Multi-state mobile thermal storage vehicle scheduling method based on central collision and strategy feedback

Abstract

The application discloses a kind of multi-state mobile heat storage car scheduling methods based on central collision and strategy feedback, mobile heat storage car multi-state characteristics are throughout the whole process of scheduling modeling, candidate scheme generation and scheme optimization, obtain optimal scheduling scheme satisfying heating demand, vehicle operation constraint and comprehensive evaluation index.The method is based on scheduling basic data to construct mobile heat storage car model, construct mobile heat storage car object feature library and to be optimized scheduling model library;Based on the multi-state object features of mobile heat storage car, central collision optimization is carried out, object feature embedded coding, fitness evaluation, collision center determination and double space update are completed, and a candidate scheme set is generated;Based on the multi-state heat object features, Langevin evolution screening is carried out, the feasibility of the candidate scheme is verified, strategy feedback and evolution environment selection are carried out, and finally the optimal scheduling scheme is output.The application improves the executability and adaptability of mobile heat storage car scheduling scheme.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system optimization scheduling and mobile thermal storage control technology, and to a multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback. Background Technology

[0002] With the rapid development of new power systems and integrated energy systems, heat source coordination, heat load matching, and spatiotemporal transfer of heat energy in district heating systems have become important technological directions for improving energy utilization efficiency and ensuring heating reliability. Mobile thermal storage vehicles can transfer heat between heat sources such as biomass power plants, industrial waste heat sources, and centralized heating stations and decentralized users, realizing the mobile storage, transportation, and release of heat energy. This helps alleviate the mismatch between heat sources and users in terms of spatial location, heating time, and heat quality requirements. Introducing mobile thermal storage vehicles into integrated energy systems and district heating dispatching helps improve the operational flexibility of heating systems, enhance the utilization efficiency of waste heat or renewable heat energy, and promote low-carbon operation of the heating process.

[0003] Existing mobile thermal storage vehicle scheduling methods often simplify mobile thermal storage vehicles into scheduling units with fixed thermal storage capacity, fixed charging and discharging efficiencies, and route transportation times. They focus primarily on factors such as vehicle routes, charging and discharging power, and user heating sequence, failing to fully reflect the multi-state thermal characteristics of mobile thermal storage vehicles in actual operation. During scheduling, mobile thermal storage vehicles are influenced by factors such as multi-compartment thermal storage structure, compartment temperature and heat storage capacity, thermal grade status, travel heat loss, thermal attenuation, interface matching, route time windows, and vehicle task status. If these characteristics are not incorporated into the candidate solution generation, fitness evaluation, constraint correction, and screening feedback processes, deviations between the scheduling scheme and the actual executable state of the mobile thermal storage vehicle can easily occur. Therefore, constructing a mobile thermal storage vehicle scheduling model oriented towards the multi-state thermal characteristics, and embedding these characteristics into the optimization solution and strategy feedback screening process, has significant engineering application value. Summary of the Invention

[0004] The purpose of this invention is to overcome the problems in existing mobile thermal storage vehicle scheduling methods, which simplify mobile thermal storage vehicles into ordinary transportation energy storage units, fail to fully reflect the object characteristics such as compartmentalized heat storage, heat quality changes, driving heat loss, interface matching, path time window and task status, and lack integration with the multi-state object characteristics of mobile thermal storage vehicles in the candidate scheme generation, optimization update and final screening process. This invention provides a multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback.

[0005] The objective of this invention is achieved by the following technical measures: A multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback is proposed, comprising the following steps: S1: Construct a multi-state mobile thermal storage vehicle model based on scheduling basic data, determine the multi-compartment thermal storage structure, compartment thermal storage status, heat quality status, driving heat loss status, thermal storage attenuation status, interface matching status, and vehicle task status of the mobile thermal storage vehicle, and form a mobile thermal storage vehicle object feature library; construct a scheduling model library to be optimized based on the thermal balance of biomass power plant-mobile thermal storage vehicle-user. S2: Based on the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized formed in S1, read the parameters of the mobile thermal storage vehicle object feature library and the CCO algorithm parameters. Based on the mobile thermal storage vehicle object feature library, perform variable encoding on the scheduling model library to be optimized to generate an initial candidate population. Perform candidate scheme decoding, multi-state fitness evaluation and collision center determination on the initial candidate population. Based on the mobile thermal storage vehicle object feature constraint mapping and dual-space collision update, generate a dual-space candidate scheme set. S3: Based on the dual-space candidate scheme set output by S2, the feasibility of the mobile thermal storage vehicle object features is verified. After judgment and linkage correction, candidate schemes are obtained, and the object feature states of the candidate schemes are collected. Based on the object feature states, Langevin policy feedback is performed to determine adaptive screening parameters and update the critic feedback and flexible actor network. Based on the critic feedback and flexible actor network, the object feature evaluation matrix and evolutionary environment selection are performed. After local quality and crowding sorting, the evolutionary environment selection is updated to obtain the filtered candidate set. After adaptive normalization and trade-off selection, the termination condition is judged. If it is not satisfied, the adaptive neighborhood analysis is returned. If it is satisfied, the optimal scheduling scheme is output.

