A virtual power plant optimal scheduling method considering renewable energy

Abstract

The application discloses a kind of virtual power plant optimization scheduling methods considering renewable energy, it is related to energy management technical field, including, construct multi-energy collaborative control model analysis operating characteristic dataset, dynamically adjust the resource allocation of virtual power plant;Power in power generation control scheme is carried out time period load alignment, and combined with intelligent demand response mechanism, generate user response strategy;Power generation control scheme and user response strategy are jointly scheduled, and output joint scheduling report of generation and use;Dispatching center executes power generation control and electricity guidance according to joint scheduling report of generation and use to virtual power plant, and according to the scheduling target in operating data after execution and joint scheduling report of generation and use, multi-energy collaborative control model is fed back and updated.The application realizes the accurate identification to the complementarity and coordination between different energy forms by constructing multi-energy collaborative control model analysis operating characteristic dataset, dynamically adjusts the resource allocation of virtual power plant.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a virtual power plant optimization scheduling method that takes renewable energy into account. Background Technology

[0002] With the increasing penetration rate of renewable energy, virtual power plants, as an important means of aggregating distributed energy resources and enhancing grid regulation capabilities, are playing an increasingly crucial role in modern power equipment. Traditional virtual power plant optimization scheduling methods typically construct characteristic variables based on historical operating data and employ mathematical modeling techniques such as mixed-integer linear programming to coordinate the scheduling of multiple energy types. Through load forecasting and demand response mechanisms, coordinated control between generation and consumption sides is achieved, thereby forming a complete scheduling scheme.

[0003] However, existing methods still have certain limitations when dealing with complex and ever-changing energy coordination relationships, especially in terms of the accuracy of the mapping between operating characteristics and scheduling decisions, which has room for improvement. On the one hand, conventional models often rely on static or semi-dynamic historical scheduling samples, making it difficult to accurately reflect the evolution of scheduling behavior caused by changes in equipment status; on the other hand, the lack of a refined identification and dynamic adjustment mechanism for periods of load mismatch affects the overall flexibility and adaptability of scheduling. Existing technologies typically model based on historical scheduling samples and static features to improve the mapping relationship between operating characteristics and scheduling decisions and the matching accuracy of user response strategies. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a virtual power plant optimization scheduling method that considers renewable energy to address the problem of room for improvement in the mapping accuracy between operating characteristics and scheduling decisions.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a virtual power plant optimization scheduling method considering renewable energy, comprising: collecting and preprocessing the operating data of each distributed energy node within the virtual power plant to generate an operating characteristic dataset; constructing a multi-energy collaborative control model to analyze the operating characteristic dataset, dynamically adjusting the resource allocation of the virtual power plant, and obtaining a power generation control scheme; aligning the power in the power generation control scheme with time-period loads, and generating a user response strategy by combining it with an intelligent demand response mechanism; jointly scheduling the power generation control scheme and the user response strategy, and outputting a joint power generation and consumption scheduling report; the dispatch center performing power generation control and consumption guidance on the virtual power plant according to the joint power generation and consumption scheduling report, generating control signals, and feeding back and updating the multi-energy collaborative control model based on the operating data after execution and the scheduling objectives in the joint power generation and consumption scheduling report.

[0008] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the operating data includes power generation, energy storage status, user load, environmental parameters, and operating status of distributed energy node equipment.

[0009] The preprocessing includes data cleaning, normalization, timestamp alignment, and feature extraction.

[0010] The preprocessed runtime data is integrated to generate a runtime feature dataset.

[0011] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the specific steps for constructing a multi-energy collaborative control model to analyze the operating characteristic dataset are as follows:

[0012] Extract the operational feature vector from the operational feature dataset, and extract the scheduling record corresponding to the operational feature vector from the historical scheduling results of the virtual power plant to generate scheduling response variables;

[0013] The running feature vector and scheduling response variable are divided into a cooperative scheduling training set, and modeling training is performed based on the cooperative scheduling training set to output a multi-energy cooperative control model.

[0014] The operational feature dataset is mapped to a multi-energy collaborative control model for field alignment and time synchronization, generating mapped operational feature data.

[0015] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the specific steps of dynamically adjusting the resource allocation of the virtual power plant to obtain a power generation control scheme are as follows:

[0016] Analyze the mapped operational characteristic data to identify the complementarity and coordination relationships between different energy forms and generate optimized input parameters;

[0017] Based on optimized input parameters, virtual power plant operation constraints are set in the multi-energy collaborative control model, and a cost function is set in the multi-energy collaborative control model to adjust the operating cost of the virtual power plant and generate an optimized objective function.

[0018] The standard optimization solver is used to solve the multi-energy coordinated control model containing virtual power plant operation constraints and optimization objective function, outputting a power generation regulation time series dataset, which is then integrated to output a power generation regulation scheme.

