Multi-zone mobile energy storage resource cross-region coordinated dispatching method

Abstract

This invention relates to a cross-regional collaborative scheduling method for mobile energy storage resources in multiple transformer substations. First, a real-time load monitoring and secure data exchange communication system for multiple transformer substations is constructed. Then, a supply-demand collaborative matching model for multiple transformer substations is built based on multi-agent collaborative theory. Next, a cross-regional dynamic scheduling path optimization algorithm for mobile energy storage in multiple transformer substations is constructed. Finally, a theoretical model of typical operating conditions for the transformer substations is constructed, and operational indicators under multiple scenarios are compared using a power system joint simulation platform. This allows for a quantitative analysis of the mechanism by which scheduling strategies improve grid operation and energy storage utilization efficiency, completing strategy optimization and theoretical verification. In summary, this invention has the advantages of being intelligent, efficient, highly adaptable, and capable of achieving full-domain perception.

Description

Technical Field

[0001] This invention belongs to the field of mobile energy storage technology for power grids, and specifically relates to a method for cross-regional collaborative scheduling of mobile energy storage resources in multiple distribution areas. Background Technology

[0002] With the continuous improvement of rural electrification and the widespread adoption of distributed photovoltaic power generation, rural power distribution networks exhibit completely different operating characteristics from urban power grids. These characteristics include dispersed distribution areas, large power supply radii, long road distances between distribution areas, weak wireless communication environments, significant seasonal load migration, large day-night peak-valley differences, and strong randomness in photovoltaic power generation. Furthermore, rural electricity consumption exhibits typical characteristics such as concentrated electricity consumption during holidays, sudden load increases during the busy farming season, reverse power transmission from surplus photovoltaic power in winter, and continuous overload due to high temperatures in summer. These factors result in numerous technical shortcomings that are difficult to address with traditional fixed power grid control methods. Currently, existing technologies generally adopt an operating mode of fixed capacity expansion of distribution areas, local installation of fixed energy storage, independent control of single distribution areas, and simple local dispatch. However, this mode is no longer suitable for the spatiotemporal differences, strong fluctuations, weak communication, long distances, and complex operating scenarios of multiple distribution areas in rural areas. Therefore, it is necessary to develop a cross-regional collaborative dispatch method for mobile energy storage resources in multiple distribution areas that can achieve full-domain perception, intelligent matching, dynamic path optimization, multi-scenario adaptation, and hardware and software synergy. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a cross-regional collaborative scheduling method for multiple mobile energy storage resources that can achieve full-domain perception, intelligent matching, dynamic path optimization, multi-scenario adaptation, and software and hardware collaboration.