[0006] Furthermore, step S1 specifically includes the following steps: S1.1: Collect basic scheduling data, which includes biomass power plant heating boundary parameters, mobile thermal storage vehicle structure and interface parameters, compartment temperature and heat storage parameters, vehicle route and travel time parameters, thermal storage material and insulation parameters, and user heat load and heat grade requirements; wherein, the biomass power plant heating boundary parameters include fuel calorific value, fuel supply, boiler steam parameters, cogeneration unit output boundary, and available heat power boundary; the mobile thermal storage vehicle structure and interface parameters include the number of compartments in the mobile thermal storage vehicle, the capacity of each compartment, and the charging and discharging interfaces. Location, interface power boundaries, and interface temperature range; the compartment temperature and heat storage parameters include the initial temperature of each compartment, initial heat storage, maximum heat storage, minimum usable temperature, and maximum safe temperature; the vehicle route and travel time parameters include the power plant location, user location, road distance, estimated travel time, service sequence, and waiting time; the heat storage material and insulation parameters include heat storage medium parameters, insulation layer parameters, and environmental heat exchange parameters; the user heat load and heat grade requirements include the user's time-of-use heat load, required temperature level, service window, and allowable heat shortage range; S1.2: Based on the structural and interface parameters of the mobile thermal storage vehicle, the compartment temperature and heat storage parameters, the vehicle path and travel time parameters, and the thermal storage material and insulation parameters, a multi-state mobile thermal storage vehicle model is constructed. The multi-state mobile thermal storage vehicle model includes a thermal storage state line and a transportation loss line. The thermal storage state line is used to sequentially divide the multi-compartment thermal storage structure, determine the compartment temperature and heat storage capacity, and identify the thermal grade status. The transportation loss line is used to sequentially read the vehicle path and travel time, calculate the heat loss during travel, and calculate the remaining heat after arrival. Based on the thermal grade status identification results and the remaining heat calculation results after arrival, the thermal storage attenuation status is updated, and the charging and discharging interface is matched and the vehicle task status is updated, forming a mobile thermal storage vehicle object feature library. The mobile thermal storage vehicle object feature library is a data set used to characterize the compartment thermal storage status, thermal grade status, heat loss status, interface matching status, and vehicle task status of the mobile thermal storage vehicle. S1.3: Construct a biomass power plant heating output model based on the heating boundary parameters of the biomass power plant to determine the time-of-use heat supply and heating power boundaries within the scheduling cycle; construct a user heat load demand model based on user heat load and heat grade demand to determine user time-of-use heat load, heat grade demand, and service windows; and construct a biomass power plant-mobile heat storage vehicle-user heat balance based on the mobile thermal storage vehicle object feature library, biomass power plant heating output, and user heat load demand. S1.4: Based on the thermal balance of biomass power plant-mobile thermal storage vehicle-user, construct heat loss constraints, path and service window constraints, interface security constraints, and vehicle task timing constraints to form a scheduling model library to be optimized; the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized together serve as inputs for subsequent mobile thermal storage vehicle object feature embedded center collision optimization.

