Optimized scheduling method for virtual power plant

Abstract

The invention discloses an optimal scheduling method for a virtual power plant, and relates to the technical field of power regulation, and the method comprises the steps: obtaining multi-modal data of a detection region, calculating a regional load fluctuation index and a resource mobility index, and generating a dynamic feature data set; performing real-time clustering division on the power grid nodes through a clustering algorithm based on the dynamic feature data set to generate a dynamic sub-region set; if the sub-region feature fluctuation exceeds a boundary stability threshold value, triggering boundary redivision; according to the dynamic sub-region set, comparing a building power consumption predicted value with the available load of the charging pile in real time to calculate a demand difference, and generating an elastic scheduling instruction through multi-objective optimization; uncertain factors are monitored in real time, a risk index is calculated, and the priority and the resource allocation strategy of the elastic scheduling instruction are dynamically adjusted based on the risk index; and generating a risk avoidance scheduling scheme through a robust optimization algorithm.

Description

Technical Field

[0001] This invention relates to the field of power regulation technology, and in particular to an optimized scheduling method for a virtual power plant. Background Technology

[0002] With the large-scale integration of renewable energy and the rapid popularization of electric vehicles, the operation mode of the power system is shifting from traditional centralized power generation to a distributed and diversified energy structure. As a technical means of aggregating distributed energy resources, virtual power plants can achieve flexible support for the power grid and efficient energy utilization through coordinated control and optimized scheduling.

[0003] Current virtual power plant (VPS) optimization scheduling methods primarily rely on static partitioning and deterministic optimization models. They employ a fixed geographical area-based scheduling approach, dividing the monitored area into static sub-regions and allocating resources based on historical load data or simple rules. Deterministic optimization algorithms, such as linear programming or traditional multi-objective optimization, are used to minimize scheduling costs or maximize resource utilization. However, this method has significant limitations: static sub-regions cannot adapt to dynamic changes in load and resources, leading to supply-demand mismatches and low resource utilization; deterministic optimization models struggle to handle uncertainties such as renewable energy fluctuations or equipment malfunctions, resulting in high load loss rates in high-risk areas; most schemes rely on single-type data, neglecting the fusion and analysis of multimodal data, leading to biased scheduling decisions; and scheduling command generation cycles are long and lack real-time risk monitoring mechanisms, failing to meet the rapid response requirements in complex environments. These limitations highlight the inadequacies of existing technologies in handling dynamism, uncertainty, and data diversity, necessitating a new method that can integrate multimodal data, achieve dynamic adaptive partitioning, and introduce risk-driven optimization mechanisms to improve the accuracy and reliability of VPS scheduling. Summary of the Invention

[0004] This application provides an optimized scheduling method for virtual power plants, which solves the problems of inflexible resource allocation, insufficient scheduling reliability, and lagging response mechanisms caused by rigid static partitioning in existing technologies. It achieves the following: constructing an adaptive scheduling foundation through multimodal data fusion and dynamic feature analysis; realizing dynamic optimization of sub-regions by using real-time clustering and boundary stability threshold triggering mechanisms; improving resource utilization accuracy by combining demand difference calculation and multi-objective optimization to generate elastic scheduling instructions; and generating risk avoidance schemes by relying on risk index quantitative assessment and robust optimization algorithms. The application achieves the technical effect of reducing load loss rate in high-risk areas and improving overall resource utilization.

[0005] This application provides an optimized scheduling method for a virtual power plant, including: S1: Acquire multimodal data of the detection area, calculate the regional load fluctuation index and resource mobility index, and generate a dynamic feature dataset; S2: Based on the dynamic feature dataset, the power grid nodes are clustered in real time using a clustering algorithm to generate a dynamic set of sub-regions; if the feature fluctuation of a sub-region exceeds the boundary stability threshold, the boundary is re-divided. S3: Based on the dynamic sub-region set, compare the predicted building power consumption with the available load of charging piles in real time to calculate the demand difference, and generate elastic scheduling instructions through multi-objective optimization; S4: Monitor uncertainties in real time and calculate the risk index, and dynamically adjust the priority of elastic scheduling instructions and resource allocation strategies based on the risk index; generate risk avoidance scheduling schemes through robust optimization algorithms to minimize the load loss rate in high-risk areas and maximize the overall resource utilization rate; the uncertainties include renewable energy output fluctuations, equipment status anomalies and external events.

