Substation shared energy storage control method and system based on shared energy storage demand prediction

Abstract

The present invention provides a method and system for controlling shared energy storage in a substation based on shared energy storage demand prediction, which relates to the field of smart grid and new energy technology. The method includes collecting a data set of shared energy storage demand in a substation, screening the data set for similar days, and obtaining a data set of similar days; using the data set of similar days to predict the shared energy storage demand, and obtaining the predicted value of the shared energy storage demand at each moment; establishing a multi-objective control strategy for maximizing the total operating benefit of the substation system and minimizing the predicted deviation of the shared energy storage demand, and setting the operating constraints of the substation system; solving and obtaining the optimal shared energy storage control result based on the predicted value of the shared energy storage demand, the multi-objective control strategy, and the operating constraints, and controlling the substation system according to the optimal shared energy storage control result. The present invention improves energy utilization efficiency, ensures economic benefits, and realizes the optimized operation of the substation shared energy storage system through a multi-objective control strategy and an efficient prediction model.

Description

Technical Field

[0001] The present invention relates to the field of smart grid and new energy technology, and in particular to a method and system for controlling shared energy storage in an area based on shared energy storage demand prediction. Background Art

[0002] With the advancement of the "dual carbon" goals, the rapid development of new energy sources, mainly wind and photovoltaic power generation, and the large-scale access of distributed power sources, the volatility and intermittent characteristics of their output have increased the uncertainty and randomness of the planning and operation of new distribution stations, increasing the risk and difficulty of grid operation and management. Shared energy storage, as a high-quality flexible regulation resource for the power grid, can solve the problems of rapid fluctuations and difficulty in absorbing new energy output, and is a key technical means to support the access and absorption of a high proportion of new energy. At present, shared energy storage regulation in substations is one of the hot research directions in the energy field. It refers to the regulation of energy storage output power and energy through intelligent control to achieve energy sharing and balance energy supply and demand among different buyers.

[0003] Traditional control strategies are usually optimized with the goal of system economy, assuming that all resources within the substation are controllable. However, since the distributed power sources and loads within the substation currently do not participate in the unified control of the power grid, the substation's demand for shared energy storage must be considered when determining the shared energy storage control strategy. When shared energy storage participates in charging and discharging responses, it must have sufficient response power and capacity within the response period. Its optimized control strategy should meet the needs of shared energy storage at all times while meeting its needs and economy, and be achieved through day-ahead prediction and scheduling optimization. However, there is currently a lack of accurate prediction methods for the demand for shared energy storage in different scenarios, and the scheduling strategy has not fully considered the prediction results of shared energy storage demand, resulting in the substation's shared energy storage control strategy being unable to fully tap its regulation potential, thereby affecting the distribution station's ability to absorb new energy. Summary of the invention

[0004] To this end, an embodiment of the present invention provides a substation shared energy storage control method and system based on shared energy storage demand prediction, which is used to solve the problem that traditional control strategies in the prior art lack accurate demand prediction, do not fully consider the prediction results, and affect the new energy absorption capacity.

[0005] In order to solve the above problems, an embodiment of the present invention provides a method for controlling shared energy storage in a substation area based on shared energy storage demand prediction, the method comprising:

[0006] Step S1: Collect the shared energy storage demand data set of the substation area, and filter the data set for similar days to obtain a similar day data set;

[0007] Step S2: using similar day data sets to predict the shared energy storage demand, and obtain the predicted value of the shared energy storage demand at each moment;

[0008] Step S3: Establish a multi-objective control strategy to maximize the total operating benefit of the substation system and minimize the prediction deviation of the shared energy storage demand, and set the substation system operation constraints;

[0009] Step S4: Based on the shared energy storage demand forecast value, the multi-objective control strategy and the operation constraints, the optimal shared energy storage control result is obtained, and the substation system is controlled according to the optimal shared energy storage control result.

[0010] Preferably, in step S1, a shared energy storage demand data set of the substation area is collected, and the data set is screened for similar days to obtain a similar day data set, which specifically includes:

[0011] Step S11: Considering the influencing factors including the distributed power sources and load fluctuations in the substation area, collecting and collating the substation area shared energy storage demand data set;

[0012] Step S12: The weighted grey correlation coefficient method based on the maximum information coefficient method is used to screen the historical days to obtain the historical days similar to the day to be tested, and the maximum information coefficient method coefficient is used to assign a weight coefficient to each piece of information, and then the weighted grey correlation coefficient is calculated as the similarity of the historical days;

[0013] Step S13: sorting the calculated historical daily similarities, and selecting the corresponding shared energy storage demand data within the sorting set time to form a similar day data set.

[0014] Preferably, in step S11, the influence factors including the distributed power supply and load fluctuation in the substation area are taken into consideration, and the substation area shared energy storage demand data set is collected and sorted, specifically including:

[0015] The weather data including wind speed and sunshine, the day-ahead forecast value of distributed power output, the day-ahead forecast value of load, and the market clearing electricity price are sorted and the data format is unified as follows:

[0016] ;

[0017] Among them, k is the number of time nodes; n is the number of related features; is the nth feature sequence, A sequence of variables for predicting the demand for shared energy storage; For the entire dataset collected; are the relevant characteristic variables of shared energy storage demand; To share the demand for energy storage.

