Roadside energy management and load coordination control method and system based on large model

Abstract

The application discloses a kind of based on big model's roadside energy management and load coordination control method and system, including obtaining data, constructs roadside energy state vector;Based on roadside energy state vector, obtain load demand prediction result;Roadside energy state vector, load demand prediction result, safety constraint condition and operation and maintenance instruction are input into big model decision module, by big model decision module, scene understanding and reasoning are carried out, and current dispatching scene, load priority sequence, multi-objective weight vector and initial scheduling strategy are output;Initial scheduling strategy is input into constraint optimization module, and by constraint optimization module, with safety constraint condition as target, initial scheduling strategy is corrected and solved to be feasible, and generates joint control scheme;Joint control scheme is issued to corresponding roadside power consumption object and executes.The application can realize the collaborative optimization of multiple roadside loads, improve power supply reliability and energy utilization efficiency, and balance safety and economy.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and energy management technology, specifically to a roadside energy management and load coordination control method and system based on a large model. Background Technology

[0002] With the popularization of new energy vehicles and the advancement of smart road construction, the types and quantities of roadside electrical equipment continue to grow. Typical examples include roadside energy storage devices, charging piles, streetlights and smart lighting equipment, traffic sensing devices, cameras, and communication base stations. These devices exhibit significantly different load characteristics: charging pile loads are highly volatile due to the random arrival and dwell time of vehicles; lighting loads are affected by sunrise and sunset, weather visibility, and traffic flow; and sensing and communication devices require high power supply continuity. Energy storage devices, acting as buffer units, are constrained by factors such as electricity prices, grid constraints, and their own state of charge.

[0003] Existing roadside energy management solutions mostly employ fixed rules or optimization models oriented towards a single load type. For example, lighting equipment is switched on or dimmed at fixed times, energy storage devices are charged and discharged according to peak and off-peak electricity pricing periods, and charging piles adopt a first-come, first-served strategy. Such solutions have significant shortcomings: they cannot incorporate multiple types of roadside loads into a unified framework for collaborative optimization, and it is difficult to adaptively adjust load priorities based on dynamic scenarios such as real-time traffic, weather, electricity prices, and grid status; under complex operating conditions such as peak charging times, grid power rationing, and severe weather, problems such as local optima, decreased power supply reliability, and even overload of distribution facilities are prone to occur; at the same time, the decision-making process lacks interpretability, making it difficult for operation and maintenance personnel to review and manually intervene.

[0004] Therefore, how to achieve scenario-adaptive and collaborative control of multiple types of roadside power loads, while ensuring power supply safety and taking into account both economy and service quality, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology. To achieve the above objectives, a roadside energy management and load coordination control method and system based on a large model is adopted to solve the problems mentioned in the background technology.

[0006] To address the aforementioned technical problems, this invention provides a roadside energy management and load coordination control method based on a large model. It includes: Obtain operational data and external data of roadside power users, and construct a roadside energy state vector; The load demand forecasting results are obtained based on the roadside energy state vector. The roadside energy state vector, the load demand prediction results, the safety constraints, and the operation and maintenance instructions are input into the large model decision module. The large model decision module performs scenario understanding and reasoning, and outputs the current scheduling scenario, load priority sequence, multi-objective weight vector, and initial scheduling strategy. The initial scheduling strategy is input into the constraint optimization module, which aims to satisfy the safety constraints, and performs feasible modification and solution on the initial scheduling strategy to generate a joint control scheme. The joint control scheme is then distributed to the corresponding roadside power users for execution. When the deviation between the execution feedback and the expected execution result corresponding to the joint control scheme exceeds a preset threshold, the roadside energy state vector is reconstructed or updated based on the execution feedback and the deviation, triggering the large model decision module to regenerate the initial scheduling strategy, or adjust the multi-objective weight vector, and update at least one of the prompt template, case library or scheduling parameters of the large model decision module.

[0007] Preferably, the load demand forecast results include forecasts of charging demand, road lighting demand, power supply demand for low-voltage traffic sensing and communication equipment, traffic flow trends, electricity price trends, and renewable energy output.

[0008] Preferably, the large model decision module adopts a generative large language model and generates the scheduling scenario, load priority, target weight and initial scheduling strategy based on system prompt words, roadside energy domain knowledge base or retrieval enhancement generation mechanism.