[0007] Furthermore, step S2 specifically includes the following steps: S2.1: Read the feature library of the mobile thermal storage vehicle object formed in S1.2, and read the center collision optimization algorithm parameters, i.e., CCO algorithm parameters; the center collision optimization algorithm parameters include population size, maximum number of iterations, collision center selection ratio, collision update step size and perturbation decay coefficient; wherein, the population size is used to determine the number of initial candidate schemes, the maximum number of iterations is used to limit the number of center collision update cycles, the collision center selection ratio is used to determine the center individual from the comprehensive fitness ranking results, and the collision update step size and perturbation decay coefficient are used to control the change range of candidate schemes in the dual-space update process; S2.2: Based on the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized, embedded variable encoding of mobile thermal storage vehicle object features is performed. The embedded variable encoding of mobile thermal storage vehicle object features includes compartment thermal storage state encoding, driving heat loss correction encoding, thermal quality service matching encoding, and path-time window-charge and release joint encoding. Among them, compartment thermal storage state encoding is used to characterize the temperature, heat storage capacity, and capacity boundary of each compartment; driving heat loss correction encoding is used to characterize the heat attenuation during vehicle transportation; thermal quality service matching encoding is used to characterize the matching relationship between the thermal quality of the mobile thermal storage vehicle and the thermal quality required by the user; path-time window-charge and release joint encoding is used to characterize the coupling relationship between vehicle path, service sequence, arrival time, charging capacity, and releasing capacity; an initial candidate population is generated based on the encoding results and the population size. S2.3: Decode the candidate schemes of the initial candidate population to obtain candidate mobile thermal storage vehicle scheduling schemes; perform multi-state fitness evaluation based on the candidate mobile thermal storage vehicle scheduling schemes, including heat loss fitness calculation, heat grade matching fitness calculation, compartment capacity utilization fitness calculation, path time window executability calculation, and interface constraint violation calculation; perform comprehensive fitness ranking based on the multi-state fitness evaluation results, and determine the collision center and ranking samples according to the collision center selection ratio; S2.4: Based on the collision center, sorted samples, and center collision optimization algorithm parameters, perform feature constraint mapping and dual-space collision update for the mobile thermal storage vehicle object; the dual-space collision update includes original spatial collision update and decorrelational spatial variable transformation, wherein the original spatial collision update is used to update the compartment thermal storage state, thermal quality service matching relationship, driving heat loss correction amount, path time window, and charging and discharging heat variables, and the decorrelational spatial variable transformation is used to reduce the coupling influence between the above variables; after each round of collision update, perform compartment thermal state boundary correction, thermal quality matching relationship correction, heat loss compensation update, and interface path constraint mapping correction until the preset iteration conditions are reached, and output the dual-space candidate scheme set.

[0008] Furthermore, step S3 specifically includes the following steps: S3.1: Read the dual-space candidate scheme set output by S2.4, and perform feasibility verification of the mobile thermal storage vehicle object features on the dual-space candidate scheme set; the feasibility verification of the mobile thermal storage vehicle object features includes heat grade feasibility judgment, compartment thermal storage boundary judgment, heat supply satisfaction judgment after heat loss judgment, and interface and path time window feasibility judgment; when the candidate scheme meets the above judgment conditions, the candidate scheme is archived as a feasible candidate scheme, and an object feature status aggregation result is formed; S3.2: When a candidate solution does not meet at least one of the preset constraints of heat quality, compartment heat storage, heat loss, interface matching, or path window, an infeasible solution linkage correction is performed. The infeasible solution linkage correction includes adjusting the service vehicle, service sequence, and compartment heat supply allocation according to heat quality requirements; adjusting the charging and releasing heat according to the compartment heat storage boundary; adjusting the path, arrival time, or service window according to the remaining heat after heat loss; and adjusting the charging and releasing heat interface matching relationship according to interface constraints. If the preset constraints are still not fully met after correction, the corresponding constraint violation amount is recorded and the constraint violation amount is input into the subsequent object feature evaluation matrix. S3.3: Based on the aggregated results of object feature states, calculate the thermal grade deviation state, thermal loss sensitive state, compartmentalized heat storage utilization state, and interface and time window conflict state, and input the above states into a state vector into the flexible actor network; output the strategy action through the Langevin perturbation action generation process to determine the adaptive screening parameters, which include adaptive neighborhood scale, search perturbation intensity, and candidate solution retention ratio; construct the critic value assessment and reward feedback based on the evolutionary screening results, and update the flexible actor network parameters according to the reward feedback, so that the subsequent screening parameters are adjusted as the candidate solution state changes; S3.4: Based on adaptive screening parameters, adaptive neighborhood analysis, local quality and congestion ranking, and evolutionary environment selection updates are performed to form a set of candidate solutions after screening. An object feature evaluation matrix is ​​constructed based on the set of candidate solutions. This object feature evaluation matrix includes at least operating costs, carbon emissions, heating satisfaction, vehicle utilization, heat quality matching degree, heat loss level, and constraint violation. The object feature evaluation matrix undergoes adaptive normalization and trade-offs, and it is determined whether a preset termination condition is met. If the preset termination condition is not met, the process returns to the evolutionary environment selection process to continue screening, and feedback is provided to S2.2 for candidate population updates when necessary. When the preset termination condition is met, this candidate solution is output as the optimal scheduling solution. The optimal scheduling solution includes the biomass power plant time-sharing heating arrangement, the mobile thermal storage vehicle charging and discharging heat arrangement, vehicle routes and arrival times, user service order, and corresponding evaluation index results.