[0006] Furthermore, the multimodal data includes real-time load data, electric vehicle GPS trajectory data, renewable energy output data, equipment status data, and external event data.

[0007] Furthermore, the regional load fluctuation index is obtained by calculating the load change rate between adjacent time periods, and a sliding time window mechanism is used in the calculation: , in, For time points According to the dynamic window Calculated load fluctuation index; Indicates a point in time Perform a traversal summation, the range being from... arrive ; This represents the relative absolute value of load changes between adjacent time periods. For time points The load value, For time points The load value, It is the normalization factor, calculated by dividing the summation result by the window size. .

[0008] Furthermore, the resource mobility index is calculated using the spatiotemporal variance of charging pile utilization, and a weighting factor is introduced to prioritize high mobility areas. , in, For utilization rate Within a given time period and spatial regions Spatiotemporal variance within, The average utilization rate of all spatiotemporal points. Represent each location point In time utilization rate Compared with the overall average utilization rate The square of the deviation, As the normalization factor, It is the total number of time periods within the time period. It refers to the size of the area within the spatial region; The resource mobility index introduces weighting factors: , in, This is a resource mobility index. The weighting of resource movement is adjusted using GPS trajectory data from electric vehicles.

[0009] Furthermore, the dynamic sub-region set is a spatiotemporally adaptive sub-region division result generated through real-time data analysis, characterized by: the validity period of each sub-region being synchronized with the data update cycle; the sub-region boundaries dynamically changing according to load distribution, with the smallest division unit being a single grid node; and the calculation of boundary stability thresholds. , in, The boundary stability threshold, The area involved in the change of the sub-region boundary. The total area; when When the value falls below a preset threshold, the system triggers a boundary re-division. The preset threshold is obtained based on statistical analysis of historical data.

[0010] Furthermore, the demand difference is the relative difference between the predicted building power consumption and the available load of charging stations: , in, For the demand difference, This is the predicted electricity consumption value for the building. This refers to the available load for the charging station.

[0011] Furthermore, the elastic scheduling instruction is a resource allocation instruction set generated by a multi-objective optimization algorithm. When the demand difference exceeds a preset threshold, the instruction is triggered to generate, including: target sub-region identifier, resource allocation direction, scheduling amount, time parameter and priority flag.

[0012] Furthermore, the risk index is a quantitative indicator calculated by weighted fusion of multi-dimensional uncertainty factors, used to assess the scheduling risk of a sub-region. , in, As a risk index, As a component of the volatility risk of renewable energy, As a component of equipment condition risk, As a component of external event risk, For the corresponding weights, and .

[0013] Furthermore, the risk avoidance scheduling scheme includes: allocating reserve capacity to sub-regions where the risk index exceeds a preset threshold; simultaneously optimizing and minimizing the load loss rate and maximizing resource utilization; dynamically managing grid security constraints, equipment physical constraints, and spatiotemporal coupling constraints; and using a multi-stage optimization algorithm to generate the scheme, triggering real-time updates when the risk index change rate exceeds the threshold.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages: By employing multimodal data fusion and dynamic feature analysis, real-time clustering algorithms and boundary stability threshold triggering mechanisms, demand difference calculation models and multi-objective optimization algorithms, as well as risk index quantitative assessment systems and robust optimization algorithms, we have achieved comprehensive perception and accurate prediction of the power grid operation status, dynamic adaptive sub-region division, significant improvement in resource allocation accuracy and utilization efficiency, and effective response to uncertainties such as renewable energy fluctuations. This has resulted in reduced load loss rates in high-risk areas, improved overall resource utilization, and shortened dispatch response time, significantly enhancing the economy and reliability of virtual power plant operation. Attached Figure Description

[0015] Figure 1 This is a flowchart of an optimized scheduling method for a virtual power plant according to an embodiment of the present invention. Detailed Implementation

[0016] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0018] Example 1: As Figure 1 As shown, an optimized scheduling method for a virtual power plant is proposed.