[0009] Preferably, the constraint optimization module uses the initial scheduling strategy as the initial value or soft constraint. When the solution is not feasible, it relaxes non-critical constraints according to a preset priority or calls back the large model decision module to regenerate the strategy.

[0010] Preferably, the execution feedback includes at least one of the following: the actual power of the key equipment, alarm events, illuminance monitoring values, and charging completion rate; The deviation includes at least one of the following: actual power deviation, illuminance deviation, overload rate, and charging completion rate lower than the planned value; When the deviation exceeds a preset threshold, the large model decision module marks the corresponding device as normal, derated, unavailable, or restricted based on the cause of the deviation, and sends the marking result to the constraint optimization module to update the upper limit of device power, lower limit of lighting, or load priority in the subsequent joint control scheme.

[0011] Preferably, the operation and maintenance instructions include planned maintenance windows, temporary power supply targets, priority roads for major events, and manually set load reduction ratios or equipment shutdown ranges.

[0012] This invention also provides a roadside energy management and load coordination control system based on a large model. The system includes a data acquisition module, a state vector construction module, and a load forecasting module, and further includes: The large model decision module is used to receive the roadside energy state vector, the load demand prediction results, safety constraints and operation and maintenance instructions, and perform scenario understanding and reasoning to output the current scheduling scenario, load priority sequence, multi-objective weight vector and initial scheduling strategy. The constraint optimization module is used to receive the initial scheduling strategy, and with the goal of satisfying the security constraints, to perform feasibility modification and solution on the initial scheduling strategy to generate a joint control scheme. The instruction issuing module is used to issue the joint control scheme to the corresponding roadside power users for execution; The feedback update module is communicatively connected to the large model decision module. It is used to collect the execution feedback of roadside power users and send the execution feedback and deviation information to the large model decision module when the deviation between the execution feedback and the expected deviation of the joint control scheme exceeds a threshold. This triggers rescheduling, adjusts the multi-objective weight vector, or updates the prompt template, case library, or scheduling parameters of the large model decision module.

[0013] Preferably, it includes: The feedback update module includes a feedback acquisition unit, a deviation detection unit, and a strategy update unit. The feedback acquisition unit is used to collect at least one of the following: actual power of key equipment, alarm events, illuminance monitoring values, and charging completion rate. The deviation detection unit is used to determine whether the deviation between the execution feedback and the expected execution result of the joint control scheme exceeds a preset threshold. The strategy update unit is used to send the cause of the deviation, the equipment status marking result, and the updated scheduling parameters to the large model decision module and the constraint optimization module when the deviation exceeds a preset threshold, so that the large model decision module can regenerate the initial scheduling strategy, or the constraint optimization module can update the equipment power limit, load priority, or safety constraints.

[0014] The present invention also provides an electronic device, including a processor and a memory, the memory storing a computer program, wherein the processor executes the computer program to implement the method described in any of the preceding claims.

[0015] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method described in any of the preceding claims.

[0016] Compared with the prior art, the present invention has the following technical advantages: (1) In view of the shortcomings of existing technologies that multiple types of roadside loads cannot be optimized in a coordinated manner, this invention incorporates multiple types of roadside loads such as energy storage, charging, lighting and sensing communication into a unified framework. Through unified analysis of large models and joint solution of constraint optimization, global coordinated optimization is achieved, which alleviates the local optimality problem caused by independent control of single devices and improves the overall operating efficiency of the system.

[0017] (2) In view of the shortcomings of the existing technology in adaptability to complex working conditions, the present invention uses a large model to perform comprehensive semantic understanding of traffic, environment, electricity price, equipment status and operation and maintenance intentions. It can identify complex scenarios such as charging peak, severe weather, and power grid rationing, and dynamically adjust load priority and scheduling strategy, significantly improving the scenario adaptability of the strategy.

[0018] (3) In view of the shortcomings of existing technologies that are difficult to balance safety and economy, this invention couples the target weight output by the large model with the hard constraint verification of the constraint optimization module. While reducing the cost of electricity purchase and peak load, it strictly ensures lighting safety, charging service quality and uninterrupted power supply for key weak electrical equipment, thus achieving a dynamic balance between safety and economy.

