Construction method and evaluation method of multi-dimensional evaluation model based on vehicle network interaction

Abstract

The invention discloses a construction method and evaluation method based on a vehicle network interaction multi-dimensional evaluation model, and the method comprises the steps: automatically recognizing all influence factors in a region through a machine learning algorithm based on intelligent power grid big data and geographic information data, and determining region-level evaluation parameters; fusing the evaluation parameters by a multi-dimensional data fusion method based on deep learning to obtain a comprehensive evaluation index; then, the number of input variables is reduced through feature selection and dimension reduction technologies, meanwhile, an approximation algorithm is adopted to replace complex mathematical operation, and a lightweight assessment model is constructed; and constructing a vehicle network interaction multi-dimensional evaluation model based on the lightweight evaluation model in combination with resource value information and multi-scene requirements. According to the method, the functions of differentiated scheduling, peak clipping and valley filling, frequency adjustment and emergency response are realized, the problems of resource mismatching, scheduling lag and poor regional adaptability caused by traditional static assessment are solved, the power grid stability, the scheduling efficiency and the user participation enthusiasm are improved, and the application of the V2G technology is promoted.

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-everything (V2X) interaction, and specifically to a method for constructing and evaluating a multi-dimensional evaluation model of V2X interaction. Background Technology

[0002] With the increasing prevalence of electric vehicles (EVs), the interaction between the power grid and EVs (V2G) has become a crucial component of modern energy management systems. V2G technology allows EVs to be used not only as transportation vehicles but also as distributed energy storage devices, supplying power to the grid during peak hours, thereby helping to balance grid load and improve the stability and reliability of the power system. However, achieving this goal requires a comprehensive and efficient evaluation model to manage the interaction between EVs and the grid. Existing solutions typically rely on single-dimensional data analysis or simple weighted averaging methods for evaluation, which often fail to adequately reflect the complexities of real-world operating conditions.

[0003] Traditional assessment models employ static parameter configuration, applying the same assessment weights and standards to all regions and time periods. This ignores dynamic factors such as regional grid differences, changes in user behavior, and external environmental fluctuations. High-load areas may go unidentified and continue operating under conventional dispatch strategies, potentially leading to grid overload; low-load areas may be underutilized, missing peak-shaving and frequency regulation opportunities; and untimely responses to emergencies may result in dispatch delays. Static parameter configuration leads to irrational resource allocation, low dispatch efficiency, and may even exacerbate grid fluctuations. Summary of the Invention

[0004] To address the problems of unreasonable resource allocation, low scheduling efficiency, and even potential exacerbation of power grid fluctuations caused by static parameter configuration in existing technologies, this invention proposes a construction method and evaluation method based on a multi-dimensional evaluation model of vehicle-grid interaction, including: Based on smart grid big data and geographic information data, machine learning algorithms are used to automatically identify various influencing factors in the region and determine regional-level assessment parameters. Under the constraints of regional-level evaluation parameters, a deep learning-based multidimensional data fusion method is used to evaluate the micro-performance of each electric vehicle and obtain a comprehensive evaluation index. Based on comprehensive evaluation indicators, the number of input variables is reduced through feature selection and dimensionality reduction techniques, while approximation algorithms are used to replace complex mathematical operations to construct a lightweight evaluation model. A multi-dimensional evaluation model for vehicle-to-everything (V2X) interaction is constructed based on a lightweight evaluation model that combines resource value information and multi-scenario requirements.

[0005] Optionally, the automatic identification of various influencing factors within the region based on smart grid big data and geographic information data, and the determination of regional-level evaluation parameters through machine learning algorithms, includes: By integrating the location and capacity information of charging facilities within a region from geographic information data, integrated data is obtained; Machine learning algorithms are used to identify various influencing factors within the region from the integrated data, including: the level of infrastructure development and the power grid supply capacity. Each influencing factor is used as an evaluation parameter at the regional level.

[0006] Optionally, under the constraints of evaluation parameters at the regional level, the deep learning-based multidimensional data fusion method is used to evaluate the microscopic performance of each electric vehicle to obtain a comprehensive evaluation index, including: Under the constraints of regional-level evaluation parameters, the weight allocation of various influencing indicators of electric vehicles is adaptively adjusted; Based on deep learning algorithms and the weight allocation of each influencing indicator, the comprehensive evaluation index is effectively integrated to generate a comprehensive evaluation index. The influencing indicators include: discharge operation health index, dispatchability index, discharge efficiency index, charging operation health index, grid load regulation index, and charging efficiency index.

[0007] Optionally, based on comprehensive evaluation indicators, the number of input variables is reduced through feature selection and dimensionality reduction techniques, while approximation algorithms are used to replace complex mathematical calculations to construct a lightweight evaluation model, including: Select key features; Simplify data structure using dimensionality reduction techniques based on key features; Use approximation algorithms to replace complex mathematical operations; A lightweight evaluation model is built based on simplified data structures and approximation algorithms.

[0008] Optionally, the lightweight evaluation model combines resource value information and multi-scenario requirements to construct a multi-dimensional evaluation model for vehicle-to-network interaction, including: Based on a lightweight assessment model combined with resource value information, the value assessment of electric vehicles participating in grid interaction is dynamically adjusted. Based on a lightweight evaluation model and combined with the needs of multiple scenarios, the charging and discharging strategies of electric vehicles are dynamically adjusted. A multi-dimensional evaluation model for vehicle-to-grid interaction is constructed based on the adjusted value assessment and charging / discharging strategy.

[0009] Optionally, it also includes: The business maturity of electric vehicles participating in grid interaction in different regions and of different types is assessed by combining data analysis and expert scoring. The aforementioned business maturity level can be used as an enable switch for the multi-dimensional evaluation model of vehicle-to-network interaction or as a moderating factor for the comprehensive score.

[0010] Optionally, using the business maturity level as an enable switch for the multi-dimensional evaluation model of vehicle-to-network interaction or as a moderating factor in the comprehensive score includes: The maturity of electric vehicles participating in grid interaction in different regions and of different types is divided into high-maturity areas, medium-maturity areas, and low-maturity areas. In highly mature areas, a multi-dimensional evaluation model for vehicle-to-grid interaction will be fully implemented and high-frequency scheduling will be used; in medium-maturity areas, scheduling intensity will be limited; and in low-maturity areas, only monitoring functions will be enabled.

[0011] Optionally, it also includes: By leveraging real-time user feedback on scheduling strategies, we can optimize the multi-dimensional evaluation model for vehicle-to-network interaction.

[0012] Furthermore, this invention also provides a system for constructing a multi-dimensional evaluation model based on vehicle-to-grid interaction, comprising: The evaluation parameter determination module is used to automatically identify various influencing factors within a region and determine regional-level evaluation parameters based on smart grid big data and geographic information data through machine learning algorithms. The multi-dimensional fusion module is used to evaluate the micro-performance of each electric vehicle based on a deep learning-based multi-dimensional data fusion method under the constraints of regional-level evaluation parameters, and obtain a comprehensive evaluation index. Lightweight evaluation models are used to reduce the number of input variables based on comprehensive evaluation indicators through feature selection and dimensionality reduction techniques, while using approximation algorithms to replace complex mathematical calculations, thus constructing lightweight evaluation models. The multi-dimensional building module is used to construct a multi-dimensional evaluation model for vehicle-to-network interaction based on a lightweight evaluation model, combined with resource value information and multi-scenario requirements.

[0013] Optionally, the multi-dimensional fusion module includes: A multi-dimensional data input module is used to receive various evaluation parameters; The adaptive weight allocation module is used to adaptively adjust the weight allocation of various influencing indicators of electric vehicles under the constraints of regional-level evaluation parameters. The data fusion module is used to effectively fuse various evaluation parameters based on deep learning algorithms and the weight allocation of each evaluation parameter to generate a comprehensive evaluation index.

[0014] Optionally, the lightweight evaluation model includes: The feature selection module is used to select key features; The dimensionality reduction module is used to simplify data structures based on key features using dimensionality reduction techniques. The approximation algorithm module is used to replace complex mathematical operations with approximation algorithms.

