Method, device and equipment for predicting output power of multiple new energy stations and storage medium

Abstract

The invention discloses a method, a device and equipment for predicting output power of multiple new energy stations, and a storage medium, and is used for solving the technical problem that the scheduling and stability of a power grid are affected due to the fact that the output power of the new energy stations is difficult to predict. The method comprises the following steps: acquiring historical environmental parameters and historical electrical parameters of a new energy station; preprocessing the historical environmental parameters and the historical electrical parameters to obtain modeling effective data; based on the modeling effective data, combining a Bayesian neural network and a bidirectional long-short-term memory neural network to construct a multi-step advanced power generation prediction model; constructing a refined physical model of the new energy station; fusing the multi-step advanced power generation prediction model and the refined physical model to generate a fusion model; acquiring current environmental parameters and current electrical parameters of the new energy station in real time; and inputting the current environment parameters and the current electrical parameters into the fusion model, and outputting a power prediction result.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy stations, and in particular to a method, device, equipment and storage medium for predicting output power of multiple new energy stations. Background Art

[0002] With the transformation of the global energy structure and the rapid development of renewable energy, the proportion of new energy sites in the power system is increasing. New energy power systems, especially wind farms and photovoltaic power stations, are valued for their clean and renewable characteristics.

[0003] However, these new energy sites are characterized by intermittency, volatility and uncertainty, which poses a huge challenge to the stable operation of the power system. For example, the power generation of wind farms is affected by wind speed and direction, while photovoltaic power generation is affected by light intensity and temperature. The uncertainty of these factors makes the output power of new energy sites difficult to predict, which in turn affects the dispatch and stability of the power grid. Summary of the invention

[0004] The present invention provides a method, device, equipment and storage medium for predicting the output power of multiple renewable energy stations, which are used to solve the technical problem that the output power of renewable energy stations is difficult to predict, which affects the scheduling and stability of the power grid.

[0005] The present invention provides a method for predicting output power of multiple renewable energy stations, comprising:

[0006] Obtain historical environmental parameters and historical electrical parameters of new energy stations;

[0007] Preprocessing the historical environmental parameters and the historical electrical parameters to obtain effective data for modeling;

[0008] Based on the modeling effective data, a multi-step advance power generation prediction model is constructed by combining a Bayesian neural network and a bidirectional long short-term memory neural network;

[0009] Constructing a refined physical model of the new energy station;

[0010] Fusion of the multi-step advance power generation prediction model and the refined physical model to generate a fusion model;

[0011] Real-time collection of current environmental parameters and current electrical parameters of the new energy station;

[0012] The current environmental parameters and the current electrical parameters are input into the fusion model, and a power prediction result is output.

[0013] Optionally, the step of preprocessing the historical environmental parameters and the historical electrical parameters to obtain valid modeling data includes:

[0014] Performing data cleaning on the historical environmental parameters and the historical electrical parameters to obtain cleaned data;

[0015] Perform missing value filling and outlier detection on the cleaned data to obtain target data;

[0016] The target data is normalized to obtain effective data for modeling.

[0017] Optionally, the step of constructing a multi-step advance power generation prediction model based on the modeling effective data in combination with a Bayesian neural network and a bidirectional long short-term memory neural network includes:

[0018] Calculate the weight parameters of each modeling effective data according to the Bayesian neural network;

[0019] Generate an initial bidirectional long short-term memory neural network based on the weight parameters;

[0020] The initial bidirectional long short-term memory neural network is trained using the modeling effective data to obtain a multi-step advance power generation prediction model.

[0021] Optionally, the step of constructing the refined physical model of the new energy station includes:

[0022] Obtaining the physical characteristics and operating rules of the new energy station;

[0023] The physical characteristics and the operating rules are used to construct a refined physical model of the new energy station.

