Dispatching method of new energy power station with storage under dual modes of self-consumption and power grid dispatching

Abstract

The present application discloses a method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and power grid dispatching, which relates to the field of dispatching new energy distribution and storage power stations. The method comprises: inputting a set of new energy output prediction data corresponding to the new energy distribution and storage power station in the current time period into a trained neural network model to obtain a set of power generation plans and an energy storage operation mode of the new energy distribution and storage power station in the current time period; if the energy storage operation mode in the current time period is a power grid dispatching energy storage operation mode, solving the dispatching model of the new energy distribution and storage power station to obtain a set of dispatching plans for the new energy distribution and storage energy storage in the current time period, and obtaining a dispatching scheme for the current time period based on the dispatching plan set and the power generation plan set in the current time period; if the energy storage operation mode in the current time period is a self-consumption energy storage operation mode, determining the set of power generation plans for the new energy distribution and storage power station in the current time period as the dispatching scheme for the current time period. The present application can solve the problems of tight peak-shaving resources and low energy storage utilization.

Description

Technical Field

[0001] The present application relates to the field of scheduling of new energy distribution and storage power stations, and in particular to a scheduling method of new energy distribution and storage power stations under dual modes of self-consumption and power grid scheduling. Background Art

[0002] As the proportion of renewable energy power generation in the power system continues to increase, its volatility and uncertainty have brought huge challenges to the stable operation and scheduling of the power system. The demand for peak-shaving resources at the scheduling level has become increasingly urgent. The traditional scheduling method of renewable energy power stations has resulted in tight peak-shaving resources and low energy storage utilization. Summary of the Invention

[0003] The purpose of this application is to provide a scheduling method for a new energy distribution storage power station under the dual modes of self-consumption and grid scheduling, which can solve the problems of tight peak-shaving resources and low energy storage utilization.

[0004] To achieve the above objectives, this application provides the following solutions.

[0005] The present application provides a method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatching, comprising: inputting a set of new energy output forecast data corresponding to the new energy distribution and storage power station in the current time period into a trained neural network model, and obtaining a set of power generation plans of the new energy distribution and storage power station in the current time period and an energy storage operation mode of the new energy distribution and storage power station.

[0006] If the energy storage operation mode of the new energy power station with storage in the current time period is the grid dispatching energy storage operation mode, the scheduling model of the new energy power station with storage is solved to obtain the scheduling plan set of the new energy power station with storage in the current time period, and based on the scheduling plan set of the new energy power station with storage in the current time period and the power generation plan set of the new energy power station with storage in the current time period, the scheduling plan of the new energy power station with storage in the current time period is obtained.

[0007] If the energy storage operation mode of the new energy distribution and storage power station in the current time period is the self-consumption energy storage operation mode, then the power generation plan set of the new energy distribution and storage power station in the current time period is determined, which is the scheduling plan of the new energy distribution and storage power station in the current time period.

[0008] In one embodiment, the training process of the neural network model specifically includes: obtaining historical new energy output data of the new energy distribution and storage power station; the historical new energy output data includes a new energy output prediction data set and a new energy output actual data set corresponding to multiple historical time periods.

[0009] Construct a power generation planning decision model for new energy storage power stations.

[0010] Based on the historical new energy output data of the new energy distribution and storage power station, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain the power generation plan set of the new energy distribution and storage power station in each historical time period and the energy storage operation mode of the new energy distribution and storage power station.

[0011] A training data set is constructed based on the power generation plan set of new energy distribution and storage power stations in each historical time period, the energy storage operation mode of new energy distribution and storage power stations, and the new energy output forecast data set corresponding to each historical time period.

[0012] The neural network model is trained using the training data set to obtain a trained neural network model.

[0013] In one embodiment, based on the historical new energy output data of the new energy distribution and storage power station, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain a set of power generation plans of the new energy distribution and storage power station and an energy storage operation mode of the new energy distribution and storage power station in each historical time period, specifically including: according to the new energy output prediction data set and the new energy output actual data set corresponding to each historical time period, obtaining a set of new energy output deviation values ​​corresponding to each historical time period.

[0014] Cluster the sets of new energy output deviation values ​​corresponding to all historical time periods to obtain clustering results.

[0015] Calculate the probability of each cluster center based on the clustering results.

[0016] Based on the new energy output prediction data set corresponding to each historical time period and each cluster center, a new energy output value set corresponding to each cluster center is obtained.

[0017] Based on the probability of each cluster center and the set of new energy output values ​​of each cluster center, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain the power generation plan set of the new energy distribution and storage power station and the energy storage operation mode of the new energy distribution and storage power station in each historical time period.

[0018] In one embodiment, the sets of new energy output deviation values ​​corresponding to all historical time periods are clustered to obtain clustering results. Specifically, based on the Hopkins statistic, a clustering algorithm is used to cluster the sets of new energy output deviation values ​​corresponding to each historical time period to obtain clustering results.