[0015] Optionally, the multi-dimensional construction module is specifically used for: Based on a lightweight assessment model combined with resource value information, the value assessment of electric vehicles participating in grid interaction is dynamically adjusted. Based on a lightweight evaluation model and combined with the needs of multiple scenarios, the charging and discharging strategies of electric vehicles are dynamically adjusted. A multi-dimensional evaluation model for vehicle-to-grid interaction is constructed based on the adjusted value assessment and charging / discharging strategy.

[0016] Optionally, it also includes: a business maturity analysis system for: The business maturity of electric vehicles participating in grid interaction in different regions and of different types is assessed by combining data analysis and expert scoring. The aforementioned business maturity level can be used as an enable switch for the multi-dimensional evaluation model of vehicle-to-network interaction or as a moderating factor for the comprehensive score.

[0017] Optionally, it also includes: a user feedback system, which records users’ actual experience information on scheduling strategies through a decentralized ledger, and optimizes the multi-dimensional evaluation model of vehicle-to-network interaction based on the actual experience information.

[0018] In another aspect, this application also provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the construction method and evaluation method of the multi-dimensional evaluation model based on vehicle-to-everything (V2X) interaction as described above are implemented.

[0019] Furthermore, this application also provides a readable storage medium on which an executable program is stored. When the executable program is executed, it implements the construction method and evaluation method of the multi-dimensional evaluation model based on vehicle-to-everything (V2X) interaction as described above.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method for constructing a multi-dimensional evaluation model based on vehicle-to-grid (V2G) interaction, comprising: automatically identifying various influencing factors within a region using machine learning algorithms based on smart grid big data and geographic information data, and determining regional-level evaluation parameters; fusing the evaluation parameters using a deep learning-based multi-dimensional data fusion method to obtain a comprehensive evaluation index; reducing the number of input variables through feature selection and dimensionality reduction techniques, and replacing complex mathematical calculations with approximation algorithms to construct a lightweight evaluation model based on the comprehensive evaluation index; and constructing a multi-dimensional evaluation model for V2G interaction based on the lightweight evaluation model combined with resource value information and multi-scenario requirements. This invention fuses various evaluation parameters and constructs a lightweight evaluation model to achieve differentiated scheduling, peak shaving and valley filling, frequency regulation, and emergency response functions, effectively solving the problems of resource mismatch, scheduling lag, and poor regional adaptability caused by traditional static evaluation, improving grid stability, scheduling efficiency, and user participation, and promoting the application of V2G technology. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the construction method of the multi-dimensional evaluation model based on vehicle-to-grid interaction of the present invention. Figure 2 This is a flowchart illustrating the construction method of a multi-dimensional evaluation model based on vehicle-to-network interaction in a specific embodiment of the present invention. Figure 3 This is a schematic diagram of an electronic device structure according to the present invention. Detailed Implementation

[0022] To better understand the present invention, the following description, in conjunction with the accompanying drawings and embodiments, will further illustrate the content of the present invention.

[0023] This invention provides a method for constructing a multi-dimensional evaluation model based on vehicle-to-grid (V2G) interaction. This model enables comprehensive, dynamic, and intelligent evaluation and scheduling decision support for electric vehicles participating in grid interaction. By integrating multi-source data such as vehicle status, grid load, geographic information, and user behavior, it comprehensively considers the feasibility, value, and social benefits of the technology, generating a unified scheduling priority score and optimization strategy. The model's input data includes multi-dimensional heterogeneous information such as battery SOC, health status, regional grid load, electricity price signals, charging pile distribution, and user travel habits. The output data includes comprehensive evaluation indicators, resource value assessments, optimal charging and discharging strategy suggestions, and regional collaborative scheduling instructions. Based on these output data, the system can achieve differentiated scheduling, peak shaving and valley filling, frequency regulation, and emergency response functions. It effectively solves the problems of resource mismatch, scheduling lag, and poor regional adaptability caused by traditional static evaluation, improving grid stability, scheduling efficiency, and user participation, and promoting the application of V2G technology.

[0024] Example 1: This invention provides a method for constructing a multi-dimensional evaluation model based on vehicle-to-grid interaction, such as... Figure 1 As shown, it includes: Step 1: Based on smart grid big data and geographic information data, machine learning algorithms are used to automatically identify various influencing factors within the region and determine regional-level assessment parameters; Step 2: Under the constraints of regional-level evaluation parameters, the micro-performance of each electric vehicle is evaluated using a deep learning-based multi-dimensional data fusion method to obtain a comprehensive evaluation index; Step 3: Based on the comprehensive evaluation indicators, reduce the number of input variables through feature selection and dimensionality reduction techniques, and at the same time use approximation algorithms to replace complex mathematical calculations to construct a lightweight evaluation model; Step 4: Based on a lightweight evaluation model, combine resource value information and multi-scenario requirements to build a multi-dimensional evaluation model for vehicle-to-network interaction.

[0025] A method for constructing a multi-dimensional evaluation model based on vehicle-to-grid interaction is described below, with each step further explained: Step 1: Based on smart grid big data and geographic information data, automatically identify various influencing factors within the region using machine learning algorithms, and determine regional-level assessment parameters, including: By integrating the location and capacity information of charging facilities within a region from geographic information data, integrated data is obtained; Machine learning algorithms are used to identify various influencing factors within the region from the integrated data, including: the level of infrastructure development and the power grid supply capacity. Each influencing factor is used as an evaluation parameter at the regional level.

[0026] Step 1 specifically includes: S1: Develop an automated data acquisition tool that combines a cloud computing platform to achieve real-time data acquisition and preprocessing; Preprocessing includes: data cleaning, outlier detection and correction, missing data imputation, data format standardization, data normalization (such as Z-score or Min-Max), and data time alignment.

[0027] S2: Introducing Geographic Information System (GIS) data and smart grid big data, machine learning algorithms are used to automatically identify influencing factors such as the level of charging facility construction and grid supply capacity within a region, and dynamically adjust regional-level assessment parameters. This step combines GIS and smart grid big data, and uses machine learning algorithms to identify regional-level macro-characteristics, such as the level of charging facility construction, grid supply capacity, load capacity, and areas with frequent historical faults. Its purpose is to achieve dynamic adjustment of regional assessment parameters and provide spatial context constraints for subsequent assessments.

[0028] Step 2: Under the constraints of regional-level evaluation parameters, a multi-dimensional data fusion method based on deep learning is used to evaluate the micro-performance of each electric vehicle, obtaining comprehensive evaluation indicators, including: Under the constraints of regional-level evaluation parameters, the weight allocation of various influencing indicators of electric vehicles is adaptively adjusted; Based on deep learning algorithms and the weight allocation of each influencing indicator, the comprehensive evaluation index is effectively integrated to generate a comprehensive evaluation index. The influencing indicators include: discharge operation health index, dispatchability index, discharge efficiency index, charging operation health index, grid load regulation index, and charging efficiency index.

[0029] Step 2 specifically includes: Step S3: Based on a deep learning-based multi-dimensional data fusion method, data from multiple dimensions, including discharge operation health index, dispatchability index, discharge efficiency index, charging operation health index, grid load regulation index, and charging efficiency index, are effectively fused to form a comprehensive evaluation index. The discharge operation health index, dispatchability index, and discharge efficiency index fused in this step focus on the micro-performance of a single electric vehicle or aggregation unit, which belongs to the individual-level technical and behavioral evaluation dimensions.

[0030] The relationship between steps S2 and S3 is as follows: the output of S2 can be used as a prior weight or bias term in the deep learning model of step S3, affecting the fusion method of indicators of each dimension or directly used as one of the input features to participate in the generation of comprehensive evaluation indicators, thereby achieving "macro and micro" classification evaluation.

[0031] The deep learning methods (such as CNN, RNN, and Transformer) used in step S3 belong to supervised learning and are used for multi-dimensional data fusion, with the goal of generating comprehensive evaluation indicators.

[0032] Step 3: Based on the comprehensive evaluation indicators, reduce the number of input variables through feature selection and dimensionality reduction techniques, and use approximation algorithms to replace complex mathematical calculations to construct a lightweight evaluation model, including: Select key features; Simplify data structure using dimensionality reduction techniques based on key features; Use approximation algorithms to replace complex mathematical operations; A lightweight evaluation model is built based on simplified data structures and approximation algorithms.

[0033] Step 3 specifically includes: S4: Construct a lightweight evaluation model, reduce the number of input variables through feature selection and dimensionality reduction techniques, and replace complex mathematical operations with approximation algorithms to reduce computational costs.