[0009] This invention constructs a multi-state mobile thermal storage vehicle model, incorporating the vehicle's compartmentalized heat storage, heat grade, heat loss, interface matching, path window, and task status into the scheduling modeling process, thus preventing the mobile thermal storage vehicle from being simplified into a single-capacity transportation unit. Through embedded center collision optimization of mobile thermal storage vehicle object features, these features are integrated into variable encoding, fitness evaluation, collision center determination, and dual-space collision update processes, ensuring that the candidate solution generation process corresponds to the actual operating state of the mobile thermal storage vehicle. Using Langevin evolutionary screening driven by multi-state thermal object features, screening parameters are adaptively adjusted based on heat grade deviation, heat loss sensitivity, compartmentalized heat storage utilization, and interface window conflict status, ensuring that the final mobile thermal storage vehicle scheduling scheme simultaneously considers operating costs, carbon emissions, heating satisfaction, vehicle utilization, heat grade matching, heat loss level, and constraint violation. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating the overall process structure of the present invention. Detailed Implementation

[0011] The following is a detailed description of the embodiments of the present invention with reference to the accompanying drawings. The purpose of the present invention is to provide a multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback.

[0012] like Figure 1 As shown, the multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback proceeds in the following steps: S1: Construct a multi-state mobile thermal storage vehicle model based on scheduling foundation data, determining the multi-compartment thermal storage structure, compartment thermal storage status, heat quality status, driving heat loss status, thermal storage attenuation status, interface matching status, and vehicle task status of the mobile thermal storage vehicle, forming a mobile thermal storage vehicle object feature library; construct a scheduling model library to be optimized based on the thermal balance of biomass power plant-mobile thermal storage vehicle-user; the specific steps are as follows: S1.1: Collect basic scheduling data, which includes biomass power plant heating boundary parameters, mobile thermal storage vehicle structure and interface parameters, compartment temperature and heat storage parameters, vehicle route and travel time parameters, thermal storage material and insulation parameters, and user heat load and heat grade requirements; wherein, the biomass power plant heating boundary parameters include fuel calorific value, fuel supply, boiler steam parameters, cogeneration unit output boundary, and available heat power boundary; the mobile thermal storage vehicle structure and interface parameters include the number of compartments in the mobile thermal storage vehicle, the capacity of each compartment, and the charging and discharging interfaces. Location, interface power boundaries, and interface temperature range; the compartment temperature and heat storage parameters include the initial temperature of each compartment, initial heat storage, maximum heat storage, minimum usable temperature, and maximum safe temperature; the vehicle route and travel time parameters include the power plant location, user location, road distance, estimated travel time, service sequence, and waiting time; the heat storage material and insulation parameters include heat storage medium parameters, insulation layer parameters, and environmental heat exchange parameters; the user heat load and heat grade requirements include the user's time-of-use heat load, required temperature level, service window, and allowable heat shortage range; S1.2: Based on the structural and interface parameters of the mobile thermal storage vehicle, the compartment temperature and heat storage parameters, the vehicle path and travel time parameters, and the thermal storage material and insulation parameters, a multi-state mobile thermal storage vehicle model is constructed. The multi-state mobile thermal storage vehicle model includes a thermal storage state line and a transportation loss line. The thermal storage state line is used to sequentially divide the multi-compartment thermal storage structure, determine the compartment temperature and heat storage capacity, and identify the thermal grade status. The transportation loss line is used to sequentially read the vehicle path and travel time, calculate the heat loss during travel, and calculate the remaining heat after arrival. Based on the thermal grade status identification results and the remaining heat calculation results after arrival, the thermal storage attenuation status is updated, and the charging and discharging interface is matched and the vehicle task status is updated, forming a mobile thermal storage vehicle object feature library. The mobile thermal storage vehicle object feature library is a data set used to characterize the compartment thermal storage status, thermal grade status, heat loss status, interface matching status, and vehicle task status of the mobile thermal storage vehicle. S1.3: Construct a biomass power plant heating output model based on the heating boundary parameters of the biomass power plant to determine the time-of-use heat supply and heating power boundaries within the scheduling cycle; construct a user heat load demand model based on user heat load and heat grade demand to determine user time-of-use heat load, heat grade demand, and service windows; and construct a biomass power plant-mobile heat storage vehicle-user heat balance based on the mobile thermal storage vehicle object feature library, biomass power plant heating output, and user heat load demand. S1.4: Based on the thermal balance of biomass power plant-mobile thermal storage vehicle-user, construct heat loss constraints, path and service window constraints, interface security constraints, and vehicle task timing constraints to form a scheduling model library to be optimized; the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized together serve as inputs for subsequent mobile thermal storage vehicle object feature embedded center collision optimization.