[0024] The present invention also provides a device for predicting output power of multiple renewable energy stations, comprising:

[0025] A historical environmental parameter and historical electrical parameter acquisition module is used to obtain historical environmental parameters and historical electrical parameters of new energy stations;

[0026] A preprocessing module, used for preprocessing the historical environmental parameters and the historical electrical parameters to obtain effective data for modeling;

[0027] A multi-step advance power generation prediction model construction module is used to construct a multi-step advance power generation prediction model based on the modeling effective data in combination with a Bayesian neural network and a bidirectional long short-term memory neural network;

[0028] A refined physical model building module, used to build a refined physical model of the new energy station;

[0029] A fusion module, used for fusing the multi-step advance power generation prediction model and the refined physical model to generate a fusion model;

[0030] A current environmental parameter and current electrical parameter acquisition module, used for real-time acquisition of the current environmental parameters and current electrical parameters of the new energy station;

[0031] The power prediction module is used to input the current environmental parameters and the current electrical parameters into the fusion model and output a power prediction result.

[0032] Optionally, the preprocessing module includes:

[0033] A data cleaning submodule, used for performing data cleaning on the historical environmental parameters and the historical electrical parameters to obtain cleaned data;

[0034] A missing value filling and outlier detection submodule is used to perform missing value filling and outlier detection on the cleaned data to obtain target data;

[0035] The normalization submodule is used to normalize the target data to obtain effective data for modeling.

[0036] Optionally, the multi-step advance power generation prediction model construction module includes:

[0037] The weight parameter calculation submodule is used to calculate the weight parameters of each modeling effective data according to the Bayesian neural network;

[0038] An initial bidirectional long short-term memory neural network generation submodule, used to generate an initial bidirectional long short-term memory neural network based on the weight parameters;

[0039] The multi-step advance power generation prediction model construction submodule is used to train the initial bidirectional long short-term memory neural network with the modeling effective data to obtain the multi-step advance power generation prediction model.

[0040] Optionally, the refined physical model building module includes:

[0041] A refined physical model building module acquisition submodule is used to obtain the physical characteristics and operation rules of the new energy station;

[0042] The refined physical model construction submodule is used to construct a refined physical model of the new energy station using the physical characteristics and the operating rules.

[0043] The present invention also provides an electronic device, the device comprising a processor and a memory:

[0044] The memory is used to store program code and transmit the program code to the processor;

[0045] The processor is used to execute the multi-new energy station output power prediction method as described in any one of the above items according to the instructions in the program code.

[0046] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the method for predicting the output power of multiple renewable energy stations as described in any one of the above items.

[0047] It can be seen from the above technical scheme that the present invention has the following advantages: the present invention obtains the historical environmental parameters and historical electrical parameters of the new energy station; pre-processes the historical environmental parameters and historical electrical parameters to obtain effective data for modeling; based on the effective data for modeling, combines the Bayesian neural network and the bidirectional long short-term memory neural network to build a multi-step advance power generation prediction model; obtains the physical characteristics and operation rules of the new energy station to build a refined physical model of the new energy station; integrates the multi-step advance power generation prediction model and the refined physical model to generate a fusion model; collects the current environmental parameters and current electrical parameters of the new energy station in real time; inputs the current environmental parameters and current electrical parameters into the fusion model, and outputs the power prediction result. The output power prediction accuracy of the new energy station is improved to ensure the normal dispatch and stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] 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.

[0049] Figure 1 A flowchart of a method for predicting output power of multiple renewable energy stations provided by an embodiment of the present invention;

[0050] Figure 2 A flowchart of a method for predicting output power of multiple renewable energy stations provided by another embodiment of the present invention;

[0051] Figure 3 A structural block diagram of a multi-new energy station output power prediction device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The embodiments of the present invention provide a method, device, equipment and storage medium for predicting the output power of multiple renewable energy stations, which are used to solve the technical problem that the output power of renewable energy stations is difficult to predict, which affects the scheduling and stability of the power grid.

[0053] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions 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 embodiments described below 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.

[0054] See also Figure 1 , Figure 1 A flowchart of the steps of a method for predicting output power of multiple renewable energy stations provided in an embodiment of the present invention.