[0018] Preferably, in step S12, the weighted grey correlation coefficient method based on the maximum information coefficient method is used to screen the historical days to obtain the historical days similar to the day to be tested, and the maximum information coefficient method coefficient is used to assign a weight coefficient to each piece of information, and then the weighted grey correlation coefficient is calculated as the historical day similarity, which specifically includes:

[0019] Step S121: Calculate the mutual information MI between each characteristic sequence data and the shared energy storage demand:

[0020] ;

[0021] in, is the feature sequence, ; y is the sequence of shared energy storage demand prediction variables; is the joint probability density of each characteristic factor and shared energy storage demand; and are the marginal probability densities of the two respectively;

[0022] Step S122: Set the feature sequence The data set composed of the sequence y of the shared energy storage demand prediction variable is, Divide the scatter plot of its variables into grids. The range of is divided into segment a, and the range of y is divided into segment b to obtain the grid set , traverse all grids and find the maximum mutual information value of the variables in the grid:

[0023] in, is the division of the data set P under the grid G, For The mutual information value in the case; is the maximum mutual information value of the variables in the grid;

[0024] Step S123: The maximum mutual information value in each grid is normalized to obtain a value in the range of (0,1), and the maximum value is taken as the maximum information coefficient method coefficient value MIC of the influencing factor and the shared energy storage demand, and the calculation formula is:

[0025] ;

[0026] in, is a variable;

[0027] Step S124: Calculate the weighted grey correlation coefficient as the historical day similarity, including:

[0028] Construct weight vector: Use the maximum information coefficient of each feature and the predicted variable as the weight factor to calculate the weight coefficient of each feature in the selection of similar days , the calculation formula is:

[0029] ;

[0030] The obtained weight vector is ,in is the weight coefficient;

[0031] Calculate the grey correlation coefficient matrix: Use the grey correlation analysis method to obtain the grey correlation coefficient between the test day and similar days , the calculation formula is:

[0032] ;

[0033] in, , , are the values ​​of the i-th feature in the test day and the historical day respectively; is the resolution coefficient, which takes a value between 0 and 1; the grey correlation coefficient matrix between the day and the day to be tested is obtained ,in is the grey correlation coefficient;

[0034] Calculate the weighted grey correlation coefficient: Calculate the weighted grey correlation between the historical date and the predicted date based on the weight vector and the grey correlation coefficient matrix , the calculation formula is:

[0035] .

[0036] Preferably, in step S2, the shared energy storage demand is predicted using a similar day data set to obtain a predicted value of the shared energy storage demand at each moment, specifically including:

[0037] A CNN-GRU-Self-attention combined prediction model is constructed, which includes a CNN part, a GRU part, and a Self-attention mechanism part. The similar day data set is used as the input of the model. The CNN part uses two layers of convolution and pooling to more fully extract the time series features, which are input to the connection layer to obtain useful data features through weight distribution. Its output is used as the input of the GRU layer to capture the internal feature mapping relationship of the data. Since the GRU cannot fully learn the complete time series during the model training process, the Self-attention mechanism is applied to the output of the GRU. The relationship of long-distance data is obtained by learning the information at different positions, which is independent of the time problem. The attention weight of each time step is calculated, and the weighted sum of the GRU output is obtained to obtain the final output.

[0038] The principle formula of CNN is as follows:

[0039] ;

[0040] ;

[0041] ;

[0042] In the formula, is the convolutional layer; is the activation function; is the weight vector; is the convolution operation; is the bias parameter; is the pooling layer; k is the number of data channels; z is the input of the activation function;

[0043] The principle formula of GRU is as follows:

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] In the formula, To update the gate, To reset the gate, is the sigmoid function; is the weight parameter; is the final hidden state at this time step; is the candidate hidden state for this time step; is the input for this time step;

[0049] The principle formula of the self-attention mechanism is as follows:

[0050] ;

[0051] ;

[0052] ;

[0053] ;

[0054] In the formula, x is the input matrix, and three initialization weight matrices are obtained after different matrix transformations, namely, the query vector sequence Q, the key vector sequence K, and the value vector sequence V; is the core formula; is the activation function.

[0055] Preferably, in step S3, a multi-objective control strategy is established to maximize the total operating benefit of the substation system and minimize the prediction deviation of the shared energy storage demand, specifically including:

[0056] The multi-objective control strategy includes the total operating income of the substation system Maximum objective function and shared energy storage demand forecast deviation Minimum objective function;

[0057] Among them, the total operating income of the substation system The maximum objective function is:

[0058] ;

[0059] In the formula, They are the conventional unit income, shared energy storage income and distributed power generation income respectively; M is the number of conventional units in the area; T is the total number of dispatch cycles; is the power generation capacity of the i-th conventional unit at time t, is the electricity selling price of the i-th conventional unit; is the unit fuel cost electricity price of the i-th conventional unit; It is the normal start and stop state of the unit. is the scheduling time interval; are the actual charging and discharging power of the shared energy storage at time t, are the actual charging and discharging state variables of the shared energy storage, are the charging and discharging prices of shared energy storage respectively; Q is the number of distributed power sources participating in the dispatch, are the power generation and abandoned wind and solar power of the jth distributed generation at time t, are the unit electricity sales price of the jth distributed power source and the penalty price for wind and solar power abandonment respectively;

[0060] Among them, the forecast deviation of shared energy storage demand The minimum objective function is:

[0061] ;

[0062] In the formula, They are respectively the SOC at time t and the real-time SOC predicted by the shared energy storage; are the charging and discharging efficiencies of shared energy storage, respectively; Predict charging and discharging state variables for shared energy storage, respectively; They are respectively the charging and discharging power predicted for the shared energy storage at time t a day before; Rated capacity of shared energy storage.