[0013] S2: Based on the mobile thermal storage vehicle object feature library formed in S1.2 and the scheduling model library to be optimized formed in S1.4, read the parameters of the mobile thermal storage vehicle object feature library and the CCO algorithm parameters. Based on the mobile thermal storage vehicle object feature library, perform variable encoding on the scheduling model library to be optimized to generate an initial candidate population. Decode the candidate schemes, evaluate multi-state fitness, and determine the collision center for the initial candidate population. Based on the mobile thermal storage vehicle object feature constraint mapping and dual-space collision update, generate a dual-space candidate scheme set. The specific steps are as follows: S2.1: Read the feature library of the mobile thermal storage vehicle object formed in S1.2, and read the center collision optimization algorithm parameters, i.e., CCO algorithm parameters; the center collision optimization algorithm parameters include population size, maximum number of iterations, collision center selection ratio, collision update step size and perturbation decay coefficient; wherein, the population size is used to determine the number of initial candidate schemes, the maximum number of iterations is used to limit the number of center collision update cycles, the collision center selection ratio is used to determine the center individual from the comprehensive fitness ranking results, and the collision update step size and perturbation decay coefficient are used to control the change range of candidate schemes in the dual-space update process; S2.2: Based on the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized, embedded variable encoding of mobile thermal storage vehicle object features is performed. The embedded variable encoding of mobile thermal storage vehicle object features includes compartment thermal storage state encoding, driving heat loss correction encoding, thermal quality service matching encoding, and path-time window-charge and release joint encoding. Among them, compartment thermal storage state encoding is used to characterize the temperature, heat storage capacity, and capacity boundary of each compartment; driving heat loss correction encoding is used to characterize the heat attenuation during vehicle transportation; thermal quality service matching encoding is used to characterize the matching relationship between the thermal quality of the mobile thermal storage vehicle and the thermal quality required by the user; path-time window-charge and release joint encoding is used to characterize the coupling relationship between vehicle path, service sequence, arrival time, charging capacity, and releasing capacity; an initial candidate population is generated based on the encoding results and the population size. S2.3: Decode the candidate schemes of the initial candidate population to obtain candidate mobile thermal storage vehicle scheduling schemes; perform multi-state fitness evaluation based on the candidate mobile thermal storage vehicle scheduling schemes, including heat loss fitness calculation, heat grade matching fitness calculation, compartment capacity utilization fitness calculation, path time window executability calculation, and interface constraint violation calculation; perform comprehensive fitness ranking based on the multi-state fitness evaluation results, and determine the collision center and ranking samples according to the collision center selection ratio; S2.4: Based on the collision center, sorted samples, and center collision optimization algorithm parameters, perform feature constraint mapping and dual-space collision update for the mobile thermal storage vehicle object; the dual-space collision update includes original spatial collision update and decorrelational spatial variable transformation, wherein the original spatial collision update is used to update the compartment thermal storage state, thermal quality service matching relationship, driving heat loss correction amount, path time window, and charging and discharging heat variables, and the decorrelational spatial variable transformation is used to reduce the coupling influence between the above variables; after each round of collision update, perform compartment thermal state boundary correction, thermal quality matching relationship correction, heat loss compensation update, and interface path constraint mapping correction until the preset iteration conditions are reached, output the dual-space candidate scheme set; and input it into S3.

[0014] S3: Based on the dual-space candidate scheme set output by S2.4, the feasibility of the mobile thermal storage vehicle object features is verified. After judgment and linkage correction, candidate schemes are obtained, and the object feature states of the candidate schemes are collected. Based on the object feature states, Langevin policy feedback is performed to determine adaptive screening parameters, and the critic feedback and flexible actor network are updated. Based on the critic feedback and flexible actor network, the object feature evaluation matrix and evolutionary environment selection are performed. After local quality and crowding sorting, the evolutionary environment selection is updated to obtain the filtered candidate set. After adaptive normalization and trade-off selection, the termination condition is judged. If it is not satisfied, the adaptive neighborhood analysis is returned. If it is satisfied, the optimal scheduling scheme is output. The specific steps are as follows: S3.1: Read the dual-space candidate scheme set output by S2.4, and perform feasibility verification of the mobile thermal storage vehicle object characteristics on the dual-space candidate scheme set; the feasibility verification of the mobile thermal storage vehicle object characteristics includes heat grade feasibility judgment, compartment thermal storage boundary judgment, heat supply satisfaction judgment after heat loss judgment, and interface and path time window feasibility judgment; when a candidate scheme meets the above judgment conditions, the candidate scheme is archived as a feasible candidate scheme, and an object characteristic status aggregation result is formed; the candidate schemes that meet the above conditions are archived as feasible candidate schemes; the candidate schemes that do not meet the above conditions are classified as infeasible candidate schemes, and the corresponding infeasibility reasons are recorded.