[0019] (4) In view of the shortcomings of insufficient interpretability of the decision-making process in the existing technology, the large model decision-making module of the present invention can output scene tags, priority ranking reasons and natural language descriptions related to the decision. Combined with the case library and feedback closed-loop mechanism, it can significantly improve the interpretability and maintainability of the decision-making process, and facilitate manual review and intervention.

[0020] (5) In view of the lack of adaptive evolution capability in the existing technology, the present invention enables the system to continuously learn from the operating data, dynamically correct strategy preferences, and maintain long-term adaptability to changes in equipment characteristics and load mode drift by executing a closed-loop mechanism of feedback acquisition, deviation detection and model configuration update. Attached Figure Description

[0021] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings: Figure 1 This is a schematic diagram illustrating the steps of the coordination control method according to an embodiment of this application; Figure 2 A detailed flowchart of the coordination control method according to an embodiment of this application is provided. Figure 3 This is a flowchart of a roadside energy management and load coordination control method according to an embodiment of this application; Figure 4 This is a schematic diagram of the roadside energy management and load coordination control system architecture according to an embodiment of this application. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1 This embodiment uses an energy management scenario on a main road in a city as an example. This scenario is for illustrative purposes only and does not constitute a limitation on the scope of protection of this invention.

[0024] Please refer to Figure 1 and Figure 2 In this embodiment of the invention, a roadside energy management and load coordination control method based on a large model includes the following steps: Step S1: Obtain the operating data and external data of roadside power users, and construct the roadside energy state vector; Step S2: Obtain the load demand prediction result based on the roadside energy state vector; Step S3: Input the roadside energy state vector, the load demand prediction result, the safety constraints and the operation and maintenance instructions into the large model decision module. The large model decision module performs scenario understanding and reasoning, and outputs the current scheduling scenario, load priority sequence, multi-objective weight vector and initial scheduling strategy. Step S4: Input the initial scheduling strategy into the constraint optimization module. The constraint optimization module aims to satisfy the safety constraints and performs feasible modification and solution on the initial scheduling strategy to generate a joint control scheme. like Figure 3 As shown, the diagram is a flowchart of the roadside energy management and load coordination control method; In a specific implementation, the constraint optimization module generates a joint control scheme using model predictive control. Specifically, in each control cycle, the system constructs a rolling optimization problem based on the current roadside energy state vector, the load demand forecast results within a preset future time window, and the load priority sequence and multi-objective weight vector output by the large model decision module.

[0025] The control variables of the rolling optimization problem include one of the following: energy storage charging and discharging power, charging pile power allocation value, street light dimming ratio, and power supply level of low-voltage traffic sensing and communication equipment. The optimization objectives include one or more of the following: reducing electricity purchase costs, reducing peak load in the distribution area, reducing charging wait time or charging delay, reducing illuminance deviation, maintaining the state of charge of energy storage within the target range, and reducing frequent switching of control strategies.

[0026] The model predictive control method solves for the joint control sequence of multiple future time periods within each control cycle, but only issues and executes the control variables corresponding to the current control cycle. In the next control cycle, the system re-collects execution feedback and external data, updates the roadside energy state vector, and performs rolling optimization again. Thus, dynamic coordinated control of multiple types of roadside loads is achieved under continuously changing traffic flow, electricity prices, weather, and equipment status.

[0027] When the rolling optimization problem is not feasible, the constraint optimization module relaxes the service quality constraints of non-critical loads according to the constraint relaxation priority output by the large model decision module, while keeping critical constraints such as lighting safety, power supply continuity of critical sensing equipment, power distribution capacity limit and energy storage safety boundary intact; if it is still not feasible, the large model decision module is called back to regenerate the load priority sequence and the initial scheduling strategy.

[0028] In a specific implementation, the constraint optimization module employs a robust optimization approach to handle uncertainties in load forecasting and external environment forecasting. Specifically, the system constructs uncertainty sets for charging demand, road lighting demand, traffic flow, renewable energy output, and electricity prices based on historical forecast errors, real-time sampled data, and the fluctuation range of external data.