[0063] Preferably, in step S3, setting the operation constraints of the substation system specifically includes:

[0064] The substation system operation constraints include load balance constraints, conventional unit constraints, distributed power output constraints, shared energy storage charging and discharging power and charge state constraints, and shared energy storage charge state constraints;

[0065] The load balancing constraint is expressed as:

[0066] ;

[0067] In the formula, is the load power at time t;

[0068] The conventional unit constraint is expressed as:

[0069] ;

[0070] ;

[0071] ;

[0072] In the formula, They represent the maximum and minimum allowable output of the i-th conventional unit respectively; Respectively represent the maximum upward and downward climbing rates of the i-th conventional unit; They represent the minimum continuous start and stop time of the i-th conventional unit respectively;

[0073] The distributed generation output constraint is expressed as:

[0074] ;

[0075] In the formula, The maximum permissible output for distributed power generation;

[0076] The shared energy storage charging and discharging power and charge state constraints are expressed as:

[0077] ;

[0078] In the formula, is probability; are the confidence levels for meeting the charging and discharging power requirements, respectively; Rated power of shared energy storage;

[0079] The shared energy storage charge state constraint is expressed as:

[0080] ;

[0081] In the formula, Indicates the upper and lower limits of the shared energy storage SOC.

[0082] Preferably, in step S4, based on the shared energy storage demand forecast value, the multi-objective control strategy and the operation constraints, the method for solving the optimal shared energy storage control result is:

[0083] Based on the predicted value of shared energy storage demand, multi-objective control strategy and operation constraints, the NSGA-II algorithm and fuzzy membership function are used to screen the Pareto frontier and obtain the optimal shared energy storage control result.

[0084] Preferably, the method further comprises:

[0085] Use the prediction model evaluation indicators, including root mean square error RMSE, mean absolute percentage error MAPE, mean absolute error MAE and determination coefficient R 2 , test the prediction accuracy of the model;

[0086] The root mean square error RMSE is expressed as:

[0087] ;

[0088] The mean absolute percentage error MAPE is expressed as:

[0089] ;

[0090] The mean absolute error MAE is expressed as:

[0091] ;

[0092] The coefficient of determination R 2 It is expressed as:

[0093] ;

[0094] In the formula, is the predicted value of shared energy storage demand at the i-th time point on the test day; is the actual value of shared energy storage demand at the i-th time point on the test day; is the average value of shared energy storage demand at all time points on the test day, and N represents the total number of prediction points.

[0095] The embodiment of the present invention further provides a substation shared energy storage control system based on shared energy storage demand prediction, which is used to implement the above-mentioned substation shared energy storage control method based on shared energy storage demand prediction, specifically including:

[0096] The data collection and processing module is used to collect the shared energy storage demand data set of the substation area, screen the data set for similar days, and obtain the similar day data set;

[0097] The demand forecasting module is used to forecast the shared energy storage demand using similar day data sets to obtain the forecast value of the shared energy storage demand at each moment;

[0098] The strategy establishment and constraint setting module is used to establish a multi-objective control strategy to maximize the total operating benefits of the substation system and minimize the forecast deviation of shared energy storage demand, and set the operating constraints of the substation system;

[0099] The optimization control and execution module is used to solve the optimal shared energy storage control result based on the shared energy storage demand forecast value, multi-objective control strategy and operation constraints, and control the substation system according to the optimal shared energy storage control result.

[0100] It can be seen from the above technical solutions that the present invention has the following beneficial effects:

[0101] (1) Dual optimization of economic efficiency and regulation advantages: The regulation strategy proposed in the present invention not only considers the maximization of the total operating benefits of the substation system, but also considers the minimization of the forecast deviation of the shared energy storage demand, forming a multi-objective regulation strategy. This strategy not only ensures the economical and efficient operation of the substation system, but also ensures that the shared energy storage can meet the necessary power and capacity requirements when responding to regulation. Compared with the traditional regulation strategy that only considers a single economy, the present invention helps to further give play to the regulation advantages of shared energy storage, improve energy utilization efficiency, and thus enhance the operating benefits of shared energy storage in the substation.

[0102] (2) Accurate selection of similar days and demand forecasting: When forecasting shared energy storage demand, the present invention fully considers various factors that affect the demand for shared energy storage, such as load and new energy fluctuations, market clearing electricity prices, etc. By calculating the correlation coefficient between each influencing factor and the shared energy storage demand, and using it as the weight coefficient of each factor when selecting similar days, the selected similar days are more accurate and can better reflect the similarity between the test day and the historical day. This accurate selection of similar days provides a solid foundation for subsequent demand forecasting and helps improve the accuracy of the forecast.

