Energy efficiency monitoring optimization control method and system for electric vehicle charging system

Abstract

An electric vehicle charging system energy efficiency monitoring optimization control method comprises the steps that cloud equipment predicts a user portrait of an electric vehicle owner based on a CNN-LSTM hybrid model, the user portrait is used for indicating charging behaviors of a user, and the charging behaviors comprise the battery capacity, the arrival time, the departure time, the arrival electric quantity and the required electric quantity of an electric vehicle; based on the user portrait, an ordered charging strategy is solved through a rolling optimization algorithm, the charging strategy is issued to the edge device, and the charging strategy is used for indicating the charging terminal to adjust the charging power; the edge device fuses multi-source data, predicts a temperature trend through an LSTM model, and executes predictive adaptive control on a charging terminal cooling system; and online energy efficiency post-evaluation is carried out, and an evaluation report is generated and fed back to an optimization link. According to the method, an ordered charging strategy can be generated, and predictive self-adaptive adjustment is carried out on the temperature of the terminal.

Description

Technical Field

[0001] This invention relates to the field of electrical energy efficiency analysis technology, and in particular to a method and system for energy efficiency monitoring and optimization control of electric vehicle charging systems. Background Technology

[0002] In recent years, with the continuous increase in the number of electric vehicles, the number of charging infrastructures has grown rapidly and is developing towards higher power and intelligence. Energy efficiency has become a key factor affecting operational efficiency, attracting numerous researchers to study various aspects of the charging process. The comprehensive energy efficiency of charging stations is related to multiple indicators, including long-term operational indicators such as daily energy efficiency and short-term transient indicators such as energy conversion efficiency. Optimizing charging strategies can effectively improve the operational energy efficiency of charging stations, serving more car owners; employing intelligent control algorithms can ensure operational safety by rationally controlling the internal temperature parameters of the devices. However, methods for monitoring and optimizing the energy efficiency control of charging systems are still lacking. Summary of the Invention

[0003] To address the problems existing in the prior art, embodiments of this application provide a method, system, computing device, computer storage medium, and product containing a computer program for energy efficiency monitoring and optimization control of an electric vehicle charging system, which can generate an ordered charging strategy and make predictive adaptive adjustments to the terminal temperature.

[0004] In a first aspect, embodiments of this application provide a method for energy efficiency monitoring and optimization control of an electric vehicle charging system, comprising: a cloud device predicting a user profile of an electric vehicle owner based on a CNN-LSTM hybrid model, the user profile indicating the user's charging behavior, including the electric vehicle's battery capacity, arrival time, departure time, arrival charge, and demand charge; based on the user profile, solving an ordered charging strategy using a rolling optimization algorithm, and distributing the charging strategy to an edge device, the charging strategy instructing the charging terminal to adjust the charging power; the edge device fusing multi-source data, predicting temperature trends using an LSTM model, and performing predictive adaptive control on the charging terminal cooling system; performing an online energy efficiency post-evaluation, generating an evaluation report, and feeding it back to the optimization stage.

[0005] In some possible implementations, cloud devices predict user profiles of electric vehicle owners based on a CNN-LSTM hybrid model, including: using convolutional neural networks to extract local spatial features from the input multi-dimensional feature data; inputting the local spatial features into a long short-term memory network to capture the long-term temporal dependencies of user charging behavior; and introducing an attention mechanism at the output of the long short-term memory network to dynamically calculate the importance weights of different historical time steps for the current prediction.

[0006] In some possible implementations, based on user profiles, an ordered charging strategy is solved using a rolling optimization algorithm. This includes: modeling the charging pile power allocation problem as a mixed integer linear programming problem; and using a rolling optimization method to dynamically solve the problem based on predicted user profile data and actual operating data, so as to maximize the energy efficiency of the charging system and balance grid load and user satisfaction.

