AI-based virtual power plant energy supply and demand precise matching method and system

Abstract

The application relates to the technical field of power planning, and provides an AI-based virtual power plant energy supply and demand accurate matching method and system, which solves the problems of low accuracy and overall operation efficiency of virtual power plant energy supply and demand matching. The method comprises the following steps: collecting output data sequences of power supply equipment in a virtual power plant and electricity price data sequences of a power market; dynamically adjusting initial charging and discharging power of energy storage equipment in the virtual power plant to obtain target charging and discharging power; using a federal learning mode to fuse and optimize load prediction models of each energy participating node in the virtual power plant to generate a global demand regulation plan; generating a power dispatching plan based on the output data sequences and the electricity price data sequences; and applying the power dispatching plan and the global demand regulation plan to energy supply and demand matching of the virtual power plant in a cooperative manner to realize AI-based virtual power plant energy supply and demand accurate matching. The application improves the accuracy and overall operation efficiency of virtual power plant energy supply and demand matching.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power planning, in particular to a virtual power plant energy supply and demand accurate matching method and system based on artificial intelligence (AI). BACKGROUND

[0002] In the AI-driven virtual power plant scenario, energy supply and demand accurate matching faces complex technical requirements. Virtual power plants integrate distributed power sources, energy storage devices, adjustable loads and other multi-energy nodes. The output and demand of these nodes are affected by multiple factors such as weather, time period, user behavior, etc., showing strong volatility and uncertainty. To achieve efficient use and stable supply of energy, AI technology is needed to real-time perceive the state of each node, dynamically balance energy production and consumption, and ensure that user demand is met while reducing energy loss and improving overall economic efficiency. Especially in the scenario of high proportion of renewable energy access and dynamic changes in the electricity market, the real-time and accuracy requirements of supply and demand matching are more stringent.

[0003] At present, a typical solution to the above technical requirements is a virtual power plant energy supply and demand matching solution based on distributed AI models. This solution deploys local AI models on each energy node, autonomously predicts local supply and demand conditions and generates preliminary scheduling recommendations, and uploads key parameters to the central node. The central node forms a global coordination strategy through summary analysis and feedback to each node for execution to achieve the combination of distributed decision-making and centralized coordination.

[0004] However, the existing solution has obvious defects. On the one hand, in the parameter interaction process between each node and the central node, there is a lack of targeted protection mechanism for data privacy, which easily leads to leakage of node sensitive information; on the other hand, the integration and optimization logic of the central node for each node parameter is relatively simple, and the characteristics of different nodes and dynamic changes are not fully considered. When some nodes appear abnormal fluctuations, the adaptability of the global coordination strategy is insufficient, and it is difficult to ensure the accuracy and stability of the supply and demand matching. SUMMARY

[0005] The present application provides a virtual power plant energy supply and demand accurate matching method and system based on AI to solve the problem of low accuracy and overall operation efficiency of virtual power plant energy supply and demand matching caused by the lack of privacy protection, simple integration logic and insufficient adaptability in the prior art.

[0006] In a first aspect, the present application provides a virtual power plant energy supply and demand accurate matching method based on AI, comprising:

[0007] Collecting output data sequences of power supply devices in the virtual power plant and electricity price data sequences of the electricity market;

[0008] According to the output data sequence and the electricity price data sequence, the initial charging and discharging power of the energy storage device in the virtual power plant is dynamically adjusted to obtain a target charging and discharging power;

[0009] Based on the target charging and discharging power, the load prediction models of each energy participating node in the virtual power plant are fused and optimized in a federated learning manner to generate a global demand adjustment plan;

[0010] Based on the output data sequence and the electricity price data sequence, a pre-trained scheduling model is used to generate a power scheduling plan;

[0011] The power scheduling plan and the global demand adjustment plan are used in the energy supply and demand matching process of the virtual power plant to realize accurate matching of energy supply and demand in the virtual power plant based on AI.

[0012] Optionally, based on the output data sequence and the electricity price data sequence, a pre-trained scheduling model is used to generate a power scheduling plan, comprising:

[0013] The output data sequence and the electricity price data sequence are input into a pre-trained scheduling model, and the output data sequence is divided into multiple data segments according to a preset time length by a division unit in the scheduling model;

[0014] The output value fluctuation range of each data segment is determined by a determination unit in the scheduling model, and based on the output value fluctuation range, the output fluctuation characteristics corresponding to each data segment are obtained;

[0015] The electricity price data sequence is calculated by a calculation unit in the scheduling model to obtain a price change amplitude value, and based on the price change amplitude value, a price fluctuation trend is generated;

[0016] The output fluctuation characteristics and the price fluctuation trend are associated and processed by a processing unit in the scheduling model to obtain the association between the output fluctuation characteristics and the price fluctuation trend, and according to the association, a power scheduling plan is generated.

[0017] Optionally, the electricity price data sequence is calculated to obtain a price change amplitude value, and based on the price change amplitude value, a price fluctuation trend is generated, comprising:

[0018] The electricity price data sequence is divided into multiple electricity price data intervals according to the preset time length;

[0019] The electricity price data corresponding to adjacent time points in each electricity price data interval is compared in value to obtain the price change amplitude value in each electricity price data interval;

[0020] determine a price fluctuation direction of each of the electricity price data intervals based on the price change amplitude value of each of the electricity price data intervals;

[0021] integrate the price change amplitude value and the price fluctuation direction of all the electricity price data intervals in chronological order to generate a price fluctuation trend.

[0022] Optionally, based on the target charging and discharging power, the load prediction models of each energy participating node in the virtual power plant are fused and optimized in a federated learning manner to generate a global demand regulation plan, including:

[0023] adjust the optimization parameters of the load prediction model based on the target charging and discharging power;

[0024] update the calculation results of the corresponding load prediction model according to the adjusted optimization parameters through each energy participating node in the virtual power plant, extract key parameters of the load prediction model based on the updated calculation results;

[0025] send the key parameters to a central coordination node of the virtual power plant through each energy participating node, comprehensively adjust the key parameters of all load prediction models through the central coordination node based on a federated learning manner, and generate a final unified parameter applicable to the global;

[0026] send the unified parameter to each energy participating node, so that each energy participating node optimizes the corresponding load prediction model according to the unified parameter;

[0027] integrate the output results of all optimized load prediction models based on the central coordination node to generate a global demand regulation plan.

[0028] Optionally, the sending of the key parameters to the central coordination node of the virtual power plant through each energy participating node, the comprehensive adjustment of the key parameters of all load prediction models through the central coordination node based on a federated learning manner, and the generation of a final unified parameter applicable to the global include:

[0029] determine a difference parameter based on a federated learning manner, using the central coordination node to judge whether a difference value of each key parameter exceeds a preset range, and determining the key parameter whose difference value exceeds the preset range as the difference parameter;

[0030] based on the central coordination node, gradient adjustment is performed on the difference parameter, and based on the difference parameter after the gradient adjustment, a non-difference parameter is combined for merging processing to form a preliminary unified parameter, the non-difference parameter being a key parameter whose difference value does not exceed the preset range;

[0031] distributing the preliminary unified parameters to the energy participation nodes, each node calculating a model performance index corresponding to the preliminary unified parameters using a local verification data set, and feeding back a performance change between the model performance index and an original performance index to the central coordination node;

[0032] correcting the preliminary unified parameters based on the performance change by the central coordination node to generate final unified parameters applicable to the whole.

[0033] Optionally, the energy supply and demand matching process of the virtual power plant by the coordinated action of the power dispatch plan and the global demand regulation plan comprises:

[0034] dividing the power dispatch plan and the global demand regulation plan respectively according to the preset time length to obtain a plurality of time units corresponding to each of the plans;

[0035] corresponding the time units of the power dispatch plan and the time units of the global demand regulation plan according to time sequence to form time correlation between the contents of the plans in each time unit group, and obtaining time correlation plans corresponding to each time unit group respectively;

[0036] identifying, based on the time correlation plans corresponding to the same time unit group, parts of the power dispatch plan and the global demand regulation plan that are inconsistent in the same time unit as plan conflict points;

[0037] determining adjustment rules of the plan conflict points according to real-time energy supply and demand information of the virtual power plant;

[0038] correcting the plan conflict points in each of the time correlation plans based on the adjustment rules to form a coordinated plan, and applying the coordinated plan to the energy supply and demand matching process of the virtual power plant.

[0039] Optionally, the determining of the adjustment rules of the plan conflict points according to the real-time energy supply and demand information of the virtual power plant comprises:

[0040] obtaining real-time energy supply and demand information of the virtual power plant, the real-time energy supply and demand information comprising real-time total energy supply, real-time total energy demand and real-time state data of energy storage devices of the virtual power plant;

[0041] determining a real-time supply and demand state of the virtual power plant according to a comparison result of the real-time total energy supply and the real-time total energy demand, the real-time supply and demand state comprising sufficient supply, balanced supply and insufficient supply;

[0042] if the real-time supply and demand state is in sufficient supply, adjusting unnecessary load demand in the global demand regulation plan to obtain a first adjustment rule;

[0043] Or, if the real-time supply and demand state is in supply shortage, the non-critical power generation output in the power dispatch plan is adjusted to obtain a second adjustment rule;

[0044] Or, if the real-time supply and demand state is in supply balance, a third adjustment rule is generated based on the real-time state data of the energy storage device.

[0045] In a second aspect, the present application provides an AI-based virtual power plant energy supply and demand accurate matching system, comprising:

[0046] A collection module is configured to collect output data sequences of power supply devices in a virtual power plant and electricity price data sequences of a power market;

[0047] An adjustment module is configured to dynamically adjust initial charging and discharging power of an energy storage device in the virtual power plant according to the output data sequences and the electricity price data sequences, to obtain target charging and discharging power;

[0048] An optimization module is configured to fuse and optimize load prediction models of each energy participating node in the virtual power plant by using a federated learning method based on the target charging and discharging power, to generate a global demand adjustment plan;

[0049] A generation module is configured to generate a power dispatch plan by using a pre-trained dispatch model based on the output data sequences and the electricity price data sequences;

[0050] A matching module is configured to apply the power dispatch plan and the global demand adjustment plan to an energy supply and demand matching process of the virtual power plant to realize AI-based virtual power plant energy supply and demand accurate matching.