[0029] The uncertainty set may include an interval-type uncertainty set, a scenario-type uncertainty set, or a confidence interval set based on historical error distribution. When generating a joint control scheme, the constraint optimization module ensures that key safety constraints are met in any possible scenario within the uncertainty set. These key safety constraints include one or more of the following: total load of the transformer area does not exceed the capacity limit, energy storage state of charge is within a safe range, road lighting is not lower than the safe illuminance limit, and key traffic sensing and communication equipment maintains continuous power supply.

[0030] Under this robust optimization approach, the constrained optimization module aims to minimize operating costs, load reduction, or service quality loss under the most unfavorable scenario, generating a joint control scheme with safety margins. When the system determines that the prediction uncertainty exceeds a preset threshold, the large model decision module increases the weight of the power supply reliability target and decreases the weight of the economic target, and instructs the constrained optimization module to reserve energy storage capacity or reduce the power of non-critical charging loads to improve the system's operational stability under scenarios such as sudden increases in traffic flow, severe weather, or grid power rationing.

[0031] After execution, the system corrects the boundary or scenario weights of the uncertainty set based on the actual feedback results, and writes the correction results into the case library or scheduling parameters for retrieval and optimization in subsequent control cycles.

[0032] Step S5: Distribute the joint control scheme to the corresponding roadside power users for execution; Step S6: When the deviation between the execution feedback and the expected execution result corresponding to the joint control scheme exceeds a preset threshold, the roadside energy state vector is reconstructed or updated based on the execution feedback and the deviation, triggering the large model decision module to regenerate the initial scheduling strategy, or adjust the multi-objective weight vector, and update one or more of the prompt template, case library or scheduling parameters of the large model decision module.

[0033] After the joint control scheme is implemented, the system also collects the execution feedback from roadside power users. When the deviation between the execution feedback and the expected result exceeds the threshold, it triggers rescheduling or updates the prompt template, case library or scheduling parameters of the large model decision module, thereby forming a closed-loop self-optimizing control mechanism.

[0034] In this embodiment, the load demand forecast results include forecasts of charging demand, road lighting demand, power supply demand for low-voltage traffic sensing and communication equipment, traffic flow trends, electricity price trends, and renewable energy output, providing sufficient forward-looking information for decision-making.

[0035] In this embodiment, the large model decision module adopts a generative large language model and generates the scheduling scenario, load priority, target weight and initial scheduling strategy based on system prompt words, roadside energy domain knowledge base or retrieval enhancement generation mechanism.

[0036] In this embodiment, the constraint optimization module uses the initial scheduling strategy as the initial value or soft constraint. When the solution is not feasible, it relaxes non-critical constraints according to a preset priority or calls back the large model decision module to regenerate the strategy.

[0037] This embodiment also includes: Collect execution feedback from roadside power users, including the actual power of key equipment, alarm events, and illuminance monitoring values; When the deviation between the execution feedback and the expected outcome of the joint control scheme exceeds a threshold, a rescheduling or adjustment of the target weight is triggered, and one or more of the prompt template, case library, or scheduling parameters of the large model decision module are updated accordingly.

[0038] In this embodiment, the operation and maintenance instructions include planned maintenance windows, temporary power supply objects, priority roads for major events, and manually set load reduction ratios or equipment shutdown ranges, thereby seamlessly integrating human management intentions into the automatic decision-making process.

[0039] Example 2 This embodiment, based on the technical solution described in Embodiment 1, further adds a closed-loop feedback and self-optimization mechanism, as well as a specific implementation method for integrating operation and maintenance instructions. Unlike Embodiment 1, which focuses on the basic method and process, this embodiment focuses on demonstrating how the system continuously optimizes decision-making strategies through execution feedback during operation, and how to seamlessly integrate manual operation and maintenance intentions into the automatic scheduling process.

[0040] The application scenario of this embodiment is a section of expressway ramps in a city. The load characteristics of this section are different from those of the main road in Embodiment 1: the traffic flow exhibits a significant tidal phenomenon, with the morning and evening peaks occurring in opposite directions; the accident risk at the ramp intersection is higher, placing more stringent requirements on the reliability of lighting and the continuity of power supply to traffic sensing equipment; in addition, this section frequently undertakes major event support tasks, requiring frequent responses to temporary power supply commands.