[0103] (3) Efficient prediction model and parameter optimization: The present invention uses a CNN-GRU combined prediction model based on the Self-attention mechanism in the prediction model. By introducing the attention mechanism, this model can highlight the role and influence of key factors on the shared energy storage demand, thereby improving the prediction accuracy. In addition, the GRU model has fewer training parameters than other models, which to a certain extent reduces the time consumed by model training and improves the prediction efficiency. This efficient prediction model provides strong support for the real-time regulation of shared energy storage in the substation area. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] In order to more clearly illustrate the implementation cases of the present invention or the technical solutions in the prior art, the following is a brief description of the drawings required for use in the embodiments. By referring to the drawings, the features and advantages of the present invention will be more clearly understood. The drawings are schematic and should not be understood as limiting the present invention in any way. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0105] Figure 1 A flowchart of a method for controlling shared energy storage in a substation based on shared energy storage demand prediction provided in an embodiment;

[0106] Figure 2 The structural diagram of the CNN-GRU-Self-Attention prediction model in the embodiment

[0107] Figure 3 A comparison chart of the day-ahead predicted output and the actual output of the shared energy storage in a power grid of a substation area optimized by the method proposed in the present invention in the embodiment;

[0108] Figure 4 A block diagram of a substation shared energy storage control system based on shared energy storage demand prediction provided in an embodiment. DETAILED DESCRIPTION

[0109] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Embodiment 1

[0110] In order to solve the problem that the traditional control strategy in the existing technology lacks accurate demand forecasting, does not fully consider the forecast results, and affects the new energy consumption capacity. Figure 1 As shown, an embodiment of the present invention proposes a method for controlling shared energy storage in a substation area based on shared energy storage demand prediction, the method comprising:

[0111] Step S1: Collect the shared energy storage demand data set of the substation area, and filter the data set for similar days to obtain a similar day data set;

[0112] Step S2: using similar day data sets to predict the shared energy storage demand, and obtain the predicted value of the shared energy storage demand at each moment;

[0113] Step S3: Establish a multi-objective control strategy to maximize the total operating benefit of the substation system and minimize the prediction deviation of the shared energy storage demand, and set the substation system operation constraints;

[0114] Step S4: Based on the shared energy storage demand forecast value, the multi-objective control strategy and the operation constraints, the optimal shared energy storage control result is obtained, and the substation system is controlled according to the optimal shared energy storage control result.

[0115] It can be seen from the above technical solution that the present invention provides a method for controlling shared energy storage in a substation based on shared energy storage demand prediction. By collecting and analyzing the shared energy storage demand data set of the substation, similar daily data sets are screened out, and then these data are used to accurately predict the shared energy storage demand. On the basis of the prediction, a multi-objective control strategy is constructed to maximize the total operating income of the substation system and minimize the prediction deviation of the shared energy storage demand. The advantage of this method is that it not only improves the energy utilization efficiency and operating benefits, but also ensures that the shared energy storage system in the substation can meet the regulation needs and maximize the economic benefits through accurate prediction and scientific control strategies. In particular, through the CNN-GRU combined prediction model of the Self-attention mechanism, the present invention has achieved significant improvements in prediction accuracy and efficiency, providing strong technical support for the optimized operation of the shared energy storage system in the substation.

[0116] In this embodiment, in step S1, a shared energy storage demand data set of the substation area is collected, and the data set is screened for similar days to obtain a similar day data set, which specifically includes:

[0117] Step S11: Considering the influencing factors including the distributed power generation and load fluctuation in the substation area, collect and organize the substation area shared energy storage demand data set.

[0118] Step S12: The weighted grey correlation coefficient method based on the maximum information coefficient method is used to screen the historical days to obtain the historical days similar to the day to be tested, and the maximum information coefficient method coefficient is used to assign a weight coefficient to each information, and then the weighted grey correlation coefficient is calculated as the similarity of the historical days.

[0119] Step S13: Sort the calculated historical daily similarities, and select the shared energy storage demand data corresponding to the sorting setting time (in this embodiment, the 50 days before the sorting are selected) to form a similar day data set.

[0120] Specifically, in step S11, weather data such as wind speed and sunshine, day-ahead forecast values ​​of distributed power generation output, day-ahead forecast values ​​of load, market clearing electricity prices and other historical data are sorted and the data format is unified as follows:

[0121] ;

[0122] Among them, k is the number of time nodes; n is the number of related features; is the nth feature sequence, A sequence of variables for predicting the demand for shared energy storage; For the entire dataset collected; are the relevant characteristic variables of shared energy storage demand; To share the demand for energy storage.

[0123] Furthermore, the variable information related to the shared energy storage demand can provide an effective basis for accurately judging future fluctuations. When the related data are similar, the demand also has a certain similarity. For this, the weighted grey correlation coefficient method (GRA) based on the maximum information coefficient method (MIC) is used to screen the historical days to obtain historical days similar to the day to be tested. The MIC coefficient is used to assign a weight coefficient to each piece of information and then the weighted grey correlation coefficient is calculated as the similarity of the historical days, including:

[0124] Step S121: Calculate the mutual information MI between each characteristic sequence data and the shared energy storage demand:

[0125] ;

[0126] in, is the feature sequence, ; y is the sequence of shared energy storage demand prediction variables; is the joint probability density of each characteristic factor and shared energy storage demand; and are the marginal probability densities of the two respectively.

[0127] Step S122: Set the feature sequence The data set composed of the sequence y of the shared energy storage demand prediction variable is, Divide the scatter plot of its variables into grids. The range of is divided into segment a, and the range of y is divided into segment b to obtain the grid set , traverse all grids and find the maximum mutual information value of the variables in the grid:

[0128] in, is the division of the data set P under the grid G, For The mutual information value in the case; is the maximum mutual information value of the variables in the grid.

[0129] Step S123: The maximum mutual information value in each grid is normalized to obtain a value in the range of (0,1), and the maximum value is taken as the maximum information coefficient method coefficient value MIC of the influencing factor and the shared energy storage demand, and the calculation formula is:

[0130] ;

[0131] in, is a variable, usually .