[0019] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the specific steps for aligning the power in the power generation control scheme with time-period loads are as follows:

[0020] Obtain the historical load curves on the user side of the virtual power plant and extract the total power supply from the power generation control scheme;

[0021] The historical load curves of the user side and the total power supply are superimposed on a unified time axis, and the supply and demand difference in each time period is calculated as the result of time period load alignment.

[0022] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the specific steps for generating user response strategies by combining an intelligent demand response mechanism are as follows:

[0023] To acquire the operational characteristics of virtual power plants and user electricity consumption behavior, and to build an intelligent demand response mechanism;

[0024] Identify load mismatch periods in the time-period load alignment results and combine them with an intelligent demand response mechanism to generate demand response control instructions;

[0025] The time-period load alignment results and demand response control instructions are integrated to generate user response strategies.

[0026] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the specific steps of jointly scheduling the power generation control scheme and user response strategy to output a joint power generation and user scheduling report are as follows:

[0027] The power generation control scheme and user response strategy are analyzed on a time-by-time supply and demand matching basis to generate a set of dispatch instructions.

[0028] Integrate the scheduling instruction set and output a joint scheduling report.

[0029] As a preferred embodiment of the virtual power plant optimization scheduling method considering renewable energy described in this invention, the scheduling center performs power generation regulation and power consumption guidance on the virtual power plant according to the joint power generation and consumption scheduling report, generates regulation signals, and feeds back and updates the multi-energy collaborative control model based on operating data and scheduling objectives in the joint power generation and consumption scheduling report. The specific steps are as follows:

[0030] Extract power generation control instructions and power consumption guidance signals from the joint power generation and consumption dispatch report, and issue and execute them to distributed energy resources and aggregated users within the virtual power plant to generate control signals;

[0031] Collect and execute control signals to generate post-execution operation data of the virtual power plant;

[0032] The execution data and the scheduling targets set in the joint report are compared to generate the control execution results;

[0033] The multi-energy coordinated control model is updated and fed back based on the results of the regulation and control implementation.

[0034] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the virtual power plant optimization scheduling method considering renewable energy as described in the first aspect of the present invention.

[0035] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the virtual power plant optimization scheduling method considering renewable energy as described in the first aspect of the present invention.

[0036] The beneficial effects of this invention are as follows: By constructing a multi-energy collaborative control model to analyze the operational characteristic dataset, the resource allocation of the virtual power plant is dynamically adjusted, achieving accurate identification of the complementarity and coordination relationships between different energy forms, and generating optimized input parameters and power generation control schemes. The LSTM load forecasting alignment method is used to align the power in the power generation control scheme with time-period loads, and combined with an intelligent demand response mechanism to generate user response strategies, achieving refined identification and dynamic adjustment of user-side load mismatch periods. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 A flowchart for a virtual power plant optimization scheduling method that takes renewable energy into account.

[0039] Figure 2 Optimize the scheduling architecture diagram for virtual power plants.

[0040] Figure 3 A flowchart for constructing a multi-energy collaborative control model.

[0041] Figure 4 Generate a graph for LSTM load prediction alignment and response strategy. Detailed Implementation

[0042] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0043] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0045] Reference Figures 1-4 This is one embodiment of the present invention, which provides a virtual power plant optimization scheduling method considering renewable energy, including the following steps:

[0046] S1. Collect and preprocess the operational data of each distributed energy node in the virtual power plant to generate an operational feature dataset.

[0047] Operational data includes power generation, energy storage status, user load, environmental parameters, and the operational status of distributed energy node equipment;

[0048] It should be noted that the data collection includes: wind farm and photovoltaic power station power generation capacity, measured by smart meters at different times (e.g., every 15 minutes, in kilowatts); energy storage status data, recorded by battery management devices including state of charge percentage, current charge / discharge power, and battery temperature (e.g., 65% state of charge, 20 kW charge / discharge power, 30°C battery temperature); user load data, obtained from user-side smart meters at different times (e.g., 45 kW user load at a certain time); environmental parameters, measured by meteorological sensors including wind speed, irradiance, and ambient temperature (e.g., 6 m / s wind speed, 800 W / m² irradiance, 25°C ambient temperature); and the operational status of distributed energy node devices, read by programmable logic controllers (PLCs) to determine online status, fault status flags, and operating mode indicators (e.g., device in operation with no fault alarms).

[0049] Preprocessing includes data cleaning, normalization, timestamp alignment, and feature extraction;

[0050] Specifically, the process begins with data cleaning. This involves identifying and processing outliers and missing values ​​in the collected data on power generation, energy storage status, user load, environmental parameters, and the operational status of distributed energy nodes. For example, power generation records exceeding the rated power range are removed, and missing energy storage status data is filled using linear interpolation. Next, normalization is performed, mapping the cleaned power generation, energy storage status, user load, and environmental parameter values ​​to a unified dimensional range. For instance, wind farm power generation is converted from kilowatts to standardized values ​​between 0 and 1 to facilitate comparison and calculation between different dimensions of data. Then, timestamp alignment is performed to align time-series data on power generation, energy storage status, user load, environmental parameters, and the operational status of distributed energy nodes from different sources with a unified time granularity. For example, all data is adjusted to a sampling point every 15 minutes, and time linear interpolation is used to fill in gaps caused by differences in sampling frequency. Finally, feature extraction is performed to extract statistical features and trends from the aligned power generation, energy storage status, user load, and environmental parameters. For example, the moving average of photovoltaic power generation or the daily periodic fluctuation of user load is statistically analyzed.