[0051] In a third aspect, the present application provides a computing device, comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to realize the AI-based virtual power plant energy supply and demand accurate matching method according to any one of the first aspect.

[0052] In a fourth aspect, the present application provides a computer storage medium storing a computer program, which is executed by a computer to realize the AI-based virtual power plant energy supply and demand accurate matching method according to any one of the first aspect.

[0053] In the present application, an AI-based virtual power plant energy supply and demand precise matching method is provided, which comprises the following steps: collecting the output data sequence of power supply equipment in the virtual power plant and the electricity price data sequence of the electricity market; dynamically adjusting the initial charging and discharging power of the energy storage equipment in the virtual power plant according to the output data sequence and the electricity price data sequence to obtain the target charging and discharging power; based on the target charging and discharging power, using the federated learning method to fuse and optimize the load prediction model of each energy participating node in the virtual power plant to generate a global demand adjustment plan; based on the output data sequence and the electricity price data sequence, using the pre-trained scheduling model to generate a power scheduling plan; and applying the power scheduling plan and the global demand adjustment plan to the energy supply and demand matching process of the virtual power plant to realize AI-based virtual power plant energy supply and demand precise matching.

[0054] The present application has the following advantages:

[0055] By collecting the output data sequence of power supply equipment in the virtual power plant and the electricity price data sequence of the electricity market, basic data support can be provided for subsequent charging and discharging power adjustment, scheduling plan generation, etc.; by dynamically adjusting the initial charging and discharging power of the energy storage equipment in the virtual power plant according to the output data sequence and the electricity price data sequence to obtain the target charging and discharging power, the charging and discharging power of the energy storage equipment can be adapted to the actual energy output and market price changes, and the rationality of energy storage utilization can be improved; by using the federated learning method to fuse and optimize the load prediction model of each energy participating node in the virtual power plant based on the target charging and discharging power to generate a global demand adjustment plan, the global adaptability of the load prediction model can be improved while protecting the data privacy of each node, and a more suitable overall demand adjustment plan can be generated; by using the pre-trained scheduling model to generate a power scheduling plan based on the output data sequence and the electricity price data sequence, a scientific power generation and distribution strategy can be generated in combination with energy output fluctuations and market price changes; by applying the power scheduling plan and the global demand adjustment plan to the energy supply and demand matching process of the virtual power plant, AI-based virtual power plant energy supply and demand precise matching can be realized, and energy utilization efficiency and supply stability can be improved.

[0056] Further, the output data sequence and the electricity price data sequence are input into the scheduling model, the output data segment is divided and the output fluctuation characteristics are determined, the electricity price data interval is divided, the price change amplitude value and the fluctuation direction are calculated to generate the price fluctuation trend, and the power scheduling plan is generated by correlating the two. The technical effect is that by refining the time dimension processing of output data and electricity price data, the output fluctuation characteristics and the price fluctuation trend are accurately captured, and the power scheduling plan generated by combining the correlation between the two can better adapt to the actual energy supply capacity of the virtual power plant and market dynamics, improving the scientificity and adaptability of the scheduling plan.

[0057] These and other aspects of the present application will become more apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings required to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some of the embodiments of the present application, and do not represent all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0059] Figure 1 A flow chart of an AI-based virtual power plant energy supply and demand accurate matching method provided by an embodiment of the present application;

[0060] Figure 2 A structural schematic diagram of an AI-based virtual power plant energy supply and demand accurate matching system provided by an embodiment of the present application;

[0061] Figure 3 A structural schematic diagram of a computing device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0062] In order to enable persons skilled in the art to better understand the technical solutions of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0063] In some of the descriptions in the specification and claims of the present application and the above-mentioned drawings, a plurality of operations appearing in a specific order are included, but it should be clearly understood that these operations can be executed or in parallel without the order in which they appear in this text. The serial numbers of the operations, such as 11, 12, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this text are used to distinguish different messages, devices, modules, etc., and do not represent the order of sequence, nor do "first" and "second" represent different types.

[0064] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0065] To solve the problems of lack of privacy protection, simple integration logic and insufficient adaptability in the prior art, an AI-based virtual power plant energy supply and demand accurate matching method is provided in the embodiments of the present application, which adopts the following concept: by collecting power output data of power supply equipment and electricity price data of the electricity market, a basis is provided for subsequent adjustment and plan generation; by adjusting the charge and discharge power of the energy storage equipment according to these data, the operation of the energy storage equipment is more in line with the actual demand; by optimizing the load prediction model of each node in a federated learning manner and generating a global demand adjustment plan, the model optimization can be realized without directly sharing the original data of each node, thereby protecting information security; by generating a power dispatching plan through a dispatching model combined with the fluctuation of output and price, the plan is more in line with actual changes; by matching energy supply and demand through coordination of the two plans, the overall adaptability is improved. Further, the core step of dividing data segments and intervals, analyzing the output fluctuation characteristics and price fluctuation trend and generating a dispatching plan in association makes the plan generation process more detailed and better adapts to different situations. Through these steps, the privacy protection problem in information interaction is solved, and the plan generation and integration are more in line with the actual situation, thereby improving the adaptability to dynamic changes, so as to realize the accurate matching of energy supply and demand.

[0066] Figure 1 A flowchart of an AI-based virtual power plant energy supply and demand accurate matching method provided in the embodiments of the present application is shown in Figure 1 , which comprises the following steps:

[0067] S11, collecting power output data sequences of power supply equipment in a virtual power plant and electricity price data sequences of the electricity market.

[0068] The virtual power plant is an energy system that integrates solar energy, wind energy and other various distributed energy equipment and uniformly coordinates the operation, the power supply equipment is a device that generates electricity in the system, such as a solar panel and a wind turbine, the power output data sequence is a series of data arranged in time sequence according to the power generation of these devices at different time points, which contains the specific power generation value at each time point, the electricity market is a place for electricity trading, and the electricity price will fluctuate with changes in supply and demand, the electricity price data sequence is a series of data arranged in time sequence according to the electricity price of the electricity market at different time points, which contains the specific price at each time point, and the generation result of this step is to obtain the above two types of data sequences to provide basic information for subsequent steps.

[0069] In the embodiments of the present application, the monitoring device installed on the power supply device records the power generation of the device at fixed time intervals (such as every hour), arranges these data in chronological order to form a power output data sequence, and obtains the electricity price at the same time interval from the information platform of the electricity market to arrange the electricity price data sequence in chronological order, for example, the solar panel of virtual power plant A generates 100 degrees at 8 o'clock in the morning and 120 degrees at 9 o'clock, and the wind turbine generates 50 degrees at 8 o'clock and 40 degrees at 9 o'clock, which are arranged in chronological order as 8 o'clock solar energy 100 degrees and wind energy 50 degrees, 9 o'clock solar energy 120 degrees and wind energy 40 degrees. At the same time, the electricity price of 8 o'clock is 0.5 yuan / degree and 9 o'clock is 0.55 yuan / degree, which are arranged as 8 o'clock 0.5 yuan / degree and 9 o'clock 0.55 yuan / degree.

[0070] S12, according to the power output data sequence and the electricity price data sequence, dynamically adjusting the initial charging and discharging power of the energy storage device in the virtual power plant to obtain the target charging and discharging power.

[0071] Wherein, the energy storage device is a device capable of storing electricity (charging) and releasing electricity (discharging) such as a battery pack, the initial charging and discharging power is the power size of the energy storage device in unit time before adjustment, and the target charging and discharging power is the charging and discharging power suitable for the current actual situation after adjustment according to the power output data sequence and the electricity price data sequence. The generated result of this step is to obtain the target charging and discharging power that adapts to the actual demand.

[0072] In the embodiments of the present application, first analyze the power output data sequence to determine whether the power generation of the power supply device is sufficient, if there is excess power that can be stored when the power generation is more in a certain period, then consider increasing the charging power, and then judge the electricity price according to the electricity price data sequence, if the electricity price is low in a certain period, it is suitable to charge more, then further determine the amplitude of increasing the charging power, if the electricity price is high, it is suitable to discharge more, then adjust the discharging power, and then dynamically adjust the initial charging and discharging power to obtain the target charging and discharging power through the above analysis, for example, the solar energy of virtual power plant A generates 200 degrees per hour from 12 o'clock to 14 o'clock, and the demand is only 100 degrees, with 100 degrees of excess power, and the electricity price is low at 0.4 yuan / degree, the excess power can support the charging power to increase from the initial 40kW to 90kW, so the charging power is adjusted to 90kW.

[0073] S13, based on the target charging and discharging power, using federated learning to fuse and optimize the load prediction model of each energy participating node in the virtual power plant to generate a global demand adjustment plan.

[0074] The federated learning mode refers to a method in which multiple participating units (such as energy nodes) jointly optimize a model without directly sharing original data, the energy participating node is a unit that uses or adjusts energy in the virtual power plant, such as a residential area, a factory, a shopping mall, etc., the load prediction model is a model used to predict the electricity consumption of each node in the future, and the global demand regulation plan is a whole plan for coordinating the electricity demand of each node, including the electricity consumption period and electricity allocation of each node. The generation result of this step is to generate the global plan by optimizing the model.