[0132] Step S124: Calculate the weighted grey correlation coefficient as the historical day similarity, including:

[0133] Construct weight vector: Use the maximum information coefficient of each feature and the predicted variable as the weight factor to calculate the weight coefficient of each feature in the selection of similar days , the calculation formula is:

[0134] ;

[0135] The obtained weight vector is ,in is the weight coefficient.

[0136] Calculate the grey correlation coefficient matrix: Use the grey correlation analysis method to obtain the grey correlation coefficient between the test day and similar days , the calculation formula is:

[0137] ;

[0138] in, , , are the values ​​of the i-th feature in the test day and the historical day respectively; is the resolution coefficient, which takes a value between 0 and 1, and is generally 0.5; the grey correlation coefficient matrix between the day and the day to be measured is obtained ,in is the grey correlation coefficient.

[0139] Calculate the weighted grey correlation coefficient: Calculate the weighted grey correlation between the historical date and the predicted date based on the weight vector and the grey correlation coefficient matrix , the calculation formula is:

[0140] .

[0141] In this embodiment, in step S2, the shared energy storage demand is predicted using the similar day data set to obtain the shared energy storage demand prediction value at each moment, specifically including:

[0142] Construct a CNN-GRU-Self-attention combined prediction model, such as Figure 2As shown in the figure, the model includes CNN, GRU and Self-attention. The similar day data set is used as the input of the model. The CNN part uses two layers of convolution and pooling to more fully extract the time series features. The input is input to the connection layer to obtain useful data features through weight distribution. Its output is used as the input of the GRU layer to capture the internal feature mapping relationship of the data. Since GRU cannot fully learn the complete time series during the model training process, the Self-attention mechanism is applied to the output of GRU to obtain the relationship of long-distance data by learning the information at different positions, which is independent of the time problem. By calculating the attention weight of each time step, the weighted sum of the output of GRU is obtained to obtain the final output.

[0143] The principle formula of CNN is as follows:

[0144] ;

[0145] ;

[0146] ;

[0147] In the formula, is the convolutional layer; is the activation function; is the weight vector; is the convolution operation; is the bias parameter; is the pooling layer; k is the number of data channels; z is the input of the activation function.

[0148] The principle formula of GRU is as follows:

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] In the formula, To update the gate, To reset the gate, is the sigmoid function; is the weight parameter; is the final hidden state at this time step; is the candidate hidden state for this time step; is the input for this time step;

[0154] The principle formula of the self-attention mechanism is as follows:

[0155] ;

[0156] ;

[0157] ;

[0158] ;

[0159] In the formula, x is the input matrix, and three initialization weight matrices are obtained after different matrix transformations, namely, the query vector sequence Q, the key vector sequence K, and the value vector sequence V; is the core formula; is the activation function.

[0160] In this embodiment, in step S3, a multi-objective control strategy is established to maximize the total operating benefit of the substation system and minimize the prediction deviation of the shared energy storage demand, and the substation system operation constraints are set.

[0161] Specifically, the multi-objective control strategy includes the total operating benefit of the substation system Maximum objective function and shared energy storage demand forecast deviation Minimum objective function.

[0162] Among them, the total operating income of the substation system The maximum objective function is:

[0163] ;

[0164] In the formula, They are the conventional unit income, shared energy storage income and distributed power generation income respectively; M is the number of conventional units in the area; T is the total number of dispatch cycles; is the power generation capacity of the i-th conventional unit at time t, is the electricity selling price of the i-th conventional unit; is the unit fuel cost electricity price of the i-th conventional unit; It is the normal start and stop state of the unit. is the scheduling time interval; are the actual charging and discharging power of the shared energy storage at time t, are the actual charging and discharging state variables of the shared energy storage, are the charging and discharging prices of shared energy storage respectively; Q is the number of distributed power sources participating in the dispatch, are the power generation and abandoned wind and solar power of the jth distributed generation at time t, They are the unit electricity sales price of the j-th distributed power source and the penalty price for wind and solar power abandonment respectively.

[0165] Among them, the forecast deviation of shared energy storage demand The minimum objective function is:

[0166] ;

[0167] In the formula, They are respectively the SOC at time t and the real-time SOC predicted by the shared energy storage; are the charging and discharging efficiencies of shared energy storage, respectively; Predict charging and discharging state variables for shared energy storage, respectively; They are respectively the charging and discharging power predicted for the shared energy storage at time t a day before; Rated capacity of shared energy storage.

[0168] Furthermore, the substation system operation constraints are set, which include load balance constraints, conventional unit constraints, distributed power output constraints, shared energy storage charging and discharging power and charge state constraints, and shared energy storage charge state constraints.

[0169] The substation system operation constraints include load balance constraints, conventional unit constraints, distributed power output constraints, shared energy storage charging and discharging power and charge state constraints, and shared energy storage charge state constraints;

[0170] The load balancing constraint is expressed as:

[0171] ;

[0172] In the formula, is the load power at time t.

[0173] The conventional unit constraints are expressed as:

[0174] ;

[0175] ;

[0176] ;

[0177] In the formula, They represent the maximum and minimum allowable output of the i-th conventional unit respectively; Respectively represent the maximum upward and downward climbing rates of the i-th conventional unit; They respectively represent the minimum continuous start-up and shutdown time of the i-th conventional unit.

[0178] The output constraint of distributed generation is expressed as:

[0179] ;

[0180] In the formula, The maximum allowable output of distributed power generation.

