A distributed power distribution optimization method and system based on a virtual power plant

Abstract

The present invention discloses a distributed power distribution optimization method and system based on a virtual power plant. The method first generates estimated power consumption distribution data according to the power consumption characteristic values, climate and holiday information of the distribution area, and at the same time generates estimated power generation distribution data in combination with target historical power generation records, etc. Then, it determines the peak power consumption period and the total power demand by judging the estimated power consumption distribution, and determines the buffer capacity of the energy storage device based on this and the estimated power generation distribution data. When the difference between the buffer capacity and the flow of the bottleneck point of the distribution network transmission is lower than the preset margin threshold, the power plant supplementary power ratio is generated by using the distribution ratio adjustment parameter and the real-time load data, and then the power plant scheduling frequency and the access density ratio of distributed power sources are adjusted, and finally the power distribution scheduling scheme for the target distribution period is generated. By adopting the present invention, the grid load can be effectively balanced, the power distribution efficiency can be improved, the access of distributed power sources and renewable energy can be adapted, and the intelligent scheduling and optimal operation of the grid can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid distribution safety, and particularly to a distributed power distribution optimization method and system based on a virtual power plant. Background Art

[0002] In a distribution system based on a virtual power plant, the distribution link, as a key link connecting power generation and power consumption, the efficiency and stability of its operation are crucial. Traditional distribution methods are mostly based on static load forecasting and fixed distribution strategies. When facing the new challenges of increasingly complex and variable power consumption demands and the access of a large number of distributed power sources, many limitations are exposed.

[0003] Firstly, the traditional distribution system lacks accurate grasp of real-time power consumption data and is difficult to accurately estimate the power consumption distribution in advance, resulting in overloading of the power grid in some areas during peak hours, frequent overload warnings and even power outages; while during off-peak hours, there is a large amount of idle waste of power resources, greatly affecting the economy and reliability of the power grid operation. Secondly, existing distribution methods mostly rely on static historical data and fixed distribution strategies, and cannot flexibly adjust the distribution plan according to the real-time changing power consumption demands and power generation situations. In addition, when a large number of distributed power sources are connected to the distribution network, the traditional distribution system cannot effectively coordinate the power flow between them and the power grid. This not only easily causes power quality problems such as voltage fluctuations and harmonic pollution, but also may lead to reverse power flow, threatening the safe and stable operation of the distribution network, making it difficult to fully exert the advantages of distributed power sources and renewable energy. In summary, existing distribution technologies have serious deficiencies in dealing with power grid load balancing, improving distribution efficiency, and adapting to the access of distributed power sources. Summary of the Invention

[0004] The object of the present invention is to provide a distributed power distribution optimization method and system based on a virtual power plant, which can effectively balance the power grid load, improve the distribution efficiency, adapt to the access of distributed power sources and renewable energy, and realize the intelligent scheduling and optimized operation of the power grid.

[0005] To achieve the above object, an embodiment of the present invention provides a distributed power distribution optimization method based on a virtual power plant, including:

[0006] Generating estimated power consumption distribution data according to the power consumption characteristic values in the distribution area, the climate information and holiday information of the target distribution period;

[0007] Generating estimated power generation distribution data according to the target historical power generation records, the proportion of renewable energy, the climate information of the target distribution adjustment period and the power quality detection standards;

[0008] Judge the peak power consumption period of the target power distribution cycle according to the predicted power consumption distribution data, and generate the total power demand during the peak power consumption period;

[0009] Generate the cache capacity of the energy storage device according to the total power demand during the peak power consumption period and the predicted power generation distribution data;

[0010] When the difference between the cache capacity of the energy storage device and the flow rate of the transmission bottleneck point of the distribution network is lower than the preset margin threshold, generate the power plant supplementary power ratio according to the power distribution ratio adjustment parameter and the real-time load data;

[0011] Adjust the matching ratio of the power plant dispatching frequency and the distributed power source access density according to the power plant supplementary power ratio, and generate the power distribution dispatching plan for the target power distribution cycle.

[0012] In an alternative embodiment, after adjusting the matching ratio of the power plant dispatching frequency and the distributed power source access density according to the power plant supplementary power ratio to generate the power distribution dispatching plan for the target power distribution cycle, it further includes:

[0013] Analyze the power distribution network loss under the power distribution dispatching plan to obtain the power distribution network loss evaluation value;

[0014] Generate the power distribution dispatching optimization plan for the target power distribution cycle according to the power distribution network loss evaluation value and the cache capacity of the energy storage device.

[0015] In an alternative embodiment, the generating the predicted power consumption distribution data according to the power consumption characteristic value in the power distribution area, the climate information and holiday information of the target power distribution cycle includes:

[0016] Obtain the target historical power consumption record set according to the climate information and holiday information of the target power distribution cycle;

[0017] Use the clustering algorithm to perform feature aggregation processing on the target historical power consumption record set to obtain the power consumption characteristic value in the power distribution area;

[0018] Analyze the power consumption characteristic value in the power distribution area according to the preset peak power demand threshold to obtain the peak power consumption trend;

[0019] Obtain the real-time power consumption data of the distribution network, input the peak power consumption trend and the real-time power consumption data into a preset linear regression model, and generate the predicted power consumption distribution data within the target power distribution cycle through the linear regression model.

[0020] In an alternative embodiment, the generating the predicted power generation distribution data according to the target historical power generation record, the renewable energy occupancy ratio, the climate information of the target power distribution adjustment cycle and the power quality detection standard includes:

[0021] Perform time series analysis on the target historical power generation records to obtain the power generation trend;

[0022] According to the power generation trend, calculate the power generation ratios of various energy sources to obtain a preliminary estimated power generation distribution;

[0023] Correct the preliminary estimated power generation distribution according to the climate information of the target power distribution adjustment cycle to obtain a corrected estimated power generation distribution;

[0024] Extract the data that meets the preset power quality threshold from the corrected estimated power generation distribution according to the power quality detection standard as the power quality compliance range;

[0025] Generate an intermediate result of the estimated power generation distribution according to the power quality compliance range and the target historical power generation records;

[0026] Use the linear regression algorithm to compare the intermediate result with the target historical power generation records to obtain the estimated power generation distribution data.

[0027] In an alternative embodiment, the determining the peak power consumption period of the target power distribution cycle based on the estimated power consumption distribution data and generating the total power demand for the peak power consumption period includes:

[0028] Generate an estimated power consumption distribution curve according to the estimated power consumption distribution data, and obtain the peak threshold of the same type as the current date from the historical power consumption database;

[0029] On the estimated power consumption distribution curve, calculate the first derivative at a preset time interval, and judge the candidate points of the time period boundary according to the absolute value and continuity of the first derivative;

[0030] Sort the areas between the candidate points according to the median power consumption, and mark the continuous time periods higher than the peak threshold as peak periods;

[0031] When the power demand during the peak period exceeds a preset ratio of the peak threshold, use the sliding window method to re-divide the time period until the threshold condition is met or the maximum number of iterations is reached;

[0032] Calculate the Euclidean distance between the current time period division result and the historical power consumption data. If the distance exceeds the preset tolerance, fine-tune the time period boundary by the sliding window mean method to make the error from the historical power consumption data less than the preset value to obtain the final time period division result;

[0033] Calculate the total power demand for the peak period according to the final time period division result.

[0034] In an alternative embodiment, generating the buffer capacity of the energy storage device according to the total power demand during the peak power consumption period and the predicted power generation distribution data includes:

[0035] Determine the change trend of the power demand according to the total power demand during the peak power consumption period;

[0036] Input the change trend of the power demand into a pre-established linear regression model, and calculate the output distribution of the power generation equipment according to the predicted power generation distribution data;

[0037] According to the output distribution of the power generation equipment, use the sliding window algorithm to determine the charge and discharge frequency of the energy storage device within the response duration;

[0038] When the charge and discharge frequency exceeds the preset charge and discharge frequency threshold, use the quadratic programming method to adjust the buffer capacity of the energy storage device;

[0039] According to the adjusted buffer capacity of the energy storage device, use the proportional-integral controller to judge the influence range of the power limit on the response of the power generation equipment;

[0040] According to the influence range, use the dynamic programming algorithm to determine the dynamic adjustment demand result of the buffer capacity of the energy storage device;

[0041] Judge the final buffer capacity of the energy storage device according to the dynamic adjustment demand result.

[0042] In an alternative embodiment, when the difference between the buffer capacity of the energy storage device and the flow rate of the transmission bottleneck point of the distribution network is lower than the preset margin threshold, generating the power plant supplementary power ratio according to the distribution ratio adjustment parameter and the real-time load data includes:

[0043] Calculate the difference between the buffer capacity of the energy storage device and the transmission bottleneck flow rate of the distribution network to obtain the energy storage capacity margin;

[0044] Use the random forest algorithm to analyze the historical load data and establish a load prediction model;

[0045] Use the load prediction model to predict the future load to obtain the predicted load value;

[0046] Generate the power plant supplementary power ratio according to the energy storage capacity margin and the predicted load value.

[0047] In an alternative embodiment, adjusting the ratio of the power plant dispatching frequency and the distributed power source access density according to the power plant supplementary power ratio to generate a power distribution dispatching plan for the target power distribution period includes:

[0048] When the supplementary power ratio of the power plant is greater than a preset supplementary power ratio threshold, calculate the initial power plant dispatching frequency;

[0049] Use the least squares method to analyze the matching change trend between the initial power plant dispatching frequency and the distributed power access density data;

[0050] If the matching change trend exceeds a preset change trend threshold, use the dynamic programming algorithm to update the power distribution rules, optimize the initial power plant dispatching frequency until the supplementary power ratio of the power plant is less than the preset supplementary power ratio threshold, and use the logistic regression algorithm to predict the power distribution rule adjustment plan to determine the power distribution rules;

[0051] According to the power distribution rules, obtain the operation status data from the virtual power plant and generate a power distribution dispatching plan for the target power distribution period.