[0034] Specifically, this refers to high-order mathematical operations in traditional deep learning or optimization models that are computationally expensive and difficult to execute in real time. These include, but are not limited to: large-scale matrix multiplication and self-attention mechanism calculations in deep neural networks (such as DNN and Transformer); recursive forward and backward propagation processes in recurrent neural networks; quadratic programming solutions and kernel function calculations in support vector machines; Bellman equation iteration and experience replay training in reinforcement learning; and solving NP-hard optimization problems such as mixed integer programming. These operations typically require high-performance computing resources and are difficult to execute in real time at the edge. Therefore, this embodiment uses approximate algorithms as alternatives, such as using K-nearest neighbors (KNN), decision trees, or linear regression to replace complex SVM or LSTM models, and using principal component analysis (PCA) to reduce data dimensionality to reduce input dimensions, thereby significantly reducing computational complexity and enabling lightweight and edge-deployment of the evaluation model.

[0035] The comprehensive evaluation index formed in S3 and the lightweight evaluation model built in S4 are directly related. The comprehensive evaluation index is the output of the multi-dimensional fusion of deep learning in S3, representing a unified quantitative score for the V2G capability of electric vehicles. It is the core intermediate result of the entire evaluation system. Building upon this, S4 constructs a lightweight evaluation model through feature selection, dimensionality reduction techniques, and approximation algorithms. This model aims to fit an approximate output that is highly consistent with the original comprehensive evaluation index, thereby significantly reducing computational costs while maintaining evaluation accuracy. Therefore, the lightweight model can be seen as a simplification and approximation of the complex deep learning process, enabling evaluation capabilities to be deployed on edge devices or in-vehicle terminals and supporting real-time decision-making. This comprehensive evaluation index is also closely related to other modules: its results can be used to guide resource value assessment and multi-scenario scheduling priority ranking, and it is continuously optimized through user feedback, serving as a key bridge for achieving global collaborative evaluation.

[0036] Step 4: Based on a lightweight evaluation model, and combining resource value information and multi-scenario requirements, construct a multi-dimensional evaluation model for vehicle-to-network interaction, including: Based on a lightweight assessment model combined with resource value information, the value assessment of electric vehicles participating in grid interaction is dynamically adjusted. Based on a lightweight evaluation model and combined with the needs of multiple scenarios, the charging and discharging strategies of electric vehicles are dynamically adjusted. A multi-dimensional evaluation model for vehicle-to-grid interaction is constructed based on the adjusted value assessment and charging / discharging strategy.

[0037] Step 4 specifically includes: A dynamic resource value assessment module is constructed, which is one of the core sub-modules of the multi-dimensional assessment model for vehicle-grid interaction. This module is responsible for converting the discharge behavior of electric vehicles into quantifiable value. It can calculate the economic value and social benefits of electric vehicles participating in grid interaction in real time, and dynamically adjust the value assessment of electric vehicles participating in grid interaction by utilizing real-time data and predictive algorithms (such as machine learning), combined with factors such as electricity market price fluctuations and carbon emission reductions. A multi-scenario regulation demand analysis module is designed to dynamically adjust the charging and discharging strategies of electric vehicles based on the needs of different scenarios (such as peak load periods, off-peak load periods, and emergencies). Combining reinforcement learning algorithms and real-time data, intelligent scheduling is achieved to maximize grid stability and economic benefits. This multi-scenario regulation demand analysis module serves as the strategy execution terminal for the vehicle-grid interaction multi-dimensional evaluation model, responsible for converting evaluation results into specific charging and discharging control commands. Based on the needs of different operating scenarios (such as peak load, off-peak charging, and emergency power outages), this module dynamically adjusts the charging and discharging strategies of electric vehicles by combining real-time data and reinforcement learning algorithms (such as DQN). The reinforcement learning (such as Q-learning and DQN) used in this step belongs to sequential decision learning and is used to learn the optimal scheduling strategy in a dynamic environment.

[0038] The present invention provides a method for constructing a multi-dimensional evaluation model based on vehicle-grid interaction, which further includes: evaluating the business maturity of electric vehicles of different regions and types participating in grid interaction by combining data analysis and expert scoring; The aforementioned business maturity level can be used as an enable switch for the multi-dimensional evaluation model of vehicle-to-network interaction or as a moderating factor for the comprehensive score.

[0039] The specific business maturity level is obtained through analysis by the business maturity analysis system. The business maturity analysis system and the lightweight assessment model here have a pre-screening and access control relationship. The business maturity analysis system will be further introduced below.

[0040] This business maturity analysis system assesses the maturity of electric vehicles (EVs) participating in grid interaction across different regions and types through a combination of data analysis and expert scoring. Specifically, the system uses the Analytic Hierarchy Process (AHP) and entropy weighting to comprehensively evaluate factors such as policy support, infrastructure development, and user acceptance, calculating a business maturity score for EV participation in V2G in each region and classifying it into high, medium, and low levels. This score serves as an "enabling switch" or adjustment factor for the lightweight assessment model: fully enabling the model and high-frequency dispatching in high-maturity areas; limiting dispatching intensity in medium-maturity areas; and enabling only monitoring functions in low-maturity areas. This mechanism avoids forcibly promoting V2G technology in areas with insufficient infrastructure or low user acceptance, reducing system operational risks and ensuring the robustness and efficiency of technology implementation. It is a crucial support for the regionally differentiated deployment of the lightweight model.

[0041] The method combines the Analytic Hierarchy Process (AHP) and the entropy weight method, taking into account a variety of factors (such as policy support, infrastructure construction, and user acceptance).

[0042] This embodiment also includes a power grid impact simulation platform. This power grid impact simulation platform and the lightweight evaluation model or the vehicle-grid interaction multi-dimensional evaluation model are related through offline verification and parameter optimization. The power grid impact simulation platform is described below: This platform simulates the impact of large-scale electric vehicle (EV) integration on grid stability and provides optimization suggestions. Specifically, it uses PowerWorld or PSCAD to build high-precision grid models to simulate stability indicators such as voltage fluctuations, frequency deviations, and power flow changes after large-scale EV integration. The scheduling schemes generated by the evaluation model can be imported into the platform for pre-simulation to verify their impact on grid security. If the simulation reveals potential problems (such as voltage exceeding limits), the system will feed back to the evaluation model to adjust parameters such as the "grid load regulation index" or optimize the scheduling strategy. The platform is also used for extreme scenario testing and boundary analysis to ensure the safety and reliability of the evaluation model in real-world operation. It is a key infrastructure for achieving closed-loop verification of "evaluation, simulation, and optimization."

[0043] By leveraging high-precision simulation technology and big data analysis, potential problems can be identified in advance and optimization solutions can be provided.

[0044] This embodiment also includes a cross-regional collaborative evaluation system. This cross-regional collaborative evaluation system has a horizontal expansion and global coordination relationship with the lightweight evaluation model or the vehicle-to-grid interaction multi-dimensional evaluation model. The cross-regional collaborative evaluation system is described below: This system enables data sharing and joint evaluation between different regions, improving the overall efficiency of the power grid's interaction. By leveraging distributed databases and cloud computing technologies, cross-regional data synchronization and joint evaluation can be achieved, thereby improving the overall scope and coordination of the evaluation.

[0045] This embodiment also includes a user behavior prediction module. This user behavior prediction module has a forward-looking input and policy triggering relationship with the lightweight evaluation model or the multi-dimensional evaluation model of vehicle-to-everything (V2X) interaction. The user behavior prediction module will be further described below: Leveraging big data analytics and behavioral psychology models, the module predicts users' charging and discharging behaviors over a future period. Specifically, it uses LSTM or Transformer deep learning models, combined with historical charging / discharging data, travel patterns, electricity prices, weather, and other external factors, to predict the user's charging and discharging direction over the next 24 hours. The prediction results serve as a crucial input feature for the evaluation model, directly impacting the "schedulability index" and overall score. For example, if a user is predicted to be taking a long trip, their discharging priority is automatically lowered; if a user is predicted to be parking for an extended period overnight, off-peak charging is scheduled in advance. Furthermore, the prediction results can trigger a "pre-response" mode, locking in available vehicle resources in advance, enhancing the predictability and proactivity of power grid dispatching, and providing key support for the evaluation model to achieve intelligent and forward-looking dispatching.