[0181] Affected by the randomness of system source and load and the error of shared energy storage prediction, the real-time power of shared energy storage may not always meet the day-ahead prediction results. Considering the uncertain factors, the shared energy storage power can meet the day-ahead prediction power at a certain confidence level and does not exceed the rated power of shared energy storage. The shared energy storage charging and discharging power and charge state constraints are introduced and expressed as:

[0182] ;

[0183] In the formula, is probability; are the confidence levels for meeting the charging and discharging power requirements, respectively; Rated power of shared energy storage.

[0184] The shared energy storage state of charge constraint is expressed as:

[0185] ;

[0186] In the formula, Indicates the upper and lower limits of the shared energy storage SOC.

[0187] In this embodiment, in step S4, based on the shared energy storage demand forecast value, the multi-objective control strategy and the operation constraints, the optimal shared energy storage control result is solved and the substation system is controlled according to the optimal shared energy storage control result.

[0188] This embodiment uses the NSGA-II algorithm and the fuzzy membership function to screen the Pareto frontier and obtain the optimal shared energy storage control result.

[0189] In addition, in order to verify the prediction effect of the model, the present invention uses prediction model evaluation indicators, including root mean square error RMSE, mean absolute percentage error MAPE, mean absolute error MAE and determination coefficient R 2 , test the prediction accuracy of the model.

[0190] The root mean square error RMSE is expressed as:

[0191] ;

[0192] The mean absolute percentage error MAPE is expressed as:

[0193] ;

[0194] The mean absolute error MAE is expressed as:

[0195] ;

[0196] Coefficient of determination R 2 It is expressed as:

[0197] ;

[0198] In the formula, is the predicted value of the shared energy storage demand at the i-th time point on the test day, i.e., the predicted charging and discharging power of the shared energy storage in step S3; is the actual value of the shared energy storage demand at the i-th time point on the test day, that is, the actual charging and discharging power of the shared energy storage in step S3; is the average value of shared energy storage demand at all time points on the test day, and N represents the total number of prediction points.

[0199] Comparison of the predicted output and actual output of the shared energy storage in a power grid of a substation area optimized by the method of the present invention Figure 3 The prediction error evaluation index of the result is shown in Table 1.

[0200] Table 1 Prediction error evaluation indicators

[0201] Evaluation indicators MAE(MW) RMSE(MW) MAPE(%) <![CDATA[R 2 ]]> result 0.181 0.246 6.926 0.994

[0202] From the above results, it can be seen that the shared energy storage output demand result curve obtained by the prediction model constructed by the present invention is basically consistent with the trend of the actual shared energy storage demand curve, with good fitting effect and high prediction accuracy. The prediction error of shared energy storage demand at most time points is very small, and only a few time points with strong jumps have slightly larger error values. The prediction effect of the model was further quantitatively calculated using the prediction error evaluation index, among which the results of MAE, RMSE, and MAPE are very small, indicating high prediction accuracy, and R 2 A result close to 1 indicates that the predicted curve is very close to the actual curve. Embodiment 2

[0203] like Figure 4 As shown, the present invention provides a substation shared energy storage control system based on shared energy storage demand prediction, which is used to implement the substation shared energy storage control method based on shared energy storage demand prediction of the above embodiment 1, specifically including:

[0204] The data collection and processing module 100 is used to collect the shared energy storage demand data set of the substation area, and screen the data set for similar days to obtain a similar day data set;

[0205] The demand prediction module 200 is used to predict the shared energy storage demand using similar day data sets to obtain the shared energy storage demand prediction value at each moment;

[0206] The strategy establishment and constraint setting module 300 is used to establish a multi-objective control strategy to maximize the total operating benefit of the substation system and minimize the prediction deviation of the shared energy storage demand, and set the substation system operation constraints;

[0207] The optimization control and execution module 400 is used to solve the optimal shared energy storage control result based on the shared energy storage demand prediction value, multi-objective control strategy and operation constraints, and control the substation system according to the optimal shared energy storage control result.

[0208] A substation shared energy storage control system based on shared energy storage demand prediction in this embodiment is used to implement the aforementioned substation shared energy storage control method based on shared energy storage demand prediction. Therefore, the specific implementation method of the substation shared energy storage control system based on shared energy storage demand prediction can be seen in the implementation method part of the substation shared energy storage control method based on shared energy storage demand prediction in the previous text, for example, the data collection and processing module 100, the demand prediction module 200, the strategy establishment and constraint setting module 300, the optimization control and execution module 400, are respectively used to implement steps S1, S2, S3, and S4 in the above-mentioned substation shared energy storage control method based on shared energy storage demand prediction. Therefore, its specific implementation method can refer to the description of the corresponding each part of the implementation method. In order to avoid redundancy, it will not be repeated here.

[0209] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0210] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0211] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0212] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from them are still within the protection scope of the invention.

