Load optimization scheduling method, system and equipment of virtual power plant and medium

Abstract

The invention discloses a load optimization scheduling method, system and device for a virtual power plant and a medium, and the method comprises the steps: obtaining the target output data of a power grid system and the target electricity price data of the virtual power plant; a double-layer model is constructed, the double-layer model comprises a lower-layer uncertainty prediction model and an upper-layer multi-load optimization scheduling model, and the lower-layer uncertainty prediction model is used for predicting probability distribution of electricity price and renewable energy source output of the virtual power plant in a future time period; the upper-layer multi-load optimal scheduling model is used for generating a multi-load optimal scheduling strategy of the virtual power plant; inputting the target electricity price data and the target output data into the double-layer model to obtain a multi-load optimization scheduling strategy output by the double-layer model; and performing multi-load scheduling control on the virtual power plant according to the multi-load optimal scheduling strategy. The method can effectively improve the virtual power plant load optimization scheduling effect. The invention relates to the technical field of power dispatching.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power dispatching, and in particular to a load optimization dispatching method, system, device and medium of a virtual power plant. BACKGROUND

[0002] With the promotion of global energy transformation, a large number of renewable energy is connected to the power grid, which significantly increases the uncertainty of the power system. As a new form of power system integrating distributed energy, adjustable load and other resources, the virtual power plant plays an important role in improving the renewable energy consumption capacity and optimizing the power resource configuration, so that the load optimization dispatching of the virtual power plant becomes one of the focuses of relevant staff.

[0003] At present, the related technology usually uses a machine learning method to perform load prediction of the virtual power plant, and takes the load prediction result as a deterministic input for optimization dispatching. This method does not fully consider the influence of prediction error on optimization dispatching, and the effect of load optimization dispatching is limited.

[0004] Therefore, the problems of the related technology still need to be solved and optimized. SUMMARY

[0005] The purpose of the present application is to at least partially solve one of the technical problems in the related art.

[0006] To this end, one purpose of an embodiment of the present application is to provide a load optimization dispatching method, system, device and medium of a virtual power plant, wherein the method can effectively improve the effect of load optimization dispatching of the virtual power plant.

[0007] In order to achieve the above technical purpose, the technical solutions adopted by the embodiments of the present application include: In a first aspect, the present application provides a load optimization dispatching method of a virtual power plant, comprising: obtaining target output data of a power grid system and target price data of a virtual power plant; constructing a double-layer model, the double-layer model comprising a lower-layer uncertainty prediction model and an upper-layer multi-load optimization dispatching model, the lower-layer uncertainty prediction model being used to predict the probability distribution of the price and renewable energy output of the virtual power plant in a future period, and the upper-layer multi-load optimization dispatching model being used to generate a multi-load optimization dispatching strategy of the virtual power plant; inputting the target price data and the target output data into the double-layer model to obtain a multi-load optimization dispatching strategy output by the double-layer model; performing multi-load dispatching control on the virtual power plant according to the multi-load optimization dispatching strategy.

[0008] In addition, the method according to the above-mentioned embodiment of the present application can further have the following additional technical features. Further, in an embodiment of the present application, the lower-layer uncertainty prediction model is obtained through the following training steps: obtaining output training data and electricity price training data of the virtual power plant, as well as a gradient decision model and a lower-layer objective function of the gradient decision model; performing feature engineering processing on the electricity price training data and the output training data to obtain first training features and second training features, the first training features being used to represent feature information of the electricity price training data in a certain time period, and the second training features being used to represent feature information of the output training data in a certain time period; performing information gain training on the gradient decision model according to the lower-layer objective function, the first training features and the second training features to obtain the lower-layer uncertainty prediction model.

[0009] Further, in an embodiment of the present application, the upper-layer multi-load optimal scheduling model is obtained through the following steps: obtaining several different types of decision variables of a power grid system virtual power plant; each of the decision variables corresponds to one adjustable load of the virtual power plant; determining an upper-layer objective function and an upper-layer constraint condition set according to all the decision variables, the upper-layer constraint condition set including a power balance constraint, a load adjustment range constraint and an energy storage constraint; performing linear regularization on the upper-layer objective function to obtain a regularized upper-layer objective function; constructing the upper-layer multi-load optimal scheduling model according to the upper-layer constraint condition set and the regularized upper-layer objective function.