[0052] In an alternative implementation, the generating a power distribution dispatching optimization execution plan for the target power distribution period according to the power distribution network loss evaluation value and the cache capacity of the energy storage device includes:

[0053] When the power distribution network loss evaluation value is greater than a preset network loss threshold, obtain the updated data of the cache capacity of the energy storage device, perform a first-order difference processing on the updated data to obtain a cache capacity change trend value;

[0054] Input the capacity change trend value and the real-time current fluctuation data of the power distribution network into a preset support vector regression model to obtain an electricity consumption accuracy adjustment coefficient;

[0055] According to the electricity consumption accuracy adjustment coefficient, adjust the real-time electrical parameter data of the power distribution network and calculate the adjusted power distribution network loss evaluation value;

[0056] If the adjusted power distribution network loss evaluation value is still greater than the network loss threshold, generate a power distribution network dispatching optimization plan including the charge and discharge rate and the switching time sequence based on the current energy storage device parameters of the power distribution network;

[0057] Extract the charge and discharge rate threshold and the maximum capacity ratio from the power distribution network dispatching optimization plan, and generate a first execution plan including the feasible time period according to the current cache capacity of the energy storage device, the charge and discharge rate threshold and the maximum capacity ratio requirements;

[0058] Input the first execution plan and the real-time current fluctuation data of the power distribution network into a preset random forest classifier to predict the load distribution probability of each time period;

[0059] Screen the time period combinations with a load distribution probability lower than a preset probability threshold to obtain a second execution plan;

[0060] Extract the node load balancing degree data in the second implementation plan. When the sampling error of the load balancing degree data is less than a preset error threshold, generate an optimized implementation plan for power distribution scheduling in the target power distribution cycle.

[0061] To achieve the above objectives, an embodiment of the present invention also provides a distributed power distribution optimization system based on a virtual power plant, including:

[0062] An estimated power consumption distribution module, configured to generate estimated power consumption distribution data according to the power consumption characteristic values in the power distribution area, the climate information and holiday information in the target power distribution cycle;

[0063] An estimated power generation distribution module, configured to generate estimated power generation distribution data according to the target historical power generation records, the proportion of renewable energy, the climate information in the target power distribution adjustment cycle, and the power quality detection standards;

[0064] A power demand calculation module, configured to judge the peak power consumption period in the target power distribution cycle according to the estimated power consumption distribution data, and generate the total power demand in the peak power consumption period;

[0065] A cache capacity calculation module, configured to generate the cache capacity of the energy storage device according to the total power demand in the peak power consumption period and the estimated power generation distribution data;

[0066] A power plant supplementary power ratio generation module, configured to generate a power plant supplementary power ratio according to the power distribution ratio adjustment parameters and real-time load data when the difference between the cache capacity of the energy storage device and the flow of the transmission bottleneck point of the power distribution network is lower than a preset margin threshold;

[0067] A power distribution scheduling plan generation module, configured to adjust the power plant scheduling frequency and the ratio of the distributed power source access density according to the power plant supplementary power ratio, and generate a power distribution scheduling plan for the target power distribution cycle.

[0068] Compared with the prior art, a distributed power distribution optimization method and system based on a virtual power plant provided by an embodiment of the present invention, first, the method first generates estimated power consumption distribution data based on the power consumption characteristic values in the power distribution area, climate and holiday information, and at the same time generates estimated power generation distribution data in combination with target historical power generation records, etc., and then judges the peak power consumption period through the estimated power consumption distribution and obtains the total power demand. Based on this, the cache capacity of the energy storage device is determined in combination with the estimated power generation distribution data. When the difference between the cache capacity and the flow of the transmission bottleneck point of the power distribution network is lower than the preset margin threshold, a power plant supplementary power ratio is generated using the power distribution ratio adjustment parameters and real-time load data, and then the power plant scheduling frequency and the ratio of the distributed power source access density are adjusted, and finally a power distribution scheduling plan for the target power distribution cycle is generated. Using the present invention can effectively balance the grid load, improve the power distribution efficiency, adapt to the access of distributed power sources and renewable energy, and realize the intelligent scheduling and optimized operation of the power grid. Brief Description of the Drawings

[0069] To more clearly illustrate the technical solutions of the present invention, the drawings to be used in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0070] Figure 1 is a schematic flowchart of a distributed power distribution optimization method based on a virtual power plant provided by an embodiment of the present invention;

[0071] Figure 2 is another schematic flowchart of a distributed power distribution optimization method based on a virtual power plant provided by an embodiment of the present invention;

[0072] Figure 3 is a structural block diagram of a distributed power distribution optimization system based on a virtual power plant provided by an embodiment of the present invention;

[0073] Figure 4 is a structural block diagram of a distributed power distribution optimization device based on a virtual power plant provided by an embodiment of the present invention. Detailed Embodiments

[0074] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0075] See Figure 1 , Figure 1 is a schematic flowchart of a distributed power distribution optimization method based on a virtual power plant provided by an embodiment of the present invention. The distributed power distribution optimization method based on a virtual power plant includes steps S1 to S6:

[0076] Step S1: Generate estimated power consumption distribution data according to the power consumption characteristic values in the distribution area, the climate information and holiday information of the target distribution period;

[0077] In an alternative embodiment, step S1 includes steps S101 to S104:

[0078] Step S101: Obtain a set of target historical power consumption records according to the climate information and holiday information of the target distribution period;

[0079] Step S102: Perform feature aggregation processing on the target historical electricity consumption record set using a clustering algorithm to obtain the electricity consumption characteristic values within the distribution area;

[0080] Step S103: Analyze the electricity consumption characteristic values within the distribution area according to a preset peak electricity demand threshold to obtain the electricity consumption trend during peak periods;

[0081] Step S104: Obtain the real-time electricity consumption data of the distribution network, input the peak-period electricity consumption trend and the real-time electricity consumption data into a preset linear regression model, and generate the estimated electricity consumption distribution data within the target distribution period through the linear regression model.

[0082] Exemplarily, obtain the climate information and holiday information of the target distribution adjustment period, use the climate information and holiday information as retrieval conditions, obtain the electricity consumption records of historical periods similar to the retrieval conditions from a preset database to obtain the target historical electricity consumption record set; perform feature aggregation processing on the target historical electricity consumption record set using the K-means clustering algorithm to obtain the aggregated electricity consumption characteristic value set; obtain the preset peak electricity demand threshold, and for the aggregated electricity consumption characteristic value set, determine whether each characteristic value in the set is greater than the peak electricity demand threshold. If so, mark the electricity consumption record corresponding to this characteristic value as peak-period electricity consumption data; perform trend analysis on the marked peak-period electricity consumption data using the ARIMA model to obtain the peak-period electricity consumption trend; obtain the real-time electricity consumption data of the current distribution network, input the peak-period electricity consumption trend and the real-time electricity consumption data into a preset linear regression model, and generate the estimated electricity consumption distribution data within the target distribution adjustment period through the linear regression model.

[0083] In an alternative implementation, before the step S101, the method further includes:

[0084] A1. Obtain the real-time electricity consumption data and distributed power source access density within the distribution area;

[0085] A2. Perform aggregation analysis and eigenvalue extraction on the real-time electricity consumption data to obtain the first electricity consumption characteristic value;

[0086] A3. Construct an electricity consumption pattern recognition model according to the first electricity consumption characteristic value and the power source access density information.

[0087] Exemplarily, obtain the real-time power consumption data and distributed power source access information of each monitoring point in the distribution area; according to the real-time power consumption data, use a clustering algorithm to aggregate the load data to obtain an aggregation range covering small power source access points; obtain historical scheduling data, and calculate the scheduling frequency of the virtual power plant; obtain the meteorological data corresponding to the scheduling frequency period; use the correlation analysis method to calculate the matching degree between the scheduling frequency and the meteorological data to obtain a matching result; for the aggregated load data, use the principal component analysis method to extract the first power consumption eigenvalue; according to the first power consumption eigenvalue and the power source access density information, construct a power consumption pattern recognition model; use the power consumption pattern recognition model to dynamically evaluate the load distribution and power source utilization situation in the distribution area to obtain an evaluation result; according to the matching result and the evaluation result, optimize the scheduling strategy of the virtual power plant.

[0088] Exemplarily, in a distribution area of a certain city, the monitoring points include 10 residential communities and 5 small photovoltaic power stations. The real-time power consumption data can be collected by smart meters every 15 minutes, and the recorded power values are such as 100 kW, 80 kW, etc.; the distributed power source access information includes the 50 kW rated power and the access location of the photovoltaic power station. This data provides the original basis for subsequent analysis. According to the real-time power consumption data, use a clustering algorithm to aggregate the load data to obtain an aggregation range covering small power source access points. Exemplarily, the K-means algorithm can be used to cluster the power consumption loads of residential communities according to the peak values (such as 90 kW, 60 kW) and time periods (8 am, 6 pm), and the result may form 3 load clusters, and each cluster covers the nearby photovoltaic power station. This can clarify which areas' loads match the power sources and help optimize resource allocation. Obtain historical scheduling data and calculate the scheduling frequency of the virtual power plant.

[0089] Exemplarily, assume that within the past 30 days, the virtual power plant is scheduled 4 times a day, and each time the photovoltaic output or load distribution is adjusted, then the scheduling frequency is once every 6 hours. This frequency reflects the dynamic response ability of the system. Obtain the meteorological data corresponding to the scheduling frequency period, such as the solar radiation intensity (500 W / m², 700 W / m²) and temperature (25 °C, 30 °C) at 8 am, 14 pm, and 20 pm every day. Use the correlation analysis method to calculate the matching degree between the scheduling frequency and the meteorological data. If it is found that the scheduling time is consistent with the peak radiation intensity, the correlation coefficient may reach 0.8, indicating that the scheduling is highly matched with the power generation capacity. This matching result can guide scheduling optimization.

[0090] Furthermore, for the aggregated load data, the principal component analysis method is adopted to extract the electricity consumption characteristic values. Specifically, characteristics such as peak load (90 kW) and daily average electricity consumption (600 kWh) are extracted from the load clusters. These characteristic values condense the electricity consumption patterns and facilitate pattern recognition. According to the electricity consumption characteristic values and the power access density information, an electricity consumption pattern recognition model is constructed. For example, by combining the characteristic values and the photovoltaic density (2 access points per square kilometer), a machine learning model such as a support vector machine is used to identify the "peak electricity consumption - high photovoltaic output" pattern. This model can predict the load distribution trend.

[0091] The electricity consumption pattern recognition model is used to dynamically evaluate the load distribution and power utilization in the distribution area, and the evaluation results are obtained. Exemplarily, the evaluation shows that the photovoltaic utilization rate reaches 90% at the peak load of a certain community, while it is only 50% in another area, revealing the imbalance of resource utilization. This evaluation provides data support for scheduling optimization. According to the matching results and the evaluation results, the scheduling strategy of the virtual power plant is optimized. It can be understood that if the meteorological matching degree is high but the utilization rate is low, the scheduling frequency can be increased to once every 4 hours, and at the same time, the photovoltaic output is preferentially allocated to the area with low utilization rate. This can improve the overall efficiency and reduce energy waste. Preferably, through dynamic adjustment, the peak pressure on the power grid can also be reduced, achieving a win-win situation of economic benefits and stability.