[0007] In some possible implementations, an LSTM model is used to predict temperature trends and perform predictive adaptive control on the charging terminal cooling system. This includes: collecting historical and real-time operating time-series data of the charging terminal, including AC / DC converter temperature, charging gun head temperature, charging current, voltage, ambient temperature, and humidity; preprocessing the collected data, including data cleaning, missing value handling, outlier correction, and data normalization; training an LSTM model using the processed data to learn the mapping relationship between operating parameters and future temperature changes; deploying the trained LSTM model on edge devices to predict short-term temperature trends based on real-time sensor data; and linearly adjusting the cooling fan speed or liquid cooling pump power based on the predicted temperature trend to achieve predictive temperature control.

[0008] In some possible implementations, online post-energy efficiency assessments are conducted, generating assessment reports and feeding them back to the optimization process. This includes: edge devices periodically calculating short-term energy efficiency indicators, including instantaneous charging efficiency, power consumption per unit time, power consumption of the temperature control system, and standby power consumption, and performing local rapid adjustments; cloud devices aggregating data uploaded by edge devices and calculating long-term energy efficiency indicators, including peak-valley difference rate and renewable energy absorption rate; combining short-term and long-term energy efficiency indicators, conducting multi-dimensional energy efficiency analysis, generating post-assessment reports, and feeding the assessment results back to the user profile prediction model and optimization scheduling process for model correction and strategy adjustment.

[0009] Secondly, embodiments of this application provide an energy efficiency monitoring and optimization control system for an electric vehicle charging system, comprising: a charging terminal for providing charging services to electric vehicles and collecting operational data; an edge device for communicating with the charging terminal and for receiving data from the charging terminal, performing short-term energy efficiency index calculation, temperature trend prediction, and predictive adaptive control; and a cloud device for communicating with the edge device and for predicting user profiles, solving ordered charging strategies, conducting long-term energy efficiency assessments, and generating assessment reports.

[0010] In some possible implementations, the charging terminal includes: a data acquisition module for real-time acquisition of current, voltage, AC / DC converter temperature, and charging gun head temperature data via built-in sensors; and an adjustment module for receiving and executing power allocation commands and cooling control commands from edge devices.

[0011] In some possible implementations, the edge device includes: an adaptive control module for running a lightweight LSTM model to predict temperature trends and sending control commands to the cooling system of the charging terminal; and a short-term target prediction module for calculating short-term energy efficiency indicators and uploading the calculation results to the cloud device.

[0012] In some possible implementations, the cloud device includes: a user profile prediction module for running a CNN-LSTM hybrid model to predict user charging behavior parameters; an optimization scheduling module for running a rolling optimization algorithm based on the prediction results to generate a globally optimal ordered charging strategy; and an evaluation module for aggregating data uploaded by edge devices, performing long-term energy efficiency index calculations and multi-dimensional comprehensive energy efficiency analysis, and generating a post-evaluation report.

[0013] Thirdly, embodiments of this application provide a computer-readable storage medium including computer-readable instructions that, when read and executed by a computer, cause the computer to perform the method as described in any of the first aspects.

[0014] Fourthly, embodiments of this application provide a computing device, including a processor and a memory, wherein the memory stores computer program instructions, which, when executed by the processor, perform the method as described in any of the first aspects.

[0015] Fifthly, embodiments of this application provide a product comprising a computer program that, when the computer program product is run on a processor, causes the processor to perform the method as described in any of the first aspects. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of an energy efficiency monitoring and optimization control system for an electric vehicle charging system provided in an embodiment of this application; Figure 2 This is a flowchart illustrating an energy efficiency monitoring and optimization control method for an electric vehicle charging system provided in an embodiment of this application. Figure 3 This is a schematic diagram of an LSTM network structure provided in an embodiment of this application; Figure 4 This is a schematic diagram of a user profile prediction process based on CNN-LSTM provided in an embodiment of this application; Figure 5 This is a schematic diagram of an adaptive charging pile cooling control method provided in an embodiment of this application; Figure 6 This is a schematic diagram of an online post-evaluation method for energy efficiency provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In this article, the term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The symbol " / " in this article indicates that the related objects are in an "or" relationship; for example, A / B means A or B.

[0020] The terms "first" and "second," etc., used in the specification and claims herein are used to distinguish different objects, not to describe a specific order of objects. For example, "first response message" and "second response message," etc., are used to distinguish different response messages, not to describe a specific order of response messages.

[0021] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0022] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.