[0075] In the embodiments of the present application, each energy participating node first optimizes its load prediction model locally using local electricity consumption data and the target charging and discharging power obtained in S12, and this process does not share original data but only adjusts the parameters affecting prediction in the model. Then, the optimized model parameters (such as calculation coefficients) are sent to the central coordination unit, the central coordination unit processes the parameters (such as calculating the average value) to obtain global model parameters and distributes them to each node, each node updates the model with global parameters and optimizes it again, and the above process is repeated until the model is accurate. Finally, based on the optimized model, the electricity consumption of each node is predicted and integrated to generate a global demand regulation plan. For example, residential area B uploads the parameter 0.3 after optimizing the model with local electricity consumption records and target charging and discharging power, factory C uploads the parameter 0.4, the center calculates the average value 0.35 and distributes it, and after updating the model, B needs 200 degrees per day and C needs 500 degrees per day, and the plan is that B uses electricity in the morning and C uses electricity in the afternoon.

[0076] S14, based on the output data sequence and the electricity price data sequence, generating a power dispatching plan by using a pre-trained dispatching model.

[0077] The dispatching model is a model that is trained in advance by historical data and can generate a power distribution scheme according to input data, the power dispatching plan is a plan for guiding power generation and distribution in the virtual power plant, including the power generation, storage and supply object of each period, and the generation result of this step is to generate the dispatching plan by the model in combination with the output and electricity price data.

[0078] In the embodiment of the present application, the output data sequence and the electricity price data sequence obtained by S11 are input into the pre-trained scheduling model. The model first analyzes the fluctuation characteristics of the output data to determine which time period has more power generation and which time period has less power generation, and then analyzes the fluctuation trend of the electricity price data to determine which time period has high electricity price and which time period has low electricity price. Then, the output fluctuation and the electricity price fluctuation are associated (for example, more power is stored when power generation is more and the electricity price is low, and high-demand nodes are preferentially supplied when power generation is more and the electricity price is high). Based on the above analysis, a power scheduling plan is generated. For example, the model analyzes that the wind power of virtual power plant A is 150 degrees per hour (more) from 2 a.m. to 6 a.m. and the electricity price is 0.3 yuan / degree (lower), and plans to store more power. The solar power is 200 degrees per hour (more) from 12 noon to 2 p.m. and the electricity price is 0.6 yuan / degree (higher), and plans to preferentially supply the mall.

[0079] S15, the power scheduling plan and the global demand regulation plan are used in the energy supply and demand matching process of the virtual power plant to realize AI-based precise matching of energy supply and demand of the virtual power plant.

[0080] Wherein, the synergistic effect refers to combining the power scheduling plan and the global demand regulation plan to jointly guide the energy supply and demand matching. The energy supply and demand matching process is a process of adapting the power generation of the virtual power plant to the power consumption of each node. The precise matching refers to the power generation that just meets the power consumption without waste or shortage. The generation result of this step is to realize the AI-based precise matching of energy supply and demand of the virtual power plant.

[0081] In the embodiment of the present application, the power scheduling plan generated by S14 is compared with the global demand regulation plan generated by S13 to check whether the power generation suggestion and the power consumption demand in the same period are consistent. If there is a difference, the difference is adjusted according to the real-time situation of the virtual power plant (such as the state of the energy storage device). Finally, the adjusted synergistic plan is applied to the energy supply and demand matching process to realize precise matching. For example, the virtual power plant A has a scheduling plan to generate 700 degrees of power at 3 p.m. The demand plan shows that the total demand is 650 degrees, which has a 50-degree difference. The real-time situation shows that the energy storage device still has 50 degrees of storage space. Therefore, the power generation is adjusted to 650 degrees to just meet the demand.

[0082] The application provides the following specific examples: a virtual power plant A includes solar energy, wind energy equipment, energy storage equipment, and energy participating nodes such as residential area B, factory C, and shopping mall D. The staff records the power generation and electricity market price at different time points, and arranges them into a time-arranged data sequence. According to these data, whether the power generation is sufficient and the electricity price is high or low in different time periods is analyzed, the charging and discharging power of the energy storage equipment is adjusted, and the actual situation is made more suitable. Each energy participating node optimizes its own power consumption prediction model locally using local power consumption data and the adjusted charging and discharging power, and then sends the model parameters to the central coordination unit. The center integrates and then issues the optimized parameters, and each node updates the model to generate a global demand plan for coordinating the power consumption period of each node. At the same time, combined with the power generation and electricity price data, the power dispatching plan is generated through the dispatching model, and the power storage and supply arrangement in different time periods is clear. Finally, the two plans are compared, and the real-time situation such as the state of the energy storage equipment is adjusted to make the power generation just meet the demand of each node, so as to realize the accurate matching of energy supply and demand.

[0083] By performing S11-S15, the embodiment of the application collects power generation and electricity market price data to provide accurate basic information for the entire process, ensuring that all adjustments and plans are based on actual situations. Dynamically adjusting the charging and discharging power of the energy storage equipment makes it run more suitable for changes in power generation and electricity price, reduces power waste and improves economy, and lays a reasonable energy storage foundation for subsequent steps. The federated learning is used to optimize the load prediction model, which protects data privacy while improving prediction accuracy, and the generated global demand adjustment plan can effectively coordinate the power consumption of each node and avoid local power consumption concentration. The dispatching model generates a power dispatching plan combined with the power generation capacity and price changes, making the generation, storage and distribution of electricity more reasonable, and taking into account supply stability and economy. Finally, through the cooperation of the two plans, the power generation and power consumption are dynamically adapted, the accurate matching of energy supply and demand of the virtual power plant is realized, and the overall energy utilization efficiency and supply stability are improved.

[0084] In one possible embodiment, S14, based on the output data sequence and the electricity price data sequence, a pre-trained dispatching model is used to generate a power dispatching plan, including:

[0085] Step 141, input the output data sequence and the electricity price data sequence into the pre-trained dispatching model, and divide the output data sequence into multiple data segments according to the preset time length through the division unit in the dispatching model.

[0086] The output data sequence is a data string of power generation of the power supply device at different time points arranged by time, the electricity price data sequence is a data string of electricity prices at different time points of the power market arranged by time, the pre-trained scheduling model is a model trained in advance with historical data and capable of processing data to generate a power plan, the division unit is a part of the scheduling model for dividing data, the preset time length is a fixed time (such as several hours) set in advance, the data segment is a segment of power generation data divided according to the preset time, and the generation result of this step is to divide the output data sequence into multiple data segments.

[0087] In the embodiment of the application, the output data sequence and the electricity price data sequence are input into the pre-trained scheduling model, the division unit in the model divides the output data sequence into multiple continuous data segments according to the preset time length (such as 4 hours), for example, the output data of the virtual power plant for 24 hours is divided into 6 data segments according to 4 hours, 24÷4=6, each of which contains 4 hours of power generation data.

[0088] Step 142, determining, by a determination unit in the scheduling model, the output value fluctuation range of each data segment, and obtaining the output fluctuation feature corresponding to each data segment based on the output value fluctuation range.

[0089] The determination unit is a part of the scheduling model for determining the power generation fluctuation range, the output value fluctuation range is the difference between the maximum and minimum values of power generation in each data segment, and the output fluctuation feature is a feature (such as large or small fluctuation) reflecting the change size of power generation in the data segment. The generation result of this step is the output fluctuation feature corresponding to each data segment.

[0090] In the embodiment of the application, the determination unit of the scheduling model finds the maximum and minimum values of power generation in each data segment, obtains the output value fluctuation range by subtracting the minimum value from the maximum value, and determines the output fluctuation feature (such as large or small fluctuation) according to the range size, for example, the maximum value of a certain data segment is 80 degrees and the minimum value is 50 degrees, 80-50=30 degrees, and the feature is “moderate fluctuation”.

[0091] Step 143, calculating, by a calculation unit in the scheduling model, the electricity price data sequence to obtain a price change amplitude value, and generating a price fluctuation trend based on the price change amplitude value.

[0092] The calculation unit is a part of the scheduling model for calculating the electricity price change, the price change amplitude value is the difference between the electricity prices at adjacent time points, and the price fluctuation trend is a change condition reflecting the overall rise, fall or stability of the electricity price in a period of time. The generation result of this step is the price fluctuation trend.

[0093] In the embodiment of the present application, the calculation unit of the scheduling model compares the electricity prices of adjacent time points in the electricity price data sequence, and uses the electricity price of the latter time point to reduce the price change amplitude value of the former time point, integrates these values to judge the overall rise, fall or stability, generates the price fluctuation trend, for example, the electricity price rises from 0.5 yuan to 0.6 yuan and then to 0.7 yuan, and the amplitude value is positive, and the trend is "continuous rise".

[0094] In step 144, the processing unit in the scheduling model performs correlation processing on the output fluctuation characteristics and the price fluctuation trend to obtain the correlation relationship between the output fluctuation characteristics and the price fluctuation trend, and generates the power scheduling plan according to the correlation relationship.

[0095] Among them, the processing unit is the part of integrated analysis in the scheduling model, the correlation processing is the analysis of the relationship between the output fluctuation characteristics and the price fluctuation trend, the correlation relationship is the corresponding situation of the two, and the power scheduling plan is the power storage and supply arrangement according to the correlation relationship. The generation result of this step is the power scheduling plan.

[0096] In the embodiment of the present application, the processing unit of the scheduling model correlates the output fluctuation characteristics of each data segment with the price fluctuation trend of the corresponding time period, analyzes the correlation relationship between the two in the same period, and then formulates the power storage and supply arrangement to generate the power scheduling plan, for example, "when the power fluctuation is large, the electricity price rises", the plan preferentially supplies the high demand node and reduces the storage.