[0092] It should be noted that the first electricity consumption characteristic values in steps A1 and A2 above and the electricity consumption characteristic values in the distribution area in step S102 both originate from the electricity consumption-related data in the distribution area. Specifically, the aggregated electricity consumption characteristic values in the former are obtained by acquiring the real-time electricity consumption data of each monitoring point in the distribution area, the distributed power access information, etc., aggregating the load data through the clustering algorithm, and extracting them by the principal component analysis method; the electricity consumption characteristic values in S102 are also based on the distribution area electricity consumption data. First, the climate and holiday information of the target distribution adjustment cycle is obtained, and the electricity consumption records of similar historical periods are retrieved from the preset database based on this as the retrieval condition, and then the characteristic aggregation process is carried out by the K-means clustering algorithm.

[0093] Specifically, the electricity consumption characteristic values in the distribution area in steps S101 and S102 are screened and aggregated from the historical electricity consumption records on the basis of the first electricity consumption characteristic values in steps A1 and A2, combined with the climate and holiday information of the target distribution adjustment cycle. It is mainly used to screen the historical electricity consumption records that match the peak threshold of electricity demand, determine the estimated electricity consumption distribution data, and focus more on predicting the electricity consumption situation in the future distribution adjustment cycle. By analyzing the historical electricity consumption characteristic values and combining the current real-time electricity consumption data, the estimated electricity consumption distribution in the future cycle is generated, providing forward-looking guidance for scheduling, which is an extended application of the electricity consumption characteristic values in the time dimension on the basis of the above embodiments.

[0094] Specifically, in step S101, climate information and holiday information for the target power distribution adjustment period are obtained. Exemplarily, the climate information may include key indicators such as temperature, humidity, wind speed, etc. For example, the average temperature in a certain area in summer is 32 degrees Celsius and the humidity is 70%. These data directly affect the air-conditioning load demand. The holiday information involves the type of holiday and the duration. For example, the 7-day National Day holiday usually comes with a sharp increase in residential electricity consumption.

[0095] Furthermore, using these climate information as retrieval conditions, electricity consumption records for similar historical periods are extracted from the database. For example, assuming the target adjustment period is from July 1st to July 7th, historical electricity consumption data for the same period in July and including holidays in the past 5 years can be retrieved to form a set of target historical electricity consumption records. Specifically, from July 1st to July 7th, 2019, the average temperature was 31 degrees Celsius and there were 3 holidays, and the total electricity consumption was 5 million kWh. Such records can be included in the set.

[0096] Specifically, in step S102, when using the K-means clustering algorithm to perform feature aggregation processing on the set. Preferably, clusters are divided based on dimensions such as electricity consumption and time period. For example, the daily electricity consumption is divided into three categories of clusters: low, medium, and high. The cluster centers may be 1 million, 3 million, and 6 million kWh respectively, and finally a set of aggregated electricity consumption characteristic values is obtained. This method can effectively summarize the electricity consumption patterns and facilitate subsequent analysis.

[0097] Specifically, in step S103, first, a preset peak threshold of electricity demand is obtained, for example, set to 4 million kWh, and then each characteristic value in the set of characteristic values is judged one by one. If a certain characteristic value is 4.5 million kWh, it is marked as peak-period electricity consumption data. Exemplarily, peak-period data may be concentrated from 13:00 to 15:00 in the afternoon, reflecting the cooling demand driven by high temperature. Such marking lays the foundation for trend analysis. When using the ARIMA (Autoregressive Integrated Moving Average Model) model for trend analysis on peak-period electricity consumption data. It can be understood that this model predicts future trends through the stationarity analysis of historical data. For example, based on the peak-period data in July in the past 3 years, it is analyzed that the electricity consumption increases by 5% daily. This provides a reference for load forecasting during the adjustment period.

[0098] Specifically, in step S104, the real-time power consumption data of the current distribution network is obtained. For example, the real-time load on July 1 is 3.5 million kWh. This data is combined with the peak period trend and input into a linear regression model. The linear regression model can generate the daily power consumption distribution data within the adjustment period based on the historical slope and real-time deviation. Exemplarily, it is predicted that the load on July 3 will reach 4.2 million kWh. Such estimated data can guide the scheduling optimization and reduce the risk of supply-demand imbalance. Exemplarily, from a climate perspective, high-temperature weather may push up the predicted value, while holidays extend the peak period, and the combination of the two enhances the comprehensiveness of the prediction. Specifically, if the holiday factor is ignored, the predicted value may be underestimated by 20%, affecting the scheduling accuracy. In addition, this method can also dynamically adjust the resource allocation and improve the distribution efficiency.

[0099] Step S2: Generate the estimated power generation distribution data according to the target historical power generation records, the proportion of renewable energy, the climate information of the target distribution adjustment period, and the power quality detection standard;

[0100] In an alternative embodiment, step S2 includes steps S201 to S206:

[0101] Step S201: Perform time series analysis on the target historical power generation records to obtain the power generation trend;

[0102] Step S202: Calculate the power generation proportions of various energy sources according to the power generation trend to obtain the preliminary estimated power generation distribution;

[0103] Step S203: Correct the preliminary estimated power generation distribution according to the climate information of the target distribution adjustment period to obtain the corrected estimated power generation distribution;

[0104] Step S204: Extract the data that meets the preset power quality threshold from the corrected estimated power generation distribution as the power quality compliance range;

[0105] Step S205: Generate an intermediate result of the estimated power generation distribution according to the power quality compliance range and the target historical power generation records;

[0106] Step S206: Use the linear regression algorithm to compare the intermediate result with the target historical power generation records to obtain the estimated power generation distribution data.

[0107] Exemplarily, obtain the target historical power generation records, and use time series analysis to determine the power generation trend; according to the power generation trend, calculate the power generation ratios of various energy sources to obtain a preliminary distribution; obtain climate prediction data, and if the climate prediction data shows fluctuations, combine a preset adjustment period to correct the preliminary distribution to obtain a corrected distribution; for a preset power quality detection standard, extract data that meets the preset threshold from the corrected distribution to determine the quality compliance range; generate an intermediate result of the estimated distribution according to the quality compliance range and the historical power generation records; use a linear regression algorithm to compare the intermediate result with the historical power generation records to obtain the final estimated distribution.

[0108] In an alternative embodiment, before the step S201, the method further includes:

[0109] Obtain the historical power generation data of power generation equipment and the climate of the power generation area, and match the length of the power transmission path and the climate conditions through the access density of distributed power sources to obtain the target historical power generation records.

[0110] Specifically, the obtaining of the historical power generation data of power generation equipment and the climate of the power generation area, and matching the length of the power transmission path and the climate conditions through the access density of distributed power sources to obtain the target historical power generation records includes:

[0111] Obtain the historical records of power generation equipment and the regional climate data;

[0112] Determine the number N of distributed power source equipment of photovoltaic or wind turbine type and the regional area A within the region according to the historical records of the power generation equipment; calculate the access density D using the formula D = N / A;

[0113] Match the pre-stored length L of the power transmission path according to the access density D;

[0114] If D is greater than 5 devices per square kilometer, determine whether the length L of the power transmission path is less than 10 kilometers;

[0115] Extract the matching results that meet the path length conditions and associate the temperature field T in the regional climate data;

[0116] If T is higher than 35 degrees Celsius, adjust the length L of the power transmission path to L×(1 + 2);

[0117] Input the adjusted length L of the power transmission path and the historical power generation data into a linear regression model to obtain the regression coefficient and the intercept term;

[0118] Generate a power generation prediction value for the target area based on the regression coefficient and the intercept term;

[0119] Calculate the correlation between the power generation prediction value and the actual power generation using the Pearson correlation coefficient;

[0120] When the correlation coefficient is lower than 7, eliminate the actual power generation data to obtain the target historical power generation record.

[0121] Exemplarily, obtaining the historical records of power generation equipment and regional climate data is the basis for analyzing the operation of distributed power sources. For example, in a certain coastal area, the historical records may show the installation quantities of photovoltaic equipment and wind turbines in the past five years, while the climate data includes the annual average wind speed and sunshine duration in this area.

[0122] It is worth noting that these data provide the original basis for subsequent analysis. After determining the number N of distributed power source equipment of photovoltaic or wind turbine type and the regional area A within the region according to the historical records of power generation equipment, the distribution of the equipment can be further clarified.

[0123] Exemplarily, in a region with an area of 100 square kilometers, if 50 photovoltaic equipment and 30 wind turbines are counted, then N is 80 units and A is 100 square kilometers. When calculating the access density D using the formula D = N / A, D is 0.8 units per square kilometer. This density reflects the concentration degree of distributed power sources.

[0124] In a possible implementation manner, match the pre-stored power transmission path length L according to the access density D. The path length under a similar density can be found through the historical database. For example, when D is 0.8, the database may show that L is usually 15 kilometers. This matching method utilizes the existing empirical data and improves the efficiency of path planning. If D is greater than 5 equipment per square kilometer, it is necessary to judge whether the power transmission path length L is less than 10 kilometers. Specifically, if D in a certain region is 6 units per square kilometer and L is 8 kilometers, then the condition is met. This judgment helps to identify the transmission bottlenecks in high-density regions. After extracting the matching results that meet the path length conditions, associate the temperature field T. For example, when L is 8 kilometers, if T is 38 degrees Celsius, then proceed to the next step of adjustment. If T is higher than 35 degrees Celsius, adjust the power transmission path length L to L×(1 + 2). For example, if L was originally 8 kilometers, it becomes 24 kilometers after adjustment. This adjustment takes into account the influence of high temperature on the cable transmission efficiency and ensures that the prediction is closer to the actual situation. After inputting the adjusted L and the historical power generation data into the linear regression model, the regression coefficient and the intercept term can be obtained.

[0125] Exemplarily, the historical data may show the linear relationship between power generation and path length, with a positive coefficient and the intercept being the basic power generation. When generating the predicted power generation value for the target region based on the regression coefficient and the intercept term. Preferably, fine-tuning can be performed in combination with the climate conditions of specific months. Exemplarily, the predicted value in summer may be relatively high due to enhanced sunshine. It is worth noting that this method improves the seasonal adaptability of the prediction.

[0126] Exemplarily, when calculating the correlation between the predicted value and the actual power generation using the Pearson correlation coefficient, if the predicted value for a certain month is 1000 kWh and the actual value is 950 kWh, the correlation coefficient may be 0.9. When the correlation coefficient is lower than 7, the data is excluded. For example, if the coefficient is only 0.6 in a certain case, then this outlier is excluded to optimize the model.