[0019] In one embodiment, the neural network model includes: an improved convolutional neural network and a bidirectional gated recurrent unit connected in sequence; the improved convolutional neural network includes a deep convolution module and a convolution branch module; the input ends of the deep convolution module and the convolution branch module are connected, and the output ends of the deep convolution module and the convolution branch module are connected through a residual mechanism; the deep convolution module includes two convolutional neural networks connected in series.

[0020] In one embodiment, the new energy output prediction data set corresponding to the historical time period includes the new energy output prediction data collected at each sampling moment in the historical time period; the new energy output actual data set corresponding to the historical time period includes the new energy output actual data collected at each sampling moment in the historical time period; based on the new energy output prediction data set and the new energy output actual data set corresponding to each historical time period, a new energy output deviation value set corresponding to each historical time period is obtained, specifically including: for any historical time period, calculating the deviation value between the new energy output prediction data collected at the t-th sampling moment in the new energy output prediction data set corresponding to the historical time period and the new energy output actual data collected at the t-th sampling moment in the new energy output actual data set corresponding to the historical time period, to obtain the new energy output deviation value at the t-th sampling moment in the historical time period.

[0021] A new energy output deviation value set corresponding to the historical time period is obtained according to the new energy output deviation value at each sampling moment in the historical time period.

[0022] In one embodiment, the probability of each cluster center is calculated based on the clustering results, specifically: according to the formula Calculate the probability of the xth cluster center, where is the probability of the xth cluster center; is the number of new energy output deviation value sets assigned to the x-th cluster center; It is the total number of new energy output deviation value sets corresponding to each historical time period.

[0023] In one embodiment, the new energy distribution and storage power station scheduling model includes: a new energy distribution and storage power station scheduling objective function, a thermal power output model constraint, an independent energy storage model constraint, a new energy distribution and storage power station model constraint in a grid dispatching and storage power operation mode, and a power balance constraint.

[0024] In one embodiment, the power generation plan decision model of the new energy storage power station is solved, specifically by using a tornado optimization algorithm to solve the power generation plan decision model of the new energy storage power station.

[0025] In one embodiment, the scheduling model of the new energy distribution and storage power station is solved, specifically by using an exponential trigonometric optimization algorithm to solve the scheduling model of the new energy distribution and storage power station.

[0026] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a method for dispatching a new energy distribution storage power station under the dual modes of self-consumption and grid dispatching. The utilization hours of the internal energy storage in the new energy distribution storage power station are relatively low, making it difficult to utilize this part of the energy storage resources, resulting in a situation where peak-shaving resources are tight and energy storage utilization is low. In order to give full play to the supporting role of energy storage in the new energy distribution storage power station for the power grid, the method for dispatching a new energy distribution storage power station under the dual modes of self-consumption and grid dispatching proposed in this application adopts different dispatching schemes for different energy storage operation modes, opens up channels for new energy distribution storage to participate in unified dispatching, and while acting as a peak-shaving resource at the dispatching level, can improve its own energy storage utilization rate, solve the problem of tight peak-shaving resources and low energy storage utilization rate, and obtain higher benefits for the new energy distribution storage power station. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 A schematic diagram of the general flow of a method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatching provided in one embodiment of the present application.

[0029] Figure 2 This is a schematic diagram of a specific process of a method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatching provided in one embodiment of the present application.

[0030] Figure 3 This is a structural diagram of the improved CNN-BiGRU neural network model provided in one embodiment of the present application. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0032] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0033] In an exemplary embodiment, Figure 1As shown, a method for dispatching a new energy distribution storage power station under the dual modes of self-consumption and grid dispatch is provided, and the method includes the following steps.

[0034] Step 201: Input the set of new energy output forecast data corresponding to the new energy power station with storage in the current time period into a trained neural network model to obtain a set of power generation plans and energy storage operation modes for the new energy power station with storage in the current time period. The set of power generation plans for the new energy power station with storage includes the power generation plans for the new energy power station with storage corresponding to each piece of new energy output forecast data in the set of new energy output forecast data.

[0035] Step 202: If the energy storage operation mode of the new energy power station with energy storage in the current time period is the grid dispatching energy storage operation mode, the scheduling model of the new energy power station with energy storage is solved to obtain the scheduling plan set of the new energy power station with energy storage in the current time period, and based on the scheduling plan set of the new energy power station with energy storage in the current time period and the power generation plan set of the new energy power station with energy storage in the current time period, the scheduling plan of the new energy power station with energy storage in the current time period is obtained.

[0036] Step 203: If the energy storage operation mode of the new energy distribution and storage power station in the current time period is the self-consumption energy storage operation mode, determine the power generation plan set of the new energy distribution and storage power station in the current time period, which is the scheduling plan of the new energy distribution and storage power station in the current time period.