[0015] S3.2: When a candidate solution does not meet at least one of the preset constraints of heat quality, compartment heat storage, heat loss, interface matching, or path window, an infeasible solution linkage correction is performed. The infeasible solution linkage correction includes adjusting the service vehicle, service sequence, and compartment heat supply allocation according to heat quality requirements; adjusting the charging and releasing heat according to the compartment heat storage boundary; adjusting the path, arrival time, or service window according to the remaining heat after heat loss; and adjusting the charging and releasing heat interface matching relationship according to interface constraints. If the preset constraints are still not fully met after correction, the corresponding constraint violation amount is recorded and the constraint violation amount is input into the subsequent object feature evaluation matrix. S3.3: Based on the aggregated results of object feature states, calculate the thermal grade deviation state, thermal loss sensitive state, compartmentalized heat storage utilization state, and interface and time window conflict state, and input the above states into a state vector into the flexible actor network; output the strategy action through the Langevin perturbation action generation process to determine the adaptive screening parameters, which include adaptive neighborhood scale, search perturbation intensity, and candidate solution retention ratio; construct the critic value assessment and reward feedback based on the evolutionary screening results, and update the flexible actor network parameters according to the reward feedback, so that the subsequent screening parameters are adjusted as the candidate solution state changes; S3.4: Based on adaptive screening parameters, adaptive neighborhood analysis, local quality and congestion ranking, and evolutionary environment selection updates are performed to form a set of candidate solutions after screening. An object feature evaluation matrix is ​​constructed based on this set, including at least operating costs, carbon emissions, heating satisfaction, vehicle utilization, heat quality matching, heat loss level, and constraint violation. The object feature evaluation matrix undergoes adaptive normalization and trade-offs, and a determination is made as to whether a preset termination condition is met. If the preset termination condition is not met, the process returns to the evolutionary environment selection process for further screening, and feedback is provided to S2.2 for candidate population updates when necessary. When the preset termination condition is met, this candidate solution is output as the optimal scheduling solution. The optimal scheduling solution includes the biomass power plant's time-sharing heating arrangement, the mobile thermal storage vehicle's charging and discharging arrangement, vehicle routes and arrival times, user service order, and corresponding evaluation index results. The preset termination conditions include reaching the maximum number of iterations and the change in the optimal candidate solution during continuous iterations being less than a preset threshold.

Claims

1. A multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback, characterized in that... Follow these steps: S1: Construct a multi-state mobile thermal storage vehicle model based on scheduling basic data, determine the multi-compartment thermal storage structure, compartment thermal storage status, heat quality status, driving heat loss status, thermal storage attenuation status, interface matching status, and vehicle task status of the mobile thermal storage vehicle, and form a mobile thermal storage vehicle object feature library; construct a scheduling model library to be optimized based on the thermal balance of biomass power plant-mobile thermal storage vehicle-user. S2: Based on the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized formed in S1, read the parameters of the mobile thermal storage vehicle object feature library and the CCO algorithm parameters. Based on the mobile thermal storage vehicle object feature library, perform variable encoding on the scheduling model library to be optimized to generate an initial candidate population. Perform candidate scheme decoding, multi-state fitness evaluation and collision center determination on the initial candidate population. Based on the mobile thermal storage vehicle object feature constraint mapping and dual-space collision update, generate a dual-space candidate scheme set. S3: Based on the dual-space candidate scheme set output by S2, the feasibility of the mobile thermal storage vehicle object features is verified. After judgment and linkage correction, candidate schemes are obtained, and the object feature status of the candidate schemes is collected. Based on the object feature state, Langevin policy feedback is performed to determine adaptive screening parameters and update the critic feedback and flexible actor network. Based on the critic feedback and flexible actor network, object feature evaluation matrix and evolutionary environment selection are performed. After local quality and crowding sorting, the evolutionary environment selection is updated to obtain the candidate set after screening. After adaptive normalization and trade-off selection, termination condition judgment is performed. If not satisfied, adaptive neighborhood analysis is returned. If satisfied, the optimal scheduling scheme is output.