[0127] It should be noted that the predicted power generation values of the target area generated through a series of operations in the above steps and the actual power generation data retained after correlation analysis screening provide an important data basis for step S201. These data reflect the historical power generation situation and the relationship with the predicted values, and the target historical power generation records in step S201 are further determined based on the data processed in these steps, which is one of the key bases for generating the estimated power generation distribution data. The process of generating the power generation predicted value in the above steps considers various factors such as the historical data of power generation equipment, climate conditions, and the length of the power transmission path, and establishes a preliminary prediction model. When generating the estimated power generation distribution data in step S2, this logic of comprehensively considering multiple factors is continued, and more information such as the proportion of renewable energy, the climate prediction for the next power distribution adjustment cycle, and the power quality detection standard is further combined to make a more comprehensive and detailed prediction of the power generation distribution situation, which is a further expansion and improvement based on the prediction in the above steps.

[0128] Specifically, in step S201, after obtaining the historical power generation records, time series analysis is used to determine the power generation trend. This process aims to capture the regularity of the power generation changing over time. For example, by analyzing the monthly power generation data in the past five years, the peak trend of photovoltaic power generation in summer or the fluctuation characteristics of wind power generation in autumn can be identified. Assuming that the photovoltaic power generation in a certain area shows an increasing trend year by year in three years, it can be preliminarily judged that it is affected by equipment expansion or improved lighting conditions.

[0129] Furthermore, in step S202, according to the power generation trend, the power generation proportions of various energy sources are calculated to obtain a preliminary distribution. Exemplarily, the historical data of a certain area shows that photovoltaic accounts for 60%, wind power accounts for 30%, and other energy sources account for 10%. This proportion reflects the characteristics of the energy structure.

[0130] Specifically, in step S203, it can be understood that if the photovoltaic proportion is relatively high, it may be related to sufficient sunlight in this area. Then, climate prediction data is obtained, such as the temperature and wind speed forecasts for the next three months. If the prediction shows fluctuations, such as frequent changes in wind speed, the preliminary distribution is corrected in combination with the preset adjustment cycle. Preferably, the adjustment cycle can be set once a month. By comparing the historical fluctuations with the prediction differences, the wind power proportion is dynamically adjusted to 35% and the photovoltaic is adjusted down to 55% to form a corrected distribution.

[0131] Specifically, in step S204, according to the power quality detection standard, data meeting the preset threshold is extracted from the corrected distribution to determine the quality compliance range. Exemplarily, if the standard requires the voltage fluctuation to be less than 5%, the monthly data that remains stable after the increase in the proportion of wind power can be screened out. Assuming that the compliance range accounts for 80% of the whole year. This screening helps to identify reliable power generation periods.

[0132] Specifically, in step S205, the quality compliance range directly affects the stability of subsequent power distribution planning. Therefore, according to the quality compliance range and historical power generation records, an intermediate result of the estimated distribution is generated. For example, combining the average power generation in the compliant months, it is estimated that the proportion of annual photovoltaic power generation may be stable at about 50%.

[0133] Finally, in step S206, a linear regression algorithm is used to compare the intermediate result with the historical records to obtain the final estimated distribution. Exemplarily, through analysis, it is found that the proportion of wind power is positively correlated with the increase in wind speed. The final distribution is adjusted to 48% for photovoltaic, 38% for wind power, and 14% for others. This can be closer to the actual operating state. According to the final estimated distribution and the adjustment period, a power distribution adjustment plan is generated. Exemplarily, if the adjustment period is quarterly, the plan can recommend increasing the proportion of wind power access to 40% in autumn with higher wind speed and optimizing the dormancy time of photovoltaic equipment. This way not only improves the power distribution efficiency but also balances the equipment load and extends the service life.

[0134] It should be noted that the key to time series analysis is to capture the trend characteristics rather than short-term fluctuations, which provides a reliable basis for ratio calculation. The introduction of climate prediction enhances the adaptability of the distribution, especially in a fluctuating environment, and can effectively reduce the risk of quality exceeding the standard. The dynamic adjustment of the final plan ensures the flexibility and sustainability of power distribution.

[0135] Step S3: Determine the peak power consumption period of the target power distribution cycle according to the estimated power consumption distribution data, and generate the total power demand during the peak power consumption period;

[0136] In an alternative embodiment, step S3 includes steps S301 to S306:

[0137] Step S301: Generate an estimated power consumption distribution curve according to the estimated power consumption distribution data, and obtain the peak threshold of the same date type as the current date from the historical power consumption database;

[0138] Step S302: Calculate the first derivative at a preset time interval on the estimated power consumption distribution curve, and judge the candidate points of the time period boundary according to the absolute value and continuity of the first derivative;

[0139] Step S303: Sort the areas between the candidate points according to the median power consumption, and mark the continuous time periods higher than the peak threshold as peak periods;

[0140] Step S304: When the electricity demand during the peak period exceeds a preset ratio of the peak threshold, use the sliding window method to re-divide the time period until the threshold condition is met or the maximum number of iterations is reached;

[0141] Step S305: Calculate the Euclidean distance between the current time period division result and the historical electricity consumption data. If the distance exceeds the preset tolerance, fine-tune the time period boundary by the sliding window mean method to make the error with the historical electricity consumption data less than the preset value, and obtain the final time period division result;

[0142] Step S306: Calculate the total electricity demand during the peak period according to the final time period division result.

[0143] Exemplarily, obtain the original electricity consumption data of the power grid SCADA (Supervisory Control and Data Acquisition) system, use the median filtering method to remove outliers, and perform linear normalization on the filtered data to obtain the preprocessed electricity consumption distribution data; for the preprocessed data, generate a predicted distribution curve through the moving average algorithm, and at the same time obtain the peak threshold of the same date type as the current date from the historical database; on the predicted distribution curve, calculate the first derivative at a preset time interval, and judge the candidate points of the time period boundary according to the absolute value and continuity of the derivative; sort the areas between the candidate points according to the median of electricity consumption, mark the continuous time periods higher than the peak threshold as peak periods, and the rest as flat valley periods; if the demand during the peak period exceeds the preset ratio of the peak threshold, use the sliding window method to re-divide the time period until the threshold condition is met or the maximum number of iterations is reached; calculate the Euclidean distance between the current time period division result and the historical data. If the distance exceeds the preset tolerance, fine-tune the time period boundary by the sliding window mean method to make the error with the historical data less than the preset value; calculate the total demand during the peak period according to the final time period division result.

[0144] Specifically, in step S301, first obtaining the original electricity consumption data of the power grid SCADA system is the basis for analysis. Usually, this data contains information such as voltage, current, and power, and may be affected by equipment noise or sudden failures. For example, in the power data recorded by a regional substation, an abnormal peak occurred at a certain moment due to a short circuit trip. The median filtering method can effectively handle such problems.

[0145] Then, by taking the median value within a certain period to replace the abnormal points. For example, among the power data for 5 consecutive minutes, the jump points far higher than the normal value are removed, and the trend data is retained. This method is more resistant to extreme value interference than the average value. After that, linear normalization processing maps the data to the interval from 0 to 1. For example, the total electricity consumption on a certain day ranges from 5000 kWh to 20000 kWh, and after normalization, it is convenient for subsequent algorithm processing. For the preprocessed data, when the moving average algorithm generates the estimated distribution curve, it can be understood as capturing the trend through a smoothing window. For example, with a 1-hour window, calculate the average electricity consumption per hour to obtain a curve reflecting the daily variation. When obtaining the peak threshold from the historical database. Preferably, records consistent with the current date type (such as weekdays or holidays) can be selected. For example, the historical peak on a certain weekday is 18000 kWh, which is used as a reference.

[0146] Specifically, in step S302, on the estimated electricity consumption distribution curve, calculate the first derivative at a preset time interval. The calculation of the first derivative is used to discover the rate of change. For example, during a certain period, the electricity consumption rises rapidly from 10000 kWh to 15000 kWh, and the absolute value of the derivative increases, indicating that it may enter a peak. Continuity judgment avoids misjudgment. For example, short-term fluctuations are not regarded as boundaries.

[0147] Specifically, in step S303, when sorting the candidate points and marking the time periods, assume that the median electricity consumption on a certain day is 12000 kWh. The consecutive 3 hours higher than the peak threshold of 18000 kWh can be marked as peak time periods, and the rest are flat valleys.

[0148] Specifically, in step S304, if the peak demand exceeds the standard, such as exceeding the threshold by 20%, when the sliding window is re-divided, it can be gradually adjusted with a 30-minute step length until the demand distribution is reasonable. For example, the original 3-hour peak is split into a 2-hour peak plus a 1-hour transition section.

[0149] Specifically, in step S305, the calculation of the Euclidean distance measures the rationality of the division. For example, the deviation between the current division and the historical weekday data is 500 kWh, which is lower than the tolerance of 1000 kWh, and no fine-tuning is required. Otherwise, the boundary is adjusted using the mean method.

[0150] Specifically, in step S306, after the final time period division, the total peak demand may be 36000 kWh. Combining with the transformer capacity, such as the total capacity of 50000 kWh, allocate the load upper limit proportionally. For example, if a certain node capacity accounts for 30%, the upper limit is 15000 kWh.

[0151] In an alternative embodiment, the method further includes combining the transformer capacity parameter, proportionally allocating the load upper limit of each node, generating a corresponding power distribution instruction table. Exemplarily, the power distribution instruction table may stipulate that the power of this node during peak hours does not exceed this value to avoid overload. This method can optimize resource utilization, ensure stable power supply, and facilitate dispatchers to quickly respond to demand changes. In addition, if holiday data comparison is added, the distribution plan can be further refined to adapt to diverse scenarios.

[0152] Step S4: Generate the cache capacity of the energy storage device according to the total power demand during the peak power consumption period and the estimated power generation distribution data;

[0153] In an alternative embodiment, the step S4 includes steps S401 to S407:

[0154] Step S401: Determine the change trend of the power demand according to the total power demand during the peak power consumption period;

[0155] Step S402: Input the change trend of the power demand into a pre-established linear regression model, and calculate the output distribution of the power generation device according to the estimated power generation distribution data;

[0156] Step S403: Determine the charge and discharge frequency of the energy storage device within the response duration by using a sliding window algorithm according to the output distribution of the power generation device;

[0157] Step S404: When the charge and discharge frequency exceeds a preset charge and discharge frequency threshold, adjust the cache capacity of the energy storage device by using a quadratic programming method;

[0158] Step S405: Judge the influence range of the power limit on the response of the power generation device by using a proportional-integral controller according to the adjusted cache capacity of the energy storage device;

[0159] Step S406: Determine the dynamic adjustment demand result of the cache capacity of the energy storage device by using a dynamic programming algorithm according to the influence range;

[0160] Step S407: Judge the final cache capacity of the energy storage device according to the dynamic adjustment demand result.