[0055] The present invention provides a method for predicting the output power of multiple renewable energy stations, which may specifically include the following steps:

[0056] Step 101, obtaining historical environmental parameters and historical electrical parameters of a new energy station;

[0057] Step 102, preprocessing the historical environmental parameters and historical electrical parameters to obtain effective data for modeling;

[0058] In an embodiment of the present invention, the new energy site may include a wind farm, a photovoltaic power station, and the like.

[0059] The historical environmental parameters of new energy sites may include wind speed, wind direction, light intensity, temperature, etc.

[0060] The historical electrical parameters of new energy sites may include grid voltage, current, power factor, etc.

[0061] Effective modeling data may include environmental parameters such as output power, wind speed, light intensity, temperature of new energy sites, and electrical parameters such as voltage and current of the power grid.

[0062] In an embodiment of the present invention, historical environmental data and historical electrical data of a new energy station may be collected, and the collected historical environmental data and historical electrical data may be preprocessed to obtain effective data for modeling.

[0063] Step 103, based on the modeling effective data, a multi-step advance power generation prediction model is constructed by combining the Bayesian neural network and the bidirectional long short-term memory neural network;

[0064] Bayesian Neural Networks (BNN) is a method that introduces Bayesian reasoning into neural networks. The core idea is to combine prior knowledge with observed data through Bayes' theorem to obtain the posterior distribution of model parameters. By reasoning about the posterior distribution, the confidence of the model prediction can be quantified. Neural networks are able to learn complex patterns of data due to their powerful expressive power. Combining the Bayesian method with neural networks to form a Bayesian neural network can build a model that can both express complex patterns and quantify uncertainty.

[0065] Bi-directional Long-Short Term Memory (BiLSTM) is an improved recurrent neural network specially designed for processing sequence data. BiLSTM can capture the bidirectional dependencies in the sequence by combining the outputs of the forward and backward LSTM networks. The core feature of BiLSTM is that it can process past and future information at the same time, which makes it perform well in sequence prediction tasks. BiLSTM consists of two groups of LSTM units, one group processes the forward (past to future) input sequence, and the other group processes the reverse (future to past) input sequence. The outputs of these two groups of LSTM units are merged to provide a comprehensive understanding of the entire sequence.

[0066] In an embodiment of the present invention, the weight parameters of the multi-step advance power generation prediction model can be quantified by a Bayesian neural network to improve the reliability of the prediction, and the multi-step advance power generation prediction model with the weight parameters obtained by Bayesian neural network training can be trained by BiLSTM.

[0067] Step 104, constructing a refined physical model of the new energy station;

[0068] In an embodiment of the present invention, a refined physical model of a new energy station can be constructed based on a multi-step advance power generation prediction model and the physical characteristics and operating rules of the new energy station. For a wind farm, a mathematical model of a wind turbine can be established based on the mechanical and electrical characteristics of the wind turbine; for a photovoltaic power station, a mathematical model of a photovoltaic cell can be established based on the photoelectric conversion characteristics and temperature characteristics of the photovoltaic cell. At the same time, it is also necessary to consider factors such as the grid connection mode and control strategy of the new energy station to further refine and improve the model.

[0069] Step 105, integrating the multi-step advance power generation prediction model and the refined physical model to generate a fusion model;

[0070] In an embodiment of the present invention, the output of the multi-step advance power generation prediction model is the input of the refined physical model. The power generation data of the new energy station is predicted by the multi-step advance power generation prediction model, and the data is input into the refined physical model to obtain the output power of the new energy station.

[0071] Step 106, collecting the current environmental parameters and current electrical parameters of the new energy station in real time;

[0072] Step 107: input the current environmental parameters and the current electrical parameters into the fusion model, and output the power prediction result.

[0073] In an embodiment of the present invention, the current environmental parameters and current electrical parameters of the new energy station collected in real time can be input into the refined physical model to obtain the output power prediction result of the new energy station. The dynamic response prediction result of the new energy station can also be output. According to the prediction results, the control strategy of the new energy station can be adjusted and optimized to achieve stable operation and efficient power generation of the new energy station.