[0010] Further, in an embodiment of the present application, the method of inputting the target electricity price data and the target output data into the double-layer model to obtain the multi-load optimal scheduling strategy output by the double-layer model includes: inputting the target electricity price data and the target output data into the lower-layer uncertainty prediction model to perform uncertainty prediction, to obtain electricity price prediction data and output prediction data; inputting the electricity price prediction data and the output prediction data into the upper-layer multi-load optimal scheduling model to perform scheduling analysis, to obtain the multi-load optimal scheduling strategy.

[0011] Further, in an embodiment of the present application, the method further includes: obtaining electricity price prediction data and output prediction data output by the lower-layer uncertainty prediction model; The electricity price forecast data and power output forecast data are processed to generate several uncertainty scenarios. Based on all the aforementioned uncertainty scenarios, a robustness analysis is performed on the multi-load optimization scheduling strategy to obtain robustness analysis results.

[0012] Furthermore, in one embodiment of this application, the step of performing robustness analysis on the multi-load optimization scheduling strategy based on all the aforementioned uncertainty scenarios to obtain robustness analysis results includes: Obtain the minimum acceptable benefit, and determine the robustness index based on the minimum acceptable benefit; Based on the uncertain scenario, scenario analysis is performed on the multi-load optimization scheduling strategy to obtain scenario benefits; Based on the robustness index, a benefit analysis is performed on the scenario benefits to obtain the robustness analysis results.

[0013] Furthermore, in one embodiment of this application, the objective function of the upper-layer multi-load optimization scheduling model is:

[0014] in, The objective function of the upper-level multi-load optimization scheduling model; Number of scheduling periods; For virtual power plants during dispatch periods Electricity price; Number of adjustable power load types; For the first Adjustable power load during dispatching periods Adjustable power, Number of adjustable heat load types; For the first Adjustable thermal load during scheduling period The adjustment amount, For the first The power conversion coefficient of a type of thermal load; This is the adjustment function.

[0015] Secondly, embodiments of this application provide a load optimization and dispatching system for a virtual power plant, comprising: The first processing unit is used to acquire the target power output data of the power grid system and the target electricity price data of the virtual power plant. a second processing unit configured to construct a double-layer model, the double-layer model comprising a lower-layer uncertainty prediction model and an upper-layer multi-load optimization scheduling model, the lower-layer uncertainty prediction model being configured to predict a probability distribution of electricity price and renewable energy output of the virtual power plant in a future period, and the upper-layer multi-load optimization scheduling model being configured to generate a multi-load optimization scheduling strategy of the virtual power plant; a third processing unit configured to input the target electricity price data and the target output data into the double-layer model to obtain a multi-load optimization scheduling strategy output by the double-layer model; a fourth processing unit configured to perform multi-load scheduling control on the virtual power plant according to the multi-load optimization scheduling strategy.

[0016] In a third aspect, an electronic device is provided, comprising: at least one processor; at least one memory configured to store at least one program; when the at least one program is executed by the at least one processor, the at least one processor implements the method described above.

[0017] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a program executable by a processor, and the program, when executed by the processor, is configured to implement the method described above.