[0037] In another exemplary embodiment of the present application, the training process of the neural network model specifically includes: obtaining historical new energy output data of the new energy distribution and storage power station; the historical new energy output data includes a new energy output prediction data set and a new energy output actual data set corresponding to multiple historical time periods.

[0038] Construct a power generation planning decision model for new energy storage power stations.

[0039] Based on the historical new energy output data of the new energy distribution and storage power station, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain the power generation plan set of the new energy distribution and storage power station in each historical time period and the energy storage operation mode of the new energy distribution and storage power station.

[0040] A training data set is constructed based on the power generation plan set of new energy distribution and storage power stations in each historical time period, the energy storage operation mode of new energy distribution and storage power stations, and the new energy output forecast data set corresponding to each historical time period.

[0041] The neural network model is trained using the training data set to obtain a trained neural network model.

[0042] In another exemplary embodiment of the present application, based on the historical new energy output data of the new energy distribution and storage power station, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain a set of power generation plans of the new energy distribution and storage power station and an energy storage operation mode of the new energy distribution and storage power station in each historical time period, specifically including: according to the new energy output prediction data set and the new energy output actual data set corresponding to each historical time period, a set of new energy output deviation values ​​corresponding to each historical time period is obtained.

[0043] Cluster the sets of new energy output deviation values ​​corresponding to all historical time periods to obtain clustering results.

[0044] Calculate the probability of each cluster center based on the clustering results.

[0045] Based on the new energy output prediction data set corresponding to each historical time period and each cluster center, a new energy output value set corresponding to each cluster center is obtained.

[0046] Based on the probability of each cluster center and the set of new energy output values ​​of each cluster center, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain the power generation plan set of the new energy distribution and storage power station and the energy storage operation mode of the new energy distribution and storage power station in each historical time period.

[0047] In another exemplary embodiment of the present application, the new energy output prediction data set corresponding to the historical time period includes the new energy output prediction data collected at each sampling moment within the historical time period; the new energy output actual data set corresponding to the historical time period includes the new energy output actual data collected at each sampling moment within the historical time period; based on the new energy output prediction data set and the new energy output actual data set corresponding to each historical time period, a new energy output deviation value set corresponding to each historical time period is obtained, specifically including: for any historical time period, calculating the deviation value between the new energy output prediction data collected at the tth sampling moment in the new energy output prediction data set corresponding to the historical time period and the new energy output actual data collected at the tth sampling moment in the new energy output actual data set corresponding to the historical time period, and obtaining the new energy output deviation value at the tth sampling moment in the historical time period.

[0048] A new energy output deviation value set corresponding to the historical time period is obtained according to the new energy output deviation value at each sampling moment in the historical time period.

[0049] In another exemplary embodiment of the present application, the sets of new energy output deviation values ​​corresponding to all historical time periods are clustered to obtain clustering results. Specifically, based on the Hopkins statistic, a clustering algorithm is used to cluster the sets of new energy output deviation values ​​corresponding to each historical time period to obtain clustering results.

[0050] In another exemplary embodiment of the present application, a power generation plan decision model of a new energy storage power station is solved, specifically: a tornado optimizer with Coriolis force (TOC) is used to solve the power generation plan decision model of a new energy storage power station.

[0051] In another exemplary embodiment of the present application, the neural network model is an improved CNN-BiGRU neural network model, such as Figure 3 As shown, the proposed model specifically comprises an improved convolutional neural network and a bidirectional gated recurrent unit (BiGRU) connected in sequence. The improved convolutional neural network includes a deep convolution module and a convolution branch module. The inputs of the deep convolution module and the convolution branch module are connected, and the outputs of the deep convolution module and the convolution branch module are connected via a residual mechanism. The deep convolution module comprises two convolutional neural networks connected in series. The deep convolution module extracts features between input data. Two 3×3 convolutional neural networks are operated in series to extract deep features, ensuring performance while reducing computational complexity. A new 3×3 convolutional neural network is added as a convolution branch module to capture small object information and shallow features, addressing the feature loss problem caused by fixed convolution kernels. A residual mechanism is further introduced to combine shallow and deep features, alleviating the vanishing gradient problem in deep networks. The BiGRU extracts temporal information or irregular trends in time series components. Dynamically controlling the flow of information through a gating mechanism can capture long-term dependencies in sequential data.

[0052] In another exemplary embodiment of the present application, the new energy distribution and storage power station scheduling model includes: a new energy distribution and storage power station scheduling objective function, a thermal power output model constraint, an independent energy storage model constraint, a new energy distribution and storage model constraint in a grid dispatching and storage operation mode, and a power balance constraint.