[0097] The present application provides the following specific examples: the virtual power plant A inputs the output data sequence and the electricity price data sequence of 24 hours a day into the scheduling model, the division unit of the model divides the output data into 8 data segments (24÷3=8) according to the preset time length of 3 hours. The determination unit analyzes each data segment, for example, the maximum power generation of a 3-hour segment is 120 degrees, the minimum power generation is 60 degrees, 120-60=60 degrees, and the characteristic is "larger fluctuation"; the maximum power generation of another segment is 70 degrees, the minimum power generation is 40 degrees, 70-40=30 degrees, and the characteristic is "smaller fluctuation". The calculation unit processes the electricity price data, and the electricity price of adjacent time points is from 0.5 yuan to 0.55 yuan (amplitude 0.05 yuan), and then to 0.52 yuan (amplitude -0.03 yuan), generating the trend of "first rise and then slight decrease". The processing unit correlates and finds that "the period with larger fluctuation corresponds to the electricity price first rising and then slightly decreasing", and generates the plan accordingly: when the fluctuation is larger and the electricity price rises, preferentially supply the demand nodes such as the market; when the fluctuation is smaller, store more power.

[0098] By executing steps 141-144, the embodiment of the application divides the output data by time, makes the power generation analysis more targeted, determines the output fluctuation characteristics and price fluctuation trend, clearly presents the change rule of power generation and electricity price, correlates the two to generate a scheduling plan, and makes the power storage and supply take into account the power generation fluctuation and electricity price change to avoid unreasonable plan caused by a single factor. The overall process improves the orderliness of data processing, makes the power scheduling plan more in line with the actual situation, improves the energy utilization efficiency, and provides a scientific basis for accurate matching of virtual power plant energy supply and demand.

[0099] In a possible embodiment, step 143, the electricity price data sequence is calculated to obtain a price change amplitude value, and a price fluctuation trend is generated based on the price change amplitude value, including:

[0100] a1, dividing the electricity price data sequence into a plurality of electricity price data intervals according to a preset time length.

[0101] The electricity price data sequence is a sequence of data arranged in time order according to the electricity price at different time points in the electricity market, the preset time length is a fixed time (such as several hours) set in advance, the electricity price data interval is an interval containing the electricity price data of the time period obtained by dividing according to the preset time length, and the generation result of this step is to divide the electricity price data sequence into a plurality of electricity price data intervals.

[0102] In the embodiment of the application, a preset time length (such as 4 hours) is determined, the electricity price data sequence is divided according to the time length, and a plurality of non-overlapping electricity price data intervals are obtained, each interval containing the electricity price data of all time points in the time period. For example, the electricity price data sequence of a virtual power plant records the electricity price for 24 hours a day, and is divided according to a preset time length of 4 hours. 24 divided by 4 equals 6, so 6 electricity price data intervals are obtained, each containing 4 hours of electricity price data.

[0103] a2, comparing the electricity price data corresponding to adjacent time points in each electricity price data interval to obtain a price change amplitude value in each electricity price data interval.

[0104] The electricity price data interval is a section of electricity price data divided according to the preset time length, the adjacent time points are two time points (such as 8 o'clock and 9 o'clock) continuously in time in the same interval, the electricity price data is the specific electricity price corresponding to each time point, and the price change amplitude value is the difference between the electricity prices of adjacent time points in the same interval. The generation result of this step is to obtain the price change amplitude value in each electricity price data interval.

[0105] In the embodiments of the present application, for each electricity price data interval, all adjacent time points in the interval are found, the electricity price data of these adjacent time points are compared, the electricity price of the latter time point is subtracted from the electricity price of the former time point, the price change amplitude value of each adjacent time point is obtained, and the amplitude values in the same interval are integrated as the price change amplitude value of the interval. For example, the electricity prices of 8 o'clock, 9 o'clock and 10 o'clock are included in a certain interval, the electricity price increases from 0.5 yuan to 0.55 yuan from 8 o'clock to 9 o'clock, 0.55 yuan is subtracted from 0.5 yuan to equal 0.05 yuan, the electricity price decreases from 0.55 yuan to 0.52 yuan from 9 o'clock to 10 o'clock, 0.52 yuan is subtracted from 0.55 yuan to equal -0.03 yuan, and the price change amplitude value of the interval is 0.05 yuan and -0.03 yuan.

[0106] a3, determining the price fluctuation direction of each electricity price data interval based on the price change amplitude value of each electricity price data interval.

[0107] The price change amplitude value is the difference between the electricity prices of adjacent time points, and the price fluctuation direction is the overall change direction of the electricity price in each electricity price data interval (such as rising, falling or basically stable). The generated result of this step is to determine the price fluctuation direction of each electricity price data interval.

[0108] In the embodiments of the present application, the price change amplitude values in each electricity price data interval are observed. If the amplitude value is positive, the electricity price in the interval as a whole shows a rising trend, if it is negative, the electricity price shows a falling trend, and if it is close to 0, the electricity price is basically stable. In this way, the price fluctuation direction of each interval is determined. For example, the price change amplitude values of a certain interval are 0.05 yuan and 0.03 yuan, both of which are positive values, so the price fluctuation direction of the interval is rising.

[0109] a4, integrating the price change amplitude values and the price fluctuation directions of all electricity price data intervals in time sequence to generate a price fluctuation trend.

[0110] The time sequence is the sequence of arranging the electricity price data intervals in time order, the price change amplitude value is the difference between the electricity prices of adjacent time points, the price fluctuation direction is the change direction of the electricity price in each interval, and the price fluctuation trend is the overall change of the electricity price in a period of time after integrating the amplitude values and directions of all intervals. The generated result of this step is the price fluctuation trend.

[0111] In the embodiments of the present application, the electricity price data intervals are arranged in time order, the price change amplitude values and the price fluctuation directions of all intervals are integrated, and the overall change of the electricity price in a period of time is formed, that is, the price fluctuation trend. For example, the fluctuation directions of 6 intervals are rising, rising, falling, basically stable, rising and falling in turn, and the generated price fluctuation trend after integration is “rising first, then falling, with a short period of stability in the middle, and rising again in the later period, followed by falling again”.

[0112] The application provides the following specific examples: the power price data sequence of a virtual power plant A records the power price for 24 hours a day (one data per hour). First, divide according to the preset time length of 3 hours, 24 divided by 3 equals 8, to obtain 8 power price data intervals, each interval containing 3 hours of power price data. For the first interval (8-11), compare the power prices at adjacent time points: from 8 to 9, the price rises from 0.48 yuan to 0.52 yuan, 0.52 minus 0.48 equals 0.04 yuan; from 9 to 10, the price falls from 0.52 yuan to 0.5 yuan, 0.5 minus 0.52 equals -0.02 yuan; from 10 to 11, the price rises from 0.5 yuan to 0.55 yuan, 0.55 minus 0.5 equals 0.05 yuan, to obtain the price change amplitude values of the interval, which are 0.04 yuan, -0.02 yuan and 0.05 yuan. According to these amplitude values, although there are rises and falls in the first interval, the rising amplitude is greater, and the fluctuation direction is determined to be “small rise”; the remaining 7 intervals are processed in the same way to obtain their respective fluctuation directions (such as “fall”, “basically stable” and the like). Finally, integrate the information of the 8 intervals in time sequence to generate a price fluctuation trend of “small rise in the morning, short fall at noon, basically stable in the afternoon, and continuous rise at night”.

[0113] By performing a1-a4, the embodiments of the application convert continuous power price data into units convenient for analysis by dividing the power price data sequence into intervals according to fixed time, so that the power price change of each time period can be studied individually; the amplitude values of the price changes at adjacent time points are calculated, which can accurately capture the subtle changes in the power price in each interval; the fluctuation direction is determined based on the amplitude values, so that the power price change trend of each interval is clearer; finally, the information of all intervals is integrated to generate a price fluctuation trend, which completely presents the overall change rule of the power price in a period of time. The entire process makes the power price analysis from scattered data become systematic and orderly, provides a clear price change basis for subsequent power dispatching plans combined with power generation conditions, and helps to improve the rationality and accuracy of the plans.

[0114] In a possible embodiment, S13, based on the target charging and discharging power, fuses and optimizes the load prediction models of the energy participating nodes in the virtual power plant in a federated learning manner to generate a global demand regulation plan, including:

[0115] Step 131, adjusting the optimization parameters of the load prediction model based on the target charging and discharging power.

[0116] Among them, the target charging and discharging power is the power size suitable for the charging and discharging of the energy storage device after the power generation and the power price are adjusted; the load prediction model is a model for predicting the future power consumption of each node; the optimization parameter is a adjustable value (such as a calculation coefficient) in the model that affects the prediction result; the generated result of this step is the adjusted load prediction model optimization parameter.

[0117] In the embodiments of the present application, based on the target charging and discharging power obtained in step 12, the influence on the power consumption of each node is analyzed, and the optimization parameters related to charging and discharging in the load prediction model are adjusted (such as adjusting the coefficient reflecting the influence of energy storage equipment in the model) to make the model more suitable for the actual energy storage situation. For example, if the target charging and discharging power of a node in a virtual power plant is increased, the optimization parameter related to "energy storage power" in the load prediction model of the node is adjusted from 0.2 to 0.3.

[0118] In step 132, the calculation results of the corresponding load prediction model are updated by each energy participating node in the virtual power plant according to the adjusted optimization parameters, and the key parameters of the load prediction model are extracted based on the updated calculation results.

[0119] In step 132, the calculation results of the corresponding load prediction model are updated by each energy participating node in the virtual power plant according to the adjusted optimization parameters, and the key parameters of the load prediction model are extracted based on the updated calculation results.

[0120] In the embodiments of the present application, each energy participating node uses the optimization parameters adjusted in step 131 to update its own load prediction model, recalculates the future power consumption to obtain updated results, and then extracts the key parameters (such as the weight value that determines the power consumption peak period) that have the greatest impact on the prediction results from the model. For example, residential area B updates the model using the adjusted parameters, predicts that the daily power consumption peak is in the morning, and extracts the "morning period weight" in the model as the key parameter.

[0121] In step 133, the key parameters are sent to the central coordination node of the virtual power plant by each energy participating node, and the key parameters of all load prediction models are comprehensively adjusted by the central coordination node based on the federated learning method to generate the final unified parameter suitable for the global.