[0019] S1: Acquire multimodal data of the detection area, calculate the regional load fluctuation index and resource mobility index, and generate a dynamic feature dataset; The multimodal data includes real-time load data, electric vehicle GPS trajectory data, renewable energy output data, equipment status data, and external event data.

[0020] Specifically, multimodal data is collected in real time within the monitoring area using hardware such as IoT sensors, smart meters, GPS devices, and power grid monitoring systems. This data includes real-time load data, such as power values ​​at power grid nodes; GPS trajectory data of electric vehicles, including vehicle location, speed, and direction; renewable energy output data, such as predicted and actual output values ​​of photovoltaic power plants or wind farms; equipment status data, such as the health index of charging piles and the charging and discharging status of battery storage systems; and external event data, such as power grid fault alarms, market price fluctuation signals, and sudden weather events. The data collection frequency is synchronized with the virtual power plant's dispatch cycle.

[0021] The regional load fluctuation index is derived by calculating the load change rate between adjacent time periods, and a sliding time window mechanism is used in the calculation: , in, For time points According to the dynamic window Calculated load fluctuation index; Indicates a point in time Perform a traversal summation, the range being from... arrive ; This represents the relative absolute value of load changes between adjacent time periods. For time points The load value, For time points The load value, if =0, then =0; It is the normalization factor, calculated by dividing the summation result by the window size. Such as window size =5 (i.e., 5 time points). The system iterates through the most recent 5 time periods, calculates the load change rate for each time period, and takes the average to obtain the final result. Moment .like A high value indicates that the load in the area fluctuates drastically, requiring priority resource allocation.

[0022] The resource mobility index is calculated using the spatiotemporal variance of charging pile utilization, and a weighting factor is introduced to prioritize high mobility areas. , in, For utilization rate Within a given time period and spatial regions Spatiotemporal variance within, The average utilization rate of all spatiotemporal points. Represent each location point In time utilization rate Compared with the overall average utilization rate The square of the deviation, As the normalization factor, It is the total number of time periods within the time period. It refers to the size of the area within the spatial region; The resource mobility index introduces weighting factors: , in, This is a resource mobility index. Weighting of resource movement is adjusted using electric vehicle GPS trajectory data: , For adjustment coefficients, The number of vehicles entering the country. The total number of vehicles, and not 0; otherwise... =1. If the inflow rate of electric vehicles in a certain area is high, then increase... Value, such as >1, to increase the scheduling priority of this area.

[0023] Will , By associating real-time multimodal data with time series data, a table is generated for each record, containing a timestamp, spatial coordinates, exponential value, and original data, thus creating a dynamic feature dataset. The dataset update cycle is synchronized with data acquisition to ensure real-time performance. When data anomalies are detected, data backfilling or predictive filling algorithms are triggered.

[0024] S2: Based on the dynamic feature dataset, the power grid nodes are clustered in real time using a clustering algorithm to generate a dynamic set of sub-regions; if the feature fluctuation of a sub-region exceeds the boundary stability threshold, the boundary is re-divided. Specifically, the dynamic sub-region set is the result of sub-region division generated through real-time data analysis. The validity period of each sub-region is strictly synchronized with the data update cycle to ensure the timeliness of the sub-region division. The default data update cycle is set to 5 minutes, and the validity period of the sub-region is correspondingly set to 5 minutes, automatically becoming invalid and regenerated when the data is updated. The sub-region boundaries change dynamically according to the load distribution, and a real-time clustering algorithm is used to divide the power grid nodes. The feature vector of each power grid node contains... Current value Values, latitude and longitude coordinates, and connection strength with adjacent nodes. The initial number of clusters is dynamically determined based on the region size. , in, The initial number of clusters, The total number of power grid nodes. This is an empirical constant, obtained through training with historical data, and its value ranges from [5, 10]. Clustering calculations are performed every 5 minutes, using a sliding window mechanism, and only data from the most recent hour is processed.