[0018] The advantages and beneficial effects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be understood by the practice of the present application: The virtual power plant load optimization scheduling method, system, device and medium disclosed by the embodiments of the present application, wherein the method obtains target output data of a power grid system and target electricity price data of a virtual power plant; constructs a double-layer model, the double-layer model comprising a lower-layer uncertainty prediction model and an upper-layer multi-load optimization scheduling model, the lower-layer uncertainty prediction model being configured to predict a probability distribution of electricity price and renewable energy output of the virtual power plant in a future period, and the upper-layer multi-load optimization scheduling model being configured to generate a multi-load optimization scheduling strategy of the virtual power plant; inputs the target electricity price data and the target output data into the double-layer model to obtain a multi-load optimization scheduling strategy output by the double-layer model; and performs multi-load scheduling control on the virtual power plant according to the multi-load optimization scheduling strategy. The method can fully consider the influence of prediction error on optimization scheduling by performing uncertainty prediction on electricity price and output through a lower-layer uncertainty prediction model to determine the probability distribution of electricity price and output in a future period, and then performing multi-load optimization scheduling based on an upper-layer load optimization scheduling model, thereby effectively improving the effect of load optimization scheduling. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of this application or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions in this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1 A flowchart illustrating a load optimization scheduling method for a virtual power plant provided in this application embodiment; Figure 2 A schematic diagram of the framework of a load optimization dispatching system for a virtual power plant provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0021] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0023] Currently, the relevant technologies typically employ machine learning methods to forecast the load of virtual power plants and use the load forecast results as deterministic input for optimal scheduling. However, this approach does not fully consider the impact of forecast errors on optimal scheduling, resulting in limited effectiveness of load optimization scheduling. Furthermore, since the model solution efficiency decreases rapidly with the increase in the types and quantities of adjustable loads, this approach is difficult to meet the real-time scheduling requirements of the power grid system, leading to poor efficiency in optimal scheduling.

[0024] Further, there is a part of the related art which optimizes scheduling by assuming that the electricity price and the renewable energy output are subject to a specific probability distribution. However, due to the complex characteristics of the actual power scenario, the electricity price and the renewable energy output and their probability distribution determined by this method often do not match the complex dynamic characteristics of the actual power scenario, which makes the effectiveness and reliability of the optimized scheduling unsatisfactory.

[0025] It should be noted that the above-mentioned related art is only used to assist in understanding the technical solutions of the present application and does not mean that it belongs to the prior art which has been disclosed.

[0026] Therefore, the embodiments of the present application provide a load optimization scheduling method, system, device and medium for a virtual power plant, wherein the method determines the probability distribution of the electricity price and the output in the future period (i.e. the electricity price prediction data and the output prediction data) through the lower layer uncertainty prediction model for the uncertainty prediction of the electricity price and the output, which can obtain prediction data that matches the actual power scenario more closely; and then inputs the electricity price prediction data and the output prediction data into the upper layer load optimization scheduling model for multi-load optimization scheduling, which, compared with the traditional deterministic prediction (such as outputting only a single prediction value), can fully quantify the uncertainty in the characteristics of the input data with the prediction data with probability distribution, so that the influence of the prediction error on the optimization scheduling can be fully considered, and the effect, effectiveness and reliability of the load optimization scheduling can be effectively improved.

[0027] Further, the method constructs a linear upper layer multi-load optimization scheduling model by using the linearly regularized upper layer objective function and solves it, which, compared with the solving method of the traditional intelligent optimization algorithm, requires less computing resources and is conducive to improving the efficiency of the optimization scheduling. In addition, the method accurately captures the dynamic characteristics of the electricity price and the renewable energy output based on the lower layer uncertainty prediction model and generates an uncertainty scenario close to the actual situation, analyzes the robustness of the multi-load optimization scheduling strategy, and optimizes the virtual power plant based on the analysis results and the multi-load optimization scheduling strategy, which can effectively cope with the influence of various uncertain factors, such as the sharp fluctuations of the electricity price or the significant changes of the renewable energy output, and can ensure the stable operation of the virtual power plant, and the fluctuation range of the benefits generated by the virtual power plant in the operation process is reduced compared with the traditional deterministic optimization method, which significantly enhances the adaptability to the changes of the power market environment.

[0028] Reference Figure 1 In the embodiments of the present application, a load optimization scheduling method for a virtual power plant comprises: Step 110, obtaining target output data of a power grid system and target electricity price data of a virtual power plant; In the embodiments of the present application, the target electricity price data can be the electricity price of the virtual power plant at the current time point during the operation of the power grid system, and the target output data can be the output of renewable energy such as photovoltaic output and wind power output during the operation of the power grid system.