[0122] In step 133, the key parameters are sent to the central coordination node of the virtual power plant by each energy participating node, and the key parameters of all load prediction models are comprehensively adjusted by the central coordination node based on the federated learning method to generate the final unified parameter suitable for the global.

[0123] In the embodiments of the present application, each energy participating node sends the extracted key parameters to the central coordination node, the central coordination node adopts the federated learning manner, does not process the original data, only comprehensively adjusts all the key parameters (such as calculating the average value), generates the final unified parameter suitable for the whole global, for example, the residential area B sends the key parameter 0.6, the factory C sends 0.7, and the shopping mall D sends 0.5, the center calculates the sum of 0.6, 0.7 and 0.5 equal to 1.8, 1.8 divided by 3 is equal to 0.6, and the unified parameter 0.6 is obtained.

[0124] Step 134, sending the unified parameter to each energy participating node, so that each energy participating node optimizes the corresponding load prediction model according to the unified parameter.

[0125] In the embodiments of the present application, the central coordination node sends the final unified parameter to each energy participating node, each node replaces the corresponding parameter in the model with the parameter, optimizes the load prediction model, so that the model can reflect the local power consumption characteristics and meet the global coordination requirements, for example, the unified parameter is 0.6, the residential area B updates the corresponding parameter in the model from 0.6 (the original adjustment value) to 0.6 (the unified value), and the factory C updates from 0.7 to 0.6, so as to realize global unified optimization.

[0126] In the embodiments of the present application, the central coordination node sends the final unified parameter to each energy participating node, each node replaces the corresponding parameter in the model with the parameter, optimizes the load prediction model, so that the model can reflect the local power consumption characteristics and meet the global coordination requirements, for example, the unified parameter is 0.6, the residential area B updates the corresponding parameter in the model from 0.6 (the original adjustment value) to 0.6 (the unified value), and the factory C updates from 0.7 to 0.6, so as to realize global unified optimization.

[0127] Step 135, based on the central coordination node, integrating the output results of all the optimized load prediction models to generate a global demand regulation plan.

[0128] In the embodiments of the present application, the central coordination node sends the final unified parameter to each energy participating node, each node replaces the corresponding parameter in the model with the parameter, optimizes the load prediction model, so that the model can reflect the local power consumption characteristics and meet the global coordination requirements, for example, the unified parameter is 0.6, the residential area B updates the corresponding parameter in the model from 0.6 (the original adjustment value) to 0.6 (the unified value), and the factory C updates from 0.7 to 0.6, so as to realize global unified optimization.

[0129] In the embodiments of the present application, the central coordination node collects the output results of the optimized load prediction models of each energy participating node (i.e. the predicted power consumption of each node), analyzes the power consumption peak period and the power consumption of each node, integrates according to the global supply and demand balance principle (such as staggering the peak period), generates a global demand regulation plan, for example, after integration, it is found that the residential area consumes more power in the morning and the factory consumes more power in the afternoon, the plan is to supply power to the residential area in the morning and the factory in the afternoon, so as to avoid concentrated power consumption.

[0130] The application provides the following specific examples: a virtual power plant A includes three energy participating nodes of a residential area B, a factory C and a shopping mall D. First, according to the target charging and discharging power, the residential area B is charged by 90 kW, the factory C is discharged by 70 kW, and the optimization parameters of the load prediction model of each node are adjusted: the residential area B adjusts the "electricity consumption coefficient during charging period" from 0.4 to 0.6, and the factory C adjusts the "electricity consumption coefficient during discharging period" from 0.3 to 0.5. Each node updates the model with the adjusted parameters to calculate the predicted electricity consumption: the residential area B predicts that the morning electricity consumption is 200 degrees, the factory C predicts that the afternoon electricity consumption is 500 degrees, and the shopping mall D predicts that the evening electricity consumption is 300 degrees; at the same time, the key parameters are extracted, the residential area B is "morning weight 0.6", the factory C is "afternoon weight 0.7", and the shopping mall D is "evening weight 0.5". These key parameters are sent to the central coordination node, and the center calculates the average value in the federated learning mode: 0.6 plus 0.7 plus 0.5 equals 1.8, 1.8 divided by 3 equals 0.6, and a unified parameter 0.6 is generated. Each node optimizes the model with 0.6, and the factory C adjusts the "afternoon weight" from 0.7 to 0.6, so that the prediction is more consistent with the global. Finally, the central coordination node collects the output results of the optimized model, integrates and finds that the electricity consumption of the residential area in the morning, the factory in the afternoon and the shopping mall in the evening is concentrated, and generates a global demand regulation plan: arrange the residential area B to supply 200 degrees of power in the morning, the factory C to supply 500 degrees of power in the afternoon, and the shopping mall D to supply 300 degrees of power in the evening, to avoid overlapping of electricity consumption peaks.

[0131] By performing steps 131-135, the embodiment of the application adjusts the model optimization parameters according to the target charging and discharging power, so that the load prediction model is preliminarily adapted to the actual operation of the energy storage device, providing a reasonable basis for subsequent prediction. Each node updates the model and extracts key parameters, which not only retains the local electricity consumption characteristics, but also avoids original data leakage, protecting information security. The central coordination node generates a unified parameter in the federated learning mode, realizes global optimization without obtaining original data, and balances local characteristics and overall coordination. After each node optimizes the model with the unified parameter, the prediction result is more consistent, providing a reliable basis for the global plan. The finally integrated global demand regulation plan effectively staggered the electricity consumption peaks of each node, avoided supply and demand imbalance, and improved the rationality and overall operation efficiency of virtual power plant energy distribution.

[0132] In one possible embodiment, step 133, each energy participating node sends the key parameters to the central coordination node of the virtual power plant, and based on the federated learning mode, the central coordination node adjusts all the key parameters of the load prediction model to generate a final unified parameter suitable for the global, including:

[0133] b1, based on the federated learning mode, using the central coordination node to judge whether the difference value of each key parameter exceeds the preset range, and determining the key parameter whose difference value exceeds the preset range as a difference parameter.

[0134] wherein the federated learning mode is a method in which each node does not share original data and optimizes the model only through parameter cooperation, the center coordination node is a central unit integrating information of each node in the virtual power plant, the key parameter is a core parameter in the model that plays a major role in the prediction result, the difference value is a difference value between key parameters of different nodes, the preset range is a reasonable difference interval set in advance, the difference parameter is a key parameter whose difference value exceeds the preset range, and the generation result of this step is to determine the difference parameter.

[0135] In the embodiments of the present application, based on the federated learning mode, the center coordination node first collects the key parameters of each energy participating node, calculates the difference value between the same key parameters of different nodes, and then compares the difference values with the preset range set in advance. If the difference value of a certain key parameter exceeds the preset range, the parameter is determined as the difference parameter. For example, the same key parameters of three nodes are 0.6, 0.8 and 0.5, the preset range is plus or minus 0.1, the difference value calculated by 0.8 minus 0.6 is 0.2, and the difference value calculated by 0.8 minus 0.5 is 0.3. Both of the two difference values exceed the preset range, so the key parameter corresponding to 0.8 is determined as the difference parameter.

[0136] b2, based on the center coordination node, gradient adjustment is performed on the difference parameter, based on the difference parameter after the gradient adjustment, the non-difference parameter is combined for merging processing to form a preliminary unified parameter, and the non-difference parameter is a key parameter whose difference value does not exceed the preset range.

[0137] wherein the center coordination node is a central unit integrating information of each node, the difference parameter is the parameter determined in b1, the gradient adjustment is a process of gradually adjusting the difference parameter to make it reasonable, the non-difference parameter is a key parameter whose difference value does not exceed the preset range, and the preliminary unified parameter is a parameter obtained by merging the adjusted difference parameter and the non-difference parameter. The generation result of this step is the preliminary unified parameter.

[0138] In the embodiments of the present application, the center coordination node performs gradient adjustment on the difference parameter determined in b1, that is, gradually reduces the difference between the difference parameter and other parameters. For example, the difference parameter 0.8 is gradually adjusted to 0.7, and then the adjusted difference parameter is combined with the non-difference parameter, that is, the key parameter whose difference value does not exceed the preset range, to form a preliminary unified parameter suitable for the global. For example, the adjusted 0.7 is combined with the non-difference parameters 0.6 and 0.5, the average value of 0.6 plus 0.7 plus 0.5 is calculated to be 1.8, and 1.8 divided by 3 is equal to 0.6, and the preliminary unified parameter 0.6 is obtained.

[0139] b3, distribute the preliminary unified parameter to each energy participating node, each node calculates the model performance index corresponding to the preliminary unified parameter using the local verification data set, and feeds back the performance change between the model performance index and the original performance index to the central coordination node.

[0140] Wherein, the preliminary unified parameter is the parameter obtained in b2, the energy participating node is a unit using energy in the virtual power plant, such as a residential area and a factory, the local verification data set is the local data used by each node to test the model, the model performance index is a standard for measuring the prediction effect of the model, such as accuracy, the performance change is the difference between the preliminary parameter corresponding index and the original index, the central coordination node receives the performance change fed back by each node, and the generation result of the step is the performance change fed back to the center.

[0141] In the embodiment of the application, the central coordination node distributes the preliminary unified parameter obtained in b2 to each energy participating node, each node calculates the model performance index corresponding to the parameter using its own local verification data set, that is, the local historical data used to test the model, compares the index with the original performance index before adjustment, obtains the performance change, that is, the difference between the two, and feeds back the performance change to the central coordination node, for example, the original performance index of a node is 80 points, and the performance index after using the preliminary unified parameter is 85 points, the performance change is 5 points, and the result is fed back to the center.

[0142] b4, the central coordination node corrects the preliminary unified parameter based on the performance change to generate a final unified parameter applicable to the global.

[0143] Wherein, the central coordination node is a central unit integrating information of each node, the performance change is the difference fed back in b3, the preliminary unified parameter is the parameter obtained in b2, and the final unified parameter is the parameter applicable to the global after correction. The generation result of the step is the final unified parameter applicable to the global.