Claims

1. A method for controlling shared energy storage in a substation area based on shared energy storage demand prediction, characterized in that: include: Step S1: Collect the shared energy storage demand data set of the substation area, and filter the data set for similar days to obtain the similar day data set, which specifically includes: Step S11: Considering the influencing factors including the distributed power sources and load fluctuations in the substation area, collecting and collating the substation area shared energy storage demand data set; Step S12: The weighted grey correlation coefficient method based on the maximum information coefficient method is used to screen the historical days to obtain the historical days similar to the day to be tested, and the weight coefficient is assigned to each information using the maximum information coefficient method coefficient, and then the weighted grey correlation coefficient is calculated as the similarity of the historical days, which specifically includes: Step S121: Calculate the mutual information MI between each feature data and the shared energy storage demand: Among them, x i is a characteristic sequence, i = 1, 2, ..., n; y is a sequence of shared energy storage demand prediction variables; p(x i ,y) is the joint probability density of each characteristic factor and shared energy storage demand; p(x i ) and p(y) are the marginal probability densities of the two respectively; Step S122: Set the feature sequence x i The sequence y of the shared energy storage demand prediction variable constitutes the data set P = {(x i ,y),i=1,…,n}, divide the scatter plot of its variables into grids, and i The range of y is divided into segments a and the range of y is divided into segments b to obtain the grid set G(x i ,y), traverse all grids and find the maximum mutual information value of the variables in the grid: Among them, P|G is the division of data set P under grid G, MI P|G is the mutual information value in the case of P|G; is the maximum mutual information value of the variables in the grid; Step S123: The maximum mutual information value in each grid is normalized to obtain a value in the range of (0,1), and the maximum value is taken as the maximum information coefficient method coefficient value MIC of the influencing factor and the shared energy storage demand, and the calculation formula is: Among them, B(n) is a variable; Step S124: Calculate the weighted grey correlation coefficient as the historical day similarity, including: Construct weight vector: Use the maximum information coefficient of each feature and the predicted variable as the weight factor to calculate the weight coefficient ω of each feature in the selection of similar days i , the calculation formula is: The obtained weight vector is W = [ω1, ω2, …, ω n ], where ω1,ω2,…,ω n is the weight coefficient; Calculate the grey correlation coefficient matrix: Use the grey correlation analysis method to obtain the grey correlation coefficient ξ between the day to be tested and similar days j (i), the calculation formula is: in, are the values ​​of the i-th feature in the day to be tested and the historical day respectively; ε is the resolution coefficient, which ranges from 0 to 1; the grey correlation coefficient matrix ζ of the day and the day to be tested is obtained j =[ξ j (1),ξ j (2),…,ξ j (n)], where ξ j (1),ξ j (2),…,ξ j (n) is the grey correlation coefficient; Calculate the weighted grey correlation coefficient: Calculate the weighted grey correlation degree φ between the historical day and the predicted day according to the weight vector and the grey correlation coefficient matrix j , the calculation formula is: Step S13: sorting the calculated historical daily similarities, and selecting the corresponding shared energy storage demand data within the sorting set time to form a similar day data set; Step S2: using similar day data sets to predict the shared energy storage demand, and obtain the predicted value of the shared energy storage demand at each moment; Step S3: Establish a multi-objective control strategy to maximize the total operating benefit of the substation system and minimize the prediction deviation of the shared energy storage demand, and set the substation system operation constraints; Step S4: Based on the shared energy storage demand forecast value, the multi-objective control strategy and the operation constraints, the optimal shared energy storage control result is obtained, and the substation system is controlled according to the optimal shared energy storage control result.

2. The method for controlling shared energy storage in substations based on shared energy storage demand prediction according to claim 1, characterized in that: In step S11, the influencing factors including the distributed power supply and load fluctuation in the substation area are considered, and the substation area shared energy storage demand data set is collected and sorted, specifically including: The weather data including wind speed and sunshine, the day-ahead forecast value of distributed power output, the day-ahead forecast value of load, and the market clearing electricity price are sorted and the data format is unified as follows: Among them, k is the number of time nodes; n is the number of related features; is the nth feature sequence, is the sequence of shared energy storage demand prediction variables; D is the entire collected data set; X is the relevant characteristic variable of shared energy storage demand; Y is the shared energy storage demand.

3. The method for controlling shared energy storage in substations based on shared energy storage demand prediction according to claim 1, characterized in that: In step S2, the shared energy storage demand is predicted using the similar day data set to obtain the shared energy storage demand prediction value at each moment, specifically including: A CNN-GRU-Self-attention combined prediction model is constructed, which includes a CNN part, a GRU part, and a Self-attention mechanism part. The similar day data set is used as the input of the model. The CNN part uses two layers of convolution and pooling to more fully extract the time series features, which are input to the connection layer to obtain useful data features through weight distribution. Its output is used as the input of the GRU layer to capture the internal feature mapping relationship of the data. Since the GRU cannot fully learn the complete time series during the model training process, the Self-attention mechanism is applied to the output of the GRU. The relationship of long-distance data is obtained by learning the information at different positions, which is independent of the time problem. The attention weight of each time step is calculated, and the weighted sum of the GRU output is obtained to obtain the final output. The principle formula of CNN is as follows: H i =f(H i-1 *W i +b i ); f(z)=max(z,0); P i (k)=max(H i (2k-1),H i (2k)); In the formula, H i is the convolutional layer; f() is the activation function; W i is the weight vector; * is the convolution operation; b i is the bias parameter; P i is the pooling layer; k is the number of data channels; z is the input of the activation function; The principle formula of GRU is as follows: z t =σ[W z ·(h t-1 ,x t )]; r t =σ[W r ·(h t-1 ,x t )]; In the formula, z t is the update gate, r t is the reset gate, σ is the sigmoid function; W z ,W r , is the weight parameter; h t is the final hidden state at this time step; is the candidate hidden state for this time step; x t is the input for this time step; The principle formula of the self-attention mechanism is as follows: Q=W q x; K=W k x; V=W v x; In the formula, x is the input matrix, and three initialization weight matrices are obtained after different matrix transformations, namely the query vector sequence Q, the key vector sequence K, and the value vector sequence V; Attention() is the core formula; and softmax() is the activation function.