[0046] Combining deep learning algorithms to formulate forward-looking scheduling strategies in advance, thereby improving the flexibility and response speed of the power grid, is the relationship between special operating modes and reverse optimization.

[0047] The present invention provides a method for constructing a multi-dimensional evaluation model based on vehicle-to-grid interaction, which further includes: S5: Build an intelligent interactive platform for Internet of Things (IoT) technology to support efficient communication and data exchange between electric vehicles and the power grid, monitor the status of electric vehicles in real time, and dynamically adjust charging strategies. S6: Construct a user feedback system using blockchain technology to ensure the authenticity and immutability of user feedback information. This feedback information is then used to continuously optimize the evaluation model. The "evaluation model" here includes both a lightweight evaluation model deployed at the edge and a complete evaluation system at the central end. The optimization process is typically completed on the central server and then synchronized to the lightweight model through model distillation or parameter updates, achieving collaborative evolution of the entire network model. Specifically, a decentralized ledger records users' actual experiences with scheduling strategies (such as whether it affects vehicle usage, whether compensation is reasonable, and whether operation is convenient), ensuring the authenticity and immutability of feedback information. This high-quality feedback data is fed back to the evaluation model to correct prediction biases in the "schedulability index," adjust compensation coefficients in resource value assessment, and optimize the humanized thresholds of multi-scenario control strategies, thereby improving the model's usability and user satisfaction. Simultaneously, the transparent and trustworthy feedback mechanism enhances user trust in the system, encouraging more users to participate in V2G projects and forming a positive evolutionary ecosystem of "evaluation, scheduling, feedback, and optimization," which is a crucial support for the social adaptation and continuous optimization of the evaluation model.

[0048] The evaluation is optimized in several ways: (1) Negative feedback (such as "scheduling affects travel") is marked as training samples to retrain model parameters and reduce the scheduling probability in similar situations; (2) Feedback data can be used to dynamically adjust the index weights in AHP or entropy weight method. For example, when most users value travel convenience more, the system automatically increases the weight of "scheduling index"; (3) Feedback can be used to calibrate the parameters of user behavior prediction models (such as LSTM) and improve their prediction accuracy; (4) Concentrated negative feedback can trigger an expert intervention mechanism to manually adjust the evaluation rule base.

[0049] In this embodiment, the automated data acquisition tool in S1 further includes: S1.1: Real-time data acquisition module, used to acquire data from multiple sources; S1.2: Data cleaning module, used to clean and preprocess the collected data; S1.3: Data storage module, used to store the cleaned data on a cloud computing platform.

[0050] In this embodiment, the integration of GIS and smart grid big data in S2 further includes: S2.1: Geographic Information System (GIS) data integration module, used to integrate the location and capacity information of charging facilities within the region; S2.2: Smart grid big data analysis module, which uses machine learning algorithms to identify influencing factors such as the level of charging infrastructure construction and grid supply capacity in the region; S2.3: Dynamic adjustment module, which dynamically adjusts the evaluation parameters at the regional level based on the above analysis results.

[0051] S2 combines GIS with smart grid big data and uses machine learning algorithms to identify regional macro-characteristics, such as the level of charging infrastructure construction, grid supply capacity, load capacity, and areas with frequent historical faults. Its purpose is to achieve dynamic adjustment of regional assessment parameters and provide spatial context constraints for subsequent assessments.

[0052] In this embodiment, the deep learning multidimensional data fusion method in S3 further includes: S3.1: Multi-dimensional data input module, used to receive various data such as discharge operation health index, dispatchability index, discharge efficiency index, as well as charging operation health index, grid load regulation index, and charging efficiency index; S3.2: Data fusion module, which fuses multiple types of data based on deep learning algorithms to generate comprehensive evaluation indicators; S3.3: Adaptive weight allocation module, which can adaptively adjust the weight allocation of data in each dimension.

[0053] The discharge operation health index, dispatchability index, and discharge efficiency index integrated in S3 focus on the micro-performance of a single electric vehicle or aggregation unit, and belong to the technical and behavioral evaluation dimensions at the individual level.

[0054] The relationship between steps S2 and S3 is as follows: the output of S2 can be used as a prior weight or bias term in the S3 deep learning model, affecting the fusion method of indicators of various dimensions or directly used as one of the input features to participate in the generation of comprehensive evaluation indicators, thereby achieving "macro and micro" classification evaluation.

[0055] In this embodiment, the lightweight evaluation model in S4 further includes: S4.1: Feature selection module, used to select key features and reduce the number of input variables; S4.2: Dimensionality reduction module, which further simplifies the data structure through dimensionality reduction techniques; S4.3: Approximation algorithm module, which uses approximation algorithms to replace complex mathematical operations.

[0056] This embodiment also includes a business maturity analysis system: the business maturity analysis system and the lightweight assessment model have a pre-screening and access control relationship. The business maturity analysis system will be further described below.

[0057] This system assesses the business maturity of electric vehicles (EVs) participating in grid interaction across different regions and of different types through a combination of data analysis and expert scoring. Specifically, the system uses the Analytic Hierarchy Process (AHP) and entropy weighting to comprehensively evaluate factors such as policy support, infrastructure development, and user acceptance, calculating a business maturity score for EV participation in V2G in each region and classifying it into high, medium, and low levels. This score serves as an "enabling switch" or adjustment factor for the lightweight assessment model: in high-maturity regions, the model is fully enabled and high-frequency scheduling is implemented; in medium-maturity regions, scheduling intensity is limited; and in low-maturity regions, only monitoring functions are enabled. This mechanism avoids forcibly promoting V2G technology in regions with insufficient infrastructure or low user acceptance, reduces system operational risks, ensures the robustness and efficiency of technology implementation, and is a crucial support for the regionally differentiated deployment of the lightweight model.

[0058] The method combines the Analytic Hierarchy Process (AHP) and the entropy weight method, taking into account a variety of factors (such as policy support, infrastructure construction, and user acceptance).

[0059] This embodiment also includes a power grid impact simulation platform. This power grid impact simulation platform and the lightweight evaluation model or the vehicle-grid interaction multi-dimensional evaluation model are related through offline verification and parameter optimization. The power grid impact simulation platform is described below: This platform simulates the impact of large-scale electric vehicle (EV) integration on grid stability and provides optimization suggestions. Specifically, it uses PowerWorld or PSCAD to build high-precision grid models to simulate stability indicators such as voltage fluctuations, frequency deviations, and power flow changes after large-scale EV integration. The scheduling schemes generated by the evaluation model can be imported into the platform for pre-simulation to verify their impact on grid security. If the simulation reveals potential problems (such as voltage exceeding limits), the system will feed back to the evaluation model to adjust parameters such as the "grid load regulation index" or optimize the scheduling strategy. The platform is also used for extreme scenario testing and boundary analysis to ensure the safety and reliability of the evaluation model in real-world operation. It is a key infrastructure for achieving closed-loop verification of "evaluation, simulation, and optimization."

[0060] By leveraging high-precision simulation technology and big data analysis, potential problems can be identified in advance and optimization solutions can be provided.

[0061] This embodiment also includes a cross-regional collaborative evaluation system. This cross-regional collaborative evaluation system has a horizontal expansion and global coordination relationship with the lightweight evaluation model or the vehicle-to-grid interaction multi-dimensional evaluation model. The cross-regional collaborative evaluation system is described below: This system enables data sharing and joint evaluation between different regions, improving the overall efficiency of the power grid's interaction. By leveraging distributed databases and cloud computing technologies, cross-regional data synchronization and joint evaluation can be achieved, thereby improving the overall scope and coordination of the evaluation.

[0062] This embodiment also includes a user behavior prediction module. This user behavior prediction module has a forward-looking input and policy triggering relationship with the lightweight evaluation model or the multi-dimensional evaluation model of vehicle-to-everything (V2X) interaction. The user behavior prediction module will be further described below: Leveraging big data analytics and behavioral psychology models, the module predicts users' charging and discharging behaviors over a future period. Specifically, it uses LSTM or Transformer deep learning models, combined with historical charging / discharging data, travel patterns, electricity prices, weather, and other external factors, to predict the user's charging and discharging direction over the next 24 hours. The prediction results serve as a crucial input feature for the evaluation model, directly impacting the "schedulability index" and overall score. For example, if a user is predicted to be taking a long trip, their discharging priority is automatically lowered; if a user is predicted to be parking for an extended period overnight, off-peak charging is scheduled in advance. Furthermore, the prediction results can trigger a "pre-response" mode, locking in available vehicle resources in advance, enhancing the predictability and proactivity of power grid dispatching, and providing key support for the evaluation model to achieve intelligent and forward-looking dispatching.