[0144] In the embodiment of the application, the central coordination node collects the performance changes fed back by each energy participating node, analyzes the overall trend of the changes, if the performance changes of most nodes are positive, that is, the model effect is improved, the preliminary unified parameter is fine-tuned, if there is a negative change, that is, the model effect is reduced, the parameter is adjusted according to the change amplitude, and finally a final unified parameter applicable to the global is generated, for example, the performance of most nodes is improved by about 5 points, and only one node is reduced by 2 points, so the preliminary unified parameter 0.6 is adjusted to 0.62 as the final unified parameter.

[0145] The application provides the following specific examples: the central coordination node of the virtual power plant A collects the key parameters of the residential area B, the factory C and the shopping mall D as 0.6, 0.9 and 0.5 respectively. First, the center calculates the difference value based on the federated learning mode, 0.9 minus 0.6 equals 0.3, 0.9 minus 0.5 equals 0.3, and the preset range is plus or minus 0.2, since 0.3 exceeds the range, 0.9 is determined as the difference parameter. Then, the center adjusts the gradient of the difference parameter 0.9, first to 0.8, and then to 0.7, and then combines the non-difference parameters 0.6 and 0.5, calculates the average value of 0.6 plus 0.7 plus 0.5, which equals 1.8, and 1.8 divided by 3 equals 0.6, to obtain the preliminary unified parameter 0.6. After that, the center distributes the preliminary unified parameter 0.6 to each node, and each node tests the model performance by using the local verification data set. The original performance index of the residential area B is 75 points, and after using it, it is 80 points, and the performance change is 5 points; the original performance index of the factory C is 70 points, and after using it, it is 78 points, and the change is 8 points; the original performance index of the shopping mall D is 72 points, and after using it, it is 70 points, and the change is negative 2 points. Each node feeds back these performance changes to the center. Finally, the center analyzes the feedback results, most of the nodes have performance improvement, only the shopping mall D has slight decrease, so the preliminary unified parameter 0.6 is fine-tuned to 0.62, and the final unified parameter suitable for the whole is generated.

[0146] By performing b1~b4, the embodiments of the application can correct the parameters deviating from the reasonable range by identifying the difference parameters and adjusting the gradient, and lay a foundation for generating unified parameters. The preliminary unified parameters are formed by combining the adjusted parameters and the non-difference parameters, and the transition of the local parameters to the global parameters is realized. The performance changes are tested by the local data and fed back by each node, which provides an actual basis for parameter optimization and ensures that the parameters meet the actual situation of each node. The final unified parameters generated based on the feedback correction not only take into account the global coordination, but also adapt to the local characteristics of each node, improve the overall prediction effect of the model, and at the same time, do not involve sharing of the original data in the whole process, and protect the information security.

[0147] In a possible embodiment, S15, the power dispatch plan and the global demand regulation plan are used in the energy supply and demand matching process of the virtual power plant, including:

[0148] Step 151, according to the preset time length, the power dispatch plan and the global demand regulation plan are divided respectively to obtain a plurality of time units corresponding respectively.

[0149] The preset time length is a fixed time (for example, 3 hours) set in advance, the power dispatching plan is an arrangement for guiding power generation and distribution, the global demand regulation plan is an arrangement for coordinating power demand of each node, the time unit is a unit containing plan content of the time period after the time period is divided according to the preset time length, and the generation result of the step is that the two plans are divided into multiple time units respectively.

[0150] In the embodiment of the application, a preset time length (for example, 3 hours) is determined, and the power dispatching plan and the global demand regulation plan are divided according to the time length. Each plan is divided into multiple continuous time units, and each unit contains plan content in the time period. For example, the two plans of the virtual power plant are arranged for 24 hours a day, and the 24 is divided by 3 to equal 8. Therefore, each plan is divided into 8 time units, and each unit contains 3 hours of content.

[0151] In the embodiment of the application, a preset time length (for example, 3 hours) is determined, and the power dispatching plan and the global demand regulation plan are divided according to the time length. Each plan is divided into multiple continuous time units, and each unit contains plan content in the time period. For example, the two plans of the virtual power plant are arranged for 24 hours a day, and the 24 is divided by 3 to equal 8. Therefore, each plan is divided into 8 time units, and each unit contains 3 hours of content.

[0152] In the embodiment of the application, a preset time length (for example, 3 hours) is determined, and the power dispatching plan and the global demand regulation plan are divided according to the time length. Each plan is divided into multiple continuous time units, and each unit contains plan content in the time period. For example, the two plans of the virtual power plant are arranged for 24 hours a day, and the 24 is divided by 3 to equal 8. Therefore, each plan is divided into 8 time units, and each unit contains 3 hours of content.

[0153] In the embodiment of the application, a preset time length (for example, 3 hours) is determined, and the power dispatching plan and the global demand regulation plan are divided according to the time length. Each plan is divided into multiple continuous time units, and each unit contains plan content in the time period. For example, the two plans of the virtual power plant are arranged for 24 hours a day, and the 24 is divided by 3 to equal 8. Therefore, each plan is divided into 8 time units, and each unit contains 3 hours of content.

[0154] In the embodiment of the application, a preset time length (for example, 3 hours) is determined, and the power dispatching plan and the global demand regulation plan are divided according to the time length. Each plan is divided into multiple continuous time units, and each unit contains plan content in the time period. For example, the two plans of the virtual power plant are arranged for 24 hours a day, and the 24 is divided by 3 to equal 8. Therefore, each plan is divided into 8 time units, and each unit contains 3 hours of content.

[0155] Wherein, the same time unit group is the corresponding two time units in step 152, the time correlation plan is the associated plan content in the group, the power dispatch plan is the power supply arrangement, the global demand regulation plan is the electricity demand arrangement, the plan conflict point is the inconsistent part of the two plan contents in the same time unit, and the generation result of this step is the identified plan conflict point.

[0156] In the embodiment of the application, the time correlation plan corresponding to the same time unit group is viewed, the supply amount of the power dispatch plan and the demand amount of the global demand regulation plan in the time period are compared, if the two are not equal or the allocation objects do not match, the inconsistent part is determined as the plan conflict point, for example, in a certain time unit group, the power dispatch plan arranges to supply 500 degrees, and the demand regulation plan shows that 550 degrees are needed, and the difference of 50 degrees is the plan conflict point.

[0157] Step 154, according to the real-time energy supply and demand information of the virtual power plant, determine the adjustment rule of the plan conflict point.

[0158] Wherein, the real-time energy supply and demand information of the virtual power plant is the current actual power generation, the real-time electricity consumption of each node, the state of the energy storage device and other information, the plan conflict point is the inconsistent part identified in step 153, the adjustment rule is a specific criterion for correcting the conflict point (such as priority to meet demand or adjust the supply amount), and the generation result of this step is the determined plan conflict point adjustment rule.

[0159] In the embodiment of the application, the real-time energy supply and demand information (such as the current actual power generation, the state of the energy storage device, etc.) of the virtual power plant is collected, the relationship between these information and the plan conflict point is analyzed, if the real-time power generation is sufficient and the energy storage has space, the rule of “increasing the supply amount to meet the demand” is formulated; if the power generation is insufficient but the energy storage has power, the rule of “releasing energy storage to supplement supply” is formulated, for example, the conflict point is 50 degrees less supply, and there is surplus in real-time power generation, and it is determined that the adjustment rule is to increase 50 degrees of supply amount.

[0160] Step 155, based on the adjustment rule, correct the plan conflict point in each time correlation plan to form a collaborative plan, and apply the collaborative plan to the energy supply and demand matching process of the virtual power plant.

[0161] Wherein, the adjustment rule is the criterion for correcting the plan conflict point, the time correlation plan is the associated plan content of the same time unit group, the plan conflict point is the inconsistent part of the two plans, the collaborative plan is the unified plan formed after the conflict point is corrected, the energy supply and demand matching process is the process of adapting the power generation and the electricity consumption to each other, and the generation result of this step is the collaborative plan applied to the energy supply and demand matching.

[0162] In the embodiments of the present application, the plan conflict points in each time-related plan are modified according to the adjustment rules, and the supply is increased when the supply is insufficient, and the supply is reduced when the supply is excessive, so that the supply and demand in the same time unit are consistent, the modified plans are integrated to form a collaborative plan, which is applied to the energy supply and demand matching process. For example, the supply of a certain conflict point is 50 degrees less, and the supply is increased by 50 degrees according to the rules, and the collaborative plan is formed after modification to guide the power distribution in this period.

[0163] The present application provides the following specific examples: Virtual power plant A divides the power dispatching plan and the global demand regulation plan into 8 time units (24 ÷ 3 = 8) according to the preset time length of 3 hours. The time units in the two plans are corresponded, such as the power dispatching unit and the demand regulation unit of "8-11" form a time unit group to form a time-related plan containing the supply and demand content of the period. It is found by comparison that in the "11-14" time unit group, the power dispatching plan supplies 500 degrees, and the demand regulation plan demands 550 degrees, and the difference of 50 degrees is calculated and identified as a plan conflict point. According to the real-time information, the current power generation is 60 degrees more than the plan, and the energy storage device has idle space, so the adjustment rule is determined as allocating 50 degrees of surplus power to supplement the supply. According to this rule, the conflict point is modified to make the supply consistent with the demand, and the collaborative plan is formed by integrating all the modified plans, which is applied to the energy supply and demand matching process.

[0164] By performing steps 151-155, the embodiments of the present application divide the two plans into time units according to the preset time length, divide the overall plan into time periods for comparison, and provide a basis for analyzing the differences in the same period. The time-related plan is formed by corresponding time units, so that the supply and demand content in the same period are related to each other, which is convenient for subsequent comparison. The identification of the plan conflict point clarifies the inconsistency of the two plans, and provides a specific object for adjustment. The adjustment rule is determined according to the real-time information to ensure that the modification has actual basis and conforms to the running state of the virtual power plant. The finally formed collaborative plan is applied to the supply and demand matching, so that the power supply and demand in each period remain consistent, avoiding excess or deficiency, realizing the accurate matching of energy supply and demand of the virtual power plant, and improving the overall operation efficiency.