[0023] To facilitate understanding of the embodiments of this application, the following will provide further explanation and description with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0024] To effectively monitor and optimize the energy efficiency of electric vehicle charging systems, this application provides a method and system for monitoring, optimizing, and controlling the energy efficiency of electric vehicle charging systems. The system comprises multiple charging terminals, multiple edge computing devices, and a cloud device. The charging terminals provide charging services to electric vehicles, while the edge devices acquire data collected by the charging terminals, upload the data to the cloud device, and receive the charging strategy optimization results from the cloud device. The charging terminals perform power scheduling based on the received charging strategy scheduling optimization results. For the overall control method, a neural network is used to predict the user profiles of vehicle owners at charging stations, and the results are input to the cloud side for optimization through rolling scheduling to obtain an ordered charging strategy. During operation, the edge side integrates multi-source data to predict future temperature change trends and performs predictive adaptive temperature control.

[0025] For example, Figure 1 A schematic diagram of an energy efficiency monitoring and optimization control system for an electric vehicle charging system, provided in an embodiment of this application, is shown. Figure 1 As shown, the energy efficiency monitoring and optimization control system 100 for electric vehicle charging systems includes a charging terminal 101, an edge device 102, and a cloud device 103.

[0026] The charging terminal 101, i.e., each charging pile, is responsible for data acquisition and execution of the lowest-level commands. Each charging pile is equipped with a data acquisition module 1011, which uses built-in sensors to collect raw data in real time, such as current, voltage, AC / DC converter temperature, and charging gun head temperature. An adjustment module 1012 is also deployed, which executes power allocation and cooling control commands from the edge gateway based on the adjustment commands received from the edge device 102.

[0027] Edge device 102 is deployed locally at the charging station's edge gateway or edge server, possessing certain computing, storage, and control capabilities, and is responsible for handling local tasks with high real-time requirements. As a local computing and control hub, edge device 102 is logically located between charging terminal 101 and cloud device 103, enabling data aggregation, real-time prediction, and rapid control. Edge device 102 includes an adaptive control module 1021 and a short-term target prediction module 1022. The adaptive control module 1021 enables rapid response, ensuring device safety. It runs a lightweight LSTM prediction model, predicts temperature trends based on real-time data uploaded from the charging pile, executes predictive control algorithms, and directly sends control commands to the cooling system (such as fans and liquid pumps) of charging terminal 101 to achieve precise temperature management. The short-term target prediction module 1022 performs real-time calculations of short-term indicators, such as instantaneous charging efficiency, power consumption per unit time, temperature control system power consumption, and standby power consumption. Edge device 102 sends control commands to charging terminal 101 based on the predicted short-term targets. The short-term target prediction module 1022 can also upload the calculation results to the cloud device 103.

[0028] The cloud device 103, also known as the cloud server, can be a high-performance server cluster located in a remote data center, possessing high computing and storage capabilities to handle large-scale data analysis and complex optimization calculations. Within the cloud device 103, a user profile prediction module 1031, an optimization scheduling module 1032, and an evaluation module 1033 are deployed. The user profile prediction module 1031 runs a prediction model on the cloud device, analyzes historical and real-time data from all charging stations, and generates predictions of users' future charging needs. The optimization scheduling module 1032 runs a rolling optimization algorithm on the cloud based on the prediction results to calculate the globally optimal ordered charging strategy. The evaluation module 1033 aggregates data uploaded from all edge devices, calculates long-term indicators (such as peak-valley difference rate and renewable energy absorption rate), performs multi-dimensional comprehensive energy efficiency analysis, and generates a final post-evaluation report. When generating the post-evaluation report, the calculation results uploaded by the edge device 102 can also be incorporated.

[0029] The above is an introduction to the energy efficiency monitoring and optimization control system for the electric vehicle charging system. The modules included in the hardware equipment are only a simple introduction and are not fully included. For example, the edge device 102 may also include modules for processing the collected data.

[0030] Based on the energy efficiency monitoring and optimization control system for electric vehicle charging systems, this application provides a detailed description of the energy efficiency monitoring and optimization control method for electric vehicle charging systems.