[0029] Step 120, a double-layer model is constructed, the double-layer model comprising a lower-layer uncertainty prediction model and an upper-layer multi-load optimal scheduling model, the lower-layer uncertainty prediction model being used for predicting the probability distribution of the electricity price and the renewable energy output of the virtual power plant in a future period, and the upper-layer multi-load optimal scheduling model being used for generating a multi-load optimal scheduling strategy of the virtual power plant; In the embodiments of the present application, a cascaded double-layer model can be constructed, the lower-layer uncertainty prediction model in the double-layer model being used for predicting the probability distribution of the renewable energy output and the probability distribution of the electricity price of the virtual power plant in a future period, and the upper-layer multi-load optimal scheduling model being used for generating a multi-load optimal scheduling strategy of the virtual power plant based on the output of the lower-layer uncertainty prediction model.

[0030] In some embodiments, the lower-layer uncertainty prediction model is obtained through the following training steps: obtaining output training data and electricity price training data of the virtual power plant, and a gradient decision model and a lower-layer objective function of the gradient decision model; performing feature engineering processing on the electricity price training data and the output training data to obtain first training features and second training features, the first training features being used for characterizing the feature information of the electricity price training data in a certain period of time, and the second training features being used for characterizing the feature information of the output training data in a certain period of time; performing information gain training on the gradient decision model according to the lower-layer objective function, the first training features and the second training features to obtain the lower-layer uncertainty prediction model.

[0031] In the embodiments of the present application, the gradient decision model can be a LightGBM model, the output training data can be historical renewable energy output data of the power grid system, the output training data comprising data features such as meteorological data (such as wind speed, light intensity, temperature, etc.), historical output sequence, etc., and the electricity price training data can be historical electricity price data of the virtual power plant in the power grid system, the electricity price training data comprising data features such as day type (such as weekday, weekend, holiday, etc.), time period, historical electricity price sequence, etc., and the lower-layer objective function is used for optimizing the parameters of the gradient decision model, and can be expressed as:

[0032] wherein, is the lower-layer objective function; The number of samples can be the total number of power output training data or electricity price training data. For the first The actual values ​​of each sample (such as the actual electricity price indicated by the first training feature and the actual renewable energy output indicated by the second training feature); For the first The predicted value of a sample, that is, the predicted value obtained after the first training feature and / or the second feature are input into the gradient decision model; the gradient decision model is a loss function, which can be the mean squared error function. The number of trees in the gradient decision model. For the first Regularization terms for the tree (such as penalty terms for the number of leaf nodes) are used to prevent overfitting; For the first Prediction function for trees.

[0033] It is understood that feature engineering can extract representative features from power output training data and electricity price training data, respectively. These features can be moving average features, seasonal features, trend features, etc., to improve the accuracy of the prediction model. For example, the moving average feature in this embodiment can be represented as:

[0034] Among them, if For the first training data of output The original value at time, then For the moment of output training data The moving average, also known as the moving average characteristic, This is the length of the sliding window. Or, if... For electricity price training data The original value at time, then Training data for electricity prices The moving average.

[0035] It should be noted that, compared to traditional models that only use raw data, features such as moving averages can effectively capture short-term trends in data (such as the continuous upward / downward trend of electricity prices), helping the model to better understand the patterns of data changes and reduce prediction errors.

[0036] Information gain training can involve inputting the first and second training features into the gradient decision model for training, and then optimizing model parameters (such as the number of trees, learning rate, leaf node trees, etc.) using methods like cross-validation in conjunction with the lower-level objective function to obtain a well-trained lower-level uncertainty prediction model. Furthermore, during the feature classification process of gradient decision model training, a histogram algorithm that maximizes information gain to select split points can be used to accelerate computation and ensure that the splitting of each tree in the gradient decision model is more effective, thereby improving the training speed and accuracy of the gradient decision model. Specifically, the information gain in the histogram algorithm can be expressed as:

[0037] in, , These are the loss function values ​​of the left and right child nodes of the tree in the gradient decision model after splitting; The loss function value of the parent node; Gain represents the information gain of feature splitting (the larger the value, the better the splitting effect).