[0025] The system calculates the boundary stability threshold in real time: , in, The boundary stability threshold, The area involved in the change of the sub-region boundary. The total area; when When the value falls below a preset threshold, the system triggers a boundary re-division. The preset threshold is obtained based on statistical analysis of historical data.

[0026] S3: Based on the dynamic sub-region set, compare the predicted building power consumption with the available load of charging piles in real time to calculate the demand difference, and generate elastic scheduling instructions through multi-objective optimization; Specifically, the building power consumption forecast is for each dynamic sub-region, based on historical power consumption data, weather information, and workday type, and uses a forecasting model to predict the total building power consumption in real time within a future scheduling cycle; the available load of charging piles is the current available load capacity of all charging piles in the sub-region.

[0027] The demand difference is the relative difference between the predicted building electricity consumption and the available load of charging stations: , in, For the demand difference, The larger the value, the more severe the mismatch between supply and demand. This is the predicted electricity consumption value for the building. This refers to the available load for the charging station.

[0028] The elastic scheduling instruction is a resource allocation instruction set generated by a multi-objective optimization algorithm. The instruction is triggered when the demand difference exceeds a preset threshold. It includes: target sub-region identifier, resource allocation direction, scheduling amount, time parameter and priority flag.

[0029] Specifically, a dynamic threshold is set, such as 0.1, when the demand difference in the sub-region... When the threshold is exceeded, the system automatically triggers command generation. The threshold can be adjusted based on historical data statistical analysis; for example, the threshold can be lowered to 0.05 during peak load periods to improve scheduling sensitivity. The system calculates the dynamic sub-regions every 5 minutes. The value is then compared with a threshold. If the threshold is reached, the sub-region is marked as the scheduling target, and the multi-objective optimization algorithm is started.

[0030] The elastic scheduling instruction is a resource allocation instruction set generated by a multi-objective optimization algorithm. The instruction is triggered when the demand difference exceeds a preset threshold. It includes: target sub-region identifier, resource allocation direction, scheduling amount, time parameter and priority flag.

[0031] Specifically, the target sub-region identifier adopts a unique encoding format of region ID + timestamp to ensure the real-time correspondence between the instruction and the dynamic sub-region set; records the sub-region boundary coordinates, the list of included power grid nodes and their topological connections; and marks the validity period of the sub-region division.

[0032] The direction of resource allocation is determined based on the positive or negative value of the demand difference: when When the value is greater than 0, an injection instruction is issued to allocate resources from the redundant region to the target sub-region. When the load is less than 0, an absorption command is marked to transfer excess load to an energy storage system or a nearby area; the optimal transmission path is marked in conjunction with the power grid topology to avoid line overload.

[0033] The calculation of the scheduling quantity is as follows: , in, This represents the actual scheduling amount. For the total capacity of the region, This is the loss coefficient.

[0034] The time parameters are configured with multiple time scales: immediate execution (<1 minute), for high-risk sub-regions; planned execution (5-15 minutes), which optimizes the timing of execution based on the load forecast curve; the duration is dynamically set according to the demand difference change rate, with a minimum unit of 5 minutes and a maximum not exceeding the validity period of the sub-region.

[0035] Priority flags are set based on demand differences. ≥0.2 indicates high priority, marked as urgent, and requires immediate execution; 0.1≤ A score <0.2 indicates medium priority, is marked as important, and will be scheduled within the current scheduling cycle; A value <0.1 indicates low priority, is marked as normal, and is processed according to the resource availability.