[0063] Combining deep learning algorithms to formulate forward-looking scheduling strategies in advance, thereby improving the flexibility and response speed of the power grid, is the relationship between special operating modes and reverse optimization.

[0064] This embodiment also includes an emergency response mechanism, which is integrated with the multi-dimensional evaluation model for vehicle-to-grid interaction: An emergency response mechanism is established to rapidly mobilize electric vehicles to participate in grid regulation when power grid emergencies occur. Through a distributed control system and edge computing technology, real-time monitoring and rapid response are achieved, ensuring stable grid operation in emergencies. Specifically, when a power grid emergency occurs (such as a frequency below 49.5Hz), the system immediately switches to emergency mode, activating high-response vehicles from the "emergency dispatch pool" to quickly participate in grid regulation. At this time, the evaluation model suspends its regular dispatch logic and prioritizes executing rapid discharge or charging commands to ensure grid stability. This mechanism achieves millisecond-level response through a distributed control system and edge computing technology. Simultaneously, real-time data during the emergency dispatch process (such as response speed and compensation amount) is fed back to the main evaluation model to optimize its ability to identify emergencies and its value assessment logic, forming a closed loop of "emergency response, feedback, and optimization." This is an important supplementary mechanism for the evaluation model to ensure the safe operation of the power grid.

[0065] The advantages of this invention are: 1. By combining automated data acquisition tools with a cloud computing platform, this invention enables real-time acquisition, cleaning, and preprocessing of raw data, achieving efficient and accurate data acquisition and management, and improving data quality and processing efficiency. Compared to traditional methods relying on manual or offline data collection, this invention connects to multiple sources of equipment such as charging piles, power grid monitoring systems, and vehicle OBD modules via API interfaces. Combined with sliding window algorithms and LSTM prediction completion technology, it automatically identifies and repairs outliers and missing values, thereby ensuring the integrity and reliability of the input data for subsequent evaluation models.

[0066] 2. By integrating GIS data with smart grid big data, the system can automatically identify key parameters such as charging facility density and grid load capacity within a region. This enables dynamic adjustment of regional-level assessment parameters, improving the spatial adaptability and regional precision of the assessment model. Traditional systems often use fixed assessment parameters, making it difficult to adapt to differences in power grids across different regions. This invention, however, identifies high-load areas using the K-means clustering algorithm and dynamically adjusts scheduling weights and priorities based on the analysis results, thereby enhancing the flexibility and adaptability of the assessment model.

[0067] 3. By designing a deep learning-based multi-dimensional data fusion method, multi-dimensional data such as discharge operation health index, schedulability index, and discharge efficiency index can be effectively integrated. This achieves intelligent fusion and comprehensive scoring output of multi-dimensional evaluation indicators, thereby improving the comprehensiveness and accuracy of the evaluation model. Traditional methods often use single indicators or simple weighted averages, which are difficult to reflect the true state of complex systems. This invention, however, uses DNN or Transformer models for nonlinear modeling and combines an attention mechanism to dynamically adjust the weights of each dimension, significantly enhancing the scientific rigor and robustness of the evaluation results.

[0068] Example 2 A method for constructing a multi-dimensional evaluation model based on vehicle-to-grid interaction, such as... Figure 2 As shown, it includes the following steps: S1: Develop an automated data acquisition tool that combines a cloud computing platform to achieve real-time data acquisition, cleaning, and preprocessing; collect raw data from channels such as charging piles, power grid monitoring systems, and user feedback, and use the cloud computing platform to perform preprocessing such as data cleaning, format standardization, and missing value filling. S2: Introducing Geographic Information System (GIS) data and smart grid big data, machine learning algorithms are used to automatically identify influencing factors such as the level of charging facility construction and grid supply capacity in a region, and dynamically adjust regional-level evaluation parameters; combining GIS data and smart grid big data, machine learning algorithms are used to identify parameters such as the density of charging facilities and grid load capacity in a region, and dynamically adjust regional-level evaluation parameters, such as adjusting charging priority according to grid load conditions; S3: A deep learning-based multi-dimensional data fusion method effectively integrates data from multiple dimensions, including discharge operation health index, dispatchability index, discharge efficiency index, charging operation health index, grid load regulation index, and charging efficiency index, to form a comprehensive evaluation index; it introduces multiple evaluation indicators, including: Discharge Operation Health Index : Assess the battery health status during the discharge process of an electric vehicle; Scheduling index Assess the flexibility of vehicle participation in dispatching; Discharge efficiency index : Evaluate the energy conversion efficiency during the discharge process; Charging operation health index, power grid load regulation index, charging efficiency index, etc.; The above metrics are fused using deep learning models such as CNN, RNN, or Transformer; The model outputs a comprehensive evaluation index. The calculation formula is as follows:

[0069] in The weights are automatically learned by the model, representing the degree of influence of each dimension indicator on the final evaluation result; S4: A lightweight evaluation model is constructed, reducing the number of input variables through feature selection and dimensionality reduction techniques, while using approximation algorithms to replace complex mathematical operations, thereby reducing computational costs. Feature selection is performed on the input data to remove redundant or irrelevant features; Principal Component Analysis (PCA) or t-SNE is used for data dimensionality reduction; approximation algorithms (such as Random Forest and K-Nearest Neighbors) are used to replace traditional complex models (such as SVM and LSTM), reducing computational resource consumption. Through the design of the lightweight evaluation model, the system can significantly reduce computational resource consumption while ensuring evaluation accuracy, achieving efficient evaluation and edge deployment capabilities, and improving the model's running efficiency and applicability. Traditional evaluation models are computationally complex and resource-intensive, making them difficult to deploy on vehicle or edge devices. This invention, by replacing complex models with approximation algorithms such as feature selection, PCA dimensionality reduction, and KNN, effectively reduces the computational burden, enabling the model to run on low-power devices and enhancing the system's real-time performance and scalability. S5: Construct an intelligent interactive platform based on Internet of Things (IoT) technology to support efficient communication and data exchange between electric vehicles and the power grid, monitor the status of electric vehicles in real time, and dynamically adjust charging strategies. Specifically, based on the IoT platform, bidirectional communication between electric vehicles and the power grid is realized, vehicle status (such as battery level, SOC, and temperature) is monitored in real time, and charging and discharging strategies are dynamically adjusted according to the current power grid load. User feedback information is uploaded via a blockchain system to ensure that the information is tamper-proof, and the evaluation model is iteratively optimized using the feedback information. Through the design of the IoT-based intelligent interactive platform, bidirectional communication and real-time status monitoring between electric vehicles and the power grid are achieved, realizing dynamic adjustment of charging strategies and collaborative scheduling functions, thereby improving the efficiency and response speed of power grid and vehicle collaboration. Compared with traditional one-way communication systems, this invention obtains information such as vehicle SOC, temperature, and operating status in real time through the IoT platform, and dynamically adjusts the charging and discharging plan according to the power grid load, thereby improving the accuracy and flexibility of scheduling. S6: Construct a blockchain-based user feedback system to ensure the authenticity and immutability of user feedback information, and continuously optimize the evaluation model using this feedback information. Through the design of a blockchain-based user feedback system, user feedback information is made immutable and traceable, enabling continuous optimization of the evaluation model and enhancing user trust, thereby increasing system transparency and user participation. Traditional feedback systems suffer from problems such as easy information tampering and delayed feedback. This invention, however, constructs a decentralized feedback mechanism using blockchain technology, ensuring the authenticity and integrity of user behavior data, providing a high-quality data foundation for model iterative optimization. S7: Construct a dynamic resource value assessment module that can calculate the economic value and social benefits of electric vehicles participating in grid interaction in real time. Utilize real-time data and prediction algorithms (such as machine learning), combined with factors such as electricity market price fluctuations and carbon emission reduction, to dynamically adjust the value assessment of electric vehicles participating in grid interaction. Specifically: (1) Obtain data such as electricity market prices and carbon emission coefficients in real time; (2) Construct a value assessment model to calculate the economic value and social benefits brought by unit electricity discharge. The model formula is as follows:

[0070] in: Value assessment Electricity market prices; Discharge quantity; Battery wear and tear costs; Carbon emission reduction benefits; By designing a dynamic resource value assessment module, the economic value and social benefits of electric vehicles participating in grid interaction can be quantified and assessed in real time, realizing value-driven dispatch decision-making and improving the economic efficiency and environmental friendliness of grid dispatch. Traditional systems lack a dynamic assessment mechanism for discharge value. This invention, however, models multiple factors such as electricity market prices and carbon emission coefficients, combined with an LSTM prediction algorithm, to achieve dynamic calculation of the value per unit of discharge, providing a quantitative basis for formulating efficient dispatch strategies. S8: Design a multi-scenario control demand analysis module to dynamically adjust the charging and discharging strategies of electric vehicles based on the needs of different scenarios (such as peak load periods, off-peak load periods, and emergencies); combine reinforcement learning algorithms and real-time data to achieve intelligent scheduling, maximizing grid stability and economic benefits; specifically: (1) Preset multiple control scenarios (such as peak load, off-peak load, emergency power outage); (2) Train the scheduling strategy using reinforcement learning algorithms (such as Q-learning, DQN); (3) Dynamically adjust the charging and discharging plans of electric vehicles in different scenarios to maximize grid stability and economic benefits.

[0071] The reinforcement learning algorithm module includes: a training scheduling strategy using DQN; State space: grid load, vehicle SOC; Operational space: charging / discharging / holding; Reward function:

[0072] The symbols are explained as follows:

[0073] By designing a multi-scenario demand analysis module, the system can formulate differentiated scheduling strategies based on different power grid operation scenarios (such as peak, off-peak, and emergency), achieving intelligent scheduling and power grid stability improvement functions, and enhancing the power grid's regulation capabilities and economic benefits. Traditional scheduling strategies lack scenario adaptability, while this invention trains scheduling strategies using the DQN reinforcement learning algorithm, combined with power grid load and battery SOC status, to automatically adjust charging and discharging plans under different scenarios, significantly improving the stability of power grid operation and scheduling efficiency.

[0074] In this embodiment, the automated data acquisition tool in S1 further includes: S1.1: Real-time data acquisition module, used to acquire data from multiple sources; specifically: it connects to charging piles, power grid SCADA systems, and vehicle OBD modules via API interfaces, with a acquisition frequency of 1 second / time, and the data includes voltage, current, SOC, and timestamp; S1.2: Data cleaning module, used to clean and preprocess the collected data; specifically: (1) detect outliers (such as voltage jumps, SOC jumps); (2) use the sliding window algorithm to smooth the data; (3) fill in missing values ​​(such as using linear interpolation or LSTM prediction to fill in missing values). S1.3: Data storage module, used to store the cleaned data on the cloud computing platform; specifically including: (1) stored on the cloud platform (such as AWS S3, Alibaba Cloud OSS); (2) the data format is structured (such as Parquet, JSON); (3) supporting SQL query and real-time stream processing (such as Kafka, Flink).

[0075] In this embodiment, the integration of GIS and smart grid big data in S2 further includes: S2.1: Geographic Information System (GIS) data integration module, used to integrate the location and capacity information of charging facilities within the region; specifically including: importing regional map data (such as Shapefile format), extracting information such as charging pile location, capacity, and service radius, and performing spatial matching with power grid topology data; S2.2: Smart Grid Big Data Analysis Module, which uses machine learning algorithms to identify influencing factors such as the level of charging infrastructure construction and grid supply capacity in a region; specifically: it analyzes data such as grid load, voltage stability, and historical fault records, and uses clustering algorithms (such as K-means) to identify high-load areas and determine whether the area is suitable for large-scale deployment of V2G applications; S2.3: Dynamic adjustment module, which dynamically adjusts the evaluation parameters at the region level based on the above analysis results. Specifically, this includes dynamically adjusting evaluation parameters based on the analysis results, such as: reducing charging priority in high-load areas, increasing scheduling weight in areas with newly built charging stations, and reducing evaluation scores in areas with frequent faults.

[0076] In this embodiment, the deep learning multidimensional data fusion method in S3 further includes: S3.1: Multi-dimensional data input module, used to receive various data such as discharge operation health index, dispatchability index, discharge efficiency index, charging operation health index, grid load regulation index, and charging efficiency index; specifically: Input data includes: vehicle status data (such as SOC, health status), power grid status data (such as load rate, voltage fluctuation), market data (such as electricity price, carbon emission coefficient), and user behavior data (such as charging habits, discharge intention). S3.2: Data Fusion Module. This module fuses multiple data sources using deep learning algorithms to generate a comprehensive evaluation index. Specifically, it uses deep neural networks (DNNs) or multimodal fusion models (such as Transformer) to output the comprehensive evaluation index. ,like:

[0077] in It is a non-linear function learned by deep learning models. This is a power grid state index. This is a market condition index.

[0078] S3.3: Adaptive weight allocation module, capable of adaptively adjusting the weight allocation of data across various dimensions. Specifically, it uses an attention mechanism to dynamically adjust the weights of each dimension. The weight update formula is as follows:

[0079] in, The weight update values ​​are automatically learned by the model. The model is scored on its attention to each dimension. Representative at At time , the model calculates the original attention score for the i-th input dimension (or feature). Representative at At time i, the model calculates the original attention score for the j-th input dimension (feature), where i and j are the indices of the input dimensions.

[0080] In this embodiment, the lightweight evaluation model in S4 further includes: S4.1: Feature selection module, used to select key features and reduce the number of input variables; specifically, it uses chi-square test and mutual information method to screen important features, and also needs to remove redundant features, such as repeatedly collected SOC data. S4.2: Dimensionality reduction module, which further simplifies the data structure through dimensionality reduction techniques; specifically, it uses PCA for dimensionality reduction, retaining the first k principal components, or uses t-SNE for visual dimensionality reduction.

[0081] S4.3: Approximation Algorithm Module. This module uses approximation algorithms to replace complex mathematical calculations. Specifically, it uses KNN, decision trees, etc., to replace complex models; for regression tasks, it uses linear regression to replace neural networks.

[0082] This embodiment also includes a business maturity analysis system: The system uses a combination of data analysis and expert scoring to assess the business maturity of electric vehicles of different regions and types participating in grid interaction. This study employs a combination of Analytic Hierarchy Process (AHP) and entropy weighting, comprehensively considering multiple factors (such as policy support, infrastructure development, and user acceptance). Specifically, it includes a data analysis module and an AHP module; assessing the business maturity of electric vehicles participating in grid interaction in different regions; the data analysis module includes: collecting data on policy support, infrastructure, and user acceptance; and conducting preliminary assessments using an expert scoring system. The multilevel analysis module includes: Construct a judgment matrix, compare the importance of each factor, rank them hierarchically and calculate their weights. The comprehensive score formula is as follows:

[0083] in,

[0084] By designing a business maturity analysis system, the feasibility of electric vehicles participating in grid interaction in different regions can be scientifically assessed and classified. This enables differentiated regional promotion and policy support recommendations, thereby improving promotion efficiency and resource utilization. Traditional assessment methods lack systematicity and quantitative standards. This invention, however, combines AHP (Aspect-Based Hierarchy Process) and entropy weighting to weightedly evaluate multiple factors such as policy, infrastructure, and user acceptance, achieving a scientific score for regional business maturity and providing data support for policy formulation and resource allocation.

[0085] This embodiment also includes a power grid impact simulation platform: The platform simulates the impact of large-scale electric vehicle grid integration on grid stability and provides optimization suggestions. By leveraging high-precision simulation technology and big data analytics, potential problems can be identified in advance, and optimization solutions can be provided. Specifically, this includes: Simulation platform module: Use PowerWorld or PSCAD to build a power grid simulation model and set parameters such as electric vehicle access point and discharge power.