[0038] In some embodiments, the upper-layer multi-load optimization scheduling model is obtained through the following steps: Acquire several different types of decision variables for a virtual power plant in a power grid system; each decision variable corresponds to an adjustable load of the virtual power plant. Based on all the decision variables, determine the upper-level objective function and the upper-level constraint set, which includes power balance constraints, load regulation range constraints, and energy storage constraints. The upper-level objective function is linearly normalized to obtain the normalized upper-level objective function; Based on the set of upper-level constraints and the normalized upper-level objective function, the upper-level multi-load optimization scheduling model is constructed.

[0039] In this embodiment, the decision variable can be the adjustment power or adjustment amount of the virtual power plant at different time periods and for different types of adjustable loads. For example, for adjustable electrical loads, the types can include industrial motors, residential air conditioners, etc.; while for adjustable thermal loads, the types can include heat pumps, electric heating, etc. The upper-level objective function can be constructed by maximizing the operating efficiency of the virtual power plant and comprehensively considering the coordinated adjustment of electrical load and thermal load. This upper-level objective function can be expressed as: The objective function of the upper-level multi-load optimization scheduling model is:

[0040] in, The objective function of the upper-level multi-load optimization scheduling model; Number of scheduling periods; For virtual power plants during dispatch periods Electricity price; Number of adjustable power load types; For the first Adjustable power load during dispatching periods Adjustable power, Number of adjustable heat load types; For the first Adjustable thermal load during scheduling period The adjustment amount, For the first The power conversion coefficient of a type of thermal load; This is the adjustment function.

[0041] Understandably, the upper-level constraint set is used to provide boundary constraints for the upper-level multi-load optimization scheduling model. Its power balance constraint is used to ensure the balance between the power output of the virtual power plant and the local load demand and renewable energy output, which can be expressed as:

[0042] in, For time period The interaction power between the virtual power plant and the power grid system (positive for grid connection, negative for power purchase). For time period Local load demand; For the number of renewable energy types (such as wind power, solar power); For the first The output of a renewable energy source during time period t.

[0043] Load adjustment range constraints are used to ensure that adjustable loads are adjusted within their physically feasible range, avoiding equipment damage or failure to adjust due to over-adjustment. This load adjustment range constraint can be expressed as:

[0044] in, , The first Adjustable power load during time period Minimum and maximum regulating power; , The first Adjustable power load during time period The minimum and maximum adjustment amounts.

[0045] Energy storage constraints describe the energy changes and power and capacity limitations of a virtual power plant energy storage system during charging and discharging to ensure the safe and stable operation of the system. These constraints can be expressed as:

[0046] in, For energy storage systems during time periods The energy state; For energy storage systems during time periods The energy state; , These are the charging and discharging efficiencies, respectively. The duration of the time period; , This refers to charging and discharging power. , These are the minimum and maximum values ​​of the energy storage capacity, respectively. , These are the minimum and maximum charging power, respectively; , These represent the minimum and maximum charging power, respectively.

[0047] Linear normalization can be the nonlinear part of the upper-level objective function, such as the adjustment function. Linearization is performed to satisfy the solution requirements of linear programming, thereby obtaining the normalized upper-level objective function. In this embodiment, an adjustment function is used. Taking the cost function as an example, the adjustment function can be approximated using a piecewise linearization formula. After linearization, the piecewise linearization approximation formula can be expressed as:

[0048] in, This is the normalized adjustment function; These represent the adjustment amounts for electricity and heat loads, respectively. The number of segments (e.g., dividing a nonlinear curve into 5 segments). ); , , , The first Linearization parameters of the segment.

[0049] It should be noted that after obtaining the set of upper-level constraints and the regularized upper-level objective function, a linear programming model can be constructed with the goal of maximizing the operating efficiency of the virtual power plant. This linear programming model is then designated as the upper-level multi-load optimal scheduling model. This upper-level multi-load optimal scheduling model, on the one hand, defines the adjustment boundary of adjustable loads through power balance constraints (e.g., when renewable energy output is insufficient, the reduction in adjustable loads needs to be reduced to ensure local power supply); on the other hand, by combining the dynamic adjustment of adjustable loads with renewable energy output, it is beneficial to achieve the dual objectives of maximizing the efficiency of the virtual power plant and ensuring stable power supply.