[0041] Step S1: Data Acquisition and State Vector Construction The data acquisition scope and preprocessing method of this embodiment are basically the same as those of Embodiment 1, but the following data sources are added for the expressway scenario: ramp inflow traffic detector data, road surface slippery state sensor data, and temporary control command interface from the traffic management platform.

[0042] The state vector is expanded based on Example 1 by adding elements such as ramp flow rate and road surface adhesion coefficient to characterize the special working conditions of ramp intersection sections.

[0043] Step S2: Load demand forecasting The specific implementation method for load demand forecasting is similar to that of Example 1, but the forecasting model incorporates consideration of ramp merging traffic. When the predicted ramp traffic exceeds a threshold (e.g., 800 vehicles per hour per lane), the system automatically raises the dimming lower limit in the lighting demand forecast by 10 percentage points to address higher traffic safety risks.

[0044] Step S3: Fusion of large-scale model decision-making and operation and maintenance instructions This embodiment focuses on demonstrating the specific integration method of operation and maintenance commands. Operation and maintenance commands are input through the human-machine interface (HMI) or API interface of the operation and maintenance management system. Typical types include: (1) Scheduled maintenance window: The maintenance personnel designate a certain charging pile to be in a maintenance power-off state during a specific period. When constructing the state vector, the system marks the device as "unavailable" and sets its power limit to zero in the constraint optimization module, thus removing the device from the load forecast and constraint conditions.

[0045] (2) Temporary power supply objects: Due to the need for ensuring the safety of major events, the operation and maintenance personnel designate this section of the expressway ramp as a temporary power supply object. After receiving such instructions, the large model decision module sets the lighting equipment and traffic sensing equipment of this section to uninterrupted power supply when generating the load priority sequence, and maintains the lighting brightness at more than 70%.

[0046] (3) Load reduction ratio: When the power grid company issues an orderly power consumption notice, the operation and maintenance personnel can directly set the upper limit of the total load of the transformer area through the system (such as reducing it to 75% of the rated capacity). The constraint optimization module will enforce this as a hard constraint, and the large model decision module will re-prioritize the load under this constraint, giving priority to communication and security loads.

[0047] (4) Equipment shutdown scope: Maintenance personnel can specify the shutdown of non-critical equipment (such as landscape lighting, advertising light boxes, etc.) in a certain area to free up capacity to ensure core load. The system converts this instruction into a constraint that the power limit of the corresponding equipment is zero.

[0048] The key to the integration of operation and maintenance instructions lies in the fact that when the large model decision module generates the initial scheduling strategy, it takes the operation and maintenance instructions as a prerequisite for priority fulfillment, and performs other load optimization scheduling on the basis of meeting the instruction requirements, thereby seamlessly combining human management intentions with automated decision-making.

[0049] Step S4: Constraint Optimization Solution The specific implementation of the constraint optimization module is the same as in Example 1. However, in this example, when the operation and maintenance instruction includes a load reduction ratio, the total load limit of the transformer area is added as a new hard constraint to the optimization model.

[0050] Step S5: Issuance of Joint Control Scheme The method for issuing the joint control scheme is the same as in Example 1.

[0051] Step S6: Perform feedback collection and closed-loop self-optimization After the joint control scheme is issued and executed, the system continuously collects execution feedback data from roadside power users, specifically including: The actual power of key equipment is sampled by smart meters at 15-second intervals and compared with the command value. Alarm events include: energy storage over-temperature alarm, charging pile communication interruption alarm, transformer overload alarm, etc. Illuminance monitoring values: collected by illuminance sensors deployed on the road surface, with a sampling period of 30 seconds; Charging completion rate: The ratio of the actual charging amount completed to the planned charging amount for each charging station.

[0052] The system is configured with a deviation detection mechanism that triggers rescheduling when any of the following conditions are detected: The deviation between the actual discharge power of the energy storage and the commanded value continues to exceed 10% for more than 2 minutes; The street light illuminance monitoring value is below the safety lower limit (30 lux) for more than 30 seconds; An emergency alarm is triggered when the transformer load rate exceeds 95%. The charging completion rate has been below 80% for three consecutive control cycles.