[0051] The preprocessed runtime data is integrated using the sliding window analysis method to generate a runtime feature dataset;

[0052] Furthermore, the preprocessed operational data is integrated along the time dimension using the sliding window analysis method. The preprocessed operational data for consecutive time periods is divided according to a set time window length and step size. For example, a 24-hour window length and a 1-hour step size are used to sequentially extract data segments of power generation, energy storage status, user load, environmental parameters, and the operational status of distributed energy node equipment in different time periods. The time window length represents the time range covered by each analysis, used to extract continuous preprocessed operational data; for example, a 24-hour time window length is set. The step size represents the time interval at which the window moves forward each time, used to control the overlap between adjacent windows; for example, a 1-hour step size means that after processing the data of one window, the window slides forward by 1 hour to continue extracting data for the next time period.

[0053] The system performs statistical analysis on the power generation, energy storage status, user load, and environmental parameters within each window, including mean, variance, maximum, and minimum values. It also retains the operating status of distributed energy node devices as status identifiers. The system arranges the data segments and statistical features corresponding to all windows in chronological order to generate a structured operating feature dataset.

[0054] S2. Construct a multi-energy collaborative control model to analyze the operational characteristic dataset, dynamically adjust the resource allocation of the virtual power plant, and obtain power generation control schemes.

[0055] The time-series feature method is used to extract the operation feature vector from the operation feature dataset, and the response variable generation method is used to extract the scheduling record corresponding to the operation feature vector from the historical scheduling results of the virtual power plant, and generate the scheduling response variable.

[0056] Specifically, the operational feature dataset is segmented and sampled at a fixed time granularity, for example, with an hourly sampling interval, and data segments within each time period are extracted sequentially. Within each sampling interval, statistical and trend characteristics, such as mean, variance, rate of change, and moving average, are calculated for the operational feature dataset to form a multi-dimensional numerical combination, which serves as the operational feature vector for each time period. Using the response variable generation method, scheduling records within the time period corresponding to the operational feature vector are extracted from the historical scheduling results of the virtual power plant. These records include output adjustment instructions for wind farms and photovoltaic power plants, charging and discharging arrangements for energy storage devices, and user load adjustment strategies. For example, in a certain time period, the output of wind farms is reduced by 5%, energy storage devices enter the discharge state, and user load is reduced by 30 kilowatts. Each scheduling record is converted into a numerical response variable, and finally, scheduling response variables corresponding one-to-one with the operational feature vector are generated.

[0057] Based on the sample partitioning method, the running feature vector and scheduling response variable are divided into a collaborative scheduling training set. Then, the mixed integer linear programming method is used to model and train the collaborative scheduling training set to output a multi-energy collaborative control model.

[0058] Furthermore, the sample partitioning method is used to divide the running feature vectors and scheduling response variables into training data subsets and validation data subsets according to time order. For example, 80% of the data is used for training and 20% for validation. This ensures that each running feature vector in the training set maintains temporal consistency with the corresponding scheduling response variable, forming a collaborative scheduling training set.

[0059] A mathematical optimization framework is constructed based on mixed-integer linear programming. Within this framework, the operating characteristic vector is used as the input variable, and the scheduling response variable is used as the objective variable. Constraints are set for power generation balance, the charging and discharging capacity boundaries of energy storage devices, the user load adjustment range, and the operating status limitations of distributed energy node devices. For example, wind farm output cannot exceed the maximum available power, the absolute value of energy storage device charging and discharging power cannot exceed the rated power, and the user load adjustment range cannot exceed 20% of the baseline value. Maximum available power is typically determined based on the equipment's technical parameters and real-time environmental conditions. For instance, the maximum available power of a wind turbine depends on the current wind speed, air density, and the equipment's rated capacity; for photovoltaic power plants, it depends on sunlight intensity, temperature, and the efficiency of photovoltaic modules. Rated power refers to the power output or input capacity that a device should achieve under normal operating conditions according to standards. It is a fixed value defined by the equipment manufacturer based on the equipment's parameters and technical specifications, reflecting the maximum load-bearing capacity of the equipment during long-term stable operation. The baseline value refers to the average level of a variable (such as user load) over a specific time period, used as a benchmark for assessing the magnitude of change.