[0050] Step 130: Input the target electricity price data and the target power output data into the two-layer model to obtain the multi-load optimization scheduling strategy output by the two-layer model; In this embodiment, the target motor data and target output data at the current time can be input into the two-layer model, and the multi-load optimization scheduling strategy of the virtual power plant in the future time period can be predicted through the two-layer model.

[0051] In some embodiments, the step of inputting the target electricity price data and the target power output data into the two-layer model to obtain the multi-load optimization scheduling strategy output by the two-layer model includes: The target electricity price data and the target power output data are input into the lower-level uncertainty prediction model to perform uncertainty prediction, thereby obtaining electricity price prediction data and power output prediction data. The electricity price forecast data and the power output forecast data are input into the upper-level multi-load optimization scheduling model for scheduling analysis to obtain the multi-load optimization scheduling strategy.

[0052] In this embodiment, target electricity price data and target power output data can be input into the lower-level uncertain prediction model to obtain the electricity price prediction data and power output prediction data corresponding to the target electricity price data output by the lower-level uncertain prediction model. Both the target electricity price data and the power output prediction data contain probability distributions (such as mean, variance, confidence interval, etc.). For example, the electricity price prediction data can be "there is a 90% probability that the future electricity price will fall in the range of [0.5, 0.8] yuan / kWh", etc., to provide richer inputs for the upper-level multi-load optimization scheduling model. The same applies to the power output prediction data, which will not be elaborated further in this application.

[0053] Understandably, dispatch analysis can involve inputting electricity price forecast data and power output forecast data into an upper-level multi-load optimal dispatch model, and then solving the model using a linear programming solver to obtain the optimal regulation strategy for virtual power plants in various time periods and for various types of adjustable loads in the future. This strategy is referred to as the multi-load optimal dispatch strategy.

[0054] Step 140: Perform multi-load scheduling control on the virtual power plant according to the multi-load optimization scheduling strategy.

[0055] In this embodiment of the application, the virtual power plant can perform scheduling control on its adjustable loads based on the obtained multi-load optimization scheduling strategy to ensure the stable operation of the virtual power plant in the power grid system.

[0056] In some embodiments, the method further includes: Obtain the electricity price forecast data and power output forecast data output by the lower-level uncertainty prediction model; The electricity price forecast data and power output forecast data are processed to generate several uncertainty scenarios. In this embodiment of the application, uncertain scenarios can also be generated based on the electricity price prediction data with probability distribution output by the lower uncertainty prediction model and the processed prediction data. Specifically, several uncertain scenarios can be generated based on the Latin hypercube sampling (LHS) method to ensure that the scenarios are evenly covered in each region of the probability space.

[0057] Understandably, after identifying several uncertain scenarios, the generated uncertain scenarios can be clustered using the K-means clustering algorithm. By clustering similar uncertain scenarios, the number of scenarios can be reduced and representative uncertain scenarios can be extracted, which helps to reduce the amount of computation required for subsequent robustness analysis.

[0058] Based on all the aforementioned uncertainty scenarios, a robustness analysis is performed on the multi-load optimization scheduling strategy to obtain robustness analysis results.

[0059] Furthermore, the robustness analysis of the multi-load optimization scheduling strategy based on all the aforementioned uncertainty scenarios, to obtain robustness analysis results, includes: Obtain the minimum acceptable benefit, and determine the robustness index based on the minimum acceptable benefit; Based on the uncertain scenario, scenario analysis is performed on the multi-load optimization scheduling strategy to obtain scenario benefits; Based on the robustness index, a benefit analysis is performed on the scenario benefits to obtain the robustness analysis results.