[0086] The high-precision simulation technology module performs power flow calculations and stability analysis, outputting indicators such as voltage fluctuations and frequency deviations, and providing optimization suggestions, such as adjusting connection time and power limits. Through the design of the grid impact simulation platform, the stability impact of large-scale electric vehicle grid connection can be predicted and optimized in advance, realizing grid risk early warning and dispatch strategy optimization functions, thereby reducing connection risks and improving the grid's safe operation capabilities. Traditional assessment methods lack forward-looking analysis. This invention, however, uses PowerWorld or PSCAD to build a high-precision grid simulation model, combining power flow calculations and stability analysis to identify potential problems in advance and provide optimization suggestions, thus improving the safety and controllability of grid connection.

[0087] Example 3 Based on the same inventive concept, this invention also provides a system for constructing a multi-dimensional evaluation model for vehicle-to-everything (V2X) interaction, comprising: The evaluation parameter determination module is used to automatically identify various influencing factors within a region and determine regional-level evaluation parameters based on smart grid big data and geographic information data through machine learning algorithms. The multi-dimensional fusion module is used to evaluate the micro-performance of each electric vehicle based on a deep learning-based multi-dimensional data fusion method under the constraints of regional-level evaluation parameters, and obtain a comprehensive evaluation index. Lightweight evaluation models are used to reduce the number of input variables based on comprehensive evaluation indicators through feature selection and dimensionality reduction techniques, while using approximation algorithms to replace complex mathematical calculations, thus constructing lightweight evaluation models. The multi-dimensional building module is used to construct a multi-dimensional evaluation model for vehicle-to-network interaction based on a lightweight evaluation model, combined with resource value information and multi-scenario requirements.

[0088] Optionally, the multi-dimensional fusion module includes: A multi-dimensional data input module is used to receive various evaluation parameters; The adaptive weight allocation module is used to adaptively adjust the weight allocation of various influencing indicators of electric vehicles under the constraints of regional-level evaluation parameters. The data fusion module is used to effectively fuse various evaluation parameters based on deep learning algorithms and the weight allocation of each evaluation parameter to generate a comprehensive evaluation index.

[0089] Optionally, the lightweight evaluation model includes: The feature selection module is used to select key features; The dimensionality reduction module is used to simplify data structures based on key features using dimensionality reduction techniques. The approximation algorithm module is used to replace complex mathematical operations with approximation algorithms.

[0090] Optionally, the multi-dimensional construction module is specifically used for: Based on a lightweight assessment model combined with resource value information, the value assessment of electric vehicles participating in grid interaction is dynamically adjusted. Based on a lightweight evaluation model and combined with the needs of multiple scenarios, the charging and discharging strategies of electric vehicles are dynamically adjusted. A multi-dimensional evaluation model for vehicle-to-grid interaction is constructed based on the adjusted value assessment and charging / discharging strategy.

[0091] Optionally, it also includes: a business maturity analysis system for: The business maturity of electric vehicles participating in grid interaction in different regions and of different types is assessed by combining data analysis and expert scoring. The aforementioned business maturity level can be used as an enable switch for the multi-dimensional evaluation model of vehicle-to-network interaction or as a moderating factor for the comprehensive score.

[0092] Optionally, it also includes: a user feedback system, which records users’ actual experience information on scheduling strategies through a decentralized ledger, and optimizes the multi-dimensional evaluation model of vehicle-to-network interaction based on the actual experience information.

[0093] This embodiment also includes a cross-regional collaborative evaluation system: This system enables data sharing and joint evaluation between different regions, improving the overall efficiency of the power grid's interaction. By leveraging distributed databases and cloud computing technologies, cross-regional data synchronization and joint evaluation can be achieved, improving the overall comprehensiveness and coordination of the evaluation. Specifically, this includes: Data sharing module: Uses blockchain or federated learning technology to achieve cross-regional data sharing; ensures data privacy and security. Distributed Database Module: A distributed database is built using Cassandra or MongoDB to achieve data synchronization and federated queries, improving the globality and coordination of the assessment. Through the design of a cross-regional collaborative assessment system, assessment data from different regions can be shared and jointly analyzed, achieving global scheduling optimization and resource collaborative allocation functions, thereby improving the overall power grid interaction efficiency and coordination. Traditional systems often operate independently in different regions, lacking collaborative mechanisms. This invention, however, achieves cross-regional data sharing through blockchain and federated learning technologies, combined with a distributed database for data synchronization, enhancing the globality and coordination of the assessment model.

[0094] This embodiment also includes a user behavior prediction module: Using big data analytics and behavioral psychology models, predict users' charging and discharging behaviors over a future period of time; By combining deep learning algorithms, proactive scheduling strategies are formulated in advance, improving the flexibility and response speed of the power grid. Through the design of a user behavior prediction module, the system can predict future charging and discharging behaviors of users based on historical behavior data and external variables (such as electricity prices and weather). This not only enables proactive scheduling strategy formulation but also triggers the intelligent scheduling optimization mechanism of the multi-scenario control module, achieving the dual effect of improving the flexibility and response efficiency of power grid scheduling. This module not only provides predictive basis for the scheduling system but also automatically activates the "pre-response" mode in the emergency response mechanism through the output of prediction results, deploying dispatchable vehicle resources in advance and significantly improving the system's predictability and proactivity. Specifically, this includes: I. User Behavior Data Collection and Preprocessing: (1) Collect user historical behavior data including: a. daily charging time, location, and power consumption; b. discharging time, location, and power consumption; c. driving route, dwell time, and destination; d. external influencing factors such as weather, holidays, and electricity prices.

[0095] (2) Data cleaning: a) Remove abnormal behavior (such as long-term parking without charging); b) Complete missing data (such as using linear interpolation or LSTM prediction to complete); c) Standardize the data (such as Z-score normalization).

[0096] II. Behavioral psychology modeling includes: (1) Construct a user behavior model, considering the following behavioral characteristics: a. User travel habits (e.g., early morning departure and late return, commuting, weekend travel); b. User preferences (e.g., preference for charging at night and discharging during peak hours); c. Price sensitivity (whether the user is sensitive to electricity price fluctuations); d. Social attributes (e.g., occupation, place of residence, vehicle type).

[0097] (2) Use psychological models (such as Markov decision process MDP) to model user behavior patterns.

[0098] III. Deep Learning Prediction Models: (1) Use LSTM or Transformer models to predict the user's charging / discharging behavior in the next 24 hours; (2) Input data includes: a) historical charging / discharging time series; b) external variables (such as electricity price, weather, holidays); c) user profile (such as travel patterns, vehicle SOC); (3) The output is the probability distribution of user charging / discharging within the future time window.

[0099] Example of model formula: Let Indicates time The input feature vector (including historical behavior, external variables, and user profile) Indicates time Output behavior ( No charging Charge, (discharge), then the prediction model is:

[0100] in, Nonlinear functions learned by LSTM or Transformer models The length of the historical window (e.g., 7 days).

[0101] IV. Scheduling Strategy Generation: (1) Input the forecast results into the dispatch system; (2) Based on the power grid load forecast, formulate user dispatch strategies; Example strategies: a) If it is predicted that a user will discharge during a certain period, arrange for them to participate in power grid peak shaving in advance; b) If it is predicted that a user will charge during a certain period, arrange for them to do so when the electricity price is low.

[0102] This embodiment also includes an emergency response mechanism: An emergency response mechanism is constructed to rapidly mobilize electric vehicles to participate in grid regulation when power grid emergencies occur. Through a distributed control system and edge computing technology, real-time monitoring and rapid response are achieved, ensuring stable grid operation during emergencies. The design combining the emergency response mechanism with the distributed control system enables the system to quickly mobilize electric vehicles for regulation during power grid emergencies. This not only maintains grid stability in emergency situations but also triggers a real-time value reassessment mechanism in the resource value dynamic assessment module, achieving the additional effect of improving the economy of emergency dispatch and resource utilization. During emergency dispatch, the system automatically calls upon real-time electricity price and carbon emission data to perform value assessment and compensation calculations for vehicles participating in the emergency response, thus achieving synchronous operation of emergency dispatch and incentive mechanisms. Specifically, this includes: I. Real-time monitoring of power grid status, including: (1) using IoT sensors and edge computing nodes to monitor the power grid status in real time; (2) monitoring indicators include: frequency fluctuation, voltage anomaly, load change, and critical node failure.