[0053] After a rescheduling is triggered, the system re-inputs the current actual state vector and deviation information into the large model decision module. The large model determines the cause of the deviation and takes corresponding measures: If the deviation is caused by equipment failure (such as a charging pile's actual power failing to reach the commanded value), the equipment will be marked as "derated operation", the power limit parameter in the subsequent scheduling strategy will be updated, and an alarm will be generated to notify the operation and maintenance personnel. If the deviation is caused by inaccurate prediction (such as actual traffic flow far exceeding the predicted value), the parameter weights in the load prediction model will be adjusted, and the prompt template of the large model will be updated to add decision prompts for this type of scenario. If the deviation is caused by a sudden external event (such as a severe weather attack), similar historical cases will be retrieved from the case library, and the historical response strategies will be referenced for rapid correction.

[0054] The closed-loop update mechanism specifically includes: The prompt template has been updated: The scenario characteristics and coping strategies for this deviation event have been added to the "Precautions" section of the system prompt in the form of natural language descriptions; Case library update: The state vector snapshot, decision-making strategy, execution result and deviation reason of this event will be stored as a complete case record in the vector database of the roadside energy knowledge base for subsequent RAG retrieval reference; Scheduling parameter update: If the same type of deviation event occurs multiple times in a row, the target weight corresponding to the event will be automatically adjusted (e.g., the weight of increasing power supply reliability and the weight of reducing economic efficiency).

[0055] For example, the effect of a closed-loop update is as follows: the system detects that the actual power of a charging pile cannot reach the command value within three consecutive control cycles, and diagnosis reveals that the vehicle-side BMS actively limits the power. The system automatically marks the charging pile as "restricted" and adjusts its power limit from 120kW to 80kW in the subsequent constraint optimization model, avoiding the constraint optimization module repeatedly issuing unexecutable commands.

[0056] Example 3 The present invention also provides a control system employing a roadside energy management and load coordination control method based on a large model as described above, the system comprising a data acquisition module, a state vector construction module, a load forecasting module, and further comprising: The large model decision module is used to receive the roadside energy state vector, the load demand prediction results, safety constraints and operation and maintenance instructions, and perform scenario understanding and reasoning to output the current scheduling scenario, load priority sequence, multi-objective weight vector and initial scheduling strategy. The constraint optimization module is used to receive the initial scheduling strategy, and with the goal of satisfying the security constraints, to perform feasibility modification and solution on the initial scheduling strategy to generate a joint control scheme. The instruction issuing module is used to issue the joint control scheme to the corresponding roadside power users for execution.

[0057] In this embodiment, a feedback update module is also included, which is used to collect the execution feedback of the roadside power users, and when the deviation between the execution feedback and the expected value of the joint control scheme exceeds a threshold, trigger rescheduling or adjust the target weight, and update the configuration of the large model decision module accordingly.

[0058] like Figure 4 As shown in the figure, this is a schematic diagram of the roadside energy management and load coordination control system architecture; This embodiment provides a specific architecture for a control system that implements the above method. The roadside energy management and load coordination control system can be deployed as a cloud-edge collaborative architecture.

[0059] The edge side includes roadside gateways or edge computing units, which are responsible for aggregating data from charging pile controllers, lighting concentrators, energy storage converters, smart power distribution terminals and IoT sensing terminals, and completing necessary real-time preprocessing and local security control.

[0060] Deploy load forecasting services, large model decision services, constrained optimization solvers, configuration and case libraries, and interface modules for connecting with power grid dispatching or traffic management platforms in the cloud or regional center.

[0061] Communication can be achieved through wired private networks, cellular networks, or vehicle-road cooperative networks. Redundant links and time synchronization mechanisms should be configured for critical commands and telemetry.

[0062] The present invention also provides an electronic device, including a processor and a memory, the memory storing a computer program, wherein the processor executes the computer program to implement the method described in any of the preceding claims.

[0063] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method described in any of the preceding claims.

[0064] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention. The scope of the invention is defined by the appended claims and their equivalents, all of which should be included within the scope of protection of the invention.