[0074] The present invention obtains historical environmental parameters and historical electrical parameters of new energy stations; pre-processes the historical environmental parameters and historical electrical parameters to obtain effective modeling data; constructs a multi-step advance power generation prediction model based on the effective modeling data in combination with a Bayesian neural network and a bidirectional long short-term memory neural network; constructs a refined physical model of the new energy station; integrates the multi-step advance power generation prediction model and the refined physical model to generate a fusion model; collects the current environmental parameters and current electrical parameters of the new energy station in real time; inputs the current environmental parameters and current electrical parameters into the fusion model, and outputs the power prediction result. The output power prediction accuracy of the new energy station is improved, and the normal dispatching and stability of the power grid are ensured.

[0075] See also Figure 2 , Figure 2 A flowchart of a method for predicting the output power of a multi-new energy station provided by another embodiment of the present invention. Specifically, the following steps may be included:

[0076] Step 201, obtaining historical environmental parameters and historical electrical parameters of the new energy station;

[0077] Step 202, performing data cleaning on historical environmental parameters and historical electrical parameters to obtain cleaned data;

[0078] Step 203, performing missing value filling and outlier detection on the cleaned data to obtain target data;

[0079] Step 204, normalizing the target data to obtain effective data for modeling;

[0080] In an embodiment of the present invention, data cleaning, missing value filling, outlier detection and other processing are performed on historical environmental parameters and historical electrical parameters to ensure data quality and consistency, and then the obtained target data is normalized so that data from different sources and dimensions can be used in the same model.

[0081] Step 205, based on the modeling effective data, a multi-step advance power generation prediction model is constructed by combining the Bayesian neural network and the bidirectional long short-term memory neural network;

[0082] In the embodiment of the present invention, step 205 may include the following sub-steps:

[0083] S51, calculating weight parameters of each modeling effective data according to the Bayesian neural network;

[0084] In a Bayesian neural network, weight parameters are considered as random variables with a prior distribution, which are updated to a posterior distribution through observed data. The specific implementation steps are as follows:

[0085] 1. Prior distribution setting:

[0086] Before training begins, a prior distribution is set for the weight parameters of each modeling valid data in the multi-step ahead power generation forecasting model. This is usually a normal distribution or a uniform distribution, which represents the uncertainty of the weight before observing the data.

[0087] 2. Likelihood function definition:

[0088] The likelihood function describes the probability of observing data given the model function. In Bayesian neural networks, the likelihood function is usually a loss function based on the data and model parameters.

[0089] 3. Calculation of posterior distribution:

[0090] The posterior distribution is the probability distribution of the model parameters given the observed data. According to Bayes' theorem, the posterior distribution is the product of the prior distribution and the likelihood function, normalized.

[0091] 4. Variational Inference (VI): Since directly calculating the posterior distribution is usually very computationally complex, variational inference techniques are used to approximate the posterior distribution. VI introduces an easy-to-calculate distribution (such as a normal distribution) and optimizes the parameters to make this distribution as close to the true posterior distribution as possible.

[0092] 5. Network training:

[0093] The goal of network training is to maximize the log-likelihood of the posterior distribution, or equivalently minimize the Kullback-Leibler (KL) divergence. This involves adjusting the network parameters to reduce the difference between predictions and actual observations.

[0094] 6. In the prediction stage, the weight parameters are sampled from the posterior distribution, and prediction results are generated for each sample. By analyzing the distribution of these prediction results, the uncertainty of the prediction can be quantified, thereby determining the optimal weight parameters.

[0095] S52, generating an initial bidirectional long short-term memory neural network based on the weight parameters;

[0096] S53, using modeling effective data to train an initial bidirectional long short-term memory neural network to obtain a multi-step advance power generation prediction model.