[0004] The objective of this invention is achieved as follows: a method for cross-regional collaborative scheduling of multiple mobile energy storage resources, comprising the following steps: Step 1: Using each distribution substation as the terminal sensing unit, a hierarchical distributed load monitoring network architecture is built. Then, data transmission is carried out using a hybrid networking method of long-distance low-power wireless communication and mobile network. Next, the data is verified using cyclic redundancy check, abnormal data is removed using the isolated forest data filtering method, and time series data is repaired using a two-way time series data completion model. Finally, the load fluctuation data of the substation and the distributed photovoltaic output data are cleaned and corrected, thereby achieving reliable interaction of load data, distributed photovoltaic output data, substation voltage operation data, and equipment status data of each substation. Step 2: Establish a three-layer multi-agent collaborative framework consisting of a transformer area demand agent, a mobile energy storage supply agent, and a collaborative dispatch center agent. The transformer area demand agent quantifies the comprehensive demand weight of the transformer area based on factors such as the transformer area's load level, peak-valley difference, photovoltaic surplus consumption demand, and voltage exceedance risk, using an entropy weight hierarchical combination weighting method. The mobile energy storage supply agent constructs a fuzzy comprehensive evaluation supply capacity quantification model based on the remaining power of energy storage vehicles, available adjustment capacity, geographical location, mobile energy consumption, and response speed. The collaborative dispatch center agent, based on demand weights and supply capacity, uses a contract network task bidding matching mechanism to complete dynamic matching of supply and demand across the entire region, establishing spatiotemporally differentiated matching rules for the flow of energy storage resources from low-load photovoltaic surplus transformer areas to high-load voltage anomaly transformer areas. Step 3: With the multi-objective optimization goals of maximizing the utilization rate of the entire network's energy storage resources, minimizing the peak-valley difference between distribution areas, optimizing voltage fluctuation suppression, and minimizing dispatch energy consumption, a multi-dimensional constraint matrix is ​​established, which includes constraints on road network travel distance, energy storage capacity range, distribution area transformer capacity, and road traffic limit. Then, an improved non-dominated sorting genetic algorithm combined with a reference point selection strategy is used to iteratively solve for the optimal solution set, and the final optimal dispatch scheme is selected by a superior-inferior solution distance decision model. Step 4: Construct a theoretical model of typical operating conditions for the distribution area, including summer peak overload scenario, winter light load with surplus photovoltaic power, daytime photovoltaic large fluctuation scenario, rural holiday load surge scenario, and distribution area voltage anomaly scenario. Then, relying on the power system joint simulation platform, compare the operating indicators under multiple scenarios, including energy storage resource utilization rate, photovoltaic curtailment rate, distribution area voltage qualification rate, network loss, and dispatch energy consumption. This allows for a quantitative analysis of the mechanism by which dispatch strategies improve the grid operation status and energy storage utilization efficiency, completing strategy optimization and theoretical verification.

[0005] Furthermore, the transformer area demand agent in step 2 is equipped with a time-series cyclic unit load prediction model. This model takes historical load data, ambient temperature, irradiance, and date type as input features and outputs a prediction curve of the transformer area load change trend for the next four hours.

[0006] Furthermore, the improved non-dominated sorting genetic algorithm in step 3 includes four optimization strategies: first, introducing a hybrid selection operator of convergence and crowding to improve the quality of the solution set; second, setting iterative adaptive crossover probability and mutation probability; third, adding a mechanism for archiving and retaining elite solutions from previous generations; and fourth, designing an integer-real number hybrid chromosome coding structure for discrete variables of transformer area selection and continuous variables of power duration to adapt to the hybrid decision-making characteristics of mobile energy storage across transformer areas.

[0007] Furthermore, in step 3, the road network travel distance constraint is based on the construction of a rural road model using open-source road topology data, and a dynamic travel cost function is constructed by introducing road grade, road surface condition, bridge load limit, and mountain travel speed correction coefficient; the energy storage power constraint sets the safe working power range of energy storage vehicles to 20% to 90%, and adds a discharge depth loss penalty factor in combination with the battery cycle life loss model.

[0008] Furthermore, the spatiotemporal differentiation matching rules in step 2 are respectively set with a four-season operation mode and a date operation mode.

[0009] The beneficial effects of this invention are as follows: By establishing a time-space differentiated matching rule for the flow of energy storage resources from low-load photovoltaic surplus areas to high-load voltage abnormal areas, this invention breaks through the technical limitations of traditional fixed energy storage single-point configuration and static operation. It innovatively proposes a dynamic supply and demand matching mechanism that adapts to the time-space differences of rural areas, with seasonal differentiation and dual modes of weekdays and holidays. This enables the flexible flow and reuse of mobile energy storage resources across areas, seasons, and all weather conditions, solving the industry pain points of low utilization rate, high expansion cost, and poor adaptability of fixed energy storage. This invention establishes a hierarchical distributed load monitoring network architecture and performs data verification, outlier removal, time-series completion, and cleaning and correction to achieve reliable data interaction. This solves the practical problems of scattered rural transformer substations, high interference, poor data quality, and unstable communication, and provides a reliable data foundation for subsequent intelligent scheduling. This invention constructs a three-layer multi-agent collaborative framework and utilizes combined weighting and fuzzy evaluation to achieve precise two-way quantification of demand and supply. Combined with a task bidding mechanism, it achieves globally optimal matching. Compared with traditional nearest-neighbor scheduling and average allocation scheduling, it can effectively improve matching accuracy. This invention establishes models of five typical extreme working conditions in rural areas and reveals the optimization mechanism of scheduling strategies through multi-scenario comparative simulations, enabling scheduling strategies to fully adapt to complex real-world working conditions such as summer overload, winter backhaul, cloud fluctuations, and sudden increases during holidays. In summary, this invention has the advantages of being intelligent and efficient, highly adaptable, and capable of achieving full-domain perception. Detailed Implementation

