Energy storage scheduling decision-oriented source load matching rate prediction method, system, device, medium and product

Abstract

The invention relates to the technical field of power grid data prediction, and discloses an energy storage scheduling decision-oriented source load matching rate prediction method, system and device, a medium and a product, and the method comprises the steps: constructing an initial source load matching rate prediction model based on a deep learning algorithm, and forming a mixed loss function through a prediction error function and a decision evaluation function; and parameter optimization is carried out on the initial source load matching rate prediction model based on a mixed gradient descent learning method to obtain a source load matching rate prediction model, so that the source load matching rate is accurately predicted, and the decision precision under the condition of predicting the source load matching rate is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid data prediction, and in particular to a source-load matching rate prediction method, system, device, medium and product for energy storage scheduling decision-making. Background Art

[0002] The source-load matching rate refers to the degree of matching between the power output and the load demand, which can reflect the dynamic relationship between the output power of the system power supply and the power consumption of the load. Due to the high proportion of renewable energy access to the power system, the volatility of the source-load matching rate increases. The source-load matching rate is volatile and difficult to accurately predict. The source-load matching rate prediction is widely used in the management and dispatch of smart grids, including energy storage systems.

[0003] Traditional source-load matching rate prediction focuses on source-load matching rate prediction from the perspective of prediction range and various prediction models. In order to improve the prediction accuracy, a two-stage source-load matching rate prediction scheme is proposed, which predicts the source-load matching rate peak in the first stage and the source-load matching rate in the second stage continuously. In order to improve the prediction accuracy of neural networks, traditional source-load matching rate prediction takes energy management and control into consideration, and proposes a multi-horizon source-load matching rate prediction model to improve the prediction accuracy and detect the source-load matching rate peak, and adopts an overall method to predict the source-load matching rate of the power grid system.

[0004] In recent years, methods for predicting the source-load matching rate interval have been proposed. For example, a method based on constructing the Pareto optimal prediction interval predicts the source-load matching rate interval, which achieves the reliability and clarity of the interval. Taking into account the dynamic source-load matching rate changes in neighboring areas, the online learning method perceives the source-load matching rate fluctuations in neighboring areas and provides rolling day-ahead source-load matching rate / interval predictions and intraday predictions to improve the prediction accuracy.

[0005] Existing optimization-based work all formulates the uncertainty of the predicted source-load matching rate through sampling or distribution assumptions, and rarely considers the reverse impact of the optimization model on the upstream prediction model. The above research shows that there is a gap between the upstream prediction and the downstream optimization model. The goal of the traditional prediction model is usually to minimize the difference between the predicted source-load matching rate and the actual source-load matching rate, expressed as the prediction error. In contrast, the energy storage system focuses on maximizing the benefits from the source-load matching rate fluctuations. Maximizing the benefits of the energy storage system can be equivalent to minimizing the deviation between the optimal actual decision under the predicted source-load matching rate and the optimal decision under the actual source-load matching rate. The decision error is not necessarily consistent with the prediction error. The decision error comes from the energy storage system scheduling model and is difficult to be directly used to train the prediction model, which leads to inaccurate prediction of the source-load matching rate of the energy storage system. Summary of the invention

[0006] In view of this, the present invention provides a source-load matching rate prediction method, system, device, medium and product for energy storage scheduling decision-making, which solves the technical problem of inaccurate source-load matching rate prediction of the energy storage system in the above technical background.

[0007] A first aspect of the present invention provides a source-load matching rate prediction method for energy storage scheduling decision-making, comprising:

[0008] Based on the deep learning algorithm, the historical source-load matching rate sample set is trained to build the initial source-load matching rate prediction model;

[0009] Constructing a prediction error function and a decision evaluation function, and forming a mixed loss function according to the prediction error function and the decision evaluation function;

[0010] Based on the hybrid gradient descent learning method, the hybrid loss function is used to optimize the parameters of the initial source-load matching rate prediction model to obtain a source-load matching rate prediction model, which is used to predict the source-load matching rate at a future moment.

[0011] Preferably, the prediction error function adopts a mixed mean square error function.

