Optical power combination prediction method, system, equipment and medium

Abstract

The invention provides an optical power combination prediction method, system and device and a medium, and belongs to the technical field of optical power combination prediction. According to the method, meteorological data to be predicted are respectively input into a trained XGBoosting model and a trained BiLSTM model which are improved by adopting a Bayesian optimization method to carry out optical power combination prediction, and two optical power combination prediction results are obtained. And dynamically adjusting the final output weight of the two optical power combination prediction results to obtain a final optical power combination prediction result. Compared with the prior art, the method effectively solves the problem that in the prior art, only the output result of the model is subjected to simple average weighting, so that the optical power prediction precision is not high.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical power combination prediction, and in particular relates to an optical power combination prediction method, system, equipment and medium. Background Art

[0002] As the most abundant renewable energy with the greatest development potential, the development and utilization of solar energy is an important part of my country's construction of a new power system. However, affected by meteorological factors such as temperature, cloud cover, and atmospheric aerosol levels, the power generation of photovoltaic modules is highly volatile and random. With the continuous increase in the installed capacity of photovoltaic power generation, the uncertainty of photovoltaic output has a prominent impact on the reliability and quality of power supply of the power grid. Accurately predicting the optical power will help the power dispatching department to adjust the dispatching plan and maintenance plan in a timely manner, which is of vital importance to ensure the safe and stable operation of the power grid.

[0003] The methods currently used in the prior art for predicting optical power are: 1) using physical modeling methods to achieve optical power prediction; 2) using statistical methods to achieve optical power prediction; 3) using deep learning methods to achieve optical power prediction.

[0004] Using physical modeling to predict optical power can not only obtain prediction results more directly, but also does not require reliance on much historical data. However, this method has poor robustness.

[0005] Using physical modeling to predict optical power requires collecting and counting a large amount of historical data and extracting the intrinsic correlation information of the historical data for prediction. Compared with the physical modeling method, the modeling is relatively simple, but this method not only has high requirements on data quality, but also the optical power prediction effect is not ideal.

[0006] Using deep learning methods to predict optical power can extract high-dimensional features from massive data and still perform well in complex nonlinear mappings. Typical deep learning methods include support vector regression, convolutional neural networks, long short-term memory neural networks, and graph neural networks. Although the deep learning method can directly establish the mapping relationship between independent variables and dependent variables when predicting optical power, this method suffers from overfitting and is prone to falling into local optimality.

[0007] In summary, single optical power prediction methods (physical modeling, statistical methods, and deep learning methods) have limitations in prediction accuracy and scenario generalization. The optical power combination prediction method combines and integrates different algorithms to give full play to the advantages of different methods, thereby obtaining better prediction accuracy. The existing optical power combination prediction method optimizes the neural network through various algorithms, and simply averages and weights the output results of the model, resulting in low optical power prediction accuracy. Summary of the invention

[0008] The purpose of the present invention is to provide a method, system, device and medium for optical power combination prediction, which is used to solve the problem that the output results of the model are simply average-weighted in the prior art, resulting in low optical power prediction accuracy.

[0009] In order to achieve the above object, the present invention adopts the following technical solution: In a first aspect, the present invention provides an optical power combination prediction method, comprising the following steps: Acquire historical optical power data and meteorological data, and pre-process the obtained historical optical power data and meteorological data; Clustering the pre-processed historical optical power data and meteorological data to obtain several clustering results; Decomposing several clustering results into intrinsic mode function components at different frequencies, and using the intrinsic mode function components to construct training sets and test sets; Construct an XGBoosting model and a BiLSTM model, and use the Bayesian optimization method to improve the XGBoosting model. Based on the constructed training set, train the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method. Based on the constructed test set, test the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method. The meteorological data to be predicted are respectively input into the trained XGBoosting model and BiLSTM model improved by Bayesian optimization method to perform optical power combination prediction, and two optical power combination prediction results are obtained; The weights of the final outputs of the two optical power combination prediction results are dynamically adjusted to obtain the final optical power combination prediction result.