[0010] The present invention will now be further described.

[0011] Example: A method for cross-regional collaborative scheduling of multiple mobile energy storage resources, comprising the following steps: Step 1: Using each distribution substation as the terminal sensing unit, a hierarchical distributed load monitoring network architecture is built. Then, data transmission is carried out using a hybrid networking method of long-distance low-power wireless communication and mobile network. Next, the data is verified using cyclic redundancy check, abnormal data is removed using the isolated forest data filtering method, and time series data is repaired using a two-way time series data completion model. Finally, the load fluctuation data of the substation and the distributed photovoltaic output data are cleaned and corrected, thereby achieving reliable interaction of load data, distributed photovoltaic output data, substation voltage operation data, and equipment status data of each substation. Step 2: Establish a three-layer multi-agent collaborative framework consisting of a transformer area demand agent, a mobile energy storage supply agent, and a collaborative dispatch center agent. The transformer area demand agent quantifies the comprehensive demand weight of the transformer area based on factors such as load intensity, peak-valley difference, photovoltaic surplus consumption demand, and voltage exceedance risk, using an entropy weight hierarchical combination weighting method. Furthermore, the transformer area demand agent includes a time-series cyclic unit load prediction model, which uses historical load data, ambient temperature, solar irradiance, and date type as input features to output a predicted curve of the transformer area load change trend for the next four hours. The mobile energy storage supply agent constructs a fuzzy comprehensive evaluation supply capacity quantification model based on the remaining power of energy storage vehicles, available adjustment capacity, geographical location, mobile energy consumption, and response speed. The collaborative dispatch center agent, based on demand weights and supply capacity, uses a contract network task bidding matching mechanism to achieve dynamic matching of supply and demand across the entire region, establishing spatiotemporal differentiated matching rules for the flow of energy storage resources from low-load photovoltaic surplus transformer areas to high-load voltage anomaly transformer areas. These spatiotemporal differentiated matching rules include seasonal operation modes and date-based operation modes. Step 3: With the multi-objective optimization goals of maximizing the utilization rate of the entire network's energy storage resources, minimizing the peak-valley difference between transformer substations, optimizing voltage fluctuation suppression, and minimizing dispatch energy consumption, a multi-dimensional constraint matrix is ​​established, including constraints on road network travel distance, energy storage capacity range, transformer substation capacity, and road traffic limits. Then, an improved non-dominated sorting genetic algorithm combined with a reference point selection strategy is used to iteratively solve for the optimal solution set. A superior-inferior solution distance decision model is then used to select the final optimal dispatch scheme. The improved non-dominated sorting genetic algorithm includes four optimization strategies: first, introducing a hybrid selection operator combining convergence and congestion to improve the quality of the solution set; second, setting an iterative self-... The system adapts to crossover and mutation probabilities; thirdly, it adds a mechanism for archiving and preserving elite solutions from previous generations; fourthly, it designs a mixed integer and real number chromosome coding structure for discrete variables in transformer area selection and continuous variables in power duration, adapting to the mixed decision-making characteristics of mobile energy storage across transformer areas; furthermore, the road network travel distance constraint is based on open-source road topology data to construct a rural road model, and introduces road grade, road surface condition, bridge load limit, and mountain travel speed correction coefficient to construct a dynamic travel cost function; the energy storage power constraint sets the safe operating power range for energy storage vehicles to be 20% to 90%, and adds a discharge depth loss penalty factor in conjunction with the battery cycle life loss model. Step 4: Construct a theoretical model of typical operating conditions for the distribution area, including summer peak overload scenario, winter light load with surplus photovoltaic power, daytime photovoltaic large fluctuation scenario, rural holiday load surge scenario, and distribution area voltage anomaly scenario. Then, relying on the power system joint simulation platform, compare the operating indicators under multiple scenarios, including energy storage resource utilization rate, photovoltaic curtailment rate, distribution area voltage qualification rate, network loss, and dispatch energy consumption. This allows for a quantitative analysis of the mechanism by which dispatch strategies improve the grid operation status and energy storage utilization efficiency, completing strategy optimization and theoretical verification.