[0012] Preferably, the process of constructing the decision evaluation function includes:

[0013] Determining the regret of the decision according to the gap between the optimal decision under the actual source-load matching rate and the optimal actual decision under the source-load matching rate predicted by the initial source-load matching rate prediction model;

[0014] Performing relaxation processing on the regret of the decision and determining the minimum upper bound of the regret of the decision to obtain the proxy regret;

[0015] A substitute regret loss function is determined according to the proxy regret, and the substitute regret loss function is used as a decision evaluation function.

[0016] Preferably, a hybrid loss function is formed according to the prediction error function and the decision evaluation function, including:

[0017] The prediction error function and the decision evaluation function are weighted to obtain the mixed loss function.

[0018] Preferably, the hybrid gradient descent learning method uses the hybrid loss function to optimize the parameters of the initial source-load matching rate prediction model to obtain the source-load matching rate prediction model, including:

[0019] Determining a mixed gradient for predicting a source-load matching rate according to the mixed loss function;

[0020] The mixed gradient is used to optimize the parameters of the initial source-load matching rate prediction model in a stochastic gradient descent manner to obtain a source-load matching rate prediction model.

[0021] Preferably, the using the mixed gradient to optimize the parameters of the initial source-load matching rate prediction model in a stochastic gradient descent manner to obtain the source-load matching rate prediction model includes:

[0022] Calculate the mixed loss value of the predicted source-load matching rate of the initial source-load matching rate prediction model according to the mixed loss function;

[0023] Back-propagating the combined loss value to the parameters of the initial source-load matching rate prediction model to obtain a parameter gradient;

[0024] Using the parameter gradient to update the parameters of the initial source-load matching rate prediction model in a stochastic gradient descent manner;

[0025] The updated parameters are used to optimize the parameters of the initial source-load matching rate prediction model to obtain a source-load matching rate prediction model.

[0026] In a second aspect, the embodiment of the present application further provides a source-load matching rate prediction system for energy storage scheduling decision-making, including:

[0027] A prediction model building module is used to train a historical source-load matching rate sample set based on a deep learning algorithm to build an initial source-load matching rate prediction model;

[0028] A hybrid loss construction module, used to construct a prediction error function and a decision evaluation function, and to form a hybrid loss function according to the prediction error function and the decision evaluation function;

[0029] The prediction model optimization module is used to optimize the parameters of the initial source-load matching rate prediction model based on the hybrid gradient descent learning method using the hybrid loss function to obtain the source-load matching rate prediction model, and the source-load matching rate prediction model is used to predict the source-load matching rate at a future moment.

[0030] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the source-load matching rate prediction method for energy storage scheduling decisions as described in the first aspect.

[0031] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, the source-load matching rate prediction method for energy storage scheduling decision-making as described in the first aspect is implemented.

[0032] In a fifth aspect, an embodiment of the present application further provides a computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the source-load matching rate prediction method for energy storage scheduling decisions as described in the first aspect.

[0033] It can be seen from the above technical scheme that the present invention constructs an initial source-load matching rate prediction model based on a deep learning algorithm, uses a prediction error function and a decision evaluation function to form a hybrid loss function, and optimizes the parameters of the initial source-load matching rate prediction model based on a hybrid gradient descent learning method to obtain a source-load matching rate prediction model, thereby accurately predicting the source-load matching rate and improving the decision-making accuracy under the predicted source-load matching rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 An application environment of a source-load matching rate prediction method for energy storage scheduling decision-making provided by an embodiment of the present invention;

[0036] Figure 2 A flow chart of a method for predicting source-load matching rate for energy storage scheduling decision-making provided in an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of the structure of a source-load matching rate prediction system for energy storage scheduling decision-making provided by an embodiment of the present invention;

[0038] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only 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.

[0040] The source-load matching rate prediction method for energy storage scheduling decision-making provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, each node of the energy storage system communicates with the server 102 through the network. The data storage system can store the data that the server 102 needs to process. The data storage system can be integrated on the server 102, or it can be placed on the cloud or other network servers. Server 102 trains a set of historical source-load matching rate samples based on a deep learning algorithm to construct an initial source-load matching rate prediction model; constructs a prediction error function and a decision evaluation function, and constructs a hybrid loss function based on the prediction error function and the decision evaluation function; based on a hybrid gradient descent learning method, the hybrid loss function is used to optimize the parameters of the initial source-load matching rate prediction model to obtain a source-load matching rate prediction model, and the source-load matching rate prediction model is used to predict the source-load matching rate at future times. Server 102 can be an independent physical server, or it can be a server cluster or distributed system composed of multiple physical servers, or it can be a cloud server that provides cloud computing services.