[0053] In another exemplary embodiment of the present application, a scheduling model of a new energy power station with storage is solved, specifically by using an Exponential-Trigonometric Optimization (ETO) algorithm to solve the scheduling model of a new energy power station with storage.

[0054] In another exemplary embodiment of the present application, the historical time period may be a day, and the new energy output forecast data set corresponding to the historical time period includes: 24 new energy output forecast data sets obtained in a day at a granularity of one hour. The new energy output actual data set corresponding to the historical time period includes: 24 new energy output actual data sets obtained in a day at a granularity of one hour.

[0055] In another exemplary embodiment of the present application, for any historical time period, the deviation value between the new energy output prediction data collected at the t-th sampling moment in the new energy output prediction data set corresponding to the historical time period and the new energy output actual data collected at the t-th sampling moment in the new energy output actual data set corresponding to the historical time period is calculated to obtain the new energy output deviation value at the t-th sampling moment in the historical time period, specifically, calculated according to formula (1).

[0056] (1).

[0057] in: For the The new energy output deviation value at the t-th sampling moment in the historical time period, The installed capacity of the power station for new energy storage, For the The new energy output forecast data collected at the t-th sampling moment in the new energy output forecast data set corresponding to the historical time period, For the The actual data of new energy output collected at the t-th sampling moment in the actual data set of new energy output corresponding to the historical time period.

[0058] like Figure 2 As shown, in another exemplary embodiment of the present application, based on the Hopkins statistic, a clustering algorithm is used to cluster the set of new energy output deviation values ​​corresponding to each historical time period to obtain a clustering result. Specifically, the total number of new energy output deviation data sets corresponding to each historical time period is set to A, the number of clusters is set to K, and the k-means clustering algorithm is used to generate K cluster centers. The obtained K cluster centers constitute the new energy deviation scenario set. Since the selection of the initial cluster center in the k-means algorithm is random, the clustering result and the clustering effect are unstable. In order to screen out the cluster center with excellent clustering effect, the Hopkins statistic is then used to judge the clustering effect of the clustered data object, and finally the clustering result with the best clustering effect is obtained.

[0059] The k-means clustering algorithm process is as follows: Step 1, randomly select K new energy output deviation data sets from X data as the initial cluster centers; Step 2, calculate the distance between the new energy output deviation data set corresponding to each historical time period and the cluster center, and assign the new energy output deviation data set corresponding to each historical time period to the nearest cluster center; Step 3, recalculate the average value of all new energy output deviation data sets in each cluster as the new cluster center; Step 4, repeat Steps 2 and 3 until convergence, that is, until each cluster no longer changes.

[0060] The Hopkins statistic processing process is: sample space It includes all cluster centers obtained by k-means clustering, and each cluster center includes the deviation value at each sampling time. Manually select the maximum or minimum value at each sampling moment, randomly combine them, and get n sampling points , the sampling time corresponding to the deviation value included in each sampling point is different, and the obtained sampling points are formed by sampling from the hyperrectangular space composed of all points in the sample space as boundaries. For the jth sampling point, find its position in the sample space The nearest neighbor point in the , and calculate the distance between the sampling point and the nearest neighbor point , as shown in formula (2); from space Sampling was carried out according to the above sampling method. n Sampling points , for the jth sampling point find its The nearest neighbor point in the , and calculate the distance between the sampling point and the nearest neighbor point , as shown in formula (3).

[0061] (2).

[0062] (3).

[0063] in, is the set of running time periods, including all sampling moments in the historical time period, is the nearest neighbor point selected, and is the new energy output deviation value at the t-th sampling moment in the j-th sampling point.

[0064] The calculation formulas for the Hopkins statistic are formula (4) and formula (5).

[0065] (4).

[0066] (5).

[0067] In formula (4) and formula (5): To calculate the obtained Hopkins statistic, is the threshold of the Hopkins statistic.

[0068] For the Hopkins statistic, if the data points are evenly distributed in space, the statistic is roughly equal to 0.5. If clustering exists in the data set, H will be close to 1. For data objects that have been clustered, the Hopkins statistic within the data object is required to be close to 0.5, so a certain threshold is set to require clustering effect. In the algorithm process, the k-means clustering algorithm is first used to obtain the cluster center, and then the Hopkins statistic is used to determine the clustering effect of the data objects divided into the cluster center. If the condition (Formula (5)) is not met, k-means clustering is repeated until the condition is met.

[0069] In another exemplary embodiment of the present application, the probability of each cluster center is calculated based on the clustering result, specifically: the probability of the x-th cluster center is calculated according to formula (6).

[0070] (6).

[0071] in, is the probability of the xth cluster center; is the number of new energy output deviation value sets assigned to the x-th cluster center; It is the total number of new energy output deviation value sets corresponding to each historical time period.