[0060] In the embodiments of this application, robustness analysis can be used to verify the multi-load optimization scheduling strategy under various uncertainty scenarios, so as to analyze the fluctuation of the virtual power plant operation benefits, thereby evaluating the robustness of the multi-load optimization scheduling strategy, and continuously adjusting and optimizing the two-layer model based on the determined robustness. There are already various ways to adjust and optimize the specific two-layer model, which will not be described in detail here.

[0061] Specifically, for any uncertainty scenario, its robustness index can be constructed based on conditional value at risk (CVaR), and this robustness index can be expressed as:

[0062] in, A functional representation of the robustness index; For confidence level (e.g., 95%). for Quantile benefits (e.g., the 95th quantile indicates that the benefit is higher in scenarios with 95% uncertainty). To achieve the minimum acceptable benefit; For benefits lower than The average benefit of the scenario; The scenario benefits of a virtual power plant under a certain uncertain situation.

[0063] Understandably, scenario analysis can involve substituting multi-load optimization scheduling strategies into corresponding uncertain scenarios for verification, in order to obtain the scenario benefits of multi-load optimization scheduling strategies under uncertain scenarios; while benefit analysis can be conducted by evaluating and analyzing scenario benefits through robustness indicators, thereby obtaining robustness analysis results.

[0064] The following describes in detail, with reference to the accompanying drawings, a load optimization scheduling system for a virtual power plant according to an embodiment of this application.

[0065] Reference Figure 2 The load optimization and dispatching system for a virtual power plant proposed in this application includes: The first processing unit 101 is used to acquire the target power output data of the power grid system and the target electricity price data of the virtual power plant. The second processing unit 102 is used to construct a two-layer model, which includes a lower-layer uncertainty prediction model and an upper-layer multi-load optimization scheduling model. The lower-layer uncertainty prediction model is used to predict the probability distribution of electricity price and renewable energy output of the virtual power plant in future periods, and the upper-layer multi-load optimization scheduling model is used to generate the multi-load optimization scheduling strategy of the virtual power plant. The third processing unit 103 is used to input the target electricity price data and the target power output data into the two-layer model to obtain the multi-load optimization scheduling strategy output by the two-layer model; The fourth processing unit 104 is used to perform multi-load scheduling control on the virtual power plant according to the multi-load optimization scheduling strategy.

[0066] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0067] Reference Figure 3 This application also provides an electronic device, including: At least one processor 201; At least one memory 202 is used to store at least one program; When the at least one program is executed by the at least one processor 201, the at least one processor 201 implements the above-described method embodiments.

[0068] Similarly, it can be understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0069] This application also provides a computer-readable storage medium storing a program executable by a processor 201, which, when executed by the processor 201, is used to implement the above-described method embodiments.

[0070] Similarly, the content of the above method embodiments is applicable to the present computer-readable storage medium embodiments. The specific functions implemented by the present computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0071] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0072] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods.

[0073] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this application are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0074] Furthermore, although this application is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding this application. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional technology for an engineer. Therefore, those skilled in the art can implement the application set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of this application, which is determined by the full scope of the appended claims and their equivalents.

[0075] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0077] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0078] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0079] In the foregoing description of this specification, the references to terms such as "one embodiment," "another embodiment," or "some embodiments," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0080] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

[0081] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A load optimization scheduling method for a virtual power plant, characterized in that, include: Obtain target power output data of the power grid system and target electricity price data of virtual power plants; A two-layer model is constructed, comprising a lower-layer uncertainty prediction model and an upper-layer multi-load optimization scheduling model. The lower-layer uncertainty prediction model is used to predict the probability distribution of electricity price and renewable energy output of the virtual power plant in future periods, and the upper-layer multi-load optimization scheduling model is used to generate the multi-load optimization scheduling strategy of the virtual power plant. The target electricity price data and the target power output data are input into the two-layer model to obtain the multi-load optimization scheduling strategy output by the two-layer model; Based on the multi-load optimization scheduling strategy, the virtual power plant is subjected to multi-load scheduling control.