[0041] like Figure 2 As shown, the embodiment of the present application provides a source-load matching rate prediction method for energy storage scheduling decision-making, and the method is applied to Figure 1 The server 102 in the example is used as an example to illustrate the method, which includes the following steps S1 to S3. Among them:

[0042] Step S1: Train a historical source-load matching rate sample set based on a deep learning algorithm to build an initial source-load matching rate prediction model.

[0043] Among them, the historical source-load matching rate sample set contains the mapping relationship between the source-load matching rate and operating data of the energy storage system in the past period of time, and is trained based on the deep learning algorithm with operating data as input and source-load matching rate as output to obtain the initial source-load matching rate prediction model.

[0044] Step S2: construct a prediction error function and a decision evaluation function, and form a mixed loss function based on the prediction error function and the decision evaluation function.

[0045] In the embodiment of the present application, the prediction error function adopts a mixed mean square error function, and other error functions, such as a standard deviation function, may also be adopted.

[0046] For the decision evaluation function, the embodiment of the present application determines the difference between the optimal actual decision under the measured predicted source-load matching rate and the best predicted decision under the actual source-load matching rate in the proposed decision-aware source-load matching rate prediction, and low regret loss means an asymptotically optimal oracle decision.

[0047] Specifically, the construction process of the decision evaluation function includes:

[0048] Step S201: determining the regret of the decision according to the difference between the optimal decision under the actual source-load matching rate and the optimal actual decision under the source-load matching rate predicted by the initial source-load matching rate prediction model.

[0049] Among them, the regret of the decision is the difference between the benefit under the actual optimal decision and the benefit under the optimal decision:

[0050]

[0051]

[0052] In the formula, The regret of the decision, is the base value of source-load matching rate, is the sensitivity coefficient, L is the total system load demand, G is the total system power output, is the predicted source-load matching ratio, is the actual source-to-charge matching ratio, , They are the optimal decision under the real source-load matching rate and the optimal actual decision under the predicted source-load matching rate, respectively, and T is the matrix transpose. Among them, the optimal decision uses the real source-load matching rate as the source-load matching rate input, and is the result of the optimization calculation under the optimization objective and constraint conditions of the original system. This decision depends on the structural parameters of the selected system.

[0053] Step S202: relax the regret of the decision and determine the minimum upper bound of the regret of the decision to obtain the proxy regret.

[0054] It should be noted that the energy storage operation model is a typical mixed integer programming model, and its feasible domain is a set of polyhedrons, so its optimal decision may be at the extreme point of a polyhedron. Taking a two-dimensional polyhedron as an example, when the prediction error is the same, two different prediction source-load matching rates lead to different decision errors, and the corresponding regrets are also different. Decision under The predicted source-load matching ratio Next The same as the predicted source load matching rate Next Therefore, in this case, the prediction error is not equal to the decision error.

[0055] Relative to the predicted source-load matching ratio , the regret of the decision is discontinuous and unmanageable. The embodiment of the present application constructs a manageable regret loss function , expressed as agent regret.

[0056] Among them, the regret of the decision is relaxed, and the proxy regret is derived as follows:

[0057]

[0058] In the formula, Indicates the agent's regret, is the relaxation coefficient, , They are respectively the optimal benefit under the actual source-load matching rate and the optimal benefit under the predicted source-load matching rate.

[0059] in, Relaxation of optimal decision under This will give us an upper bound for the above formula. All of them are satisfied, and the equation can be derived:

[0060]

[0061] Actually is a decreasing function. Definition, Sub-gradient It can be expressed as: .

[0062] yes The optimal decision under Less than ,For example .

[0063] is less than 0, so is a monotone decreasing function. According to the monotone convergence theorem, Can be used Instead. With The increase of The trend of the term can be ignored, and the optimal tend to . This establishes the minimum upper bound of the regret:

[0064]

[0065] The combined prediction model is:

[0066]

[0067] in is the predicted source-load matching ratio, Yes Input Represents the prediction model, from the input feature vector To output predicted source load matching ratio ,The whole process consists of five procedures: i) data preprocessing, ii) feature engineering, iii) model selection, iv) training procedure, and v) prediction target.