[0010] A further improvement of the present invention is that the preprocessing refers to first normalizing the obtained historical optical power data and meteorological data and then performing outlier replacement processing.

[0011] A further improvement of the present invention is that the calculation formula for the normalization process is:

[0012] in, x are the normalized historical optical power data and meteorological data, is the original historical optical power data and meteorological data, and are the maximum values ​​of the original historical optical power data and meteorological data and the minimum values ​​of the original historical optical power data and meteorological data respectively; The calculation formula for the outlier replacement process is:

[0013] in, is an outlier, The first k data, After the outlier k data.

[0014] A further improvement of the present invention is that the calculation formula for improving the XGBoosting model using the Bayesian optimization method is:

[0015] in, obj is the objective function, for n The loss between the sample points and the true value, is the complexity of the tree.

[0016] A further improvement of the present invention is that an adaptive weighting method is used to dynamically adjust the weights of the final output of the two optical power combination prediction results to obtain the final optical power combination prediction result; The calculation formula of the adaptive weighting method is:

[0017] in, and are the output results of the two models. is the weight of the model in the prediction results, For the forgetting factor, is the sample set data.

[0018] A further improvement of the present invention is that an improved fuzzy C-means clustering algorithm is used to cluster the pre-processed historical optical power data and meteorological data to obtain a number of clustering results.

[0019] A further improvement of the present invention is that an empirical mode decomposition method is used to decompose a plurality of clustering results into intrinsic mode function components at different frequencies.

[0020] In a second aspect, the present invention provides an optical power combination prediction system, including a data acquisition module, a data preprocessing module, a data clustering module, a data set construction module, a model training module, an optical power combination prediction module and a weight dynamic adjustment module; The data acquisition module is used to acquire historical optical power data and meteorological data; The data preprocessing module is used to preprocess the obtained historical optical power data and meteorological data; The data clustering module is used to cluster the pre-processed historical optical power data and meteorological data to obtain several clustering results; The data set construction module is used to decompose a number of clustering results into intrinsic mode function components at different frequencies, and use the intrinsic mode function components to construct a training set and a test set; The model training module is used to construct an XGBoosting model and a BiLSTM model, and improve the XGBoosting model using a Bayesian optimization method, train the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method based on the constructed training set, and test the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method based on the constructed test set, to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method; The optical power combination prediction module is used to input the meteorological data to be predicted into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and obtain two optical power combination prediction results; The weight dynamic adjustment module is used to dynamically adjust the weights of the two optical power combination prediction results finally output to obtain the final optical power combination prediction result.

[0021] In a third aspect, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the optical power combination prediction method introduced above when executing the computer program.

[0022] In a fourth aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the optical power combination prediction method introduced above.

[0023] Compared with the prior art, the present invention has the following beneficial effects: The present invention belongs to an improved invention. Compared with the existing optical power combination prediction method, the optical power combination prediction method proposed by the present invention inputs the meteorological data to be predicted into the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and obtains two optical power combination prediction results. The weights of the two optical power combination prediction results are dynamically adjusted to obtain the final optical power combination prediction result. It can be seen that the present invention uses the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method to obtain two optical power combination prediction results. On the one hand, the present invention uses the Bayesian optimization method to improve the XGBoosting model, thereby improving the efficiency and accuracy of the XGBoosting model. On the other hand, the present invention first clusters the pre-processed historical optical power data and meteorological data to obtain a number of clustering results, and then performs optical power combination prediction, thereby improving the accuracy of the prediction. Moreover, the present invention dynamically adjusts the weight of the model to reduce the prediction error of the combination model. Compared with the prior art, the present invention effectively solves the problem that the output results of the model are simply average-weighted in the prior art, resulting in low optical power prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the optical power combination prediction method of the present invention; Figure 2 is a schematic diagram of an optical power combination prediction system of the present invention; Figure 3 It is a schematic diagram of the structure of the electronic device of the present invention. DETAILED DESCRIPTION

[0025] In order to further understand the content of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0026] The optical power combination prediction method proposed in the present invention inputs the meteorological data to be predicted into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and obtains two optical power combination prediction results. The weights of the final output of the two optical power combination prediction results are dynamically adjusted to obtain the final optical power combination prediction result. Compared with the prior art, the present invention effectively solves the problem that the output results of the model are simply average-weighted in the prior art, resulting in low optical power prediction accuracy.