[0060] The objective function is solved on the collaborative scheduling training set to make the predicted scheduling response variables as close as possible to the actual scheduling records, such as minimizing the weighted sum of output deviation and regulation cost. The objective function is a mathematical expression used to measure the merits of the scheduling scheme. In the multi-energy collaborative control model, it consists of multiple weighted terms, such as power generation deviation, regulation cost, and energy storage loss. By minimizing the objective function, the generated scheduling response variables are made close to the actual records. Finally, the multi-energy collaborative control model is output.

[0061] The operational feature dataset is mapped to a multi-energy collaborative control model using a data mapping method to perform field alignment and time synchronization, thereby generating mapped operational feature data.

[0062] It should be noted that, according to the field naming rules, the running feature dataset is matched one by one with the input fields required by the multi-energy collaborative control model. For example, the wind farm power generation is mapped to the corresponding wind farm output active power field in the multi-energy collaborative control model.

[0063] Time synchronization is performed on the time series data corresponding to each field according to the time granularity set by the multi-energy collaborative control model. For example, all time series data are standardized to a sampling point every 15 minutes, and linear interpolation is used to handle time offsets and missing data in the fields. Given the data of adjacent time points, approximate values ​​at missing time points are calculated using a linear function to achieve alignment and integrity restoration of the time series. The set time granularity is usually determined according to the requirements and scheduling cycle of the multi-energy collaborative control model. For example, a shorter time granularity is used in minute-level rapid response scenarios, while a longer time granularity can be used in day-ahead scheduling. It is ensured that all operational feature data are consistent with the input format of the multi-energy collaborative control model in both time and field dimensions, ultimately generating mapped operational feature data.

[0064] The collaborative analysis method is used to analyze the mapped operational characteristic data, identify the complementarity and coordination relationships between different energy forms, and generate optimized input parameters.

[0065] Specifically, multivariate correlation analysis is performed on the mapped operational characteristic data, and the Pearson correlation coefficient is used to calculate the correlation strength between the output of each energy source. For example, the negative correlation coefficient between wind power output and solar intensity is calculated to be -0.73. Through time series cross-correlation analysis, the output fluctuation patterns and response delay characteristics of various energy sources at different times are identified. For example, it is identified that photovoltaic output is at a high level from 10 am to 4 pm every day, while wind power output is mainly concentrated at night to early morning.

[0066] The correlation strength between the outputs of various energy sources is calculated using the Pearson correlation coefficient, expressed as follows:

[0067] ;

[0068] in, This represents the strength of the correlation between the outputs of various energy sources; This represents an index representing a specific type of distributed energy node within a virtual power plant; Indicates an index for another type of energy node, used for comparison with... Perform correlation analysis; Indicates energy node The sequence of operational characteristics; Indicates energy node The sequence of operational characteristics; T represents the total number of time points; Indicates the corresponding point in time and The sum of products; express Summation over the entire time period; express Summation over the entire time period;

[0069] Based on the correlation strength between various energy outputs, statistical analysis is used to extract key parameters characterizing energy synergy. These key parameters include the wind-solar complementarity rate, energy storage regulation response time, and load-new energy output matching degree. The wind-solar complementarity rate is calculated based on the power difference and superposition value of wind and solar power within the same time period in historical output data. The energy storage regulation response time is derived from the charging and discharging records of energy storage devices, for example, statistically analyzing the time interval from receiving an instruction to its actual execution, with an average response time of 5 minutes. The load-new energy output matching degree is calculated based on the overlap area ratio between the user load curve and the new energy output curve. The user load curve refers to the record of the actual power consumption of the user side changing over time within a certain time range, reflecting the temporal characteristics of electricity consumption behavior. The new energy output curve refers to the changing trend of the actual output power of renewable energy equipment such as wind farms within a certain time range, reflecting the volatility and intermittent characteristics of new energy power generation. These key parameters are used as optimization input parameters for the multi-energy synergistic control model.

[0070] Based on optimized input parameters, constraint modeling is used to set virtual power plant operation constraints in the multi-energy collaborative control model, and particle swarm optimization is used to set cost functions in the multi-energy collaborative control model to adjust the operating cost of the virtual power plant and generate optimization objective functions.

[0071] It should be noted that the optimized input parameters are used as the basic input for constraint modeling. The constraint modeling method is used to set boundaries for the power generation balance, energy storage charging and discharging capacity, user load adjustment range, and operating status of distributed energy node equipment in the multi-energy collaborative control model, which serve as constraints for the operation of the virtual power plant. For example, the total output power of wind farms and photovoltaic power plants must not be lower than the minimum power supply requirement, the charging and discharging power of energy storage equipment must not exceed the rated power, and the user load adjustment range must not exceed 20% of the baseline value. The minimum power supply requirement means that during the operation of the virtual power plant, the combined power generation of wind farms and photovoltaic power plants must meet a minimum power supply level to ensure the basic stability and reliability of power supply. It is usually set based on load forecasting or dispatching plans. For example, the virtual power plant must provide at least 10 MW of renewable energy output during a certain period to ensure the electricity demand of key users. The rated power means that the charging of energy storage equipment at any time cannot exceed the maximum value of the rated power, which is used to prevent the equipment from overloading and protect the battery life. For example, if the rated charging and discharging power of an energy storage device is 500 kW, then it can only charge or release a maximum of 500 kWh of energy per hour. The baseline value is usually a reference load level derived from historical electricity consumption data. For example, if a user's baseline load is 100 kilowatts during a specific period, the load after adjustment must not be lower than 80 kilowatts or higher than 120 kilowatts.