[0068]

[0069] The linear regression model is and Map the input feature vector to the predicted source-load matching ratio.

[0070] The regret minimization of the prediction model can be further generalized as follows:

[0071]

[0072] In the formula, represents the system model parameters, and n represents the number of scenarios.

[0073] Among them, the formula for further generalization of the regret minimization of the prediction model gives the generalized form of the objective function when α takes different values. The second line of the objective function corresponds to the case of any α≥0; the third line shows that when α approaches infinity, all α can be replaced by the same parameter; the fourth line obtains the upper bound estimate by setting α to 2; the fifth line relaxes the objective function under the feasible decision value.

[0074] Step S203: determining a substitute regret loss function according to the proxy regret, and using the substitute regret loss function as a decision evaluation function.

[0075] Among them, given the predicted source-load matching rate Source-load matching ratio , the alternative regret loss function can be defined as:

[0076]

[0077] Given the predicted source-load matching rate and the actual source-load matching rate, the alternative regret loss function has the following properties: ; is the predicted source-load matching ratio Convex function of .

[0078] Among them, according to the definition of the alternative regret loss function, the gradient of the alternative regret loss function to the predicted source-load matching rate can be derived as:

[0079]

[0080] In the formula, is the gradient of the alternative regret loss function with respect to the predicted source-load matching rate.

[0081] Among them, the hybrid mean square error and the proxy regret focus on the prediction error and the decision error respectively, and look at the prediction mismatch from the perspective of the prediction model and the energy storage operation model respectively. Therefore, the embodiment of the present application combines the prediction error function and the decision evaluation function to form a hybrid loss function to train the source-load matching rate prediction model.

[0082] Among them, the energy storage operation model is:

[0083] When the source-charge matching ratio is low, ESS usually works in the charging state, and when the source-charge matching ratio is high, it is in the discharging state. The operation of ESS is usually as follows:

[0084] The source-load matching rate signal maximizes its benefits to construct the objective function of the energy storage operation model. The objective function of the energy storage operation model is:

[0085]

[0086] Where P is the operating power, is the feasible region of operating power, is the source-load matching ratio at time t, is the operating power at time t.

[0087] Among them, the operating constraints of the energy storage operation model include:

[0088] The feasible domain of is subject to a set of operations and technical constraints as follows:

[0089]

[0090] In the formula, Indicates that ESS is charged from the grid, Indicates that ESS discharges to the grid

[0091]

[0092]

[0093]

[0094] In the formula, , are the energy stored in the ESS at time t-1 and time t respectively. is among the allowed values, , Indicates the minimum and maximum values ​​of the battery system capacity.

[0095]

[0096]

[0097]

[0098]

[0099]

[0100] To prevent simultaneous charging and discharging, the embodiment of the present application adopts a big-M method to charge the ESS; M is a large positive number. and is a binary variable indicating the charge and charge status, where 1 indicates the state is on and 0 indicates the opposite. and is the maximum value of charging and discharging power. The entire ESS scheduling model is expressed as a mixed integer linear programming.

[0101] Specifically, the prediction error function and the decision evaluation function are weighted to obtain a hybrid loss function. The hybrid loss function is expressed as:

[0102]

[0103] In the formula, is the mixed loss function, is the weighting coefficient, which indicates the importance of minimizing the prediction error. For total time.

[0104] Step S3: Based on the hybrid gradient descent learning method, the hybrid loss function is used to optimize the parameters of the initial source-load matching rate prediction model to obtain the source-load matching rate prediction model, which is used to predict the source-load matching rate at a future moment.

[0105] Specifically, in step S3, based on the hybrid gradient descent learning method, the hybrid loss function is used to optimize the parameters of the initial source-load matching rate prediction model to obtain the source-load matching rate prediction model, including:

[0106] Step S301: Determine a mixed gradient for predicting a source-load matching rate according to a mixed loss function.

[0107] The mixed gradient is:

[0108]

[0109] In the formula, is a mixed gradient.