2. The method according to claim 1, characterized in that, The lower-level uncertainty prediction model is obtained through the following training steps: Acquire power output training data and electricity price training data of the virtual power plant, as well as the gradient decision model and the lower-level objective function of the gradient decision model; The electricity price training data and the power output training data are subjected to feature engineering processing to obtain a first training feature and a second training feature. The first training feature is used to characterize the feature information of the electricity price training data within a certain time period, and the second training feature is used to characterize the feature information of the power output training data within a certain time period. Based on the lower-level objective function, the first training feature, and the second training feature, the gradient decision model is trained with information gain to obtain the lower-level uncertainty prediction model.

3. The method according to claim 1, characterized in that, The upper-level multi-load optimization scheduling model is obtained through the following steps: Acquire several different types of decision variables for a virtual power plant in a power grid system; each decision variable corresponds to an adjustable load of the virtual power plant. Based on all the decision variables, determine the upper-level objective function and the upper-level constraint set, which includes power balance constraints, load regulation range constraints, and energy storage constraints. The upper-level objective function is linearly normalized to obtain the normalized upper-level objective function; Based on the set of upper-level constraints and the normalized upper-level objective function, the upper-level multi-load optimization scheduling model is constructed.

4. The method according to claim 1, characterized in that, The step of inputting the target electricity price data and the target power output data into the two-layer model to obtain the multi-load optimization scheduling strategy output by the two-layer model includes: The target electricity price data and the target power output data are input into the lower-level uncertainty prediction model to perform uncertainty prediction, thereby obtaining electricity price prediction data and power output prediction data. The electricity price forecast data and the power output forecast data are input into the upper-level multi-load optimization scheduling model for scheduling analysis to obtain the multi-load optimization scheduling strategy.

5. The method according to any one of claims 1-4, characterized in that, The method further includes: Obtain the electricity price forecast data and power output forecast data output by the lower-level uncertainty prediction model; The electricity price forecast data and power output forecast data are processed to generate several uncertainty scenarios. Based on all the aforementioned uncertainty scenarios, a robustness analysis is performed on the multi-load optimization scheduling strategy to obtain robustness analysis results.

6. The method according to claim 5, characterized in that, The robustness analysis of the multi-load optimization scheduling strategy is performed based on all the aforementioned uncertainty scenarios to obtain robustness analysis results, including: Obtain the minimum acceptable benefit, and determine the robustness index based on the minimum acceptable benefit; Based on the uncertain scenario, scenario analysis is performed on the multi-load optimization scheduling strategy to obtain scenario benefits; Based on the robustness index, a benefit analysis is performed on the scenario benefits to obtain the robustness analysis results.

7. The method according to claim 1, characterized in that, The objective function of the upper-level multi-load optimization scheduling model is: in, The objective function of the upper-level multi-load optimization scheduling model; Number of scheduling periods; For virtual power plants during dispatch periods Electricity price; Number of adjustable power load types; For the first Adjustable power load during dispatching periods Adjustable power, Number of adjustable heat load types; For the first Adjustable thermal load during scheduling period The adjustment amount, For the first The power conversion coefficient of a type of thermal load; This is the adjustment function.

8. A load optimization dispatching system for a virtual power plant, characterized in that, include: The first processing unit is used to acquire the target power output data of the power grid system and the target electricity price data of the virtual power plant. The second processing unit is used to construct a two-layer model, which includes a lower-layer uncertainty prediction model and an upper-layer multi-load optimization scheduling model. The lower-layer uncertainty prediction model is used to predict the probability distribution of electricity price and renewable energy output of the virtual power plant in future periods, and the upper-layer multi-load optimization scheduling model is used to generate the multi-load optimization scheduling strategy of the virtual power plant. The third processing unit is used to input the target electricity price data and the target power output data into the two-layer model to obtain the multi-load optimization scheduling strategy output by the two-layer model. The fourth processing unit is used to perform multi-load scheduling control on the virtual power plant according to the multi-load optimization scheduling strategy.

9. An electronic device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor performs the method as described in any one of claims 1-7.

10. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to implement the method as described in any one of claims 1-7.

Images