Claims

1. A roadside energy management and load coordination control method based on a large model, characterized in that, Includes the following steps: Obtain operational data and external data of roadside power users, and construct a roadside energy state vector; The load demand forecasting results are obtained based on the roadside energy state vector. The roadside energy state vector, the load demand prediction results, the safety constraints, and the operation and maintenance instructions are input into the large model decision module. The large model decision module performs scenario understanding and reasoning, and outputs the current scheduling scenario, load priority sequence, multi-objective weight vector, and initial scheduling strategy. The initial scheduling strategy is input into the constraint optimization module, which aims to satisfy the safety constraints, and performs feasible modification and solution on the initial scheduling strategy to generate a joint control scheme. The joint control scheme is then distributed to the corresponding roadside power users for execution. When the deviation between the execution feedback and the expected execution result corresponding to the joint control scheme exceeds a preset threshold, the roadside energy state vector is reconstructed or updated based on the execution feedback and the deviation, triggering the large model decision module to regenerate the initial scheduling strategy, or adjust the multi-objective weight vector, and update at least one of the prompt template, case library or scheduling parameters of the large model decision module.

2. The method according to claim 1, characterized in that, The load demand forecast results include forecasts of charging demand, road lighting demand, power supply demand for low-voltage traffic sensing and communication equipment, traffic flow trends, electricity price trends, and renewable energy output.

3. The method according to claim 1, characterized in that, The large model decision module adopts a generative large language model and generates the scheduling scenario, load priority, target weight and initial scheduling strategy based on system prompt words, roadside energy domain knowledge base or retrieval enhancement generation mechanism.

4. The method according to claim 1, characterized in that, The constraint optimization module uses the initial scheduling strategy as the initial value or soft constraint. When the solution is not feasible, it relaxes non-critical constraints according to a preset priority or calls back the large model decision module to regenerate the strategy.

5. The method according to claim 1, characterized in that, The execution feedback includes at least one of the following: the actual power of the key equipment, alarm events, illuminance monitoring values, and charging completion rate; The deviation includes at least one of the following: actual power deviation, illuminance deviation, overload rate, and charging completion rate lower than the planned value; When the deviation exceeds a preset threshold, the large model decision module marks the corresponding device as normal, derated, unavailable, or restricted based on the cause of the deviation, and sends the marking result to the constraint optimization module to update the upper limit of device power, lower limit of lighting, or load priority in the subsequent joint control scheme.

6. The method according to claim 1, characterized in that, The operation and maintenance instructions include planned maintenance windows, temporary power supply targets, priority roads for major events, and manually set load reduction ratios or equipment shutdown ranges.

7. A roadside energy management and load coordination control system based on a large model, the system comprising a data acquisition module, a state vector construction module, and a load forecasting module, characterized in that, Also includes: The large model decision module is used to receive the roadside energy state vector, the load demand prediction results, safety constraints and operation and maintenance instructions, and perform scenario understanding and reasoning to output the current scheduling scenario, load priority sequence, multi-objective weight vector and initial scheduling strategy. The constraint optimization module is used to receive the initial scheduling strategy, and with the goal of satisfying the security constraints, to perform feasibility modification and solution on the initial scheduling strategy to generate a joint control scheme. The instruction issuing module is used to issue the joint control scheme to the corresponding roadside power users for execution; The feedback update module is communicatively connected to the large model decision module. It is used to collect the execution feedback of roadside power users and send the execution feedback and deviation information to the large model decision module when the deviation between the execution feedback and the expected deviation of the joint control scheme exceeds a threshold. This triggers rescheduling, adjusts the multi-objective weight vector, or updates the prompt template, case library, or scheduling parameters of the large model decision module.

8. The system according to claim 7, characterized in that, include: The feedback update module includes a feedback acquisition unit, a deviation detection unit, and a strategy update unit. The feedback acquisition unit is used to collect at least one of the following: actual power of key equipment, alarm events, illuminance monitoring values, and charging completion rate. The deviation detection unit is used to determine whether the deviation between the execution feedback and the expected execution result of the joint control scheme exceeds a preset threshold. The strategy update unit is used to send the cause of the deviation, the equipment status marking result, and the updated scheduling parameters to the large model decision module and the constraint optimization module when the deviation exceeds a preset threshold, so that the large model decision module can regenerate the initial scheduling strategy, or the constraint optimization module can update the equipment power limit, load priority, or safety constraints.

9. An electronic device comprising a processor and a memory, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method according to any one of claims 1 to 6.

Images