[0110] Among them, the mixed gradient of the mixed loss function can be explicitly calculated by the above formula. In order to improve computational efficiency, the automatic differentiation algorithm in the Autograd tool in Pytorch is used to implicitly calculate the gradient of MSE. Autograd is a reverse automatic differentiation system for calculating derivatives. It is the core of Pytorch. It records a graph of all operations performed on gradient-enabled tensors and creates an acyclic graph called a dynamic computational graph. The gradient is calculated by tracing the graph from the root to the leaf and multiplying each gradient by the chain rule. Autograd's automatic differentiation algorithm has high computational efficiency because its core logic is C++. It can implicitly calculate the MSE gradient, which is more efficient than the explicit calculation of the above formula. In contrast, the gradient of the agent's regret can be explicitly calculated by solving the mixed integer programming problem.

[0111] Step S302: Optimize the parameters of the initial source-load matching rate prediction model by using a mixed gradient in a stochastic gradient descent manner to obtain a source-load matching rate prediction model.

[0112] In some embodiments, the step S302 uses a mixed gradient to perform parameter optimization on the initial source-load matching rate prediction model in a stochastic gradient descent manner to obtain the source-load matching rate prediction model, including:

[0113] Step S3021: Calculate the mixed loss value of the predicted source-load matching rate of the initial source-load matching rate prediction model according to the mixed loss function.

[0114] Step S3022: back-propagate the combined loss value to the parameters of the initial source-load matching rate prediction model to obtain the parameter gradient.

[0115] Step S3023: Update the parameters of the initial source-load matching rate prediction model by using the parameter gradient in a stochastic gradient descent manner.

[0116] Step S3024: Optimize the parameters of the initial source-load matching rate prediction model using the updated parameters to obtain the source-load matching rate prediction model.

[0117] It should be noted that the hybrid gradient indication prediction model learning proposed in the embodiment of the present application reduces the decision error of the prediction model. The prediction error is measured by the mean squared error (MSE). Based on MSE and proxy regret, the present invention designs a hybrid loss function and proposes a hybrid stochastic gradient descent (SGD) learning method for updating the prediction model parameters, thereby improving the prediction and decision accuracy.

[0118] For example, in order to solve two different gradient calculation methods, the prediction model training process under the mixed gradient is divided into three steps:

[0119] 1) Calculate the gradient of the mixed mean square error loss function for predicting the source-load matching rate, back-propagate the gradient of the mixed mean square error loss function to the parameters of the initial source-load matching rate prediction model, and retain the gradient for subsequent parameter gradient updates.

[0120] 2) Calculate the gradient of the proxy regret for predicting the source-load matching rate, and back-propagate the gradient to the same parameters of the initial source-load matching rate prediction model;

[0121] 3) Update the parameters of the initial source-load matching rate prediction model according to the accumulated gradient. The hybrid gradient update is achieved through two back propagations and one update.

[0122] It is worth noting that the stochastic gradient descent learning method for energy storage system scheduling proposed in the embodiments of the present application can be applicable to simple linear prediction models and complex deep learning models.

[0123] To distinguish from previous studies, the innovations of the present invention are embodied in three aspects.

[0124] 1) The method of the present invention uses prediction and decision errors to train the source-load matching rate prediction model for energy storage scheduling, while existing studies only consider prediction errors.

[0125] 2) In the context of deep learning, the present invention proposes a tractable proxy regret to train a multi-layer neural network called residual neural network (ResNet).

[0126] 3) This invention proposes a hybrid loss design and learning method to improve the accuracy of energy storage operation decision-making and energy efficiency, and is applied to energy storage scheduling prediction for the first time.

[0127] The embodiment of the present application uses prediction error and decision error to learn the parameters of the prediction model, and proposes a decision-centered energy storage scheduling source-load matching rate prediction method. Compared with previous prediction methods, the decision-centered method proposed in the embodiment of the present application focuses on the impact of prediction error on the downstream optimization model, and improves the decision-making accuracy under the prediction source-load matching rate.

[0128] The embodiment of the present application proposes a tractable alternative regret for the decision error between the actual decision under the predicted source-load matching rate and the decision under the real source-load matching rate. Its gradient with the predicted source-load matching rate is further derived, which is the focus of this article. The decision regret under the predicted source-load matching rate is discontinuous and difficult to handle. Therefore, this article relaxes the feasible domain of the original problem and derives a tractable upper bound of regret to measure the decision error, expressed as a proxy regret. The proxy regret gradient of the predicted source-load matching rate is derived and passed from the downstream optimization model to the upstream prediction model to reduce decision errors.