[0165] In one possible embodiment, step 154, determining the adjustment rule of the plan conflict point according to the real-time energy supply and demand information of the virtual power plant, comprises:

[0166] c1, obtaining the real-time energy supply and demand information of the virtual power plant, the real-time energy supply and demand information comprising: the real-time total energy supply of the virtual power plant, the real-time total energy demand and the real-time state data of the energy storage device.

[0167] The virtual power plant is a system integrating dispersed energy equipment, and the real-time energy supply and demand information is the current actual energy supply and demand situation, including the total energy that the virtual power plant can currently provide, i.e., the total real-time energy supply, the total energy that each node currently needs, i.e., the total real-time energy demand, and the energy storage state of the energy storage equipment, i.e., the real-time state data. The generation result of this step is to obtain these real-time information.

[0168] In the embodiment of the application, the real-time energy supply total amount is obtained by collecting the current power generation of the power supply equipment through the monitoring device such as a sensor on each device of the virtual power plant, the real-time energy demand total amount is obtained by collecting the current power consumption of each energy participating node, and the real-time state data is obtained by collecting the current power storage and chargeable and dischargeable capacity of the energy storage equipment. The data is summarized to obtain the real-time energy supply and demand information. For example, the virtual power plant collects 500 degrees of total power supply through the sensor, i.e., the total amount of supply, 400 degrees of total power consumption of each node, i.e., the total amount of demand, and 150 degrees of power storage of the energy storage equipment, i.e., the state data.

[0169] c2, determining the real-time supply and demand state to which the virtual power plant belongs according to the comparison result of the total real-time energy supply and the total real-time energy demand, wherein the real-time supply and demand state includes supply sufficient, supply balance and supply insufficient.

[0170] The total real-time energy supply is the total amount of energy that the virtual power plant can actually provide at present, the total real-time energy demand is the total amount of energy that each node actually needs at present, and the real-time supply and demand state is the relationship between the current supply and demand, including supply sufficient, balance or insufficient. The generation result of this step is to determine the real-time supply and demand state to which the virtual power plant belongs.

[0171] In the embodiment of the application, the total real-time energy supply obtained in step c1 is compared with the total real-time energy demand. If the total supply is significantly greater than the total demand, for example, more than 20% more than the demand, it is determined to be supply sufficient. If the two are approximately equal with a difference of within 5%, it is determined to be supply balance. If the total supply is significantly less than the total demand, for example, less than 20% less than the demand, it is determined to be supply insufficient. For example, the total supply is 500 degrees, the total demand is 400 degrees, 500 minus 400 is 100 degrees, and the supply is 25% more than the demand, so it is determined to be supply sufficient.

[0172] c3, if the real-time supply and demand state is supply sufficient, adjusting the unnecessary load demand in the global demand adjustment plan to obtain the first adjustment rule.

[0173] The real-time supply-demand state is the supply-demand relationship determined in step c2, the supply sufficiency is a state in which the total supply is significantly greater than the total demand, the global demand regulation plan is a whole arrangement for coordinating power consumption of each node, the non-essential load demand is a non-essential power consumption demand in the current period, such as power consumption of entertainment equipment, and the first adjustment rule is an adjustment criterion formulated for the supply sufficiency. The generation result of the present step is to obtain the first adjustment rule.

[0174] In the embodiments of the present application, when the real-time supply-demand state is the supply sufficiency, the power consumption demand of each node in the global demand regulation plan is analyzed, the essential load such as basic life power consumption of residents and the non-essential load such as decoration power consumption of a shopping mall during non-business hours are distinguished, and the adjustment rule is formulated: appropriately delaying or reducing the power consumption time and power of the non-essential load to avoid energy waste. For example, when the supply sufficiency, the decoration lamp power consumption of the shopping mall is planned to be delayed from the current period to the power consumption valley period, thereby forming the first adjustment rule.

[0175] c4、or, if the real-time supply-demand state is in the supply deficiency, the non-critical power generation output in the power dispatching plan is adjusted to obtain the second adjustment rule.

[0176] The real-time supply-demand state is the relationship determined in step c2, the supply deficiency is a state in which the total supply is significantly less than the total demand, the power dispatching plan is an arrangement for guiding power distribution, the non-critical power generation output is the power generation amount of a current non-essential priority operation power generation device, such as a low-efficiency power generator, and the second adjustment rule is an adjustment criterion formulated for the supply deficiency. The generation result of the present step is to obtain the second adjustment rule.

[0177] In the embodiments of the present application, when the real-time supply-demand state is the supply deficiency, the power generation arrangement of each power supply device in the power dispatching plan is analyzed, the critical power generation output such as a high-efficiency wind power device and the non-critical power generation output such as a low-efficiency diesel generator are distinguished, and the adjustment rule is formulated: increasing the power generation amount of the critical power generation output and reducing or suspending the non-critical power generation output to preferentially guarantee the essential load power consumption. For example, when the supply deficiency, the high-efficiency wind power field is planned to generate more power, and the low-efficiency generator is suspended, thereby forming the second adjustment rule.

[0178] c5、or, if the real-time supply-demand state is in the supply balance, the third adjustment rule is generated based on the real-time state data of the energy storage device.

[0179] The real-time supply-demand state is the relationship determined in step c2, the supply balance is a state in which the supply and the demand are approximately equal, the real-time state data of the energy storage device is the current power storage and charging and discharging capacity of the energy storage device, and the third adjustment rule is an adjustment criterion formulated for the supply balance. The generation result of the present step is to obtain the third adjustment rule.

[0180] In the embodiment of the present application, when the real-time supply and demand state is supply shortage, the power generation arrangement of each power supply device in the power dispatching plan is analyzed, the key power generation output such as high-efficiency wind power device and the non-essential power generation output such as low-efficiency diesel generator are distinguished, and the adjustment rule is formulated: the power generation of the key power generation output is increased, the non-essential power generation output is reduced or suspended, and the necessary load power consumption is preferentially guaranteed. For example, when the supply is insufficient, the plan allows the high-efficiency wind power plant to generate more power, and the low-efficiency generator is suspended, forming a second adjustment rule.

[0181] The present application provides the following specific examples: Virtual power plant A obtains real-time energy supply and demand information through monitoring devices on each device: solar power generation of 200 degrees, wind power generation of 300 degrees, and the total real-time energy supply of 200 plus 300 equals 500 degrees; residential power consumption of 100 degrees, factory power consumption of 250 degrees, and shopping mall power consumption of 50 degrees, and the total real-time energy demand of 100 plus 250 plus 50 equals 400 degrees; the current energy storage capacity of the energy storage device is 70 degrees. After comparison, it is found that the supply is 100 degrees more than the demand, and the real-time supply and demand state is determined to be supply sufficient. The global demand regulation plan is analyzed, the residential lighting is determined to be necessary load, and the shopping mall billboard is determined to be non-essential load, and the first adjustment rule is formulated: the current power consumption of the shopping mall billboard is reduced by half and is restored in the evening. If the supply is insufficient later, for example, the supply is 300 degrees and the demand is 400 degrees, the second adjustment rule is formulated: increase wind power generation and suspend diesel generator; if the supply is balanced and the energy storage has space, the third adjustment rule is formulated: store 10% of the current power generation.

[0182] By executing c1-c5, the embodiment of the present application provides accurate basis for judging the current supply and demand relationship by obtaining real-time energy supply and demand information; after determining the real-time supply and demand state, adjustment rules are formulated for supply sufficient, insufficient or balanced, which can flexibly cope with different situations. When the supply is sufficient, the non-essential load is adjusted to avoid waste, when the supply is insufficient, the power generation output is optimized to guarantee the necessary demand, and when the supply is balanced, the energy storage is used to maintain stability, which improves the rationality of energy utilization as a whole, enhances the ability of the virtual power plant to cope with supply and demand fluctuations, and ensures more accurate energy supply and demand matching and more stable operation.

[0183] Figure 2 A structure diagram of an AI-based virtual power plant energy supply and demand accurate matching system provided by the embodiment of the present application is shown in Figure 2 as shown, the system comprises:

[0184] The acquisition module 21 is configured to acquire the output data sequence of the power supply device in the virtual power plant and the electricity price data sequence of the electricity market.

[0185] The adjustment module 22 is configured to dynamically adjust the initial charging and discharging power of the energy storage device in the virtual power plant according to the output data sequence and the electricity price data sequence to obtain the target charging and discharging power.

[0186] The optimization module 23 is configured to fuse and optimize the load prediction models of the energy participating nodes in the virtual power plant based on the target charge-discharge power in a federated learning manner to generate a global demand regulation plan.

[0187] The generation module 24 is configured to generate a power dispatch plan based on the output data sequence and the electricity price data sequence by using a pre-trained scheduling model.

[0188] The matching module 25 is configured to apply the power dispatch plan and the global demand regulation plan to the energy supply-demand matching process of the virtual power plant to realize AI-based accurate energy supply-demand matching of the virtual power plant.

[0189] Figure 2 The AI-based accurate energy supply-demand matching system of the virtual power plant can perform Figure 1 The AI-based accurate energy supply-demand matching method of the virtual power plant of the embodiments described above will not be described in detail. The specific operation of each module, unit of the AI-based accurate energy supply-demand matching system of the virtual power plant in the above embodiments has been described in detail in the embodiments related to the method, which will not be described in detail.

[0190] In one possible design, Figure 2 The AI-based accurate energy supply-demand matching system of the virtual power plant of the embodiments described above can be implemented as a computing device, such as Figure 3 As shown, the computing device can include a storage component 31 and a processing component 32.