4. The method for controlling shared energy storage in substations based on shared energy storage demand prediction according to claim 1, characterized in that: In step S3, a multi-objective control strategy is established to maximize the total operating benefit of the substation system and minimize the forecast deviation of shared energy storage demand, specifically including: The multi-objective control strategy includes the maximum objective function of the total operating benefit F1 of the substation system and the minimum objective function of the shared energy storage demand forecast deviation F2; Among them, the maximum objective function of the total operating income F1 of the substation system is: In the formula, I G ,I ES ,I NEW are the conventional unit income, shared energy storage income and distributed power generation income respectively; M is the number of conventional units in the area; T is the total number of dispatch cycles; P gi,t is the power generation capacity of the i-th conventional unit at time t, is the electricity selling price of the i-th conventional unit; is the unit fuel cost electricity price of the i-th conventional unit; gi.t is the normal unit start and stop state, Δt is the scheduling time interval; are the actual charging and discharging power of the shared energy storage at time t, are the actual charging and discharging state variables of the shared energy storage, are the charging and discharging prices of shared energy storage respectively; Q is the number of distributed power sources participating in the dispatch, P wj.t , are the power generation and abandoned wind and solar power of the jth distributed generation at time t, c wj , are the unit electricity sales price of the jth distributed power source and the penalty price for wind and solar power abandonment respectively; Among them, the minimum objective function of the shared energy storage demand forecast deviation F2 is: In the formula, They are respectively the SOC at time t and the real-time SOC predicted by the shared energy storage; are the charging and discharging efficiencies of shared energy storage, respectively; Predict charging and discharging state variables for shared energy storage, respectively; They are respectively the charging and discharging power predicted for the shared energy storage at time t a day before; Rated capacity of shared energy storage.

5. The method for controlling shared energy storage in substations based on shared energy storage demand prediction according to claim 4, characterized in that: In step S3, the operation constraints of the substation system are set, including: The substation system operation constraints include load balance constraints, conventional unit constraints, distributed power output constraints, shared energy storage charging and discharging power and charge state constraints, and shared energy storage charge state constraints; The load balancing constraint is expressed as: Where P L,t is the load power at time t; The conventional unit constraint is expressed as: i gi.t P gi.min ≤i gi.t P gi.t ≤i gi.t P gi.max ; -r d.gi ≤P gi.t -P gi.t-1 ≤r u.gi ; Where P gi.max , P gi.min Respectively represent the maximum and minimum allowable output of the i-th conventional unit; r u.gi 、r d.gi Respectively represent the maximum upward and downward climbing rates of the i-th conventional unit; T i on , T i off They represent the minimum continuous start and stop time of the i-th conventional unit respectively; The distributed generation output constraint is expressed as: 0≤P wj.t ≤P wj.max ; Where P wj.max The maximum permissible output for distributed power generation; The shared energy storage charging and discharging power and charge state constraints are expressed as: Where Pr{·} is the probability; β1 and β2 are the confidence levels of meeting the charging and discharging power requirements respectively; P ES Rated power of shared energy storage; The shared energy storage charge state constraint is expressed as: In the formula, S oc,max , S oc,min Indicates the upper and lower limits of the shared energy storage SOC.

6. The method for controlling shared energy storage in substations based on shared energy storage demand prediction according to claim 1, characterized in that: In step S4, based on the shared energy storage demand forecast value, the multi-objective control strategy and the operation constraints, the method for solving the optimal shared energy storage control result is: Based on the predicted value of shared energy storage demand, multi-objective control strategy and operation constraints, the NSGA-II algorithm and fuzzy membership function are used to screen the Pareto frontier and obtain the optimal shared energy storage control result.

7. The method for controlling shared energy storage in substations based on shared energy storage demand prediction according to claim 1, characterized in that: The method further comprises: Use the prediction model evaluation indicators, including root mean square error RMSE, mean absolute percentage error MAPE, mean absolute error MAE and determination coefficient R 2 , test the prediction accuracy of the model; The root mean square error RMSE is expressed as: The mean absolute percentage error MAPE is expressed as: The mean absolute error MAE is expressed as: The coefficient of determination R 2 It is expressed as: In the formula, is the predicted value of shared energy storage demand at the i-th time point on the test day; y i is the actual value of shared energy storage demand at the i-th time point on the test day; is the average value of shared energy storage demand at all time points on the test day, and N represents the total number of prediction points.

8. A shared energy storage control system based on shared energy storage demand forecast, characterized in that: The system is used to implement the shared energy storage control method based on shared energy storage demand prediction according to any one of claims 1 to 7, specifically comprising: The data collection and processing module is used to collect the shared energy storage demand data set of the substation area, screen the data set for similar days, and obtain the similar day data set; The demand forecasting module is used to forecast the shared energy storage demand using similar day data sets to obtain the forecast value of the shared energy storage demand at each moment; The strategy establishment and constraint setting module is used to establish a multi-objective control strategy to maximize the total operating benefits of the substation system and minimize the forecast deviation of shared energy storage demand, and set the operating constraints of the substation system; The optimization control and execution module is used to solve the optimal shared energy storage control result based on the shared energy storage demand forecast value, multi-objective control strategy and operation constraints, and control the substation system according to the optimal shared energy storage control result.

Images