[0012] In the application of this invention, firstly, a hierarchical distributed load monitoring network architecture is built, using each distribution transformer area as the terminal sensing unit. Then, data transmission is performed using a hybrid networking method combining long-distance low-power wireless communication and mobile networks. The data is then sequentially verified, anomaly removed, time-series completed, and cleaned and corrected, thereby achieving reliable data interaction. This solves the practical problems of dispersed rural transformer areas, high interference, poor data quality, and unstable communication, providing a reliable data foundation for subsequent intelligent scheduling. Next, a three-layer multi-agent collaborative framework is established, consisting of a transformer area demand agent, a mobile energy storage supply agent, and a collaborative dispatch center agent. The transformer area demand agent quantifies the comprehensive demand weight of the transformer area based on its load factor, peak-valley difference, photovoltaic surplus consumption demand, and voltage exceedance risk using an entropy weight hierarchical combination weighting method. The mobile energy storage supply agent constructs a fuzzy comprehensive evaluation supply capacity quantification model based on the remaining power of energy storage vehicles, available adjustment capacity, geographical location, mobile energy consumption, and response speed. The collaborative dispatch center agent, based on demand... By considering weights and supply capacity, a contract network task bidding matching mechanism is adopted to achieve dynamic matching of supply and demand across the entire region. A spatiotemporally differentiated matching rule is established for the flow of energy storage resources from low-load photovoltaic surplus areas to high-load voltage anomaly areas. Finally, with the optimization objectives of maximizing the utilization rate of energy storage resources across the entire network, minimizing the peak-valley difference between distribution areas, optimizing voltage fluctuation suppression, and minimizing dispatch energy consumption, a multi-dimensional constraint matrix is ​​established, including constraints on road network travel distance, energy storage capacity range, distribution area transformer capacity, and road traffic limit. Subsequently, an improved non-dominated sorting genetic algorithm combined with a reference point selection strategy is used to iteratively solve for the optimal solution set, and a superior-inferior solution distance decision model is used to select the final optimal dispatch scheme. After completing the above operations, a theoretical model of typical operating conditions in distribution areas is constructed, and the operating indicators under multiple scenarios are compared using a power system joint simulation platform. This allows for a quantitative analysis of the mechanism by which the dispatch strategy improves the grid operating status and energy storage utilization efficiency, completing strategy optimization and theoretical verification. In summary, this invention has the advantages of being intelligent, efficient, highly adaptable, and capable of achieving full-domain perception.

[0013] The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.