[0072] In another exemplary embodiment of the present application, based on the new energy output prediction data set corresponding to each historical time period and each cluster center, the new energy output value set corresponding to each cluster center is obtained, specifically: the new energy output value corresponding to the t-th sampling time under the x-th cluster center is calculated according to formula (7): .

[0073] (7).

[0074] Where: A new energy output forecast data set is selected from the new energy output forecast data sets corresponding to all historical time periods as the benchmark new energy output forecast data set, is the new energy output forecast data collected at the tth sampling time in the benchmark new energy output forecast data set, is the new energy output deviation value at the t-th sampling moment in the historical time period corresponding to the x-th cluster center.

[0075] In practical applications, the power generation plan decision model of the new energy distribution storage power station realizes the formulation of the power generation plan of the new energy distribution storage power station and the decision on the energy storage operation mode of the new energy distribution storage power station. The power generation plan of the new energy distribution storage power station is specifically the discharge power and charging power of the energy storage in the new energy distribution storage power station. The power generation plan decision model of the new energy distribution storage power station includes: the objective function of the power generation plan decision model of the new energy distribution storage power station.

[0076] (8).

[0077] (9).

[0078] (10).

[0079] in, , Respectively represent the cluster center set and the running time set, To reduce the operating costs of power stations equipped with storage for new energy, 、 、 and They are the risk cost of power curtailment, the risk cost of power shortage, the operating cost of energy storage power, and the expected additional benefits of unified dispatch; 、 、 and They are the unit power abandonment penalty cost, the unit power shortage penalty cost, the unit power operation cost of energy storage, and the unit capacity additional benefit. and is the surplus and shortage of electric power at the t-th sampling moment in the historical time period corresponding to the x-th cluster center; and is the discharge power and charging power of the energy storage in the new energy storage power station at the t-th sampling moment in the historical time period corresponding to the x-th cluster center; It is the rated capacity of energy storage in the new energy storage power station. is a binary variable that characterizes participation in the unified scheduling mode. Contribute to the planning of new energy storage power stations.

[0080] The energy storage of new energy distribution and storage power stations can choose two operating modes. One is the self-use of the new energy distribution and storage power station, which is the above-mentioned self-consumption energy storage operating mode; the other is to participate in unified dispatching, which is the above-mentioned grid dispatching energy storage operating mode. The two energy storage operating modes can improve the energy storage utilization rate of the new energy distribution and storage power station and reduce the idle rate of energy storage in the new energy distribution and storage power station.

[0081] The energy storage operation mode selection constraints are as follows.

[0082] (11).

[0083] In formula (11): It is a binary variable that represents the self-storage operation mode of a new energy storage power station.

[0084] When the energy storage operation mode of the new energy storage power station is selected to participate in unified dispatch, its constraint is only formula (11). When the energy storage operation mode of the new energy storage power station is selected for self-use, its constraint also includes formula (12).

[0085] (12).

[0086] in, and is the state of charge of the energy storage in the new energy distribution and storage power station at the t-th sampling time and the t-1-th sampling time in the historical time period corresponding to the x-th cluster center; and is the state of charge of the energy storage in the new energy distribution and storage power station during the initial period and the final period of the historical time period corresponding to the xth cluster center; and The minimum state of charge and maximum state of charge of energy storage in new energy storage power stations; and The charging and discharging efficiency of energy storage in new energy storage power stations; It is the rated energy stored in the new energy storage power station.

[0087] In practical applications, the Tornado Optimization Algorithm is used to solve the power generation planning decision model for renewable energy power stations with energy storage. Specifically, the Tornado Optimization Algorithm simulates the evolution of storms into tornadoes, storms into thunderstorms, and thunderstorms into tornadoes. By analyzing the evolution of storms and thunderstorms into tornadoes, it is possible to search for the global optimal solution near the tornado. The three evolutionary strategies of the algorithm are shown in Equation (13).

[0088] (13).

[0089] The location information of storms, thunderstorms, and tornadoes represents the solution to the optimization problem at this time. Specifically, it is the variables of the optimization problem, including the energy storage operation mode of the energy storage equipped with new energy, the discharge power and charging power of the energy storage in the energy storage equipped with new energy power stations, and is the location information of the storm at the nth and n+1th iterations, and is the location information of the thunderstorm at the nth iteration and the n+1th iteration, and is the location information of the tornado at the nth iteration and the n+1th iteration, and A random tornado location and a random thunderstorm location; is a random function; The speed of a storm to form a thunderstorm or tornado; is the random adaptive coefficient.