[0072] A cost function is constructed by introducing particle swarm optimization (PSO). The total cost objective function expression of the virtual power plant's multi-energy coordinated dispatch is incorporated into the power generation costs of wind farms, photovoltaic power plants, energy storage equipment charging and discharging losses, and user-side load regulation incentive costs. For example, the total operating cost is expressed as a weighted sum of the unit kilowatt-hour power generation cost and regulation costs. The optimal solution is iteratively searched using PSO under the premise of satisfying all operating constraints. The weight allocation is adjusted to minimize the operating cost, and finally, an optimization objective function for regulating the operation strategy of the virtual power plant is generated.

[0073] The standard optimization solver is used to solve the multi-energy coordinated control model containing virtual power plant operation constraints and optimization objective function, outputting a power generation regulation time series dataset, which is then integrated using the structured data integration method to output a power generation regulation scheme;

[0074] Furthermore, the objective function and operational constraints are loaded as inputs into the standard optimization solver. Under the premise of satisfying all operational constraints, the standard optimization solver, with the objective of minimizing operating costs, uses the interior-point method to numerically solve the multi-energy coordinated control model. Under the premise of satisfying the virtual power plant's operational constraints, it iteratively solves the constructed mixed-integer linear programming problem through mathematical optimization algorithms, outputting the time-period output adjustment values ​​of wind farms and photovoltaic power plants, the charging and discharging plans of energy storage devices, and user-side load regulation instructions, which are integrated into a structured power generation regulation time-series dataset. The time-period output adjustment values ​​of wind farms and photovoltaic power plants refer to the specific power generation values ​​determined by the optimal power generation regulation scheme obtained from solving the multi-energy coordinated control model within each scheduling period. The charging and discharging plans of energy storage devices refer to the charging or discharging states and corresponding power arrangements that energy storage devices should execute within each scheduling period. User-side load regulation instructions refer to the power consumption adjustment requirements issued to users within each scheduling period, used to participate in the supply and demand balance of power equipment.

[0075] By using structured data integration, the various types of control instructions in the power generation control time series dataset are classified and organized according to time sequence, energy type, and execution object. For example, wind farm, photovoltaic power station, energy storage equipment, and user-side load adjustment instructions are organized at the hourly granularity. Energy type refers to the different types of energy involved in power generation control, such as wind power, photovoltaic, and energy storage. Execution object refers to the specific equipment or user group affected by the power generation control instructions, such as wind farm, photovoltaic power station, and electricity users. Finally, a complete power generation control scheme is output.

[0076] S3. Align the power in the power generation control scheme with the load during the time period, and generate user response strategies by combining them with the intelligent demand response mechanism.

[0077] The load synchronization method is used to obtain the user-side historical load curve of the virtual power plant, and the multi-source power time-series superposition method is used to extract the total power supply in the power generation control scheme.

[0078] Specifically, the electricity load records of various users in the virtual power plant are organized according to a fixed time granularity. For example, with a sampling point every 15 minutes, the user-side load data of the past 30 days is extracted to form the user-side historical load curve. The multi-source power time-series superposition method is used to align the time-by-time output adjustment values ​​of wind farms and photovoltaic power plants and the charging and discharging plans of energy storage devices in the power generation control scheme, and the values ​​are superimposed at each time point. For example, if the output of wind farm is 10 MW, the output of photovoltaic power plant is 8 MW, and the discharge of energy storage device is 2 MW in a certain time period, the total power supply of that time period is 20 MW, thus completing the extraction of the total power supply.

[0079] The LSTM load forecasting alignment method is used to overlay the user-side historical load curve and total power supply on a unified time axis, and calculate the supply-demand difference in each time period as the time period load alignment result.

[0080] Furthermore, historical load data from the user side of the virtual power plant is collected and organized according to a fixed time granularity to form time series data. The time series data is normalized and divided into training and testing sets, for example, 80% of the data is used for training and 20% for validation. An LSTM neural network structure is constructed, including an input layer, several LSTM hidden layers, and an output layer. The input data is the historical load sequence and related influencing factors, such as temperature and date type features. The output is the load forecast value within the target time period. The network parameters are continuously adjusted through the backpropagation algorithm to minimize the prediction error on the training set and evaluate the performance of the LSTM load forecast model on the test set. The model performance evaluation is used to measure the prediction accuracy and stability of the LSTM load forecast model on the test data. The network parameters refer to the weights and biases used in the data calculation in the LSTM load forecast model, which determine the way the load forecast value affects the load forecast result. Finally, the LSTM load forecast model is constructed.