[0103] II. Emergency Identification and Classification: (1) Define different levels of power grid emergencies: Level 1 Incident (Severe): Main substation failure, frequency below 49.5 Hz; Level 2 Event (Moderate): Sudden drop in regional load, voltage fluctuation exceeding [a certain threshold]. ; Level 3 event (mild): partial power outage, large deviation in load forecast.

[0104] (2) Use rule engines or machine learning models (such as SVM, XGBoost) to identify and classify events.

[0105] III. Rapid response mechanism for electric vehicles, including: (1) Establish an "emergency dispatch pool" that includes electric vehicles with rapid response capabilities; (2) In the event of an emergency, these vehicles should be prioritized for dispatch to participate in power grid regulation. (3) The scheduling methods include: a) rapid discharge to make up for the load gap; b) rapid charging to absorb excess power; c) starting V2G function to participate in frequency regulation.

[0106] IV. Distributed control systems, including: (1) Adopt a distributed control architecture (such as a multi-agent system based on the Consensus algorithm); (2) Each region is equipped with an edge computing node, which has local decision-making capabilities; (3) The central dispatch system is responsible for overall coordination; (4) A hybrid control architecture of “edge computing + central coordination” is realized.

[0107] (5) Example of control model: Let For the first Vehicle control inputs (charging / discharging power). These are local power grid status observations. If the signal is for global coordination, then the distributed control law is as follows:

[0108] in, To control the gain, this model allows each node to converge quickly to a consistent state.

[0109] V. Edge computing support includes: (1) deploying edge computing devices in charging stations or substations; (2) realizing local data processing, decision-making and control; (3) reducing dependence on central servers and improving response speed.

[0110] Example 4 like Figure 3 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0111] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the construction method of a multi-dimensional evaluation model based on vehicle-to-grid interaction in the above embodiments.

[0112] Example 4 Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the method for constructing a multi-dimensional evaluation model based on vehicle-to-grid interaction in the above embodiments.

[0113] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0114] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0115] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0116] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A method for constructing a vehicle-network interaction multi-dimensional evaluation model, characterized in that, The method comprises the following steps: Based on the smart grid big data and geographic information data, the influencing factors in the region are automatically identified by machine learning algorithm, and the evaluation parameters at the regional level are determined; Under the constraint of the evaluation parameters at the regional level, the micro performance of each electric vehicle is evaluated based on the deep learning multi-dimensional data fusion method, and the comprehensive evaluation index is obtained; Based on the comprehensive evaluation index, the number of input variables is reduced by feature selection and dimension reduction technology, and the lightweight evaluation model is constructed by replacing the complex mathematical operation with the approximation algorithm; Based on the lightweight evaluation model, the multi-dimensional evaluation model of vehicle-to-grid interaction is constructed by combining resource value information and multi-scenario demand.

2. The method of claim 1, wherein, The method comprises the following steps: Integrate the location and capacity information of charging facilities in the region in the geographic information data to obtain integrated data; Identify the influencing factors in the region from the integrated data by machine learning algorithm, wherein the influencing factors include construction level and power supply capacity; Take each influencing factor as the evaluation parameter at the regional level.

3. The method of claim 2, wherein, The method comprises the following steps: Under the constraint of the evaluation parameters at the regional level, the weight distribution of each influencing index of the electric vehicle is adjusted adaptively; Based on the deep learning algorithm, the influencing indexes are effectively fused based on the weight distribution of each influencing index to generate the comprehensive evaluation index; The influencing indexes include discharge operation health index, dispatchability index, discharge efficiency index, charging operation health index, power load regulation index and charging efficiency index.

4. The method of claim 1, wherein, The method comprises the following steps: Select key features; Simplify the data structure based on the key features by dimension reduction technology; Replace the complex mathematical operation with the approximation algorithm; Based on the simplified data structure and the approximation algorithm, a lightweight evaluation model is constructed.

5. The method of claim 1, wherein, The method comprises the following steps: Based on the lightweight evaluation model and resource value information, the value evaluation of the electric vehicle participating in the power grid interaction is dynamically adjusted; Based on the lightweight evaluation model and multi-scenario demand, the charging and discharging strategy of the electric vehicle is dynamically adjusted; The multi-dimensional evaluation model of vehicle-to-grid interaction is constructed based on the adjusted value evaluation and charging and discharging strategy.

6. The method of claim 1, wherein, The method further comprises the following steps: By combining data analysis and expert scoring, the business maturity of electric vehicles of different types participating in power grid interaction in different regions is evaluated; The business maturity is used as an enabling switch or a mediation factor of comprehensive score of the multi-dimensional evaluation model of vehicle-to-grid interaction.

7. The method of claim 6, wherein, The method comprises the following steps: The business maturity of electric vehicles of different regions and different types participating in the grid interaction is divided into a high-maturity region, a medium-maturity region and a low-maturity region; In the high-maturity region, the multi-dimensional evaluation model of vehicle-grid interaction is fully enabled and high-frequency scheduling is performed; in the medium-maturity region, the scheduling intensity is limited; and in the low-maturity region, only the monitoring function is enabled.

8. The method of claim 1, wherein, Further comprising: The multi-dimensional evaluation model of vehicle-grid interaction is optimized through the real-time experience information of users fed back to the scheduling strategy. 9.A system for constructing a vehicle-network interaction multi-dimensional evaluation model, characterized in that, Comprising: An evaluation parameter determination module configured to automatically identify various influencing factors in a region and determine evaluation parameters at a regional level based on smart grid big data and geographic information data through a machine learning algorithm; A multi-dimensional fusion module configured to evaluate the micro performance of each electric vehicle based on a multi-dimensional data fusion method of deep learning under the constraint of the evaluation parameters at the regional level, and obtain a comprehensive evaluation index; A lightweight evaluation model configured to reduce the number of input variables based on the comprehensive evaluation index through feature selection and dimension reduction techniques, and replace complex mathematical operations with an approximation algorithm to construct a lightweight evaluation model; A multi-dimensional construction module configured to construct a multi-dimensional evaluation model of vehicle-grid interaction based on the lightweight evaluation model in combination with resource value information and multi-scenario demand.

10. The system of claim 9, wherein, The multi-dimensional fusion module comprises: A multi-dimensional data input module configured to receive various evaluation parameters; An adaptive weight allocation module configured to adaptively adjust the weight allocation of each influencing index of an electric vehicle under the constraint of evaluation parameters at a regional level; A data fusion module configured to effectively fuse each influencing index based on a deep learning algorithm in combination with the weight allocation of each influencing index to generate a comprehensive evaluation index.

11. The system of claim 9, wherein, The lightweight evaluation model comprises: A feature selection module configured to select key features; A dimension reduction module configured to simplify the data structure based on the key features through dimension reduction techniques; An approximation algorithm module configured to replace complex mathematical operations with an approximation algorithm.

12. The system of claim 9, wherein, The multi-dimensional construction module is specifically configured to: Dynamically adjust the value evaluation of an electric vehicle participating in grid interaction based on the lightweight evaluation model in combination with resource value information; Dynamically adjust the charging and discharging strategy of an electric vehicle based on the lightweight evaluation model in combination with multi-scenario demand; Construct a multi-dimensional evaluation model of vehicle-grid interaction from the adjusted value evaluation and charging and discharging strategy.

13. The system of claim 9, wherein, Further comprising: The business maturity analysis system is configured to: Evaluate the business maturity of electric vehicles of different regions and different types participating in the grid interaction through a combination of data analysis and expert scoring; Use the business maturity as an enabling switch or a mediation factor for the comprehensive score of the multi-dimensional evaluation model of vehicle-grid interaction.

14. The system of claim 9, wherein, Further comprising: A user feedback system configured to record the actual experience information of users on the scheduling strategy through a decentralized ledger, and optimize the multi-dimensional evaluation model of vehicle-grid interaction based on the actual experience information.

15. An electronic device, comprising: Comprising: At least one processor and a memory; The memory and the processor are connected through a bus; The memory is configured to store one or more programs; When the one or more programs are executed by the at least one processor, a construction method of a multi-dimensional evaluation model of vehicle-grid interaction is implemented as claimed in any one of claims 1 to 8.

16. A readable storage medium, characterized by, A computer program product, having a computer program stored thereon, which, when executed, implements the method of claim 1 to 8.

Images