[0036] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages: This application employs multimodal data fusion and dynamic feature analysis. By collecting real-time load data, electric vehicle GPS trajectory data, renewable energy output data, equipment status data, and external event data, and calculating regional load fluctuation index and resource mobility index, a dynamic feature dataset is generated, enabling comprehensive perception and accurate prediction of the power grid's operating status. Utilizing a real-time clustering algorithm and a boundary stability threshold triggering mechanism, power grid nodes are clustered and partitioned in real-time based on the dynamic feature dataset, generating a dynamic sub-region set. When the feature fluctuation of a sub-region exceeds a threshold, boundary re-partitioning is triggered, achieving dynamic adaptive sub-region partitioning and effectively overcoming the inflexible resource allocation problem caused by static partitioning rigidity. Combining a demand difference calculation model and a multi-objective optimization algorithm, the demand difference is calculated by comparing the predicted building power consumption with the available load of charging piles in real time. When the demand difference exceeds a preset threshold, an elastic dispatch instruction is generated, significantly improving resource allocation accuracy and utilization efficiency, reducing supply-demand mismatch and resource waste, thereby improving the overall response speed and economy of virtual power plant dispatch.

[0037] Example 2: In Example 1, scheduling decisions were mainly based on known supply and demand data, but real-time monitoring and quantitative evaluation of uncertain factors were lacking. This example further supplements the content of Example 1.

[0038] S4: Monitor uncertainties in real time and calculate the risk index, and dynamically adjust the priority of elastic scheduling instructions and resource allocation strategies based on the risk index; generate risk avoidance scheduling schemes through robust optimization algorithms to minimize the load loss rate in high-risk areas and maximize the overall resource utilization rate; the uncertainties include renewable energy output fluctuations, equipment status anomalies and external events.

[0039] Specifically, the system collects real-time data on the deviation between predicted and actual power output values ​​through photovoltaic power plants and wind farm monitoring systems; it uses smart meters and sensors to monitor the health of charging piles, energy storage devices, etc., with data streams transmitted in real time and abnormal thresholds set based on historical statistics; and it integrates external events with the power grid SCADA system, weather forecast API, and event logs to capture fault signals, price fluctuations, or extreme weather events in real time.

[0040] The risk index is a quantitative indicator calculated by weighted fusion of multi-dimensional uncertainty factors, used to assess the scheduling risk of a sub-region. , in, As a risk index, As a component of the volatility risk of renewable energy, As a component of equipment condition risk, As a component of external event risk, For the corresponding weights, and It is dynamically adjusted based on historical data.

[0041] The renewable energy volatility risk component is calculated using the power output forecast error rate: , in, As a component of the volatility risk of renewable energy, To predict the output value, and >0, This is the actual output value; if the error rate exceeds 10%, It increases linearly.

[0042] The equipment condition risk component is calculated using the equipment health index: , As a component of equipment condition risk, The normalized health score is obtained based on fault history and runtime, and its value range is [0, 1].

[0043] The external event risk component uses an event severity score: , in, As a component of external event risk, Event levels are assigned, such as power grid failure = 1.0, weather warning = 0.5; The influence range coefficient is determined based on the proportion of affected nodes, and its value ranges from [0, 1].

[0044] The risk index is used to dynamically adjust the priority and resource allocation of elastic scheduling instructions: when <0.3 indicates low risk; the instruction priority is marked as normal and executed according to plan. 0.3≤ A score <0.6 indicates medium risk; the priority is raised to important, and resource allocation is weighted. A value ≥0.6 indicates high risk; the priority is marked as urgent, and immediate action is taken, with priority allocation of standby capacity.