[0090] The specific solution process is as follows: First, variables are initialized based on their boundaries in the power generation planning decision model for renewable energy power stations with energy storage, generating a population. Individuals in the population are divided into three types: storm, thunderstorm, and tornado. Each individual includes the energy storage operation mode of the renewable energy power storage, as well as the discharge and charging power of the renewable energy power storage station. Second, individuals in the population are continuously updated using the three evolutionary process strategies of the Tornado Optimization Algorithm, with the three position update strategies being performed sequentially. Third, the current best individual and target value are updated and recorded. Finally, iterations are repeated, increasing the number of iterations until the maximum number of iterations is reached and the loop is terminated.

[0091] In another exemplary embodiment of the present application, a training data set is constructed based on the set of power generation plans and energy storage operation modes of new energy power stations in each historical time period and the set of new energy output prediction data corresponding to each historical time period. Specifically, the training data set includes multiple groups of data, each group of data corresponds to a historical time period, including each new energy output prediction data in the new energy output prediction data set corresponding to this historical time period, the power generation plan of the new energy power station corresponding to the new energy output prediction data, and the energy storage operation mode of the new energy power station in this historical time period.

[0092] In another exemplary embodiment of the present application, a set of new energy output prediction data corresponding to the new energy distribution and storage power station in the current time period is input into a trained neural network model to obtain a set of power generation plans of the new energy distribution and storage power station in the current time period and an energy storage operation mode of the new energy distribution and storage power station. Specifically, after the training is completed, the new energy output prediction data of the current time period is input into the trained neural network model, and the set of power generation plans of the new energy distribution and storage power station in the current time period and the energy storage operation mode of the new energy distribution and storage power station are output.

[0093] In another exemplary embodiment of the present application, the scheduling objective function of the new energy storage power station includes formula (14) and formula (15).

[0094] (14).

[0095] (15).

[0096] in, is the overall scheduling cost, 、 and Unify the dispatching costs for thermal power operation costs, independent energy storage operation costs, and energy storage built with new energy sources; and The unit power operating cost of thermal power and the unit power operating cost of independent energy storage; is the electric power of the i-th thermal power unit at the t-th sampling moment, and are the charging power and discharging power of the i-th independent energy storage at the t-th sampling moment. A binary variable representing the participation of the i-th renewable energy source in the unified dispatch mode when equipped with energy storage; The rated capacity of energy storage for the i-th new energy source. For thermal power units, It is an independent energy storage collection; Build energy storage systems to support new energy.

[0097] The thermal power output model constraint is formula (16).

[0098] (16).

[0099] in, and is the minimum and maximum technical output of thermal power unit No. i; and is the ramp-up power limit and ramp-down power limit of thermal power unit No. i.

[0100] The independent energy storage model is constrained by formula (17).

[0101] (17).

[0102] in, and is the rated capacity and rated energy of the i-th independent energy storage, and is the state of charge of the ith independent energy storage at the tth sampling moment and the t-1th sampling moment; and is the state of charge of the initial period and the final period of the dispatching of the i-th independent energy storage; and The minimum state of charge and maximum state of charge for independent energy storage; and It is the charging efficiency and discharging efficiency of independent energy storage.

[0103] The constraints of the new energy storage model in the grid dispatching energy storage operation mode are as follows: Formula (18).

[0104] (18).

[0105] in, and The discharge power and charging power of the energy storage for the i-th renewable energy source at the t-th sampling moment; and The rated capacity and rated energy of energy storage for the i-th new energy source, and The state of charge of the energy storage device No. i at the t-th sampling moment and the t-1-th sampling moment; and The state of charge of the i-th renewable energy storage during the initial period and the final period of the dispatch period; and The minimum and maximum states of charge for energy storage deployed with renewable energy. (Energy storage in renewable energy power stations can be referred to as renewable energy deployed with energy storage.) Model constraints for thermal power output, independent energy storage, and renewable energy deployed with energy storage in a unified dispatch mode are collectively referred to as model constraints.

[0106] The power balance constraint is formula (19).

[0107] (19).

[0108] in, Represents the collection of new energy storage power stations. represents the planned output of the i-th renewable energy storage power station at the t-th sampling moment, is the electric power demand at the tth sampling moment.

[0109] In another exemplary embodiment of the present application, the scheduling plan of the new energy storage equipped with energy under the grid dispatching energy storage operation mode includes: the electric power of the thermal power unit, the charging power and discharging power of the independent energy storage, the discharge power and charging power of the new energy storage equipped with energy under the grid dispatching energy storage operation mode, and the scheduling model of the new energy storage equipped with energy is solved. Specifically, the exponential triangle optimization algorithm is used to solve the scheduling model of the new energy storage equipped with energy. Specifically, the exponential triangle optimization algorithm is divided into an exploration stage and a development stage. The algorithm expands the exploration stage into the first exploration stage and the second exploration stage by adding additional adaptive variables, such as formula (22), and expands the development stage into the first development stage and the second development stage, such as formula (23), to prevent convergence to the local optimal value, and proposes a conversion mechanism of exploration and development to improve the efficiency of the algorithm. This algorithm design not only improves the performance of the algorithm in exploring and developing the search space, but also improves the flexibility and adaptability of the algorithm to various problems. Formula (20) realizes the conversion mechanism of exploration and development. When When , the algorithm is in exploration mode. On the contrary, when When , the algorithm switches to development mode. Formula (21) determines whether the first-stage update formula or the second-stage update formula is used during exploration or development.