[0081] An LSTM load forecasting model is used to perform time series forecasting on the historical load curves on the user side, generating load forecast values ​​with the same time granularity as the power generation control scheme. The total power supply is aligned to a unified time axis according to the same time granularity. Under the unified time axis, the total power supply at each time point is compared with the corresponding load forecast value, and the supply-demand difference for each time period is calculated. For example, if the total power supply is 20 MW and the load forecast value is 18 MW in a certain time period, then the supply-demand difference for that time period is 2 MW. Finally, the supply-demand differences for all time periods are combined to form the time period load alignment result.

[0082] By using the load engine method to obtain the operating characteristics of virtual power plants and user electricity consumption behavior, an intelligent demand response mechanism is constructed.

[0083] It should be noted that the load engine method is used to collect the operating characteristics of the virtual power plant, such as wind farm power generation, photovoltaic power generation, energy storage equipment charging and discharging status, user load and environmental parameters. Combined with the power generation control scheme and the time period load alignment results, the typical operating mode under multi-energy collaborative control is identified.

[0084] By analyzing the actual execution of user-side load adjustment commands, the electricity distribution characteristics, load fluctuations, and response behaviors of various user groups at different time periods are extracted to identify adjustment capabilities and response patterns. For example, after receiving an electricity price incentive signal, a certain type of commercial user can proactively reduce their air conditioning load during peak hours, with an adjustment range of approximately 10% to 20%. Based on the analysis results, combined with user load characteristics and response sensitivity, a user response capability classification rule is established, dividing users into three categories: high response potential, medium response potential, and low response potential. Corresponding load adjustment methods are then set according to different categories, such as guiding users to adjust autonomously based on price signals, thereby constructing an intelligent demand response mechanism.

[0085] The supply-demand gap threshold determination method is used to identify the load mismatch period in the period load alignment results, and combined with the intelligent demand response mechanism, demand response control instructions are generated.

[0086] Specifically, the supply-demand difference for each time period in the load alignment results is compared with a set threshold. For example, time periods where the absolute value of the supply-demand difference exceeds 5% of the total power supply are identified as load mismatch periods. Based on the identification results, load mismatch periods with insufficient or excessive power supply are screened out. The set threshold is determined based on the tolerance range of power supply and demand balance during the operation of the virtual power plant, and is usually set with reference to factors such as historical load fluctuation characteristics, power supply stability requirements, and control response capabilities. For example, with a total power supply of 20 MW, the supply-demand difference threshold can be set to 5% to 10%.

[0087] By combining the pre-defined user responsiveness classification rules and adjustment methods in the intelligent demand response mechanism, load reduction instructions based on price signals are sent to users with high responsiveness potential during periods of load mismatch. During periods of power surplus, users are guided to increase electricity consumption, ultimately generating differentiated demand response control instructions for different user groups. The pre-defined user responsiveness classification rules and adjustment methods refer to selecting appropriate user groups and implementing matching load adjustment measures during periods of load mismatch based on user responsiveness potential categories and corresponding control strategies. The corresponding control strategies guide how to implement effective demand response control instructions for various user groups during periods of load mismatch.

[0088] The multi-dimensional rule mapping method is used to integrate the time period load alignment results and demand response control instructions to generate user response strategies.

[0089] Furthermore, based on the supply-demand difference value and load mismatch type at each time point in the load alignment results, it is determined whether control actions are triggered in each time period; the demand response control instructions are aligned according to a unified time axis to ensure that the control actions are synchronized with the load mismatch period; the control action refers to the specific power consumption adjustment operation issued to the user in accordance with the preset control strategy during the load mismatch period; the preset control strategy refers to the control rules and execution plan formulated in the intelligent demand response mechanism for different load mismatch types and user response capabilities.

[0090] In terms of energy type, the impact of different energy forms such as wind power, photovoltaics, and energy storage on user response behavior is differentiated. For example, when wind and solar power output fluctuates significantly, users are given priority to adjust flexible loads. In terms of user, specific load adjustment ratios and incentive amounts are allocated by combining users' historical response records with current control targets. For example, an instruction to reduce load by 10% is issued to a certain type of user, along with an incentive standard of 0.3 yuan per kilowatt. Finally, information is structured and integrated through a multi-dimensional rule mapping method to form a user response strategy organized by time, energy type, and user category.

[0091] S4. Jointly schedule the power generation control scheme and user response strategy, and output a joint scheduling report for power generation and user.

[0092] By using the generation-consumption collaborative analysis method, the power generation control scheme and user response strategy are analyzed to match supply and demand on a time-by-time basis, and a set of dispatch instructions is generated.