[0045] Adjust the demand difference according to the risk index, such as when When >0.5, the scheduling quantity value Adjusted to This enhances redundancy; high-risk sub-regions are automatically associated with nearby low-risk regions to form mutual assistance clusters, thus avoiding line overload.

[0046] The risk avoidance scheduling scheme includes: allocating reserve capacity to sub-regions where the risk index exceeds a preset threshold; simultaneously optimizing and minimizing the load loss rate and maximizing resource utilization; dynamically managing power grid security constraints, equipment physical constraints, and spatiotemporal coupling constraints; and using a multi-stage optimization algorithm to generate the scheme, triggering real-time updates when the risk index change rate exceeds the threshold.

[0047] Specifically, preset risk index thresholds, such as ≥0.6, when the risk index of a sub-region exceeds the threshold, the allocation of standby capacity is automatically triggered; the standby capacity is dynamically allocated from the redundant resources of low-risk sub-regions or the shared energy storage system, for example, the idle load of charging piles and the reserved capacity of battery energy storage systems are designated as standby pools.

[0048] Minimize the high-risk area using a robust optimization algorithm. The goal is to achieve a load lapse rate of ≥0.6 and maximize overall resource utilization; the multi-objective function is: , in, This refers to the set of all sub-regions whose risk indices exceed a preset threshold. For a specific high-risk sub-region In this context, the unmet electricity load demand accounts for the proportion of total demand, i.e., the load deficit rate. For average resource utilization rate, This represents the average resource underutilization rate. This is a tradeoff coefficient, and it is greater than 0, used to balance the two optimization objectives; The scaling factor is a fixed normalization constant based on historical data.

[0049] Grid safety constraints include maintaining stable node voltages, verifying voltage deviation limits to be ±0.05 of the rated voltage through real-time power flow calculations to prevent grid instability; simultaneously, monitoring line capacity to ensure transmission power does not exceed thermal stability limits, and dynamically adjusting dispatch paths to avoid overload. Equipment physical constraints involve maximum output limits for charging piles, and energy storage systems must adhere to upper limits on charge and discharge rates to extend equipment lifespan and ensure safe operation. Spatiotemporal coupling constraints integrate spatial and temporal dimensions. Spatial constraints predict load movement trends based on electric vehicle GPS trajectory data to avoid resource allocation conflicts; temporal constraints require that the effective time and duration of dispatch commands be strictly synchronized with the validity period of sub-regions to ensure timing consistency.

[0050] The multi-stage optimization process begins with the first stage generating a scheduling plan for the next 15 minutes based on historical data and real-time characteristics. The second stage receives the latest risk index every 5 minutes and adjusts the resource allocation strategy in the plan accordingly. The third stage issues the final instruction, monitors the execution effect, and feeds back to the first stage. When the risk index changes... If the threshold is exceeded, such as 5% / minute, the current scheduling cycle is immediately interrupted, and multi-stage optimization is re-executed; the tradeoff coefficient is dynamically adjusted. If the risk rises sharply ( If >10%, then reduce Up to 0.3, prioritizing reliability.

[0051] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages: This application constructs a risk-driven scheduling enhancement mechanism by real-time monitoring of uncertainties such as renewable energy output fluctuations, equipment anomalies, and external events, and by calculating a risk index based on a weighted fusion algorithm. It dynamically adjusts the priority of elastic scheduling commands and resource allocation strategies through the risk index, and uses a robust optimization algorithm to generate risk-avoidance scheduling schemes. This multi-objective function simultaneously optimizes to minimize the load shortage rate in high-risk areas and maximize overall resource utilization. Furthermore, it dynamically manages grid security constraints, equipment physical constraints, and spatiotemporal coupling constraints through a multi-stage optimization algorithm, triggering real-time updates when the risk index change rate exceeds a threshold. This achieves adaptive response to uncertain environments, ultimately reducing the load shortage rate and improving resource utilization.

