to 7.

Images