Claims

1. A method for cross-regional collaborative scheduling of multiple mobile energy storage resources, characterized in that, Includes the following steps: Step 1: Using each distribution substation as the terminal sensing unit, a hierarchical distributed load monitoring network architecture is built. Then, data transmission is carried out using a hybrid networking method of long-distance low-power wireless communication and mobile network. Next, the data is verified using cyclic redundancy check, abnormal data is removed using the isolated forest data filtering method, and time series data is repaired using a two-way time series data completion model. Finally, the load fluctuation data of the substation and the distributed photovoltaic output data are cleaned and corrected, thereby achieving reliable interaction of load data, distributed photovoltaic output data, substation voltage operation data, and equipment status data of each substation. Step 2: Establish a three-layer multi-agent collaborative framework consisting of a transformer area demand agent, a mobile energy storage supply agent, and a collaborative dispatch center agent. The transformer area demand agent quantifies the comprehensive demand weight of the transformer area based on factors such as the transformer area's load level, peak-valley difference, photovoltaic surplus consumption demand, and voltage exceedance risk, using an entropy weight hierarchical combination weighting method. The mobile energy storage supply agent constructs a fuzzy comprehensive evaluation supply capacity quantification model based on the remaining power of energy storage vehicles, available adjustment capacity, geographical location, mobile energy consumption, and response speed. The collaborative dispatch center agent, based on demand weights and supply capacity, uses a contract network task bidding matching mechanism to complete dynamic matching of supply and demand across the entire region, establishing spatiotemporally differentiated matching rules for the flow of energy storage resources from low-load photovoltaic surplus transformer areas to high-load voltage anomaly transformer areas. Step 3: With the multi-objective optimization goals of maximizing the utilization rate of the entire network's energy storage resources, minimizing the peak-valley difference between distribution areas, optimizing voltage fluctuation suppression, and minimizing dispatch energy consumption, a multi-dimensional constraint matrix is ​​established, which includes constraints on road network travel distance, energy storage capacity range, distribution area transformer capacity, and road traffic limit. Then, an improved non-dominated sorting genetic algorithm combined with a reference point selection strategy is used to iteratively solve for the optimal solution set, and the final optimal dispatch scheme is selected by a superior-inferior solution distance decision model. Step 4: Construct a theoretical model of typical operating conditions for the distribution area, including summer peak overload scenario, winter light load with surplus photovoltaic power, daytime photovoltaic large fluctuation scenario, rural holiday load surge scenario, and distribution area voltage anomaly scenario. Then, relying on the power system joint simulation platform, compare the operating indicators under multiple scenarios, including energy storage resource utilization rate, photovoltaic curtailment rate, distribution area voltage qualification rate, network loss, and dispatch energy consumption. This allows for a quantitative analysis of the mechanism by which dispatch strategies improve the grid operation status and energy storage utilization efficiency, completing strategy optimization and theoretical verification.

2. The method for cross-regional collaborative scheduling of multiple mobile energy storage resources as described in claim 1, characterized in that: The transformer area demand agent in step 2 has a time-series cyclic unit load prediction model inside. This model takes historical load data, ambient temperature, irradiance, and date type as input features and outputs a prediction curve of the transformer area load change trend for the next four hours.

3. The method for cross-regional collaborative scheduling of multiple mobile energy storage resources as described in claim 1, characterized in that: The improved non-dominated sorting genetic algorithm in step 3 includes four optimization strategies: First, it introduces a hybrid selection operator of convergence and crowding to improve the quality of the solution set; second, it sets iterative adaptive crossover probability and mutation probability; third, it adds a mechanism for archiving and retaining elite solutions from previous generations; and fourth, it designs an integer-real number hybrid chromosome coding structure for discrete variables of transformer area selection and continuous variables of power duration to adapt to the hybrid decision-making characteristics of mobile energy storage across transformer areas.

4. The method for cross-regional collaborative scheduling of multiple mobile energy storage resources as described in claim 1, characterized in that: In step 3, the road network travel distance constraint is based on the rural road model constructed from open-source road topology data. The dynamic travel cost function is constructed by introducing road grade, road surface condition, bridge load limit, and mountain travel speed correction coefficient. The energy storage capacity constraint is set to a safe operating capacity range of 20% to 90% for energy storage vehicles, and a discharge depth loss penalty factor is added in conjunction with the battery cycle life loss model.

5. The method for cross-regional collaborative scheduling of multiple mobile energy storage resources as described in claim 1, characterized in that: The spatiotemporal differentiation matching rules in step 2 are set with four seasons operation mode and date operation mode respectively.