[0161] Exemplarily, data during peak electricity consumption periods is acquired to determine the changing trend of electricity demand; the changing trend of the electricity demand is input into a pre-established linear regression model, and in combination with the predicted power generation distribution data, the output distribution of power generation equipment is calculated; according to the output distribution of the power generation equipment, the sliding window algorithm is adopted to determine the charge and discharge frequency of the energy storage equipment within the response duration; if the charge and discharge frequency exceeds a preset threshold, the quadratic programming method is adopted to adjust the buffer capacity of the energy storage equipment; according to the adjusted buffer capacity, through the proportional-integral controller, the influence range of the rate limit on the response of the power generation equipment is judged; according to the change of the influence range, the dynamic programming algorithm is adopted to determine the dynamic adjustment requirement of the capacity margin; the result of the dynamic adjustment requirement is acquired to judge the final buffer capacity of the energy storage equipment.

[0162] Specifically, in step S401, acquiring data during peak electricity consumption periods and determining the changing trend of electricity demand is the basis for analyzing the operation of the power system. Exemplarily, in a certain urban power grid, by collecting data during peak periods on summer weekdays, it is found that the electricity demand shows an upward trend from 10 am to 3 pm, with the peak appearing at 12 noon, approximately 500 megawatts. It can be understood that this trend is highly correlated with the peak air-conditioning usage, reflecting the direct connection between temperature and electricity consumption.

[0163] Specifically, in step S402, the above-mentioned changing trend is input into the linear regression model, and in combination with the predicted power generation distribution data, the output distribution of the power generation equipment is calculated. Exemplarily, assume that there are two power generation methods, thermal power and photovoltaic power, in a certain area. The basic output of thermal power is stable at 300 megawatts, and the output of photovoltaic power reaches 150 megawatts at noon due to sunlight changes. Through model prediction, the total output during peak periods needs to reach 480 megawatts, and thermal power needs to supplement an additional 30 megawatts. This method can effectively match the relationship between power generation and demand.

[0164] Specifically, in step S403, according to the output distribution of the power generation equipment, the sliding window algorithm is adopted to determine the charge and discharge frequency of the energy storage equipment. Preferably, the window is set to 15 minutes. During the peak period from 12 noon to 1 pm, the energy storage equipment needs to charge and discharge every 30 minutes, and the total frequency is 2 times per hour.

[0165] Specifically, in step S404, if the preset threshold is 1.5 times per hour, the frequency exceeds the standard. At this time, the quadratic programming method is used to adjust the buffer capacity of the energy storage equipment. For example, the initial capacity is 50 megawatt-hours, and after adjustment, it increases to 70 megawatt-hours to smooth the losses caused by frequent charging and discharging. This adjustment can extend the equipment life and improve the response efficiency.

[0166] Specifically, in step S405, a proportional-integral controller is used to determine the influence range of power limitation on the response of the power generation equipment. For example, the ramp rate of thermal power is limited to a 5 MW per minute ramp-up capacity. When the demand suddenly increases by 50 MW, it takes 10 minutes to fully respond.

[0167] Furthermore, in step S406, if the influence range is manifested as a response delay, which may lead to short-term supply-demand imbalance, the dynamic programming algorithm determines the adjustment requirements for the capacity margin based on this range. Exemplarily, it is calculated that an additional 20 MWh of capacity needs to be reserved to cope with sudden fluctuations. This approach can optimize resource allocation.

[0168] Specifically, in step S407, the buffer capacity of the final energy storage device is determined by dynamically adjusting the requirements. Assuming an initial design capacity of 60 MWh, after dynamic programming, it increases to 80 MWh. The energy storage device can stably support the power grid during peak hours and avoid the risk of overload.

[0169] It should be noted that this capacity design can better adapt to demand fluctuations and improve power supply reliability. The above solution has a clear logical progression from core demand analysis to extended capacity adjustment. Exemplarily, if considering charging during the night valley and discharging during the day peak, the energy storage device can also reduce the power generation cost. Preferably, through multi-faceted verification, such as the collaborative optimization of frequency, capacity, and response range, the final solution is both practical and flexible, fully ensuring the stable operation of the power grid.

[0170] Step S5: When the difference between the buffer capacity of the energy storage device and the flow rate at the transmission bottleneck point of the distribution network is lower than the preset margin threshold, generate the power plant supplementary power ratio according to the distribution ratio adjustment parameter and the real-time load data;

[0171] In an alternative embodiment, step S5 includes steps S501 to S504:

[0172] Step S501: Calculate the difference between the buffer capacity of the energy storage device and the transmission bottleneck flow rate of the distribution network to obtain the energy storage capacity margin;

[0173] Step S502: Analyze the historical load data using the random forest algorithm to establish a load prediction model;

[0174] Step S503: Use the load prediction model to predict the future load to obtain the predicted load value;

[0175] Step S504: Generate the power plant supplementary power ratio according to the energy storage capacity margin and the predicted load value.

[0176] Exemplarily, obtain the real-time capacity data of the energy storage device and the flow data of the transmission bottleneck point; calculate the capacity margin according to the difference between the energy storage capacity and the transmission bottleneck flow, where the capacity margin is equal to the energy storage capacity minus the transmission bottleneck flow; obtain the historical load data, analyze the historical load data using the random forest algorithm, and establish a load prediction model; use the load prediction model to predict the load for the next 24 hours to obtain the predicted load value; determine whether the predicted load value is greater than the capacity margin, and if so, adjust the voltage and current parameters of the distribution line to optimize the load distribution; calculate the proportion of the power that needs to be supplemented by the power plant according to the adjusted distribution parameters and the real-time transmitted load data, where the supplementary power proportion is equal to (the predicted load value minus the capacity margin) divided by the predicted load value; obtain the historical load data, analyze the historical load data using the K-means clustering algorithm, determine the optimal load aggregation range, and achieve the balance of the power grid load and the efficient utilization of the energy storage device.

[0177] Specifically, in step S501, first obtain the real-time capacity data of the energy storage device and the flow data of the transmission bottleneck point, which is the basis for analyzing the operation state of the power grid. For example, in a certain urban power grid, the real-time capacity of the energy storage device may be 5000 kWh, and the flow data of the transmission bottleneck point is 4500 kWh, which reflects the current transmission limit of the power grid. The acquisition of real-time data can be completed by intelligent sensors, which are deployed on the energy storage device and key nodes and update the data every 5 minutes to ensure the timeliness of information. Calculate the capacity margin according to the difference between the energy storage capacity and the transmission bottleneck flow. This method intuitively reflects the buffering capacity of the power grid. Therefore, 5000 kWh minus 4500 kWh gives a capacity margin of 500 kWh. This margin indicates that there is still a certain adjustment space for the energy storage device under the current load. Exemplarily, if the bottleneck flow rises to 4800 kWh due to sudden demand, the margin is reduced to 200 kWh, indicating that potential overload risks need to be concerned.

[0178] Specifically, in step S502, obtaining the historical load data and establishing a load prediction model using the random forest algorithm can improve the prediction accuracy. It can be understood that the random forest analyzes historical data by integrating multiple decision trees. For example, the daily electricity consumption trend in the past year is used to identify that the peak period usually appears in the afternoon of summer.

[0179] Specifically, in step S503, the load prediction model is used to predict the future load. The inputs include features such as weather, holidays, and electricity consumption habits. The model outputs the predicted load values for the next 24 hours. For example, the predicted peak load is 5,200 kWh. This provides a basis for advance planning. Judging whether the predicted load value is greater than the capacity margin is the key to the optimization decision. Exemplarily, if the predicted load of 5,200 kWh is greater than the margin of 500 kWh, the distribution line parameters need to be adjusted.

[0180] In a possible implementation, by raising the voltage to 110 kV and adjusting the current distribution, the load is transferred from the overloaded area to the standby line. This adjustment can effectively relieve the bottleneck pressure and improve the power supply reliability.

[0181] Further, in step S504, the supplementary power ratio is calculated based on the adjusted distribution parameters and the real-time load data, which can further provide clear dispatching instructions for the power plant. Specifically, (5,200 kWh of predicted load - 500 kWh of margin) / 5,200 kWh of predicted load gives a supplementary ratio of approximately 90.4%. This means that the power plant needs to allocate most of the additional power generation resources.

[0182] Preferably, dynamically adjusting the ratio in combination with real-time data can better match the actual demand. Using the K-means clustering algorithm to analyze the historical load data and determine the optimal load aggregation range is an important means to achieve grid balance. It should be noted that K-means clusters the historical loads into three categories: high, medium, and low, and finds the typical distribution of each category of load. For example, the high-load cluster is concentrated during the summer peak electricity consumption period, with an average value of 5,100 kWh.

[0183] In one embodiment, the discharge strategy of the energy storage device can also be adjusted according to the clustering results, releasing energy intensively during high-load periods, thereby smoothing the load curve and improving the energy storage utilization rate. This method can also extend the equipment life and reduce the maintenance cost. If the predicted load surges to 5,500 kWh, far exceeding the margin, the distribution optimization and power supplementation can be implemented jointly. Exemplarily, first, 1,000 kWh of load is shared through line adjustment, and then the power plant supplements the remaining gap. Therefore, this multi-faceted collaborative solution ensures the grid stability and the efficient operation of the energy storage device, while providing data support for future capacity expansion.

[0184] Step S6: Adjust the ratio of the power plant dispatching frequency and the distributed power source access density according to the supplementary power ratio of the power plant, and generate a power distribution dispatching plan for the target power distribution period.

[0185] In an alternative embodiment, step S6 includes steps S601 to S604:

[0186] Step S601: When the power plant supplementary power ratio is greater than the preset supplementary power ratio threshold, calculate the initial power plant scheduling frequency;

[0187] Step S602: Analyze the matching change trend between the initial power plant scheduling frequency and the distributed power source access density data by using the least squares method;

[0188] Step S603: If the matching change trend exceeds the preset change trend threshold, update the power distribution rule by using the dynamic programming algorithm, optimize the initial power plant scheduling frequency until the power plant supplementary power ratio is less than the preset supplementary power ratio threshold, and use the logistic regression algorithm to predict the power distribution rule adjustment plan to determine the power distribution rule;

[0189] Step S604: According to the power distribution rule, obtain the operation status data from the virtual power plant and generate the power distribution scheduling plan for the target power distribution period.

[0190] Exemplarily, obtain the supplementary power ratio data from the power plant, judge whether the current power demand meets the requirement of the demand response participation degree according to the preset threshold. If it exceeds the threshold, calculate the initial scheduling frequency; according to the initial scheduling frequency, obtain the distributed source access density data from the virtual power plant; analyze the dynamic relationship between the scheduling frequency and the access density by using the least squares method to determine the proportional matching change trend; if the proportional matching change trend exceeds the preset threshold, update the power distribution rule by using the dynamic programming algorithm to generate the optimized scheduling frequency; according to the optimized scheduling frequency, obtain the real-time access density data from the distributed source and judge whether it meets the requirement of the demand response participation degree; if there is a difference between the real-time access density data and the supplementary power ratio, use the logistic regression algorithm to predict the power distribution rule adjustment plan to determine the final dynamic power distribution rule; according to the final dynamic power distribution rule, obtain the operation status data from the virtual power plant and generate the complete scheduling plan.