[0097] In the specific implementation, the network structure of BiLSTM is as follows:

[0098] Input layer: receives the feature vector of time series data as input.

[0099] Forward LSTM layer: processes the input sequence from front to back, capturing the positive dependencies in the sequence.

[0100] Reverse LSTM layer: processes the input sequence from back to front, capturing the reverse dependencies in the sequence.

[0101] Merge layer: Merge the outputs of the forward and backward LSTM layers, either by simple concatenation or by performing nonlinear transformations through additional layers.

[0102] Output layer: Depending on the task requirements, it can be a fully connected layer or other types of layers to generate the final prediction results.

[0103] The implementation steps of BiLSTM prediction are as follows:

[0104] Forward propagation: The input sequence is passed step by step in the forward LSTM layer, and the output of each time step depends not only on the current input, but also on the output of the previous time step.

[0105] Backward Propagation: The same input sequence is passed through the backward LSTM layer in the opposite direction to capture the reverse dependencies of the sequence.

[0106] Merge output: The outputs from both directions are combined in the merge layer to form a complete representation of the entire sequence. Prediction generation: The merged output passes through the output layer to generate the final prediction result.

[0107] In an embodiment of the present invention, the BiLSTM model can be trained by modeling valid data.

[0108] After the BiLSTM model is trained, the prediction results for multiple future time steps are generated by inputting valid modeling data. For each future time step, the model needs to consider all previous prediction results and newly input data. Then, the performance of the model is evaluated by analyzing the prediction results, and the model parameters are adjusted according to actual needs to optimize the prediction accuracy, thus obtaining a multi-step advance power generation prediction model.

[0109] Step 206, constructing a refined physical model of the new energy station;

[0110] In this embodiment of the present invention, step 206 may include the following sub-steps:

[0111] S61, obtaining the physical characteristics and operation rules of the new energy station;

[0112] S62, uses physical characteristics and operating rules to build a refined physical model of the new energy station.

[0113] In the specific implementation, for a wind farm, a mathematical model of a wind turbine is established based on the mechanical and electrical characteristics of the wind turbine. The mathematical model of a wind turbine may include an aerodynamic model of the wind turbine, a mechanical model of the transmission system, an electromagnetic model of the generator, etc. At the same time, it is also necessary to consider the control strategies of the wind turbine, such as variable pitch control, variable speed control, etc., to further refine and improve the model. For a photovoltaic power station, a mathematical model of a photovoltaic cell is established based on the photoelectric conversion characteristics and temperature characteristics of the photovoltaic cell. The mathematical model of a photovoltaic cell may include an IV characteristic curve model, a temperature effect model, etc. of the photovoltaic cell. At the same time, it is also necessary to consider factors such as the grid connection mode of the photovoltaic power station and the inverter control strategy, to further refine and improve the model.

[0114] Step 207, integrating the multi-step advance power generation prediction model and the refined physical model to generate a fusion model;

[0115] In an embodiment of the present invention, the output of the multi-step advance power generation prediction model is the input of the refined physical model. The power generation data of the new energy station is predicted by the multi-step advance power generation prediction model, and the data is input into the refined physical model to obtain the output power of the new energy station.

[0116] Step 208, collecting the current environmental parameters and current electrical parameters of the new energy station in real time;

[0117] Step 209: input the current environmental parameters and the current electrical parameters into the fusion model, and output the power prediction result.

[0118] In the specific implementation, the current environmental parameters and current electrical parameters of the new energy station can be collected in real time through sensors and measuring equipment. The collected data is preprocessed, including data cleaning, screening and normalization. Then the preprocessed data is input into the fusion model to obtain the output power and dynamic response prediction results of the new energy station. The prediction results can include the output power, voltage fluctuation, frequency fluctuation, etc. of the new energy station.

[0119] Furthermore, in the embodiment of the present invention, the control strategy of the new energy station can be adjusted and optimized according to the prediction results. The control strategy can include variable pitch control, variable speed control, inverter control, etc. By adjusting and optimizing the control strategy, the stable operation and efficient power generation of the new energy station can be achieved.