[0093] Specifically, the power output adjustment values ​​of wind farms and photovoltaic power plants in the power generation control scheme, the charging and discharging plans of energy storage devices, and the load adjustment instructions of users in the user response strategy are aligned on a unified time axis to generate power generation-side control actions and user-side response actions. The total power supply and the expected load level on the user side are summarized in each time period, and the supply and demand difference in the time period load alignment results are matched and analyzed. For example, if the total power supply is 20 MW, the expected load on the user side is 18 MW, and there is a 2 MW power supply surplus in a certain time period, then the supply is in a state of oversupply in that time period.

[0094] Based on the matching analysis results, and taking into account user response capability classification rules and corresponding control strategies, it is determined whether further adjustments to power generation output and guidance for users to increase electricity consumption are needed. Finally, the power generation-side control actions and user-side response actions for each time period are combined to form a dispatch instruction set.

[0095] The scheduling instruction set is integrated using the scheduling aggregation method, and a joint scheduling report is output.

[0096] It should be noted that the dispatch instruction sets are summarized and organized in chronological order to ensure that the time-period output adjustment values ​​of wind farms and photovoltaic power plants, the charging and discharging plans of energy storage devices, and the load adjustment instructions of users are consistent in the time dimension. The dispatch instruction sets for each time period are classified and summarized according to energy type and execution object. For example, the total adjustment and peak value changes of wind power, photovoltaic, energy storage, and different user groups in each time period of the day are statistically analyzed separately. Finally, the summary is organized in a structured form to form a complete dispatch summary that includes time distribution, energy type distribution, and user response status, which serves as a joint dispatch report for generation and consumption.

[0097] S5. The dispatch center performs power generation regulation and power consumption guidance on the virtual power plant according to the joint dispatch report of power generation and consumption, generates control signals, and feeds back and updates the multi-energy collaborative control model based on the post-execution operation data and the dispatch objectives in the joint dispatch report of power generation and consumption.

[0098] The generation control instructions and consumption guidance signals are extracted from the joint dispatch report using the control instruction parsing method, and then distributed and executed to the distributed energy and aggregated users in the virtual power plant through the control signal transmission method to generate control signals.

[0099] Furthermore, the control command parsing method is used to identify and extract the corresponding power generation control commands in each time period of the power generation and consumption joint dispatch report, including the time-by-time output adjustment values ​​of wind farms and photovoltaic power plants, the charging and discharging plans of energy storage devices, and the power consumption guidance signals; the extracted control commands are matched to the execution object categories in the virtual power plant according to the time sequence, including the corresponding distributed energy nodes and aggregated user groups;

[0100] By using a control signal transmission method, power generation control commands are sent to the local control terminals of wind farms, photovoltaic power stations, and energy storage devices via a communication protocol. At the same time, power consumption guidance signals are sent to the user-side energy management center. Finally, after the command parsing and action response are completed at each execution terminal, control signals are generated.

[0101] Collect and execute control signals to generate post-execution operation data of the virtual power plant;

[0102] Specifically, after the control signal is issued and executed, the actual output values ​​of the wind farm and photovoltaic power station of the virtual power plant, the charging and discharging power of the energy storage equipment, and the actual load changes of the aggregated users are collected in real time according to a unified time granularity. The operating status records of each distributed energy node and user group in each time period are summarized. For example, in a certain time period, the actual output of the wind farm is 18 MW, the energy storage equipment discharges 3 MW, and the load of a certain aggregated user group decreases by 2.5 MW. The collected operating status records are classified and organized to form the post-execution operating data.

[0103] The quantitative evaluation method compares the post-implementation operational data with the scheduling targets set in the joint generation and utilization report to generate control execution results. Based on the control execution results, the multi-energy collaborative control model is fed back and updated.

[0104] It should be noted that a quantitative evaluation method is used to compare the post-execution operational data with the corresponding dispatch target values ​​in the joint power generation and consumption dispatch report for each time period to obtain the execution deviation of each control action. For example, if the target output of a wind farm is 20 MW and the actual output is 19.2 MW, the deviation is 0.8 MW. Control actions refer to the specific operations that adjust the output of power generation equipment within a specific time period according to dispatch instructions.

[0105] Based on the execution deviations at all time points, key performance indicators such as mean absolute error, the period of maximum deviation, and response delay time are statistically analyzed to form the control execution results. The control execution results are used as feedback input to the multi-energy collaborative control model to identify the prediction deviations in the optimization process of the multi-energy collaborative control model, and to adjust the parameters of the multi-energy collaborative control model to complete the iterative update of the multi-energy collaborative control model.

[0106] This embodiment also provides a computer device applicable to the virtual power plant optimization scheduling method considering renewable energy, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the virtual power plant optimization scheduling method considering renewable energy as proposed in the above embodiment.

[0107] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0108] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the virtual power plant optimization scheduling method considering renewable energy proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0109] In summary, this invention achieves precise identification of the complementarity and coordination relationships between different energy forms by constructing a multi-energy collaborative control model to analyze operational characteristic datasets and dynamically adjust resource allocation in virtual power plants, thereby generating optimized input parameters and power generation control schemes. Furthermore, by using the LSTM load forecasting alignment method to align power in the power generation control scheme across time periods and combining it with an intelligent demand response mechanism to generate user response strategies, this invention enables refined identification and dynamic adjustment of user-side load mismatch periods.