[0191] Specifically, in step S601, when obtaining the supplementary power ratio data from the power plant, the power output and demand data of the current power grid can be collected through a real-time monitoring system. For example, assume that the supplementary power ratio of a power plant in a certain area is 20% and the preset threshold is set at 15%. Obviously, it has exceeded the threshold, which means that it is necessary to further evaluate whether it meets the requirement of the demand response participation degree. It can be understood that the demand response participation degree is usually related to the user's electricity consumption habits and peak loads, so it is necessary to combine historical data to judge whether the current ratio is reasonable. If it exceeds the threshold, calculate the initial scheduling frequency by analyzing the power consumption during the peak period of the power grid, such as setting it to adjust the power distribution strategy once per hour. When obtaining the distributed source access density data from the virtual power plant according to the initial scheduling frequency, the access conditions of distributed power sources such as photovoltaic and wind energy can be concerned.

[0192] Specifically, in step S602, assuming the initial scheduling frequency is once per hour and the distributed source access density is shown as 10 access points per square kilometer. When analyzing the dynamic relationship between the two using the least squares method, a trend curve can be plotted through historical scheduling frequency and access density data to determine the changing trend of the proportional ratio.

[0193] Specifically, in step S603, if the trend shows that the access density decreases as the scheduling frequency increases, it may indicate insufficient response capacity of the distributed source. At this time, if the changing trend exceeds the preset threshold, for example, the density decreases by more than 20%, then the scheduling needs to be optimized. Preferably, when updating the power distribution rules using the dynamic programming algorithm, the power can be redistributed according to the real-time load distribution of the power grid. For example, the scheduling frequency can be adjusted from once per hour to once every 30 minutes to adapt to the rapidly changing demand. After generating the optimized scheduling frequency, real-time access density data is obtained from the distributed source. For example, the density rises to 12 access points per square kilometer after adjustment. It should be noted that if the density data can match the demand for replenishing 20% of the power at this time, it indicates that the participation degree of demand response is satisfied; otherwise, further adjustment is required.

[0194] Furthermore, in step S603, when the power replenishment ratio of the power plant is less than the preset power replenishment ratio threshold, if there is a difference between the real-time access density and the power replenishment ratio, for example, the power supported by the density only accounts for 18% of the demand, then the adjustment plan is predicted through the logistic regression algorithm. Based on historical adjustment records, it can be analyzed which parameters such as voltage or current are more effective, and then the final dynamic power distribution rules can be determined. For example, it may be concluded that the access ratio of distributed sources needs to be increased by 5%.

[0195] Specifically, in step S604, according to the dynamic power distribution rules, operation status data such as equipment operation efficiency and failure rate are obtained from the virtual power plant to generate a complete scheduling plan. The advantage of this method is to improve the flexibility and stability of the power grid. For example, in a certain urban power grid, after optimization through the above method during the peak load period, the scheduling frequency is adjusted to once every 20 minutes, the access density is increased from 10 to 13, and the power replenishment ratio precisely matches the demand, significantly improving the utilization rate of distributed sources. In a possible implementation, if the future load prediction shows that the demand increases by 10%, the rules can be adjusted in advance to ensure continuous balance.

[0196] It is worth noting that this multi-level analysis and adjustment not only optimize the power distribution but also enhance the power grid's ability to respond to sudden demands.

[0197] See Figure 2 , Figure 2 which is another flowchart of a distributed power source power distribution optimization method based on a virtual power plant provided by an embodiment of the present invention. As Figure 2As shown, in an alternative embodiment, after the step S6, the method further includes steps S7 to S8:

[0198] Step S7: Analyze the power distribution network losses under the power distribution scheduling plan to obtain a power distribution network loss evaluation value;

[0199] In an alternative embodiment, the step S7 includes steps S701 to S706:

[0200] Step S701: Calculate the initial value of the power distribution network losses according to the operation parameter data of the power distribution network;

[0201] Step S702: Optimize the power distribution scheduling plan using a linear programming algorithm, calculate the power distribution network loss value of the optimized power distribution scheduling plan, and obtain a second power distribution network loss value;

[0202] Step S703: Judge the load distribution and loss change of the power distribution network according to the difference between the second power distribution network loss value and the initial value of the power distribution network losses;

[0203] Step S704: Obtain the optimization degree of the optimized power distribution scheduling plan for the power distribution network according to the load distribution and loss change of the power distribution network;

[0204] Step S705: Based on the optimization degree, use a clustering algorithm to classify the loss rate values of each power distribution scheduling plan, and determine the best power distribution scheduling plan;

[0205] Step S706: Verify the power distribution network loss value of the power distribution scheduling plan using a statistical analysis tool to obtain a power distribution network loss evaluation value.

[0206] Exemplarily, obtain the voltage, current, and power data of the power distribution network to determine the initial value of the network losses under dynamic scheduling; optimize the dynamic scheduling plan using a linear programming algorithm to obtain the adjusted operation parameters of the power distribution network; calculate the initial value of the network losses and the loss rate values before and after adjustment. If the loss rate value decreases and meets the preset threshold, judge that the adjustment plan matches the load balance; obtain the overall optimization degree of the adjustment plan for the power distribution network according to the load distribution and loss change of the power distribution network; use the K-means clustering algorithm to classify the loss rate values of multiple adjustment plans to determine the best dynamic scheduling plan; verify the evaluation result of the best plan using a statistical analysis tool to obtain a power distribution network loss evaluation value.

[0207] Specifically, in step S701, obtaining the voltage, current, and power data of the distribution network is the basis for dynamic scheduling analysis. For example, in a certain regional distribution network, assuming the voltage of a certain line is 10 kV, the current is 200 A, and the power factor is 0.9, the operating state of the network can be preliminarily estimated through these data. Exemplarily, the voltage data reflects the stability of the line, the current data indicates the load intensity, and the power data is used to calculate the initial losses. In a possible implementation, the acquisition device can record the voltage fluctuations of a certain node within a day, such as dropping from 10 kV to 9.8 kV, and combined with the current change, calculate that the initial network loss is approximately 50 kW. This method facilitates quickly locating high-loss areas and provides a basis for subsequent optimization.

[0208] Specifically, in step S702, when using the linear programming algorithm to optimize the dynamic scheduling scheme. It can be understood that its core lies in adjusting the operating parameters to reduce losses. Specifically, based on the initial data, the algorithm can adjust the power distribution through constraint conditions (such as voltage upper and lower limits). Exemplarily, in a line, if the load of a certain section is too high, resulting in increased losses, linear programming can transfer part of the load to the adjacent line. After adjustment, the voltage stabilizes at 9.9 kV and the loss drops to 40 kW. This adjustment not only reduces waste but also extends the service life of the equipment. Calculating the initial value of the network loss and the loss rate value before and after the adjustment is the key to evaluating the effectiveness of the scheme.

[0209] Specifically, in step S703, if the loss rate before adjustment is 5% and it drops to 3% after adjustment, and if the preset threshold is 4%, it indicates that the scheme is feasible. Preferably, the matching can be verified from the perspective of load balancing. For example, the load distribution changes from being concentrated on a single line to being evenly dispersed. While the loss is reduced, the line pressure decreases and the operating reliability is enhanced.

[0210] Specifically, in step S704, the overall optimization degree is evaluated according to the load distribution and loss changes of the distribution network because the uniformity of the load distribution directly affects the result. Exemplarily, if the load in a certain network is concentrated in the industrial area and is dispersed to the commercial area and residential area after adjustment, the overall loss rate drops from 5% to 2.5%, and the optimization degree is significant. This analysis helps to clarify the improvement of the scheme for the entire network.

[0211] Furthermore, in step S705, when using the K-means clustering algorithm to classify the loss rate values of multiple adjustment schemes, the best scheme can be found through historical data. For example, classifying the loss rates of the past 10 adjustments (such as 3%, 2.8%, 3.5%, etc.) into high, medium, and low categories, the clustering result shows that the scheme near 2.5% is the best. This classification method is intuitive and efficient, and can quickly screen out stable and low-loss scheduling strategies.

[0212] Specifically, in step S706, a statistical analysis tool can be used to verify the optimal solution. For example, a t-test can be used to analyze the significance of the loss evaluation value. Exemplarily, after running the optimal solution multiple times, the average loss evaluation value is 2.4% and the standard deviation is 0.1% to verify its stability.

[0213] It should be noted that this method can also reveal potential outliers. For example, if the loss suddenly increases to 3% at a certain time, it is prompted to check a specific line. This verification ensures the practicality of the solution and provides data support for long-term operation.

[0214] Step S8: Generate an optimized power distribution scheduling plan for the target power distribution period according to the power distribution network loss evaluation value and the cache capacity of the energy storage device.

[0215] In an alternative embodiment, step S8 includes steps S801 to S809:

[0216] Step S801: When the power distribution network loss evaluation value is greater than a preset network loss threshold, obtain updated data of the cache capacity of the energy storage device, and perform first-order difference processing on the updated data to obtain a cache capacity change trend value;

[0217] Step S802: Input the capacity change trend value and the real-time current fluctuation data of the power distribution network into a preset support vector regression model to obtain an electricity consumption accuracy adjustment coefficient;

[0218] Step S803: Adjust the real-time electrical parameter data of the power distribution network according to the electricity consumption accuracy adjustment coefficient, and calculate the adjusted power distribution network loss evaluation value;

[0219] Step S804: If the adjusted power distribution network loss evaluation value is still greater than the network loss threshold, generate an optimized power distribution network scheduling plan including charge and discharge rates and switching timings based on the current energy storage device parameters of the power distribution network;

[0220] Step S805: Extract the charge and discharge rate thresholds and the maximum capacity ratio from the optimized power distribution network scheduling plan, and generate a first execution plan including feasible time periods according to the cache capacity of the current energy storage device, the charge and discharge rate thresholds, and the maximum capacity ratio requirements;

[0221] Step S806: Input the first execution plan and the real-time current fluctuation data of the power distribution network into a preset random forest classifier to predict the load distribution probability for each time period;

[0222] Step S807: Screen out the time period combinations with load distribution probabilities lower than a preset probability threshold to obtain a second execution plan;

[0223] Step S808: Extract the node load balancing degree data in the second execution plan. When the sampling error of the load balancing degree data is less than a preset error threshold, generate an optimized execution plan for power distribution scheduling in the target power distribution period.