[0129] The embodiment of the present application designs a hybrid loss function to measure the prediction and decision errors based on the weighted sum of the prediction error and the agent regret. Then a hybrid stochastic gradient descent learning method for energy storage scheduling is proposed to train the prediction model to improve the prediction and decision accuracy based on the hybrid loss. Case analysis shows that the decision-centric method proposed in the present invention can more accurately capture and predict the changing trend of the source-load matching rate, thereby improving the decision-making accuracy and energy efficiency.

[0130] It should be noted that the embodiment of the present application constructs an initial source-load matching rate prediction model based on a deep learning algorithm, uses a prediction error function and a decision evaluation function to form a hybrid loss function, and optimizes the parameters of the initial source-load matching rate prediction model based on a hybrid gradient descent learning method to obtain a source-load matching rate prediction model, thereby accurately predicting the source-load matching rate and improving the decision-making accuracy under the predicted source-load matching rate.

[0131] Based on the same inventive concept, an embodiment of the present application further provides a source-load matching rate prediction system for energy storage scheduling decision-making, which is used to implement the above-mentioned source-load matching rate prediction method for energy storage scheduling decision-making.

[0132] The implementation solution for solving the problem provided by the system is similar to the implementation solution recorded in the above method. Therefore, the specific limitations in one or more embodiments of the source-load matching rate prediction system for energy storage scheduling decisions provided below can be referred to the limitations of the source-load matching rate prediction method for energy storage scheduling decisions above, and will not be repeated here.

[0133] like Figure 3 As shown, the embodiment of the present application also provides a source-load matching rate prediction system for energy storage scheduling decision-making, including:

[0134] The prediction model building module 100 is used to train a historical source-load matching rate sample set based on a deep learning algorithm to build an initial source-load matching rate prediction model;

[0135] A mixed loss construction module 200 is used to construct a prediction error function and a decision evaluation function, and to form a mixed loss function according to the prediction error function and the decision evaluation function;

[0136] The prediction model optimization module 300 is used to optimize the parameters of the initial source-load matching rate prediction model based on the hybrid gradient descent learning method and the hybrid loss function to obtain the source-load matching rate prediction model, which is used to predict the source-load matching rate at a future moment.

[0137] In some embodiments, the prediction error function employs a mixed mean square error function.

[0138] In some embodiments, the process of constructing a decision evaluation function includes:

[0139] The regret of the decision is determined according to the gap between the optimal decision under the actual source-load matching rate and the optimal actual decision under the source-load matching rate predicted by the initial source-load matching rate prediction model;

[0140] Relax the regret of the decision and determine the minimum upper bound of the regret of the decision to obtain the proxy regret;

[0141] The alternative regret loss function is determined according to the proxy regret, and the alternative regret loss function is used as the decision evaluation function.

[0142] In some embodiments, a hybrid loss function is formed according to the prediction error function and the decision evaluation function, including:

[0143] The prediction error function and the decision evaluation function are weighted to obtain a mixed loss function.

[0144] In some embodiments, the prediction model optimization module 300 is used to determine the mixed gradient of the predicted source-load matching rate according to the mixed loss function; and use the mixed gradient to optimize the parameters of the initial source-load matching rate prediction model by stochastic gradient descent to obtain the source-load matching rate prediction model.

[0145] The mixed gradient is used to optimize the parameters of the initial source-load matching rate prediction model in a stochastic gradient descent manner to obtain the source-load matching rate prediction model, including:

[0146] Calculate the mixed loss value of the predicted source-load matching rate of the initial source-load matching rate prediction model according to the mixed loss function;

[0147] Back-propagating the combined loss value to the parameters of the initial source-load matching rate prediction model to obtain the parameter gradient;

[0148] The parameter gradient is updated by stochastic gradient descent in the initial source-load matching rate prediction model;

[0149] The updated parameters are used to optimize the parameters of the initial source-load matching rate prediction model to obtain the source-load matching rate prediction model.

[0150] like Figure 4 As shown, an embodiment of the present application further provides an electronic device, the electronic device 10 includes a memory 20 and a processor 30, the memory 20 stores a computer program, and when the computer program is executed by the processor 30, the processor 30 executes the steps of the source-load matching rate prediction method for energy storage scheduling decision-making in any of the above embodiments.

[0151] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, a source-load matching rate prediction method for energy storage scheduling decision-making as described in any of the above embodiments is implemented.