[0031] For example, Figure 2This illustration shows a flowchart of an energy efficiency monitoring and optimization control method for an electric vehicle charging system provided in an embodiment of this application. Figure 2 As shown, the control method may include the following steps:

[0032] S21: Cloud devices predict user profiles of electric vehicle owners based on a CNN-LSTM hybrid model. The user profiles are used to indicate the user's charging behavior, including key parameters such as the electric vehicle's battery capacity, arrival time, departure time, arrival charge, and charge demand.

[0033] In this embodiment, the cloud device utilizes a combination of Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM) to predict the charging behavior of electric vehicles. The CNN is responsible for extracting spatial features (such as vehicle characteristics and arrival time) from the input data, while the LSTM processes temporal data, capturing the long-term dependencies of charging behavior. This hybrid model can improve the prediction accuracy of complex temporal behavior patterns, thereby providing reliable data support for charging strategy optimization.

[0034] Specifically, cloud devices receive data transmitted from terminal devices via edge devices. When transmitting data, edge devices handle missing and outlier values, performing standardization and normalization. Since electric vehicle charging behavior is influenced by various factors, the data transmitted from terminal devices can include historical travel habits (e.g., arrival and departure times for each charging session), vehicle characteristics (e.g., battery capacity), and relevant external environmental parameters (e.g., ambient temperature and humidity). This data constitutes a high-dimensional feature vector. Based on this high-dimensional feature vector, cloud devices predict user profiles, including key behavioral parameters such as battery capacity, departure time, initial charge level (i.e., charge level before charging), and required charge. Convolutional Neural Networks (CNNs) efficiently extract local spatial features from this multi-dimensional input data. CNNs, through their convolutional kernels, perform sliding window-style local connections and weight-sharing operations on the data, automatically and effectively identifying local correlation patterns and the importance of combinations among these features. The core operation of a convolutional neural network is convolution. For a time-series input that influences charging behavior features, the feature calculation of the k-th convolutional kernel at position m can be expressed as:

[0035]

[0036] In the formula, Characterize the activation function, Characterizing the offset, This represents the value of the input sequence at position m+s-1. This represents the weight of the k-th convolutional kernel at offset s. This represents the corresponding bias term. Through convolution, effective local patterns can be learned from the high-dimensional feature vector of the original input.

[0037] The data output from the convolutional neural network (CNN) enters the Long Short-Term Memory (LSTM) network. The LSTM network is specifically designed to process the feature sequences refined by the CNN, capturing long-term temporal dependencies in charging behavior. Since user charging behavior is not an isolated event but rather exhibits continuous temporal patterns—such as differences in charging patterns between weekdays and weekends, or seasonal variations in charging—the LSTM network, through its internal forget gate, input gate, and output gate structure, can selectively remember or forget historical information, thereby effectively learning feature information from temporal sequences. This allows the model to understand and predict the dynamic changes in behavior over time, rather than relying solely on judgments based on instantaneous states.

[0038] For example, Figure 3 A schematic diagram of an LSTM network structure provided in an embodiment of this application is shown. Figure 3 As shown, the computation process of an LSTM unit within a single time step (from t-1 to t) involves receiving information from the previous time step and the current input, passing through a series of gating mechanisms, and finally generating a new state and outputting the result. Its core objective is to selectively update and pass information, thereby solving the long-term memory problem. Specifically, the gating mechanisms include a forget gate, an input gate, and an output gate. Each gate uses a sigmoid function to generate a value between 0 and 1 to control the amount of information passing through, where 0 represents complete blocking and 1 represents complete passage. The forget gate receives the output from the previous time step, i.e., And the current input (not shown in the diagram), as well as the long-term memory state. The forget gate is used to decide which information to discard from long-term memory. For example, when the model learns that a user is starting a new charging cycle, it might choose to forget the exact charging time from last week, but remember the user's preference for charging at night. The input gate is used to decide which new information to store in long-term memory. It works in conjunction with a tanh layer to update the memory. The tanh layer is used to generate candidate values. The output gate outputs the next hidden state based on the updated long-term memory.

[0039] In some possible implementations, in order to further enhance the model's ability to focus on key time points, an attention mechanism can be introduced at the output of the LSTM.