[0224] Exemplarily, obtain the real-time current and voltage monitoring data of the power distribution network, and calculate the network loss evaluation value; compare the loss evaluation value with the dynamic threshold determined based on historical data. When the loss evaluation value exceeds the dynamic threshold, trigger the state detection of the energy storage device; obtain the cache capacity update data of the energy storage device, perform a first-order difference process on the cache capacity update data to obtain the capacity change trend value; input the capacity change trend value and the real-time current fluctuation data of the power distribution network into a preset support vector regression model to obtain the power consumption accuracy adjustment coefficient; according to the power consumption accuracy adjustment coefficient, adjust the real-time current and voltage monitoring data of the power distribution network, and recalculate the loss evaluation value; if the adjusted loss evaluation value still exceeds the dynamic threshold, generate an optimization plan including the charge and discharge rate and the switching time sequence based on the current energy storage device parameters; extract the charge and discharge rate threshold and the maximum capacity ratio from the optimization plan, and use them as key parameters; determine whether the current cache capacity of the energy storage device meets the requirements of the rate threshold and the capacity ratio, and generate a preliminary execution plan including the feasible time period; input the preliminary execution plan and the real-time current fluctuation data of the power distribution network into a preset random forest classifier to predict the load distribution probability in each time period; screen the time period combinations with a load probability lower than the preset probability threshold to obtain an optimized execution plan for the energy storage device; extract the node load balancing degree data in the optimized execution plan, compare it with the current sampling accuracy of the power distribution network monitoring device, and when the sampling error is less than the preset error threshold, confirm the final execution plan; according to the final execution plan, adjust the charge and discharge rate output parameters of the energy storage device converter to achieve the optimized control of the power distribution network.

[0225] Specifically, in step S801, first obtain the real-time current and voltage monitoring data of the power distribution network, which is the basis for analyzing the network state. Exemplarily, in the power distribution network of an industrial area, the monitoring device records the voltage of a certain line as 10 kV and the current as 180 A. Through these data, the load situation of the network can be initially judged. If the voltage drops to 9.7 kV during the peak period and the current rises to 210 A, it indicates that the load increases and the loss may increase accordingly. This real-time monitoring provides a reliable basis for subsequent evaluation. Combining parameters such as the power factor and the line resistance can reflect the operating efficiency of the network.

[0226] Suppose the power factor of a certain line is 0.85 and the initial loss assessment value is approximately 45 kW. Comparing it with the dynamic threshold determined based on historical data, for example, setting the threshold at 40 kW, when the loss exceeds it, it indicates that the network operation deviates from the normal range and further detection is required. After triggering the status detection of the energy storage device, further obtain the updated data of the cache capacity. Specifically, the capacity of a certain energy storage device drops from 500 kWh to 480 kWh. Through first-order difference processing, the capacity change trend value is -20 kWh / h. This trend value reflects the discharge speed of the energy storage device and provides a reference for subsequent adjustments.

[0227] Specifically, in step S802, when inputting the capacity change trend value and the real-time current fluctuation data into the support vector regression model, the electricity consumption accuracy adjustment coefficient can be predicted. For example, when the current fluctuation range is between 180 A and 220 A, the model outputs an electricity consumption accuracy adjustment coefficient of 1.05, indicating that the monitoring data needs to be slightly amplified to more accurately reflect the actual loss.

[0228] Specifically, in step S803, after recalculating the loss assessment value according to the adjusted electricity consumption accuracy adjustment coefficient, it may drop from 45 kW to 42 kW, approaching the threshold range. If the loss still exceeds the standard after adjustment, it is necessary to generate an optimization plan based on the energy storage device parameters.

[0229] Furthermore, in step S804, assuming the maximum capacity of the energy storage device is 600 kWh and the current capacity is 450 kWh, indicating that the adjusted distribution network loss assessment value is still greater than the network loss threshold, then based on the current energy storage device parameters of the distribution network, generate a distribution network scheduling optimization plan including the charge and discharge rate and the switching time sequence. Exemplarily, the plan may recommend a charge and discharge rate of 50 kW, and the switching time sequence is to discharge during peak hours and charge during off-peak hours. Such a plan can effectively smooth the load fluctuation. After extracting the charge and discharge rate threshold and the maximum capacity ratio, judge whether the current capacity meets the requirements.

[0230] Specifically, in step S805, exemplarily, the rate threshold extracted from the optimization plan generated in step S804 is 60 kW, and the capacity ratio needs to reach 70%. Since 450 kWh only accounts for 75%, which meets the condition, a first execution plan including the feasible time period can be generated, such as discharging from 8 pm to 10 pm.

[0231] Specifically, in step S806, after inputting the preliminary plan and the current fluctuation data into the random forest classifier, predict the load distribution probability for each time period.

[0232] Furthermore, in step S807, if the predicted load probability from 8 o'clock to 9 o'clock in step S806 is 30%, which is lower than the preset threshold of 40%, then after screening, the optimized execution plan is to discharge from 9 o'clock to 10 o'clock.

[0233] Specifically, in step S808, the data of the load balancing degree of each node in the optimization solution of step S807 can be extracted and compared with the current sampling accuracy of the monitoring device to ensure the credibility of the solution. For example, if the load balancing degree is 85% and the sampling error is 2%, which is less than the threshold of 3%, it is confirmed that the final solution is feasible, and an optimized execution plan for power distribution scheduling in the target power distribution period is generated.

[0234] It should be noted that the parameters of the energy storage device converter can also be adjusted according to the optimized execution plan for power distribution scheduling in the final target power distribution period. For example, setting the discharge rate to 55 kW can effectively reduce the losses during peak hours and improve the network stability. This control method can also reduce the risk of equipment overload and extend the service life.

[0235] In summary, a distributed power distribution optimization method based on a virtual power plant provided by an embodiment of the present invention first generates estimated power consumption distribution data based on the power consumption characteristic values, climate, and holiday information in the power distribution area, and at the same time generates estimated power generation distribution data in combination with target historical power generation records, etc. Then, the peak power consumption period is judged through the estimated power consumption distribution, and the total power demand is obtained. Based on this, the buffer capacity of the energy storage device is determined in combination with the estimated power generation distribution data. When the difference between the buffer capacity and the flow rate at the bottleneck point of the power distribution network is lower than the preset margin threshold, the power plant supplementary power ratio is generated by using the power distribution ratio adjustment parameter and the real-time load data, and then the power plant scheduling frequency and the access density ratio of distributed power sources are adjusted, and finally a power distribution scheduling plan for the target power distribution period is generated. By adopting the present invention, the grid load can be effectively balanced, the power distribution efficiency can be improved, the access of distributed power sources and renewable energy can be adapted, and the intelligent scheduling and optimized operation of the grid can be realized.

[0236] See Figure 3 , Figure 3 4 is a structural block diagram of a distributed power distribution optimization system 200 based on a virtual power plant provided by an embodiment of the present invention. The distributed power distribution optimization system 200 based on a virtual power plant includes:

[0237] An estimated power consumption distribution module 21, configured to generate estimated power consumption distribution data according to the power consumption characteristic values in the power distribution area, the climate information, and holiday information in the target power distribution period;

[0238] An estimated power generation distribution module 22, configured to generate estimated power generation distribution data according to the target historical power generation records, the renewable energy ratio, the climate information, and the power quality detection standard in the target power distribution adjustment period;

[0239] A power demand calculation module 23, configured to judge the peak power consumption period of the target power distribution period according to the estimated power consumption distribution data, and generate the total power demand during the peak power consumption period;

[0240] A cache capacity calculation module 24, configured to generate a cache capacity of an energy storage device according to the total power demand during the peak power consumption period and the predicted power generation distribution data;

[0241] A power plant supplementary power ratio generation module 25, configured to generate a power plant supplementary power ratio according to a power distribution ratio adjustment parameter and real-time load data when the difference between the cache capacity of the energy storage device and the flow rate of the transmission bottleneck point of the distribution network is lower than a preset margin threshold;

[0242] A distribution dispatch plan generation module 26, configured to adjust the ratio of the power plant dispatch frequency and the distributed power source access density according to the power plant supplementary power ratio, and generate a distribution dispatch plan for a target distribution period.

[0243] In an alternative embodiment, the distributed power source distribution optimization system based on a virtual power plant further includes:

[0244] A distribution network loss assessment module, configured to analyze the distribution network loss under the distribution dispatch plan to obtain a distribution network loss assessment value;

[0245] A distribution dispatch plan optimization module, configured to generate a distribution dispatch optimization plan for a target distribution period according to the distribution network loss assessment value and the cache capacity of the energy storage device.

[0246] It should be noted that the distributed power source distribution optimization system based on a virtual power plant provided in the embodiment of the present invention is used to execute all the process steps of the distributed power source distribution optimization method based on a virtual power plant in the above embodiment, and the working principles and beneficial effects of the two correspond one by one, so they will not be elaborated here.

[0247] See Figure 4 , Figure 4 is a structural block diagram of a distributed power source distribution optimization device 300 based on a virtual power plant provided in the embodiment of the present invention. The distributed power source distribution optimization device 300 based on a virtual power plant includes a processor 31, a memory 32, and a computer program stored in the memory 32 and executable on the processor 31. When the processor 31 executes the computer program, the steps in the above-mentioned embodiments of the distributed power source distribution optimization method based on a virtual power plant are implemented, such as steps S1 to S8.

[0248] Exemplarily, the computer program can be divided into one or more modules / units. The one or more modules / units are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the distributed power source distribution optimization device 300 based on a virtual power plant.

[0249] The distributed power distribution optimization device 300 based on a virtual power plant may include, but is not limited to, a processor 31 and a memory 32. Those skilled in the art can understand that the schematic diagram is only an example of the distributed power distribution optimization device 300 based on a virtual power plant, and does not constitute a limitation on the distributed power distribution optimization device 300 based on a virtual power plant. It may include more or fewer components than those shown, or combine certain components, or different components. For example, the distributed power distribution optimization device 300 based on a virtual power plant may further include input / output devices, network access devices, a bus, etc.

[0250] The processor 31 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor 31 is the control center of the distributed power distribution optimization device 300 based on a virtual power plant, and connects various parts of the entire distributed power distribution optimization device 300 based on a virtual power plant through various interfaces and lines.