[0110] (20).

[0111] (twenty one).

[0112] (twenty two).

[0113] (twenty three).

[0114] in, is the key quantity of the conversion mechanism, is a random function, and is the current number of iterations and the maximum number of iterations; is the control coefficient related to the number of iterations. To control the key quantities of the first and second stages, A function that performs a round-down operation.

[0115] in, and For the it The iteration and it +1 iteration The first individual (one individual includes the power of thermal power units, the charging power and discharging power of independent energy storage, and the discharging power and charging power of energy storage built by new energy participating in the unified dispatch mode) In the optimization process of the dispatching model of the new energy storage power station, multiple candidate solutions are used to approach the optimal solution; is the first optimal solution in the current global solution. The information of the element. is the oscillation control coefficient, whose value changes significantly in the initial iteration and then gradually stabilizes; is the decreasing control coefficient, its value gradually decreases and converges to 0; It is the rising control coefficient, and its value increases gradually. 、 、 and is a random number in the range [0, 1]. To protect the coefficient, it is determined by exponential and trigonometric functions, which helps to maintain the diversity among candidate solutions.

[0116] The specific solution process is as follows: First, the variables are randomly initialized according to the overall scheduling model to generate a population, which includes multiple individuals; second, use formula (20) to determine whether to enter the exploration phase or the development phase, and use formula (21) to determine whether to enter the first phase or the second phase. After confirmation, use the exploration phase update equation (formula (22)) and the development phase update equation (formula (23)) to update the iterative individuals; finally, update the individual's best information, that is, the variable status in the candidate solution, continuously iterate and increase the iteration count, and terminate the loop when the maximum iteration is reached.

[0117] The set of scheduling plans for new energy storage with energy storage in the current time period includes: scheduling plans for new energy storage with energy storage corresponding to each sampling moment in the current time period; the scheduling plans for new energy storage with energy storage include: the power of thermal power units, the charging power and discharging power of independent energy storage, and the discharging power and charging power of new energy storage with energy storage in the grid-dispatched energy storage operation mode. The set of power generation plans for new energy storage with energy storage stations in the current time period includes power generation plans for new energy storage with energy storage stations corresponding to each sampling moment in the current time period; the power generation plans for new energy storage with energy storage stations specifically include the discharging power and charging power of energy storage in the new energy storage with energy storage stations. In another exemplary embodiment of the present application, based on the set of scheduling plans for new energy storage with energy storage in the current time period and the set of power generation plans for new energy storage with energy storage stations in the current time period, a scheduling scheme for the new energy storage with energy storage stations in the current time period is obtained, specifically: for any sampling moment in the current time period, the discharge power of energy storage in the new energy storage with energy storage station in the power generation plan of the new energy storage with energy storage station at the sampling moment is added to the discharge power of energy storage in the grid-dispatched energy storage operation mode in the scheduling plan for new energy storage with energy storage, to obtain the added discharge power.

[0118] The charging power of the energy storage in the new energy storage station in the power generation plan of the new energy storage station at the sampling moment is added to the charging power of the new energy storage in the grid dispatching energy storage operation mode in the dispatching plan of the new energy storage to obtain the added charging power.

[0119] The electric power of the thermal power units, the charging power and discharging power of the independent energy storage, the added discharge power and the added charging power in the scheduling plan for the new energy storage system are determined as the scheduling plan for the new energy storage system at the sampling moment.

[0120] The scheduling plan of the new energy distribution and storage power station at all sampling moments in the current time period is determined as the scheduling plan of the new energy distribution and storage power station in the current time period.

[0121] The clustering algorithm based on Hopkins statistics and k-means proposed in this application can effectively determine the clustering characteristics of data distribution and significantly improve the accuracy and representativeness of the new energy deviation scenario set.

[0122] This application uses the Tornado optimization algorithm to solve the power generation plan decision model of the new energy storage power station, which can effectively avoid falling into the local optimal solution and improve the global search capability. At the same time, it is insensitive to the initial conditions and can stably converge to a high-quality solution under different initial conditions.

[0123] This application proposes an improved CNN-BiGRU neural network to rapidly acquire power generation plans and energy storage operation modes for renewable energy power plants. The improved convolutional neural network efficiently extracts spatial features from input data. By introducing a residual mechanism, it alleviates the vanishing gradient problem in deep networks and improves neural network training. The multi-scale features extracted by the improved convolutional neural network are combined with the time series features extracted by the BiGRU to improve the accuracy of power generation planning and energy storage operation mode decisions.