[0027] Embodiment 1: The flowchart of the optical power combination prediction method of the present invention is as follows Figure 1 As shown, the optical power combination prediction method of the present invention comprises the following steps: S1. Obtain historical optical power data and meteorological data, and pre-process the obtained historical optical power data and meteorological data.

[0028] S2. Cluster the pre-processed historical optical power data and meteorological data to obtain several clustering results.

[0029] S3. Decompose several clustering results into intrinsic mode function components at different frequencies, and use the intrinsic mode function components to construct training sets and test sets.

[0030] S4. Construct an XGBoosting model and a BiLSTM model, and use the Bayesian optimization method to improve the XGBoosting model. Based on the constructed training set, the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are trained, and based on the constructed test set, the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are tested to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method.

[0031] S5. The meteorological data to be predicted are respectively input into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and two optical power combination prediction results are obtained.

[0032] S6. Dynamically adjust the weights of the final outputs of the two optical power combination prediction results to obtain the final optical power combination prediction result.

[0033] Embodiment 2: The schematic diagram of the optical power combination prediction system of the present invention is as follows Figure 2 As shown, the optical power combination prediction system of the present invention includes a data acquisition module, a data preprocessing module, a data clustering module, a data set construction module, a model training module, an optical power combination prediction module and a weight dynamic adjustment module.

[0034] The data acquisition module is used to obtain historical optical power data and meteorological data.

[0035] The data preprocessing module is used to preprocess the obtained historical optical power data and meteorological data.

[0036] The data clustering module is used to cluster the pre-processed historical optical power data and meteorological data to obtain several clustering results.

[0037] The data set construction module is used to decompose several clustering results into intrinsic mode function components at different frequencies, and use the intrinsic mode function components to construct training sets and test sets.

[0038] The model training module is used to construct the XGBoosting model and the BiLSTM model, and improve the XGBoosting model using the Bayesian optimization method. The XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are trained based on the constructed training set, and the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are tested based on the constructed test set to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method.

[0039] The optical power combination prediction module is used to input the meteorological data to be predicted into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction and obtain two optical power combination prediction results.

[0040] The weight dynamic adjustment module is used to dynamically adjust the weights of the two optical power combination prediction results finally output to obtain the final optical power combination prediction result.

[0041] Embodiment 3: S1. Obtain historical optical power data and meteorological data, and pre-process the obtained historical optical power data and meteorological data.

[0042] The historical optical power data and meteorological data are obtained, and the obtained historical optical power data and meteorological data are preprocessed. The preprocessing refers to first normalizing the obtained historical optical power data and meteorological data, and then performing outlier replacement processing.

[0043] The process of determining outliers in historical optical power data and meteorological data is described as follows: The acquired historical optical power data and meteorological data anomalies are divided into intervals according to irradiance, and the optical power data in each irradiance interval is sorted. The power data in each irradiance interval is divided into four equal parts using three quartiles (respectively: Q1 (25%), Q2 (50%), Q3 (75%)), and the interquartile range and upper and lower boundaries are calculated. The calculation formula is as follows:

[0044]

[0045]

[0046] Where IQR is the interquartile range, UP is the upper limit and LOW is the lower limit.

[0047] The historical optical power data and meteorological data exceeding the upper boundary and the lower boundary are abnormal historical optical power data and meteorological data.

[0048] The calculation formula for normalization is:

[0049] in, x are the normalized historical optical power data and meteorological data, is the original historical optical power data and meteorological data, and They are the maximum value of the original historical optical power data and meteorological data and the minimum value of the original historical optical power data and meteorological data respectively.