[0040] In this embodiment, the LSTM hidden state h at each time step t Calculate weight α t This forms the context vector c, which is used for the final prediction. The calculation formula is as follows:

[0041]

[0042] In the formula, Characterizing the hidden state at time step t Importance score Representing trainable weight vectors, The weight matrix represents the trainable weight matrix. Characterizes the trainable bias vector. The normalized attention weights represent the values ​​at time step t. Characterizing sequence length, This represents the sum of scores over all time steps. The context vector obtained from the characterization calculation.

[0043] Attention mechanisms can dynamically calculate the importance weights of different historical time steps to the current prediction, enabling the model to focus more on key events (such as high charging demand in the same period last week) when making decisions, thereby effectively improving the model's ability to focus on important time segments and enhancing the interpretability of predictions.

[0044] For example, Figure 4 A complete flowchart of a CNN-LSTM-based method for predicting electric vehicle owner profiles is shown. Figure 4 As shown, based on the multi-dimensional features of a user's historical electric vehicle charging data, a CNN model is used to obtain local charging patterns, and an LSTM model is used to obtain long-term dependencies in the time-series information. Then, an attention mechanism is used to focus on important information in the user's historical charging behavior, which is ultimately mapped to the final prediction target.

[0045] S22: Based on user profiles, an ordered charging strategy is solved using a rolling optimization algorithm and then distributed to the edge side. The charging strategy is used to instruct the charging terminal to adjust the charging power.

[0046] In this embodiment, a cloud-based intelligent optimization algorithm is used to dynamically adjust the power of charging piles to maximize the energy efficiency of charging terminals. Since the power of charging piles is limited, allocating power among the needs of different electric vehicles becomes a key issue. The optimization objective is to maximize energy efficiency and balance grid load and owner satisfaction through charging pile power scheduling. Based on the predicted conditions of the charging stations, the entire problem can be modeled as a mixed integer linear programming (MILP) problem, which can be solved using a solver. However, in actual operation, the predicted traffic flow data may differ from the actual data. Therefore, the problem needs to be solved on a rolling basis, using the latest available data to recalculate at each time step to match the real-world scenario. Rolling optimization can be continuously performed over time, thus gradually solving for the energy efficiency benefit of the current period. The difference between theoretical and actual energy efficiency reflects the impact of randomness on the user prediction model, which can be used for feedback in the post-evaluation stage to correct the electric vehicle owner user profile prediction model. The predicted charging strategy is then sent to edge devices.

[0047] S23: The edge side fuses multi-source data, predicts temperature trends through an LSTM model, and performs predictive adaptive control on the charging terminal cooling system.

[0048] In this embodiment, traditional control measures are taken after the temperature exceeds the limit, while this invention intervenes in advance based on predicted trends. The edge gateway combines real-time data to predict the temperature change trend of the charging pile and controls the power of the cooling equipment. The internal temperature of the charging pile has a significant impact on energy efficiency; excessively high temperatures lead to a decrease in energy efficiency. By controlling the temperature of the charging pile through the edge gateway, multi-modal data (such as current, voltage, temperature, and humidity) is used to predict temperature change trends, and the power consumption of the cooling system is adaptively adjusted according to the prediction results. This adaptive control based on real-time data can effectively improve overall energy efficiency and reduce unnecessary power consumption.

[0049] Specifically, the internal temperature of a charging pile is a crucial factor affecting energy efficiency. By optimizing the control strategy of the charging pile's cooling system (such as controlling the cooling fan speed or liquid cooling pump power), the power consumption of the cooling system can be reduced and the overall system energy efficiency improved while ensuring that the temperature rise does not exceed the limit. Edge gateways possess excellent control capabilities, enabling closed-loop control of the charging pile's internal temperature. Excessive charging pile temperature can lead to reduced AC / DC converter efficiency, increased harmonics, and a lower power factor, resulting in safety and energy efficiency issues. Therefore, multiple sensors within the charging pile can be used to measure multi-dimensional data such as current, voltage, temperature, and humidity to predict temperature rise trends. Then, closed-loop control can be used to measure and control the power of the cooling equipment. When the temperature is too high, the power of the air-cooled or liquid-cooled equipment can be linearly increased; when the temperature is low or within a safe range, it can be linearly controlled to enter a low-power standby state, reducing unnecessary standby power consumption and ultimately improving overall energy efficiency.