2. The multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback according to claim 1, characterized in that, S1 specifically includes: S1.1: Collect basic scheduling data, which includes biomass power plant heating boundary parameters, mobile thermal storage vehicle structure and interface parameters, compartment temperature and heat storage parameters, vehicle route and travel time parameters, thermal storage material and insulation parameters, and user heat load and heat grade requirements; wherein, the biomass power plant heating boundary parameters include fuel calorific value, fuel supply, boiler steam parameters, cogeneration unit output boundary, and available heat power boundary; the mobile thermal storage vehicle structure and interface parameters include the number of compartments in the mobile thermal storage vehicle, the capacity of each compartment, and the charging and discharging interfaces. Location, interface power boundaries, and interface temperature range; the compartment temperature and heat storage parameters include the initial temperature of each compartment, initial heat storage, maximum heat storage, minimum usable temperature, and maximum safe temperature; the vehicle route and travel time parameters include the power plant location, user location, road distance, estimated travel time, service sequence, and waiting time; the heat storage material and insulation parameters include heat storage medium parameters, insulation layer parameters, and environmental heat exchange parameters; the user heat load and heat grade requirements include the user's time-of-use heat load, required temperature level, service window, and allowable heat shortage range; S1.2: Based on the structural and interface parameters of the mobile thermal storage vehicle, the compartment temperature and heat storage parameters, the vehicle path and travel time parameters, and the thermal storage material and insulation parameters, a multi-state mobile thermal storage vehicle model is constructed. The multi-state mobile thermal storage vehicle model includes a thermal storage state line and a transportation loss line. The thermal storage state line is used to sequentially divide the multi-compartment thermal storage structure, determine the compartment temperature and heat storage capacity, and identify the thermal grade status. The transportation loss line is used to sequentially read the vehicle path and travel time, calculate the heat loss during travel, and calculate the remaining heat after arrival. Based on the thermal grade status identification results and the remaining heat calculation results after arrival, the thermal storage attenuation status is updated, and the charging and discharging interface is matched and the vehicle task status is updated, forming a mobile thermal storage vehicle object feature library. The mobile thermal storage vehicle object feature library is a data set used to characterize the compartment thermal storage status, thermal grade status, heat loss status, interface matching status, and vehicle task status of the mobile thermal storage vehicle. S1.3: Construct a biomass power plant heating output model based on the heating boundary parameters of the biomass power plant to determine the time-of-use heat supply and heating power boundaries within the scheduling cycle; construct a user heat load demand model based on user heat load and heat grade demand to determine user time-of-use heat load, heat grade demand, and service windows; and construct a biomass power plant-mobile heat storage vehicle-user heat balance based on the mobile thermal storage vehicle object feature library, biomass power plant heating output, and user heat load demand. S1.4: Based on the thermal balance of biomass power plant-mobile thermal storage vehicle-user, construct heat loss constraints, path and service window constraints, interface security constraints, and vehicle task timing constraints to form a scheduling model library to be optimized; the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized together serve as inputs for subsequent mobile thermal storage vehicle object feature embedded center collision optimization.