[0052] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An optimized scheduling method for a virtual power plant, characterized in that, include: S1: Acquire multimodal data of the detection area, calculate the regional load fluctuation index and resource mobility index, and generate a dynamic feature dataset; S2: Based on the dynamic feature dataset, the power grid nodes are clustered in real time using a clustering algorithm to generate a dynamic set of sub-regions; if the feature fluctuation of a sub-region exceeds the boundary stability threshold, the boundary is re-divided. S3: Based on the dynamic sub-region set, compare the predicted building power consumption with the available load of charging piles in real time to calculate the demand difference, and generate elastic scheduling instructions through multi-objective optimization; S4: Monitor uncertainties in real time and calculate the risk index, and dynamically adjust the priority of elastic scheduling instructions and resource allocation strategies based on the risk index; A risk-avoidance scheduling scheme is generated through a robust optimization algorithm to minimize the load loss rate in high-risk areas and maximize the overall resource utilization. The uncertainties include fluctuations in renewable energy output, abnormal equipment status, and external events.

2. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The multimodal data includes real-time load data, electric vehicle GPS trajectory data, renewable energy output data, equipment status data, and external event data.

3. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The regional load fluctuation index is derived by calculating the load change rate between adjacent time periods, and a sliding time window mechanism is used in the calculation: ， in, For time points According to the dynamic window Calculated load fluctuation index; Indicates a point in time Perform a traversal summation, the range being from... arrive ; This represents the relative absolute value of load changes between adjacent time periods. For time points The load value, For time points The load value, It is the normalization factor, calculated by dividing the summation result by the window size. .

4. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The resource mobility index is calculated using the spatiotemporal variance of charging pile utilization, and a weighting factor is introduced to prioritize high mobility areas. ， in, For utilization rate Within a given time period and spatial regions Spatiotemporal variance within, The average utilization rate across all spatiotemporal points. Represent each location point In time utilization rate Compared with the overall average utilization rate The square of the deviation, As the normalization factor, It is the total number of time periods within the time cycle. It refers to the size of the area within the spatial region; The resource mobility index introduces weighting factors: ， in, This is a resource mobility index. The weighting of resource movement is adjusted using GPS trajectory data from electric vehicles.

5. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The dynamic sub-region set is a spatiotemporally adaptive sub-region division result generated through real-time data analysis. Its characteristics include: the validity period of each sub-region is synchronized with the data update cycle; the sub-region boundaries dynamically change according to load distribution, with the smallest division unit being a single grid node; and the boundary stability threshold is calculated. ， in, The boundary stability threshold, The area involved in the change of the sub-region boundary. The total area; when When the value falls below a preset threshold, the system triggers a boundary re-division. The preset threshold is obtained based on statistical analysis of historical data.

6. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The demand difference is the relative difference between the predicted building electricity consumption and the available load of charging stations: ， in, For the demand difference, This is the predicted electricity consumption value for the building. This refers to the available load for the charging station.

7. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The elastic scheduling instruction is a resource allocation instruction set generated by a multi-objective optimization algorithm. The instruction is triggered when the demand difference exceeds a preset threshold. It includes: target sub-region identifier, resource allocation direction, scheduling amount, time parameter and priority flag.

8. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The risk index is a quantitative indicator calculated by weighted fusion of multi-dimensional uncertainty factors, used to assess the scheduling risk of a sub-region. ， in, As a risk index, As a component of the volatility risk of renewable energy, As a component of equipment condition risk, As a component of external event risk, For the corresponding weights, and .

9. The optimized scheduling method for a virtual power plant as described in claim 1, characterized in that, The risk avoidance scheduling scheme includes: allocating reserve capacity to sub-regions where the risk index exceeds a preset threshold; simultaneously optimizing and minimizing the load loss rate and maximizing resource utilization; dynamically managing power grid security constraints, equipment physical constraints, and spatiotemporal coupling constraints; and using a multi-stage optimization algorithm to generate the scheme, triggering real-time updates when the risk index change rate exceeds the threshold.

Images