[0120] The present invention obtains historical environmental parameters and historical electrical parameters of new energy stations; pre-processes the historical environmental parameters and historical electrical parameters to obtain effective modeling data; constructs a multi-step advance power generation prediction model based on the effective modeling data in combination with a Bayesian neural network and a bidirectional long short-term memory neural network; constructs a refined physical model of the new energy station; integrates the multi-step advance power generation prediction model and the refined physical model to generate a fusion model; collects the current environmental parameters and current electrical parameters of the new energy station in real time; inputs the current environmental parameters and current electrical parameters into the fusion model, and outputs the power prediction result. The output power prediction accuracy of the new energy station is improved, and the normal dispatching and stability of the power grid are ensured.

[0121] See also Figure 3 , Figure 3 A structural block diagram of a multi-new energy station output power prediction device provided in an embodiment of the present invention.

[0122] An embodiment of the present invention provides a device for predicting output power of a multi-new energy station, comprising:

[0123] A historical environmental parameter and historical electrical parameter acquisition module 301 is used to acquire historical environmental parameters and historical electrical parameters of a new energy station;

[0124] A preprocessing module 302 is used to preprocess historical environmental parameters and historical electrical parameters to obtain effective data for modeling;

[0125] A multi-step advance power generation prediction model construction module 303 is used to construct a multi-step advance power generation prediction model based on modeling effective data in combination with a Bayesian neural network and a bidirectional long short-term memory neural network;

[0126] A refined physical model building module 304 is used to build a refined physical model of a new energy station;

[0127] A fusion module 305 is used to fuse the multi-step advance power generation prediction model and the refined physical model to generate a fusion model;

[0128] The current environmental parameter and current electrical parameter acquisition module 306 is used to acquire the current environmental parameters and current electrical parameters of the new energy station in real time;

[0129] The power prediction module 307 is used to input the current environmental parameters and the current electrical parameters into the fusion model and output the power prediction result.

[0130] In this embodiment of the present invention, the preprocessing module 302 includes:

[0131] A data cleaning submodule is used to clean the historical environmental parameters and historical electrical parameters to obtain cleaned data;

[0132] The missing value filling and outlier detection submodule is used to fill missing values ​​and detect outliers on the cleaned data to obtain the target data;

[0133] The normalization submodule is used to normalize the target data to obtain effective data for modeling.

[0134] In the embodiment of the present invention, the multi-step advance power generation prediction model construction module 303 includes:

[0135] The weight parameter calculation submodule is used to calculate the weight parameters of each modeling effective data according to the Bayesian neural network;

[0136] An initial bidirectional long short-term memory neural network generation submodule is used to generate an initial bidirectional long short-term memory neural network based on weight parameters;

[0137] The multi-step ahead power generation prediction model construction submodule is used to train the initial bidirectional long short-term memory neural network with effective modeling data to obtain the multi-step ahead power generation prediction model.

[0138] In the embodiment of the present invention, the refined physical model building module 304 includes:

[0139] The refined physical model construction module obtains the submodule, which is used to obtain the physical characteristics and operation rules of the new energy station;

[0140] The refined physical model construction submodule is used to construct a refined physical model of a new energy station using physical characteristics and operating rules.

[0141] An embodiment of the present invention further provides an electronic device, the device comprising a processor and a memory:

[0142] The memory is used to store the program code and transmit the program code to the processor;

[0143] The processor is used to execute the multi-new energy station output power prediction method of the embodiment of the present invention according to the instructions in the program code.

[0144] An embodiment of the present invention further provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute the multi-new energy station output power prediction method of the embodiment of the present invention.

[0145] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0146] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0147] It will be appreciated by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take 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.

[0148] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks 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 terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal 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.

[0149] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0150] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement 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.