3. The multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback according to claim 2, characterized in that, S2 specifically includes: S2.1: Read the feature library of the mobile thermal storage vehicle object formed in S1.2, and read the center collision optimization algorithm parameters, i.e., CCO algorithm parameters; the center collision optimization algorithm parameters include population size, maximum number of iterations, collision center selection ratio, collision update step size and perturbation decay coefficient; wherein, the population size is used to determine the number of initial candidate schemes, the maximum number of iterations is used to limit the number of center collision update cycles, the collision center selection ratio is used to determine the center individual from the comprehensive fitness ranking results, and the collision update step size and perturbation decay coefficient are used to control the change range of candidate schemes in the dual-space update process; S2.2: Based on the mobile thermal storage vehicle object feature library and the scheduling model library to be optimized, embedded variable encoding of mobile thermal storage vehicle object features is performed. The embedded variable encoding of mobile thermal storage vehicle object features includes compartment thermal storage state encoding, driving heat loss correction encoding, thermal quality service matching encoding, and path-time window-charge and release joint encoding. Among them, compartment thermal storage state encoding is used to characterize the temperature, heat storage capacity, and capacity boundary of each compartment; driving heat loss correction encoding is used to characterize the heat attenuation during vehicle transportation; thermal quality service matching encoding is used to characterize the matching relationship between the thermal quality of the mobile thermal storage vehicle and the thermal quality required by the user; path-time window-charge and release joint encoding is used to characterize the coupling relationship between vehicle path, service sequence, arrival time, charging capacity, and releasing capacity; an initial candidate population is generated based on the encoding results and the population size. S2.3: Decode the candidate schemes of the initial candidate population to obtain candidate mobile thermal storage vehicle scheduling schemes; perform multi-state fitness evaluation based on the candidate mobile thermal storage vehicle scheduling schemes, including heat loss fitness calculation, heat grade matching fitness calculation, compartment capacity utilization fitness calculation, path time window executability calculation, and interface constraint violation calculation; perform comprehensive fitness ranking based on the multi-state fitness evaluation results, and determine the collision center and ranking samples according to the collision center selection ratio; S2.4: Based on the collision center, sorted samples, and center collision optimization algorithm parameters, perform feature constraint mapping and dual-space collision update for the mobile thermal storage vehicle object; the dual-space collision update includes original spatial collision update and decorrelational spatial variable transformation, wherein the original spatial collision update is used to update the compartment thermal storage state, thermal quality service matching relationship, driving heat loss correction amount, path time window, and charging and discharging heat variables, and the decorrelational spatial variable transformation is used to reduce the coupling influence between the above variables; after each round of collision update, perform compartment thermal state boundary correction, thermal quality matching relationship correction, heat loss compensation update, and interface path constraint mapping correction until the preset iteration conditions are reached, and output the dual-space candidate scheme set.

4. The multi-state mobile thermal storage vehicle scheduling method based on center collision and strategy feedback according to claim 3, characterized in that, S3 specifically includes: S3.1: Read the dual-space candidate scheme set output by S2.4, and perform feasibility verification of the mobile thermal storage vehicle object features on the dual-space candidate scheme set; the feasibility verification of the mobile thermal storage vehicle object features includes heat grade feasibility judgment, compartment thermal storage boundary judgment, heat supply satisfaction judgment after heat loss judgment, and interface and path time window feasibility judgment; when the candidate scheme meets the above judgment conditions, the candidate scheme is archived as a feasible candidate scheme, and an object feature status aggregation result is formed; S3.2: When a candidate solution does not meet at least one of the preset constraints of heat quality, compartment heat storage, heat loss, interface matching, or path window, an infeasible solution linkage correction is performed. The infeasible solution linkage correction includes adjusting the service vehicle, service sequence, and compartment heat supply allocation according to heat quality requirements; adjusting the charging and releasing heat according to the compartment heat storage boundary; adjusting the path, arrival time, or service window according to the remaining heat after heat loss; and adjusting the charging and releasing heat interface matching relationship according to interface constraints. If the preset constraints are still not fully met after correction, the corresponding constraint violation amount is recorded and the constraint violation amount is input into the subsequent object feature evaluation matrix. S3.3: Based on the aggregated results of object feature states, calculate the thermal grade deviation state, thermal loss sensitive state, compartmentalized heat storage utilization state, and interface and time window conflict state, and input the above states into a state vector into the flexible actor network; output the strategy action through the Langevin perturbation action generation process to determine the adaptive screening parameters, which include adaptive neighborhood scale, search perturbation intensity, and candidate solution retention ratio; construct the critic value assessment and reward feedback based on the evolutionary screening results, and update the flexible actor network parameters according to the reward feedback, so that the subsequent screening parameters are adjusted as the candidate solution state changes; S3.4: Based on adaptive screening parameters, adaptive neighborhood analysis, local quality and congestion ranking, and evolutionary environment selection updates are performed to form a set of candidate solutions after screening. An object feature evaluation matrix is ​​constructed based on the set of candidate solutions. This object feature evaluation matrix includes at least operating costs, carbon emissions, heating satisfaction, vehicle utilization, heat quality matching degree, heat loss level, and constraint violation. The object feature evaluation matrix undergoes adaptive normalization and trade-offs, and it is determined whether a preset termination condition is met. If the preset termination condition is not met, the process returns to the evolutionary environment selection process to continue screening, and feedback is provided to S2.2 for candidate population updates when necessary. When the preset termination condition is met, this candidate solution is output as the optimal scheduling solution. The optimal scheduling solution includes the biomass power plant time-sharing heating arrangement, the mobile thermal storage vehicle charging and discharging heat arrangement, vehicle routes and arrival times, user service order, and corresponding evaluation index results.

Images