[0050] For example, Figure 5 A schematic flowchart of an adaptive charging pile cooling control method according to an embodiment of this application is shown. Please refer to... Figure 5 To train an LSTM model that can accurately predict the temperature trend of charging piles, training data is first required. The training data consists of time-series data of the charging terminal's historical operation. This includes multi-point temperatures of the AC / DC converter and charging gun head, which directly reflect the thermal state of the equipment, as well as operating parameters that affect thermal changes, such as charging current and voltage, and external environmental parameters such as ambient temperature and humidity. All data is accompanied by timestamps.

[0051] After obtaining the raw data, it is preprocessed. Since the raw data may contain missing or outlier values ​​due to occasional sensor malfunctions or communication interference, it is cleaned and corrected to ensure data integrity and reliability. Next, because features such as current, voltage, and temperature have different dimensions and numerical ranges, normalization is used to scale all features to a uniform scale, accelerating model convergence during training and improving the accuracy of the final prediction.

[0052] After data preprocessing, the next step is to construct samples that the model can recognize. Given the correlation between time series data, a suitable time window is selected, and a sliding window approach is used to segment the continuous time series into multiple overlapping short sequences. Each sequence contains historical observations from the past period, with the corresponding label being the temperature value at one or more future time points. In this way, the raw data is converted into a format that the LSTM network can process.

[0053] Next, an LSTM model structure was built using the Keras neural network framework. The model learns and analyzes samples constructed from a sliding window to understand the mapping relationship between various operating parameters and future temperature changes. This completes the training of the neural network. The trained model can then predict temperature change trends over future time periods based on real-time sensor data, providing decision-making strategies for subsequent temperature adaptive control.

[0054] Once the model is trained, it can be integrated into the control logic of the edge gateway to achieve predictive adaptive control. The trained LSTM model can be deployed on the edge gateway. The gateway continuously receives real-time data streams from sensors and inputs them into the model to predict short-term (e.g., the next few minutes) temperature trends. If the predicted temperature will rise rapidly and approach a critical point, the system can linearly increase the cooling fan speed or liquid cooling pump power in advance to achieve smooth cooling; if the predicted temperature will stabilize within a safe range, the cooling system will operate at low power to save energy.

[0055] S24: Conduct online post-energy efficiency assessment, generate an assessment report, and feed it back to the optimization process.

[0056] In this embodiment, after the system has been running for a certain period of time, the optimization effect can be verified through post-evaluation. Some traditional post-engineering evaluation methods are usually conducted offline, which lacks real-time capability and cannot adapt to the characteristics of varying charging station loads and differentiated demands. This embodiment can implement online energy efficiency post-evaluation through the collaborative work of the cloud, edge, and end-side. The edge side is mainly responsible for calculating real-time energy efficiency indicators and performing local adjustments, while the cloud performs macro-analysis and long-term data storage. Combining the advantages of both, energy efficiency evaluation can more accurately identify energy-inefficient links and provide feedback for subsequent optimization. By generating a post-evaluation report, operators can analyze various aspects affecting energy efficiency and make adjustments.