[0151] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0152] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0153] As described above, 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 thereof 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 method for predicting the output power of multiple renewable energy stations, characterized in that: include: Obtain historical environmental parameters and historical electrical parameters of new energy stations; Preprocessing the historical environmental parameters and the historical electrical parameters to obtain effective data for modeling; Based on the modeling effective data, a multi-step advance power generation prediction model is constructed by combining a Bayesian neural network and a bidirectional long short-term memory neural network; Constructing a refined physical model of the new energy station; Fusion of the multi-step advance power generation prediction model and the refined physical model to generate a fusion model; Real-time collection of current environmental parameters and current electrical parameters of the new energy station; The current environmental parameters and the current electrical parameters are input into the fusion model, and a power prediction result is output.

2. The method according to claim 1, characterized in that The step of preprocessing the historical environmental parameters and the historical electrical parameters to obtain effective data for modeling includes: Performing data cleaning on the historical environmental parameters and the historical electrical parameters to obtain cleaned data; Perform missing value filling and outlier detection on the cleaned data to obtain target data; The target data is normalized to obtain effective data for modeling.

3. The method according to claim 1, characterized in that The step of constructing a multi-step advance power generation prediction model based on the modeling effective data and combining a Bayesian neural network and a bidirectional long short-term memory neural network comprises: Calculate the weight parameters of each modeling effective data according to the Bayesian neural network; Generate an initial bidirectional long short-term memory neural network based on the weight parameters; The initial bidirectional long short-term memory neural network is trained using the modeling effective data to obtain a multi-step advance power generation prediction model.

4. The method according to claim 1, characterized in that: The step of constructing the refined physical model of the new energy station includes: Obtaining the physical characteristics and operating rules of the new energy station; The physical characteristics and the operating rules are used to construct a refined physical model of the new energy station.

5. A device for predicting output power of multiple renewable energy stations, characterized in that: include: A historical environmental parameter and historical electrical parameter acquisition module is used to obtain historical environmental parameters and historical electrical parameters of new energy stations; A preprocessing module, used for preprocessing the historical environmental parameters and the historical electrical parameters to obtain effective data for modeling; A multi-step advance power generation prediction model construction module is used to construct a multi-step advance power generation prediction model based on the modeling effective data in combination with a Bayesian neural network and a bidirectional long short-term memory neural network; A refined physical model building module, used to build a refined physical model of the new energy station; A fusion module, used for fusing the multi-step advance power generation prediction model and the refined physical model to generate a fusion model; A current environmental parameter and current electrical parameter acquisition module, used for real-time acquisition of the current environmental parameters and current electrical parameters of the new energy station; The power prediction module is used to input the current environmental parameters and the current electrical parameters into the fusion model and output a power prediction result.

6. The device according to claim 5, characterized in that The preprocessing module comprises: A data cleaning submodule, used for performing data cleaning on the historical environmental parameters and the historical electrical parameters to obtain cleaned data; A missing value filling and outlier detection submodule is used to perform missing value filling and outlier detection on the cleaned data to obtain target data; The normalization submodule is used to normalize the target data to obtain effective data for modeling.

7. The device according to claim 5, characterized in that The multi-step advance power generation prediction model building module includes: The weight parameter calculation submodule is used to calculate the weight parameters of each modeling effective data according to the Bayesian neural network; An initial bidirectional long short-term memory neural network generation submodule, used to generate an initial bidirectional long short-term memory neural network based on the weight parameters; The multi-step advance power generation prediction model construction submodule is used to train the initial bidirectional long short-term memory neural network with the modeling effective data to obtain the multi-step advance power generation prediction model.

8. The device according to claim 5, characterized in that The refined physical model building module includes: A refined physical model building module acquisition submodule is used to obtain the physical characteristics and operation rules of the new energy station; The refined physical model construction submodule is used to construct a refined physical model of the new energy station using the physical characteristics and the operating rules.

9. An electronic device, characterized in that: The device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the multi-new energy station output power prediction method described in any one of claims 1-4 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store program code, and the program code is used to execute the method for predicting the output power of multiple renewable energy stations as described in any one of claims 1-4.

Images