[0124] This application uses an exponential trigonometric optimization algorithm to solve the scheduling model of a new energy distribution and storage power station. By introducing exponential and trigonometric functions, the algorithm simulates the behavior of exploration and development, and can achieve a dynamic balance between global search and local optimization, avoiding falling into a local optimal solution.

[0125] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0126] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatching, characterized in that: The scheduling method for a new energy distribution and storage power station under the dual modes of self-consumption and grid scheduling includes: Input the new energy output forecast data set corresponding to the new energy distribution and storage power station in the current time period into the trained neural network model to obtain the power generation plan set and energy storage operation mode of the new energy distribution and storage power station in the current time period; If the energy storage operation mode of the new energy power station with storage in the current time period is the grid dispatching energy storage operation mode, then the scheduling model of the new energy power station with storage is solved to obtain the scheduling plan set of the new energy power station with storage in the current time period, and based on the scheduling plan set of the new energy power station with storage in the current time period and the power generation plan set of the new energy power station with storage in the current time period, the scheduling plan of the new energy power station with storage in the current time period is obtained; If the energy storage operation mode of the new energy distribution and storage power station in the current time period is the self-consumption energy storage operation mode, then determine the power generation plan set of the new energy distribution and storage power station in the current time period, which is the scheduling plan of the new energy distribution and storage power station in the current time period; The training process of the neural network model specifically includes: Obtaining historical new energy output data of a new energy distribution and storage power station; the historical new energy output data includes a set of new energy output prediction data and a set of new energy output actual data corresponding to multiple historical time periods; the set of new energy output prediction data corresponding to the historical time period includes the new energy output prediction data collected at each sampling time within the historical time period; the set of new energy output actual data corresponding to the historical time period includes the new energy output actual data collected at each sampling time within the historical time period; Construct a decision-making model for power generation planning of new energy storage power stations; For any historical time period, calculate the deviation between the new energy output forecast data collected at the t-th sampling moment in the new energy output forecast data set corresponding to the historical time period and the new energy output actual data collected at the t-th sampling moment in the new energy output actual data set corresponding to the historical time period, and obtain the new energy output deviation value at the t-th sampling moment in the historical time period; Obtaining a set of new energy output deviation values ​​corresponding to the historical time period according to the new energy output deviation values ​​at each sampling moment in the historical time period; Based on the Hopkins statistic, a clustering algorithm is used to cluster the set of new energy output deviation values ​​corresponding to each historical time period to obtain the clustering results; Calculate the probability of each cluster center based on the clustering results; Based on the new energy output forecast data set corresponding to each historical time period and each cluster center, the new energy output value set corresponding to each cluster center is obtained; Based on the probability of each cluster center and the set of new energy output values ​​of each cluster center, the power generation plan decision model of the new energy distribution and storage power station is solved to obtain the power generation plan set of the new energy distribution and storage power station and the energy storage operation mode of the new energy distribution and storage power station in each historical time period; Construct a training data set based on the power generation plan set and energy storage operation mode of the new energy power station in each historical time period and the new energy output forecast data set corresponding to each historical time period; The neural network model is trained using the training data set to obtain a trained neural network model.

2. The method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatch according to claim 1 is characterized in that: The neural network model includes: an improved convolutional neural network and a bidirectional gated recurrent unit connected in sequence; the improved convolutional neural network includes a deep convolution module and a convolution branch module; the input ends of the deep convolution module and the convolution branch module are connected, and the output ends of the deep convolution module and the convolution branch module are connected through a residual mechanism; the deep convolution module includes two convolutional neural networks connected in series.

3. The method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatch according to claim 1 is characterized in that: The probability of each cluster center is calculated based on the clustering results, specifically: According to the formula Calculate the probability of the xth cluster center, where is the probability of the xth cluster center; is the number of new energy output deviation value sets assigned to the x-th cluster center; It is the total number of new energy output deviation value sets corresponding to each historical time period.

4. The method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatching according to claim 1 is characterized in that: The new energy storage power station dispatching model includes: The dispatch objective function of the new energy storage power station, the thermal power output model constraints, the independent energy storage model constraints, the new energy storage model constraints in the grid dispatch energy storage operation mode, and the power balance constraints.

5. The method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatching according to claim 1 is characterized in that: The power generation plan decision model of the new energy storage power station is solved. Specifically, the tornado optimization algorithm is used to solve the power generation plan decision model of the new energy storage power station.

6. The method for dispatching a new energy distribution and storage power station under the dual modes of self-consumption and grid dispatch according to claim 1, characterized in that: The scheduling model of the new energy storage power station is solved, specifically: the exponential triangle optimization algorithm is used to solve the scheduling model of the new energy storage power station.

Images