[0050] The calculation formula for outlier replacement processing is:

[0051] in, is an outlier, The first k data, After the outlier k data.

[0052] S2. Cluster the pre-processed historical optical power data and meteorological data to obtain several clustering results.

[0053] Specifically, in this step, when clustering the pre-processed historical optical power data and meteorological data, the meteorological data with a greater impact on photovoltaic output is first screened out through MI (Mutual Information). The data in step S1 includes solar irradiance, photovoltaic backplane temperature, air relative humidity, conversion efficiency, photovoltaic output data, wind speed, wind direction and low-altitude cloud cover.

[0054] The meteorological data obtained through MI screening are weighted according to the importance of their influence, and a high-dimensional panel data model of influencing factors applied to photovoltaic output data clustering is constructed. The expression of the high-dimensional panel data model is as follows: (6) Where: For the The first key meteorological factor Samples in The value of the moment.

[0055] Specifically, in this step, an improved fuzzy C-means clustering algorithm is used to cluster the pre-processed historical optical power data and meteorological data to obtain several clustering results. The specific improvement of the fuzzy C-means clustering algorithm is to introduce a negative exponential variable into the objective function of the traditional fuzzy C-means clustering algorithm. , in order to achieve scaling of the original data and improve the robustness of the model.

[0056] The objective function of the improved fuzzy C-means clustering algorithm is as follows:

[0057] in, is the objective function of the improved fuzzy C-means clustering algorithm, , d is the dimension of the data, For the data set i and j The data points corresponding to the dimension.

[0058] Specifically, in this step, a clustering algorithm is used to cluster the pre-processed historical optical power data and meteorological data, and the process of obtaining several clustering results is described as follows: A. Input data X, cluster center C, determine the fuzzy stopping standard ɛ and fuzzy coefficient m; B. Randomly initialize the membership function value of each data; C. Calculate the cluster center. The calculation method of the cluster center is: The Lagrange multiplier method is used to solve the minimum value of the objective function of the improved fuzzy C-means clustering algorithm. The constraints are: , the minimum expression obtained and the minimum expression for c j The calculation formula for the cluster center obtained by partial derivative is as follows:

[0059]

[0060] D. Calculate the membership of each sample data point, where the membership function is the minimum value expression for u ij The calculation formula obtained by partial derivative is as follows:

[0061] E. Repeatedly adjust the cluster center matrix to update the membership matrix, obtain the best membership and cluster center, and divide the sample point clusters according to the membership. When , it is considered that the best membership is obtained.

[0062] In this step, two indicators (sum of squared errors and silhouette coefficient) are used to determine the optimal number of clusters for the clustering results. First, the sum of squared errors (SSE) is used to calculate the SSE of points and corresponding cluster centers under different numbers of clusters. The more clusters there are, the smaller the SSE value will be. The calculation formula is as follows:

[0063] in, is the sum of squared errors, C i For the i Clusters, m i For the i Cluster centers, j For the i A sample point of a cluster.

[0064] In this step, the silhouette coefficient is also used to evaluate the clustering effect. The calculation formula is as follows:

[0065] in, For the i The average distance from a sample point to other samples in the same cluster is used to indicate the cohesion of the sample points. For the i The average distance between a sample point and all samples in the adjacent cluster is used to indicate the separation degree of sample points between different clusters. The average silhouette coefficient is calculated by averaging the silhouette coefficients of all sample points. When the distance between samples in the same cluster is closer and the distance between samples in different clusters is farther, the larger the average silhouette coefficient is, indicating a better clustering effect.

[0066] S3. Decompose several clustering results into intrinsic mode function components at different frequencies, and use the intrinsic mode function components to construct training sets and test sets.

[0067] Several clustering results are decomposed into intrinsic mode function components at different frequencies, and the intrinsic mode function components are used to construct training sets and test sets.