[0057] Specifically, Figure 6 A schematic diagram of the online post-energy efficiency assessment method is shown. Figure 6As shown, after the system has been running for a period of time, the energy efficiency optimization effect under the optimized control strategy can be verified through post-evaluation. The edge side, leveraging its proximity to the data source, can perform real-time energy efficiency index calculations and closed-loop control. According to a set time interval, the edge gateway periodically calculates key energy efficiency indicators, such as standby power consumption of the charging module, charging pile utilization rate, overall charging efficiency, load rate, AC / DC converter losses, failure rate, instantaneous charging efficiency, power consumption per unit time, and power consumption of the temperature control system. Based on the real-time evaluation results, the edge gateway can independently complete rapid control loops. For example, based on predicted temperature trends, it can linearly adjust the cooling system power in advance without waiting for cloud commands, greatly reducing response latency. The edge side is mainly used to calculate short-term energy efficiency index changes; the calculation results can be uploaded to the cloud and then participate in subsequent comprehensive energy efficiency evaluations. The cloud side, utilizing its powerful computing and storage capabilities, is responsible for non-real-time, macro-level, and in-depth evaluation tasks. After receiving energy efficiency indicators uploaded from the edge segment, long-term energy efficiency indicators can be calculated through long-term data storage and analysis, such as peak-valley difference rate, renewable energy absorption rate, V2G participation, and maintenance cycle. Using historical information, operators can identify energy-inefficient links or discover existing problems. Finally, a comprehensive analysis of the charging system's functional structure and operating characteristics, considering the impact of multiple dimensions such as technical performance, operation management, power quality, and ancillary service capabilities to the grid, yields the charging station's overall energy efficiency. After the evaluation, a post-evaluation report can be generated, providing in-depth analysis of various aspects affecting overall energy efficiency and their corresponding magnitudes. This method can provide indicator trends and optimization suggestions for subsequent station operation, accurately pinpointing current problems and helping operators adjust future operating strategies and evaluate control effectiveness.

[0058] The above describes the energy efficiency monitoring and optimization control method for electric vehicle charging systems provided in this application. It combines the advantages of convolutional neural networks in spatial feature extraction with the advantages of long short-term memory networks in capturing long-term temporal dependencies. An attention mechanism is also introduced, enabling the model to dynamically focus on key historical events, thus improving the prediction accuracy of user charging behavior. Based on the predicted data, a mathematical model with multiple constraints, including power, capacity, and user satisfaction, is constructed and solved using a rolling optimization algorithm. By recalculating using the latest data at each decision point, an ordered charging power allocation strategy is dynamically generated and the instructions are sent to the edge of the charging station. The edge gateway integrates multi-source data from charging pile operation, such as current, voltage, and temperature, and uses its built-in LSTM model to predict the short-term temperature rise trend of the equipment. Based on the prediction results, predictive adaptive control is implemented on the cooling system, i.e., intervention before the temperature approaches the safety limit and maintaining low power consumption during safe periods, thereby reducing the energy consumption of the auxiliary system. Finally, the system calculates short-term indicators such as instantaneous efficiency at the edge, while the cloud performs long-term macro-analysis. The evaluation results are generated into a report and fed back, allowing for the correction of model parameters and adjustment of strategies, thereby initiating the next round of optimization cycles and achieving continuous improvement in system performance.

[0059] It is understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. Furthermore, in some possible implementations, each step in the above embodiments may be selectively executed according to actual circumstances; it may be partially or fully executed, without limitation here. All or part of any feature of any embodiment of this application can be freely and arbitrarily combined without contradiction. The combined technical solutions are also within the scope of this application.

[0060] It should be understood that the above system is used to execute the methods in the above embodiments. The corresponding program modules in the system are similar in implementation principle and technical effect to those described in the above methods. The working process of the system can be referred to the corresponding process in the above methods, and will not be repeated here.

[0061] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0062] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0063] It is understood that the processor in the embodiments of this application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor may be a microprocessor or any conventional processor.

[0064] The method steps in the embodiments of this application can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0065] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0066] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application.

Claims

1. A method for energy efficiency monitoring and optimization control of an electric vehicle charging system, characterized in that, The method comprises: The cloud device predicts the user portrait of the electric vehicle owner based on a CNN-LSTM hybrid model, and the user portrait is used to indicate the charging behavior of the user, including the battery capacity of the electric vehicle, the arrival time, the departure time, the arrival power, and the required power; Based on the user portrait, an orderly charging strategy is solved by a rolling optimization algorithm, and the charging strategy is issued to the edge device, which is used to instruct the charging terminal to adjust the charging power; The edge device fuses multi-source data, predicts the temperature trend through an LSTM model, and performs predictive adaptive control on the cooling system of the charging terminal; Online energy efficiency post-evaluation is performed, an evaluation report is generated, and the evaluation results are fed back to the optimization link.