[0251] The memory 32 can be used to store the computer programs and / or modules. The processor 31 realizes various functions of the distributed power distribution optimization device 300 based on a virtual power plant by running or executing the computer programs and / or modules stored in the memory 32, and by calling the data stored in the memory 32. The memory 32 mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0252] Among them, if the modules / units integrated in the distributed power distribution optimization device 300 based on the virtual power plant are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-described embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor 31, the steps of the above-described various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0253] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of the present technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A distributed power distribution optimization method based on a virtual power plant, characterized in that Including: Generating estimated power consumption distribution data according to the power consumption characteristic values in the distribution area, the climate information of the target distribution period, and the holiday information; Generating estimated power generation distribution data according to the target historical power generation records, the proportion of renewable energy, the climate information of the target distribution adjustment period, and the power quality detection standard; Judging the peak power consumption period of the target distribution period according to the estimated power consumption distribution data, and generating the total power demand during the peak power consumption period; Generating the buffer capacity of the energy storage device according to the total power demand during the peak power consumption period and the estimated power generation distribution data; When the difference between the buffer capacity of the energy storage device and the flow rate of the transmission bottleneck point of the distribution network is lower than the preset margin threshold, generating the power plant supplementary power ratio according to the distribution ratio adjustment parameter and the real-time load data; Adjusting the ratio of the power plant dispatching frequency and the distributed power source access density according to the power plant supplementary power ratio, and generating the distribution dispatching plan for the target distribution period; Among them, the step of judging the peak power consumption period of the target distribution period according to the estimated power consumption distribution data and generating the total power demand during the peak power consumption period includes: Generating an estimated power consumption distribution curve according to the estimated power consumption distribution data, and obtaining the peak threshold with the same date type as the current date from the historical power consumption database; On the estimated power consumption distribution curve, calculating the first derivative at a preset time interval, and judging the candidate points of the time period boundary according to the absolute value and continuity of the first derivative; Sorting the areas between the candidate points according to the median power consumption, and marking the continuous time periods higher than the peak threshold as peak periods; When the power demand during the peak period exceeds a preset ratio of the peak threshold, using the sliding window method to re-divide the time period until the threshold condition is met or the maximum number of iterations is reached; Calculating the Euclidean distance between the current time period division result and the historical power consumption data. If the distance exceeds the preset tolerance, fine-tuning the time period boundary by the sliding window mean method to make the error with the historical power consumption data less than the preset value, and obtaining the final time period division result; Calculating the total power demand during the peak period according to the final time period division result.

2. The distributed power distribution optimization method based on a virtual power plant according to claim 1, wherein After adjusting the ratio of the power plant dispatching frequency and the distributed power source access density according to the power plant supplementary power ratio to generate the distribution dispatching plan for the target distribution period, it further includes: Analyzing the distribution network loss under the distribution dispatching plan to obtain the distribution network loss evaluation value; Generating the optimized distribution dispatching plan for the target distribution period according to the distribution network loss evaluation value and the buffer capacity of the energy storage device.

3. The distributed power distribution optimization method based on a virtual power plant according to claim 1, wherein, The step of generating the estimated power consumption distribution data according to the power consumption characteristic values in the distribution area, the climate information of the target distribution period, and the holiday information includes: Obtaining the target historical power consumption record set according to the climate information and holiday information of the target distribution period; Using the clustering algorithm to perform feature aggregation processing on the target historical power consumption record set to obtain the power consumption characteristic values in the distribution area; Analyzing the power consumption characteristic values in the distribution area according to the preset peak power demand threshold to obtain the peak period power consumption trend. Obtain the real-time power consumption data of the distribution network, input the peak-period power consumption trend and the real-time power consumption data into a preset linear regression model, and generate the estimated power consumption distribution data within the target distribution period through the linear regression model.

4. The distributed power distribution optimization method based on a virtual power plant according to claim 3, wherein, Generating the estimated power generation distribution data according to the target historical power generation records, the renewable energy ratio, the climate information of the target distribution adjustment period, and the power quality detection standard, including: Perform time series analysis on the target historical power generation records to obtain the power generation trend; Calculate the power generation ratios of various energy sources according to the power generation trend to obtain the preliminary estimated power generation distribution; Correct the preliminary estimated power generation distribution according to the climate information of the target distribution adjustment period to obtain the corrected estimated power generation distribution; Extract the data that meets the preset power quality threshold from the corrected estimated power generation distribution according to the power quality detection standard as the power quality compliance range; Generate an intermediate result of the estimated power generation distribution according to the power quality compliance range and the target historical power generation records; Use a linear regression algorithm to compare the intermediate result with the target historical power generation records to obtain the estimated power generation distribution data.

5. The distributed power distribution optimization method based on a virtual power plant according to claim 1, wherein, Generating the buffer capacity of the energy storage device according to the total power demand during the peak power consumption period and the estimated power generation distribution data, including: Determine the change trend of the power demand according to the total power demand during the peak power consumption period; Input the change trend of the power demand into a pre-established linear regression model, and calculate the output distribution of the power generation equipment according to the estimated power generation distribution data; Determine the charge and discharge frequency of the energy storage device within the response duration by using a sliding window algorithm according to the output distribution of the power generation equipment; When the charge and discharge frequency exceeds the preset charge and discharge frequency threshold, use the quadratic programming method to adjust the buffer capacity of the energy storage device; Judge the influence range of the power limit on the response of the power generation equipment through a proportional-integral controller according to the adjusted buffer capacity of the energy storage device; Determine the dynamic adjustment demand result of the buffer capacity of the energy storage device by using a dynamic programming algorithm according to the influence range; Judge the final buffer capacity of the energy storage device according to the dynamic adjustment demand result.

6. The distributed power distribution optimization method based on a virtual power plant according to claim 5, characterized in that When the difference between the buffer capacity of the energy storage device and the flow rate of the transmission bottleneck point of the distribution network is lower than the preset margin threshold, generate the power plant supplementary power ratio according to the distribution ratio adjustment parameter and the real-time load data, including: Calculate the difference between the buffer capacity of the energy storage device and the transmission bottleneck flow rate of the distribution network to obtain the energy storage capacity margin; Use the random forest algorithm to analyze the historical load data and establish a load prediction model; Use the load prediction model to predict the future load to obtain the predicted load value; Generate the power plant supplementary power ratio according to the energy storage capacity margin and the predicted load value.

7. The distributed power distribution optimization method based on a virtual power plant according to claim 1, characterized in that Adjusting the ratio of the power plant dispatching frequency and the distributed power source access density according to the power plant supplementary power ratio to generate the distribution dispatching plan for the target distribution period, including: When the power plant supplementary power ratio is greater than the preset supplementary power ratio threshold, calculate the initial power plant dispatching frequency; Analyze the matching change trend between the initial power plant scheduling frequency and the distributed power access density data by using the least squares method; If the matching change trend exceeds the preset change trend threshold, use the dynamic programming algorithm to update the distribution rules and optimize the initial power plant scheduling frequency until the power plant supplementary power ratio is less than the preset supplementary power ratio threshold, and use the logistic regression algorithm to predict the distribution rule adjustment plan to determine the distribution rules; According to the distribution rules, obtain the operation status data from the virtual power plant and generate a distribution scheduling plan for the target distribution period.

8. The distributed power distribution optimization method based on a virtual power plant according to claim 2, characterized in that, The generation of the distribution scheduling optimization execution plan for the target distribution period according to the distribution network loss evaluation value and the cache capacity of the energy storage device includes: When the distribution network loss evaluation value is greater than the preset network loss threshold, obtain the updated data of the cache capacity of the energy storage device, and perform first-order difference processing on the updated data to obtain the cache capacity change trend value; Input the capacity change trend value and the real-time current fluctuation data of the distribution network into a preset support vector regression model to obtain the power consumption accuracy adjustment coefficient; According to the power consumption accuracy adjustment coefficient, adjust the real-time electrical parameter data of the distribution network and calculate the adjusted distribution network loss evaluation value; If the adjusted distribution network loss evaluation value is still greater than the network loss threshold, generate a distribution network scheduling optimization plan including the charge and discharge rate and the switching time sequence based on the current energy storage device parameters of the distribution network; Extract the charge and discharge rate threshold and the maximum capacity ratio from the distribution network scheduling optimization plan, and generate a first execution plan including the feasible time period according to the cache capacity of the current energy storage device, the charge and discharge rate threshold and the maximum capacity ratio requirements; Input the first execution plan and the real-time current fluctuation data of the distribution network into a preset random forest classifier to predict the load distribution probability of each time period; Screen the time period combinations with the load distribution probability lower than the preset probability threshold to obtain a second execution plan; Extract the load balance degree data of each node in the second execution plan, and generate a distribution scheduling optimization execution plan for the target distribution period when the sampling error of the load balance degree data is less than the preset error threshold.

9. A distributed power distribution optimization system based on a virtual power plant, characterized in that Including: An estimated power consumption distribution module for generating estimated power consumption distribution data according to the power consumption characteristic value in the distribution area, the climate information and holiday information of the target distribution period; An estimated power generation distribution module for generating estimated power generation distribution data according to the target historical power generation record, the renewable energy ratio, the climate information of the target distribution adjustment period and the power quality detection standard; A power demand calculation module for judging the peak power consumption period of the target distribution period according to the estimated power consumption distribution data and generating the total power demand during the peak power consumption period; A cache capacity calculation module for generating the cache capacity of the energy storage device according to the total power demand during the peak power consumption period and the estimated power generation distribution data; The power plant supplementary power ratio generation module is used to generate the power plant supplementary power ratio according to the power distribution ratio adjustment parameters and real-time load data when the difference between the buffer capacity of the energy storage device and the flow of the transmission bottleneck point of the distribution network is lower than the preset margin threshold; The distribution dispatch plan generation module is used to adjust the ratio of the power plant dispatch frequency and the distributed power source access density according to the power plant supplementary power ratio, and generate the distribution dispatch plan for the target distribution period; Among them, the electric energy demand calculation module is specifically used for: Judging the peak power consumption period of the target distribution period according to the predicted power consumption distribution data, and generating the total electric energy demand of the peak power consumption period, including: Generating a predicted power consumption distribution curve according to the predicted power consumption distribution data, and obtaining the peak threshold with the same date type as the current date from the historical power consumption database; On the predicted power consumption distribution curve, calculate the first derivative at preset time intervals, and judge the candidate points of the time period boundary according to the absolute value and continuity of the first derivative; Sort the areas between the candidate points according to the median power consumption, and mark the continuous time periods higher than the peak threshold as peak time periods; When the electric energy demand during the peak time period exceeds a preset ratio of the peak threshold, use the sliding window method to re-divide the time period until the threshold condition is met or the maximum number of iterations is reached; Calculate the Euclidean distance between the current time period division result and the historical power consumption data. If the distance exceeds the preset tolerance, fine-tune the time period boundary by the sliding window mean method to make the error with the historical power consumption data less than the preset value, and obtain the final time period division result; Calculate the total electric energy demand of the peak time period according to the final time period division result.

Images