[0068] Specifically, in this step, several clustering results are classified into three categories: sunny, rainy, and sunny to cloudy. The empirical mode decomposition method is used to decompose the several clustering results into intrinsic mode function components at different frequencies.

[0069] S4. Construct an XGBoosting model and a BiLSTM model, and use the Bayesian optimization method to improve the XGBoosting model. Based on the constructed training set, the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are trained, and based on the constructed test set, the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are tested to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method.

[0070] An XGBoosting model and a BiLSTM model are constructed, and the Bayesian optimization method is used to improve the XGBoosting model. The XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are trained based on the constructed training set, and the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method are tested based on the constructed test set to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method.

[0071] Specifically, the calculation formula for improving the XGBoosting model using the Bayesian optimization method in this step is:

[0072] in, obj is the objective function, for n The loss between the sample points and the true value, is the complexity of the tree.

[0073] S5. The meteorological data to be predicted are respectively input into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and two optical power combination prediction results are obtained.

[0074] The meteorological data to be predicted are respectively input into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and two optical power combination prediction results are obtained.

[0075] S6. Dynamically adjust the weights of the final outputs of the two optical power combination prediction results to obtain the final optical power combination prediction result.

[0076] Specifically, in this step, an adaptive weighting method is used to dynamically adjust the weights of the two optical power combination prediction results that are finally output, so as to obtain the final optical power combination prediction result.

[0077]

[0078]

[0079] in, and These are the output results of the two models. is the weight of the model in the prediction results, For the forgetting factor, is the sample set data, ∈[0, 1], The larger the value of, the smaller the weight of the previous moment is retained. , which means that the smaller the prediction error of the state value at time k is, the more weight of time k will be retained at time k+1.

[0080] In order to verify the effectiveness of the optical power combination prediction method proposed in the present invention, this embodiment selects 35040 data of a photovoltaic power station in Yunnan in 2023 for prediction, and uses 90% of the 35040 data of a photovoltaic power station in Yunnan in 2023 as a training set and 10% as a test set. This embodiment uses the root mean square error (RMSE) and the mean absolute error (MAE) to evaluate the optical power combination prediction results, and the evaluation results are as follows: Under the sunny day type data, the existing prediction methods (such as similar day clustering and BiLSTM) have an RMSE of 6.03 and a MAE of 4.85. The optical power combination prediction method proposed in the present invention has an RMSE of 5.36 and a MAE of 4.15.

[0081] Under cloudy data, the existing prediction methods (such as similar day clustering and BiLSTM) have an RMSE of 7.89 and a MAE of 6.73. The optical power combination prediction method proposed in this invention has an RMSE of 6.48 and a MAE of 5.26.

[0082] For rainy day data, the existing prediction methods (such as similar day clustering and BiLSTM) have an RMSE of 4.18 and a MAE of 3.19. The optical power combination prediction method proposed in the present invention has an RMSE of 2.91 and a MAE of 2.06.

[0083] According to the above evaluation results, it can be seen that under the three types of data: sunny day type, cloudy day type and rainy day type, the RMSE and MAE values ​​of the optical power combination prediction method proposed in the present invention are smaller, which verifies the effectiveness of the prediction method of the present invention.

[0084] Embodiment 4: See also Figure 3As shown, the present invention also provides an electronic device 100 for an optical power combination prediction method; the electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0085] The memory 101 can be used to store the computer program 103. The processor 102 implements the steps of the optical power combination prediction method described in Example 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data (such as audio data) created according to the use of the electronic device 100, etc. In addition, the memory 101 can include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices.

[0086] The at least one processor 102 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or any conventional processor, etc. The processor 102 is the control center of the electronic device 100, and uses various interfaces and lines to connect various parts of the entire electronic device 100.

[0087] The memory 101 in the electronic device 100 stores a plurality of instructions to implement a method for evaluating the storage life of an electronic device, and the processor 102 can execute the plurality of instructions to implement: Embodiment 5: If the module / unit integrated in the electronic device 100 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 present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory and read-only memory (ROM, Read-Only Memory).