[0191] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0192] The processing component 32 is configured to perform the following processes: collecting an output data sequence of power supply equipment in a virtual power plant and an electricity price data sequence of a power market; dynamically adjusting an initial charge-discharge power of energy storage equipment in the virtual power plant according to the output data sequence and the electricity price data sequence to obtain a target charge-discharge power; fusing and optimizing load prediction models of energy participating nodes in the virtual power plant based on the target charge-discharge power in a federated learning manner to generate a global demand regulation plan; generating a power dispatch plan based on the output data sequence and the electricity price data sequence by using a pre-trained scheduling model; and applying the power dispatch plan and the global demand regulation plan to an energy supply-demand matching process of the virtual power plant to realize AI-based accurate energy supply-demand matching of the virtual power plant.

[0193] The processing component 32 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more Application-Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Process Device (DSPD), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, for executing the above method.

[0194] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage devices or a combination thereof, such as Random Access Memory (RAM), Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0195] Of course, the computing device can also include other components, such as an input / output interface, a display component, a communication component, etc.

[0196] The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, etc.

[0197] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0198] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can be a cloud server, and the processing component, the storage component, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0199] The embodiment of the application further provides a computer storage medium, which stores a computer program, and the computer program can realize the above method when being executed by a computer. Figure 1 The embodiment of the application further provides an AI-based virtual power plant energy supply and demand accurate matching method.

[0200] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, and will not be described here.

[0201] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment scheme. Those skilled in the art can understand and implement without creative labor.

[0202] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and the necessary general hardware platform, and of course, it can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, which can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, server, or network device, etc.) execute the method described in each embodiment or some part of the embodiment.

[0203] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. An AI-based virtual power plant energy supply and demand precise matching method, characterized in that, The method comprises the following steps: Collecting output data sequence of power supply equipment in a virtual power plant and electricity price data sequence of a power market; According to the output data sequence and the electricity price data sequence, dynamically adjusting the initial charging and discharging power of the energy storage equipment in the virtual power plant to obtain the target charging and discharging power; Based on the target charging and discharging power, using a federated learning method to fuse and optimize the load prediction model of each energy participating node in the virtual power plant to generate a global demand adjustment plan; Based on the output data sequence and the electricity price data sequence, using a pre-trained scheduling model to generate a power scheduling plan; The power scheduling plan and the global demand adjustment plan are used in the energy supply and demand matching process of the virtual power plant to realize accurate matching of energy supply and demand in the virtual power plant based on AI; The method for generating a power scheduling plan based on the output data sequence and the electricity price data sequence using a pre-trained scheduling model comprises the following steps: The output data sequence and the electricity price data sequence are input into a pre-trained scheduling model, and the output data sequence is divided into multiple data segments by a division unit in the scheduling model according to a preset time length; The output value fluctuation range of each data segment is determined by a determination unit in the scheduling model, and the output fluctuation characteristics corresponding to each data segment are obtained based on the output value fluctuation range; The price change amplitude value is calculated by a calculation unit in the scheduling model based on the electricity price data sequence, and the price fluctuation trend is generated based on the price change amplitude value; The output fluctuation characteristics and the price fluctuation trend are associated and processed by a processing unit in the scheduling model to obtain the association between the output fluctuation characteristics and the price fluctuation trend, and the power scheduling plan is generated according to the association; The method for calculating the electricity price data sequence to obtain the price change amplitude value and generating the price fluctuation trend based on the price change amplitude value comprises the following steps: The electricity price data sequence is divided into multiple electricity price data intervals according to the preset time length; The price data of adjacent time points in each electricity price data interval is compared to obtain the price change amplitude value in each electricity price data interval; The price fluctuation direction of each electricity price data interval is determined based on the price change amplitude value of each electricity price data interval; The price change amplitude value and the price fluctuation direction of all the electricity price data intervals are integrated in chronological order to generate the price fluctuation trend.

2. The AI-based virtual power plant energy supply and demand precise matching method according to claim 1, characterized in that, The method for generating a global demand adjustment plan by using a federated learning method to fuse and optimize the load prediction model of each energy participating node in the virtual power plant based on the target charging and discharging power comprises the following steps: Adjusting the optimization parameters of the load prediction model based on the target charging and discharging power; Updating the calculation results of the corresponding load prediction model based on the adjusted optimization parameters by each energy participating node in the virtual power plant, and extracting the key parameters of the load prediction model based on the updated calculation results; The key parameters are sent to the central coordination node of the virtual power plant by the energy participation nodes, and the key parameters of all load prediction models are comprehensively adjusted by the central coordination node based on a federated learning manner to generate final unified parameters applicable to the global; The unified parameters are sent to the energy participation nodes, so that the energy participation nodes optimize the corresponding load prediction models according to the unified parameters; The output results of all optimized load prediction models are integrated based on the central coordination node to generate a global demand regulation plan.

3. The AI-based virtual power plant energy supply and demand precise matching method according to claim 2, characterized in that, The key parameters are sent to the central coordination node of the virtual power plant by the energy participation nodes, and the key parameters of all load prediction models are comprehensively adjusted by the central coordination node based on a federated learning manner to generate final unified parameters applicable to the global, including: Based on the federated learning manner, the central coordination node is used to judge whether the difference value of each key parameter exceeds the preset range, and the key parameter whose difference value exceeds the preset range is determined as a difference parameter; Based on the central coordination node, the difference parameters are gradient adjusted, and based on the gradient adjusted difference parameters, the non-difference parameters are combined to form preliminary unified parameters, the non-difference parameters being the key parameters whose difference values do not exceed the preset range; The preliminary unified parameters are distributed to the energy participation nodes, each node calculates the model performance index corresponding to the preliminary unified parameters using the local verification data set, and the performance change amount between the model performance index and the original performance index is fed back to the central coordination node; The central coordination node is used to correct the preliminary unified parameters based on the performance change amount to generate final unified parameters applicable to the global.

4. The AI-based virtual power plant energy supply and demand precise matching method according to claim 1, characterized in that, The power dispatch plan and the global demand regulation plan are used in the energy supply and demand matching process of the virtual power plant, including: The power dispatch plan and the global demand regulation plan are respectively divided into multiple time units according to a preset time length; The time units of the power dispatch plan and the time units of the global demand regulation plan are corresponded according to time sequence, so that the contents of the plans in each time unit group form a time correlation to obtain time correlation plans corresponding to each time unit group; Based on the time correlation plans corresponding to the same time unit group, the inconsistent parts of the power dispatch plan and the global demand regulation plan in the same time unit are identified as plan conflict points; According to the real-time energy supply and demand information of the virtual power plant, adjustment rules of the plan conflict points are determined; Based on the adjustment rules, the plan conflict points in the time correlation plans are corrected to form a collaborative plan, and the collaborative plan is applied to the energy supply and demand matching process of the virtual power plant.

5. The AI-based virtual power plant energy supply and demand precise matching method according to claim 4, characterized in that, The adjustment rules of the plan conflict points are determined according to the real-time energy supply and demand information of the virtual power plant, including: Acquire real-time energy supply and demand information of the virtual power plant, the real-time energy supply and demand information including: real-time total energy supply, real-time total energy demand and real-time state data of energy storage equipment of the virtual power plant; Determine the real-time supply and demand state of the virtual power plant according to the comparison result of the real-time total energy supply and the real-time total energy demand, the real-time supply and demand state including: sufficient supply, balanced supply and insufficient supply; If the real-time supply and demand state is in sufficient supply, adjust the non-essential load demand in the global demand regulation plan to obtain a first adjustment rule; Or, if the real-time supply and demand state is in insufficient supply, adjust the non-critical power generation output in the power dispatching plan to obtain a second adjustment rule; Or, if the real-time supply and demand state is in balanced supply, generate a third adjustment rule based on the real-time state data of the energy storage equipment.

6. An AI-based virtual power plant energy supply and demand precise matching system, characterized in that, Comprise: The acquisition module is used for collecting the output data sequence of the power supply equipment in the virtual power plant and the electricity price data sequence of the electricity market; The adjustment module is used for dynamically adjusting the initial charging and discharging power of the energy storage equipment in the virtual power plant according to the output data sequence and the electricity price data sequence to obtain the target charging and discharging power; The optimization module is used for fusing and optimizing the load prediction model of each energy participating node in the virtual power plant in a federated learning manner based on the target charging and discharging power to generate a global demand regulation plan; The generation module is used for generating a power dispatching plan by using a pre-trained dispatching model based on the output data sequence and the electricity price data sequence; The matching module is used for coordinating the power dispatching plan and the global demand regulation plan to act on the energy supply and demand matching process of the virtual power plant to realize AI-based accurate matching of energy supply and demand of the virtual power plant; The power dispatching plan is generated by using a pre-trained dispatching model based on the output data sequence and the electricity price data sequence, comprising: The output data sequence and the electricity price data sequence are input into the pre-trained dispatching model, and the output data sequence is divided into multiple data segments by a division unit in the dispatching model according to a preset time length; The output value fluctuation range of each data segment is determined by a determination unit in the dispatching model, and the output fluctuation characteristics corresponding to each data segment are obtained based on the output value fluctuation range; The price change amplitude value is obtained by calculating the electricity price data sequence by a calculation unit in the dispatching model, and the price fluctuation trend is generated based on the price change amplitude value; The output fluctuation characteristics and the price fluctuation trend are associated and processed by a processing unit in the dispatching model to obtain the association relationship between the output fluctuation characteristics and the price fluctuation trend, and the power dispatching plan is generated according to the association relationship; The electricity price data sequence is divided into multiple electricity price data intervals according to the preset time length; ​ The price change amplitude values of the electricity price data intervals are determined based on the price change amplitude values of the electricity price data intervals. The price fluctuation directions of the electricity price data intervals are determined based on the price change amplitude values of the electricity price data intervals. The price change amplitude values and the price fluctuation directions of all the electricity price data intervals are integrated in chronological order to generate a price fluctuation trend.

Images