2. The method of claim 1, wherein, The cloud device predicts the user portrait of the electric vehicle owner based on a CNN-LSTM hybrid model, including: Local spatial features are extracted from input multi-dimensional feature data using a convolutional neural network; The local spatial features are input into a long short-term memory network to capture the long-term time sequence dependence of the user's charging behavior; An attention mechanism is introduced at the output end of the long short-term memory network to dynamically calculate the importance weight of different historical time steps for the current prediction.

3. The method of claim 2, wherein, The dynamic calculation of the importance weight of different historical time steps for the current prediction is as follows: wherein a hidden state of a t-th time step, an importance score of a t-th time step, a trainable weight vector, a trainable weight matrix, a trainable bias vector, a normalized attention weight of a t-th time step, a sequence length, a sum of scores over all time steps, a computed context vector.

4. The method of claim 1, wherein, Based on the user portrait, an orderly charging strategy is solved by a rolling optimization algorithm, including: The charging pile power allocation problem is modeled as a mixed integer linear programming problem; Based on the predicted user portrait data and actual operation data, a rolling optimization method is used for dynamic solution to maximize the energy efficiency of the charging system and balance the grid load and user satisfaction.

5. The method of claim 1, wherein, The temperature trend is predicted by the LSTM model, and predictive adaptive control is performed on the cooling system of the charging terminal, including: Collecting historical and real-time operation time series data of the charging terminal, including AC / DC converter temperature, charging gun temperature, charging current, voltage, ambient temperature, and humidity; The collected data are preprocessed, including data cleaning, missing value processing, outlier correction, and data normalization; The processed data are used to train the LSTM model to learn the mapping relationship between the operating parameters and the future temperature change; The trained LSTM model is deployed on the edge device to predict the future short-term temperature trend based on real-time sensor data; Based on the predicted temperature trend, the cooling fan speed or liquid cooling pump power is linearly adjusted to realize predictive temperature control.

6. The method of claim 1, wherein, The online energy efficiency post-evaluation is performed, an evaluation report is generated, and the evaluation results are fed back to the optimization link, including: The edge device periodically calculates short-term energy efficiency indicators, including instantaneous charging efficiency, unit time power consumption, temperature control system power consumption, and standby power consumption, and performs local rapid regulation and control; The cloud device aggregates the data uploaded by the edge device, calculates long-term energy efficiency indicators, including peak-valley difference rate and renewable energy consumption rate; The short-term and long-term energy efficiency indicators are integrated for multi-dimensional energy efficiency analysis, and a post-evaluation report is generated, and the evaluation results are fed back to the user portrait prediction model and the optimization scheduling link for model correction and strategy adjustment.

7. An electric vehicle charging system energy efficiency monitoring and optimization control system, characterized by, The system comprises: A charging terminal for providing charging services for electric vehicles and collecting operation data; An edge device, in communication connection with the charging terminal, is configured to receive data from the charging terminal, perform short-term energy efficiency index calculation, temperature trend prediction, and predictive adaptive control; A cloud device, in communication connection with the edge device, is configured to predict user profile, solve ordered charging strategy, perform long-term energy efficiency evaluation, and generate evaluation report.

8. The system of claim 7, wherein, The charging terminal comprises: A data acquisition module, configured to acquire current, voltage, AC / DC converter temperature, and charging gun head temperature data in real time through built-in sensors; An adjustment module, configured to receive and execute power distribution instructions and cooling control instructions from the edge device.

9. The system of claim 7, wherein, The edge device comprises: An adaptive control module, configured to run a light-weighted LSTM model to predict temperature trend, and send control instructions to the cooling system of the charging terminal; A short-term target prediction module, configured to calculate short-term energy efficiency index, and upload the calculation result to the cloud device.

10. The system of claim 7, wherein, The cloud device comprises: A user profile prediction module, configured to run a CNN-LSTM hybrid model to predict user charging behavior parameters; An optimization scheduling module, configured to run a rolling optimization algorithm based on the prediction result, and generate a globally optimal ordered charging strategy; An evaluation module, configured to aggregate data uploaded by the edge device, perform long-term energy efficiency index calculation, and multi-dimensional comprehensive energy efficiency analysis, and generate a post-evaluation report.

Images