[0088] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, 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 code.

[0089] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to 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 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 device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0090] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing 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.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting optical power combination, characterized in that: The following steps are involved: Acquire historical optical power data and meteorological data, and pre-process the obtained historical optical power data and meteorological data; Clustering the pre-processed historical optical power data and meteorological data to obtain several clustering results; Decomposing several clustering results into intrinsic mode function components at different frequencies, and using the intrinsic mode function components to construct training sets and test sets; Construct an XGBoosting model and a BiLSTM model, and use the Bayesian optimization method to improve the XGBoosting model. Based on the constructed training set, train the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method. Based on the constructed test set, test the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method. The meteorological data to be predicted are respectively input into the trained XGBoosting model and BiLSTM model improved by Bayesian optimization method to perform optical power combination prediction, and two optical power combination prediction results are obtained; The weights of the final outputs of the two optical power combination prediction results are dynamically adjusted to obtain the final optical power combination prediction result.

2. The optical power combination prediction method according to claim 1, characterized in that: The preprocessing refers to first normalizing the obtained historical optical power data and meteorological data and then performing outlier replacement processing.

3. The optical power combination prediction method according to claim 2, characterized in that: The calculation formula for the normalization process is: in, x are the normalized historical optical power data and meteorological data, is the original historical optical power data and meteorological data, and are the maximum values ​​of the original historical optical power data and meteorological data and the minimum values ​​of the original historical optical power data and meteorological data respectively; The calculation formula for the outlier replacement process is: in, is an outlier, The first k data, After the outlier k data.

4. The optical power combination prediction method according to claim 1, characterized in that: The calculation formula for improving the XGBoosting model using the Bayesian optimization method is: in, obj is the objective function, for n The loss between the sample points and the true value, is the complexity of the tree.

5. The optical power combination prediction method according to claim 1, characterized in that: Adopting adaptive weighting method to dynamically adjust the weights of the final output of the two optical power combination prediction results to obtain the final optical power combination prediction result; The calculation formula of the adaptive weighting method is: in, and are the output results of the two models. is the weight of the model in the prediction results, For the forgetting factor, is the sample set data.

6. The optical power combination prediction method according to claim 1, characterized in that: The improved fuzzy C-means clustering algorithm is used to cluster the preprocessed historical optical power data and meteorological data, and several clustering results are obtained.

7. The optical power combination prediction method according to claim 1, characterized in that: The empirical mode decomposition method is used to decompose several clustering results into intrinsic mode function components at different frequencies.

8. An optical power combination prediction system, characterized in that: It includes a data acquisition module, a data preprocessing module, a data clustering module, a data set construction module, a model training module, an optical power combination prediction module and a weight dynamic adjustment module; The data acquisition module is used to acquire historical optical power data and meteorological data; The data preprocessing module is used to preprocess the obtained historical optical power data and meteorological data; The data clustering module is used to cluster the pre-processed historical optical power data and meteorological data to obtain several clustering results; The data set construction module is used to decompose a number of clustering results into intrinsic mode function components at different frequencies, and use the intrinsic mode function components to construct a training set and a test set; The model training module is used to construct an XGBoosting model and a BiLSTM model, and improve the XGBoosting model using a Bayesian optimization method, train the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method based on the constructed training set, and test the training results of the XGBoosting model and the BiLSTM model improved by the Bayesian optimization method based on the constructed test set, to obtain the trained XGBoosting model and the BiLSTM model improved by the Bayesian optimization method; The optical power combination prediction module is used to input the meteorological data to be predicted into the trained XGBoosting model and BiLSTM model improved by the Bayesian optimization method to perform optical power combination prediction, and obtain two optical power combination prediction results; The weight dynamic adjustment module is used to dynamically adjust the weights of the two optical power combination prediction results finally output to obtain the final optical power combination prediction result.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the optical power combination prediction method according to any one of claims 1 to 7 are implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the optical power combination prediction method according to any one of claims 1 to 7 are implemented.

Images