[0110] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A virtual power plant optimization scheduling method considering renewable energy, characterized in that: include, Collect and preprocess the operational data of each distributed energy node in the virtual power plant to generate an operational feature dataset; Construct a multi-energy collaborative control model to analyze the operational characteristic dataset, dynamically adjust the resource allocation of the virtual power plant, and obtain power generation control schemes; The power in the power generation control scheme is aligned with the load during different time periods, and combined with the intelligent demand response mechanism, a user response strategy is generated. The power generation control scheme and user response strategy are jointly dispatched, and a joint dispatch report of power generation and user is output. The dispatch center performs power generation regulation and power consumption guidance on the virtual power plant based on the joint dispatch report of power generation and consumption, generates regulation signals, and provides feedback and updates to the multi-energy collaborative control model based on the post-execution operation data and the dispatch objectives in the joint dispatch report of power generation and consumption. The specific steps for constructing a multi-energy collaborative control model and analyzing the operational characteristic dataset are as follows: Extract the operational feature vector from the operational feature dataset, and extract the scheduling record corresponding to the operational feature vector from the historical scheduling results of the virtual power plant to generate scheduling response variables; The running feature vector and scheduling response variable are divided into a cooperative scheduling training set, and modeling training is performed based on the cooperative scheduling training set to output a multi-energy cooperative control model. The operational feature dataset is mapped to the multi-energy collaborative control model, and field alignment and time synchronization are performed to generate the mapped operational feature data. The specific steps for dynamically adjusting the resource allocation of the virtual power plant to obtain a power generation control scheme are as follows: Analyze the mapped operational characteristic data to identify the complementarity and coordination relationships between different energy forms and generate optimized input parameters; Based on optimized input parameters, virtual power plant operation constraints are set in the multi-energy collaborative control model, and a cost function is set in the multi-energy collaborative control model to adjust the operating cost of the virtual power plant and generate an optimized objective function. The standard optimization solver is used to solve the multi-energy coordinated control model containing virtual power plant operation constraints and optimization objective function, outputting a power generation regulation time series dataset, which is then integrated to output a power generation regulation scheme. The specific steps for aligning the power output in the power generation control scheme with the load over time are as follows: The load synchronization method is used to obtain the user-side historical load curve of the virtual power plant, and the multi-source power time-series superposition method is used to extract the total power supply in the power generation control scheme. The LSTM load forecasting alignment method is used to overlay the user-side historical load curve and total power supply on a unified time axis, and calculate the supply-demand difference in each time period as the time period load alignment result. The specific steps for generating a user response strategy by combining an intelligent demand response mechanism are as follows: To acquire the operational characteristics of virtual power plants and user electricity consumption behavior, and to build an intelligent demand response mechanism; Identify load mismatch periods in the time-period load alignment results and combine them with an intelligent demand response mechanism to generate demand response control instructions; The time-period load alignment results and demand response control instructions are integrated to generate user response strategies.

2. The virtual power plant optimization scheduling method considering renewable energy as described in claim 1, characterized in that: The operational data includes power generation, energy storage status, user load, environmental parameters, and the operational status of distributed energy node equipment. The preprocessing includes data cleaning, normalization, timestamp alignment, and feature extraction. The preprocessed runtime data is integrated to generate a runtime feature dataset.

3. The virtual power plant optimization scheduling method considering renewable energy as described in claim 1, characterized in that: The specific steps for jointly scheduling the power generation control scheme and user response strategy, and outputting a joint power generation and user scheduling report, are as follows: The power generation control scheme and user response strategy are analyzed on a time-by-time supply and demand matching basis to generate a set of dispatch instructions. Integrate the scheduling instruction set and output a joint scheduling report.

4. The virtual power plant optimization scheduling method considering renewable energy as described in claim 1, characterized in that: The dispatch center performs power generation control and power consumption guidance on the virtual power plant based on the joint power generation and consumption dispatch report, generates control signals, and updates the multi-energy collaborative control model based on the post-execution operating data and the dispatch objectives in the joint power generation and consumption dispatch report. The specific steps are as follows: Extract power generation control instructions and power consumption guidance signals from the joint power generation and consumption dispatch report, and issue and execute them to distributed energy resources and aggregated users within the virtual power plant to generate control signals; Collect and execute control signals to generate post-execution operation data of the virtual power plant; The execution data is compared with the scheduling targets set in the joint report to generate the control execution results; The multi-energy coordinated control model is updated and fed back based on the results of the regulation and control implementation.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the virtual power plant optimization scheduling method considering renewable energy as described in any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the virtual power plant optimization scheduling method considering renewable energy as described in any one of claims 1 to 4.

Images