[0152] An embodiment of the present application also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer executes a source-load matching rate prediction method for energy storage scheduling decisions as described in any of the above embodiments.

[0153] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, electronic devices, computer storage media and computer program products can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0154] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatus.

[0155] In several embodiments provided by the present invention, it is understood that each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and a part of a module, a program segment or a code includes one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved.

[0156] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, electronic devices, computer storage media, computer program products and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0157] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0158] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0159] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for executing all or part of the steps of the method described in each embodiment of the present invention through a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk, etc. Various media that can store program codes.

[0160] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A source-load matching rate prediction method for energy storage scheduling decision-making, characterized in that: include: Based on the deep learning algorithm, the historical source-load matching rate sample set is trained to build the initial source-load matching rate prediction model; Constructing a prediction error function and a decision evaluation function, and forming a mixed loss function according to the prediction error function and the decision evaluation function; Based on the hybrid gradient descent learning method, the hybrid loss function is used to optimize the parameters of the initial source-load matching rate prediction model to obtain a source-load matching rate prediction model, which is used to predict the source-load matching rate at a future moment.

2. The source-load matching rate prediction method for energy storage scheduling decision according to claim 1 is characterized in that: The prediction error function adopts a mixed mean square error function.

3. The source-load matching rate prediction method for energy storage scheduling decision according to claim 1 is characterized in that: The construction process of the decision evaluation function includes: Determining the regret of the decision according to the gap between the optimal decision under the actual source-load matching rate and the optimal actual decision under the source-load matching rate predicted by the initial source-load matching rate prediction model; Performing relaxation processing on the regret of the decision and determining the minimum upper bound of the regret of the decision to obtain the proxy regret; A substitute regret loss function is determined according to the proxy regret, and the substitute regret loss function is used as a decision evaluation function.

4. The source-load matching rate prediction method for energy storage scheduling decision according to claim 1 is characterized in that: A hybrid loss function is formed according to the prediction error function and the decision evaluation function, including: The prediction error function and the decision evaluation function are weighted to obtain the mixed loss function.

5. The source-load matching rate prediction method for energy storage scheduling decision according to claim 1 is characterized in that: The hybrid gradient descent learning method is based on the hybrid loss function to optimize the parameters of the initial source-load matching rate prediction model to obtain the source-load matching rate prediction model, including: Determining a mixed gradient for predicting a source-load matching rate according to the mixed loss function; The mixed gradient is used to optimize the parameters of the initial source-load matching rate prediction model in a stochastic gradient descent manner to obtain a source-load matching rate prediction model.

6. The source-load matching rate prediction method for energy storage scheduling decision according to claim 1 is characterized in that: The method of optimizing the parameters of the initial source-load matching rate prediction model by using the mixed gradient in a stochastic gradient descent manner to obtain the source-load matching rate prediction model includes: Calculate the mixed loss value of the predicted source-load matching rate of the initial source-load matching rate prediction model according to the mixed loss function; Back-propagating the combined loss value to the parameters of the initial source-load matching rate prediction model to obtain a parameter gradient; Using the parameter gradient to update the parameters of the initial source-load matching rate prediction model in a stochastic gradient descent manner; The updated parameters are used to optimize the parameters of the initial source-load matching rate prediction model to obtain a source-load matching rate prediction model.

7. A source-load matching rate prediction system for energy storage scheduling decision-making, characterized in that: include: A prediction model building module is used to train a historical source-load matching rate sample set based on a deep learning algorithm to build an initial source-load matching rate prediction model; A hybrid loss construction module, used to construct a prediction error function and a decision evaluation function, and to form a hybrid loss function according to the prediction error function and the decision evaluation function; The prediction model optimization module is used to optimize the parameters of the initial source-load matching rate prediction model based on the hybrid gradient descent learning method using the hybrid loss function to obtain the source-load matching rate prediction model, and the source-load matching rate prediction model is used to predict the source-load matching rate at a future moment.

8. An electronic device, characterized in that: The electronic device includes a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the source-load matching rate prediction method for energy storage scheduling decision as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the source-load matching rate prediction method for energy storage scheduling decision-making as described in any one of claims 1-6 is implemented.

10. A computer program product, characterized in that The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer executes the source-load matching rate prediction method for energy storage scheduling decisions as described in any one of claims 1-6.

Images