New energy power generation power prediction method, device, equipment, medium and program product

Abstract

The invention relates to the technical field of electronics, and discloses a new energy power generation power prediction method, device and equipment, a medium and a program product. According to the method, a weight coefficient optimization model is solved by using first actual power generation power data of a new energy station in a first time period and first predicted power generation power data of the new energy station in the first time period output by each target power prediction model, so that optimal weight coefficients corresponding to different target power prediction models are obtained; the optimal weight coefficient enables the overall prediction error value to be minimum, and the target predicted power generation power data of the new energy station in the second time period is determined according to the weight coefficient of each target power prediction model and the predicted power generation power data of the second time period, so that the power prediction precision of the new energy station is effectively improved. The problem that the prediction results are not accurate enough due to the fact that the prediction results of different models are linearly or non-linearly combined through previous experience to obtain the fused prediction result in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of electronic technology, and in particular to a method, device, equipment, medium and program product for predicting new energy power generation. Background Art

[0002] Renewable energy sources, such as wind and solar power, are characterized by intermittent, random, and volatile performance. With the rapid development and application of renewable energy, the output of renewable energy stations is becoming unstable, posing a series of challenges to grid scheduling, peak regulation, and safe operation. Accurate power forecasting is crucial for ensuring a stable energy supply and optimized power system operation. For renewable energy stations, in the context of new power system development, power forecast errors can lead to significant assessment costs and negatively impact their earnings in the electricity spot market. To improve power forecast accuracy, research on renewable energy prediction algorithms is intensifying. Currently, mainstream algorithms include physical methods, time series methods, and artificial intelligence approaches. However, single prediction algorithms have limitations and cannot adapt well to varying terrain, climate, and seasonal variations. Combining multiple algorithms can help improve the accuracy and robustness of power forecasts.

[0003] In related technologies, the fusion of power prediction-related algorithm models should be concentrated at the underlying algorithm level, that is, the prediction results of multiple basic algorithms such as support vector machines, random forests or neural networks are integrated to determine the final prediction results. When integrating the prediction results of different algorithm models, this fusion method generally performs linear or nonlinear combinations of the prediction results of different models according to preset rules to obtain the fusion results. The preset rules are generally determined based on past experience and have a certain degree of randomness, which leads to the final prediction results being not accurate enough. Summary of the Invention

[0004] In view of this, the present invention provides a new energy power generation power prediction method, device, equipment, medium and program product to solve the problem in related technologies that the prediction results of different models are linearly or nonlinearly combined based on past experience to obtain a fused prediction result, resulting in inaccurate prediction results.

[0005] In a first aspect, the present invention provides a method for predicting renewable energy power generation, the method comprising: obtaining target power prediction models from different sources, first predicted power generation data of a renewable energy station output by each target power prediction model in a first time period, first actual power generation data of the renewable energy station in the first time period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model; inputting the first meteorological characteristic data and the first actual power generation data into each target power prediction model so that the corresponding target power prediction model outputs second predicted power generation data of the renewable energy station in a second time period, the second time period being later than the first time period; using the first actual power generation data and the first predicted power generation data output by each target power prediction model to solve a pre-constructed weight coefficient optimization model to obtain weight coefficients corresponding to different target power prediction models, the weight coefficient optimization model being constructed with the goal of minimizing the overall prediction error value, the overall prediction error value being calculated using the first actual power generation data, the weight coefficients of each target power prediction model, and the first predicted power generation data output by each target power prediction model; and calculating the target predicted power generation data of the renewable energy station in the second time period based on the weight coefficients and predicted power generation data corresponding to different target power prediction models.

[0006] The method for predicting power generation of new energy provided by the present invention solves a pre-built weight coefficient optimization model using the first actual power generation data of the new energy station in the first period and the first predicted power generation data output by each target power prediction model to obtain the weight coefficients corresponding to the different target power prediction models, and calculates the target predicted power generation data of the new energy station in the second period based on the weight coefficients and predicted power generation data corresponding to the different target power prediction models. The method provided by the present invention solves the weight coefficient optimization model using the first actual power generation data of the new energy station in the first period and the first predicted power generation data of the new energy station in the first period output by each target power prediction model to obtain the optimal weight coefficients corresponding to the different target power prediction models, the optimal weight coefficients can minimize the overall prediction error value, and the target predicted power generation data of the new energy station in the second period is determined based on the weight coefficients of each target power prediction model and the predicted power generation data of the second period, effectively improving the power prediction accuracy of the new energy station, and solving the problem in the related art of obtaining a fused prediction result by linearly or nonlinearly combining the prediction results of different models based on past experience, resulting in inaccurate prediction results.

[0007] In an optional embodiment, the target power prediction model is obtained through the following steps: obtaining power prediction models from multiple different sources and actual power generation data for a second preset time period, preset weight values ​​of multiple verification indicators, meteorological characteristic data for a first preset time period, and actual power generation data; inputting the meteorological characteristic data and actual power generation data for the first preset time period into each power prediction model so that the corresponding power prediction model outputs predicted power generation data for the second preset time period; calculating multiple verification indicator values ​​of the corresponding power prediction model based on the actual power generation data for the second preset time period and the predicted power generation data for the second preset time period output by each power prediction model; calculating an evaluation score of the corresponding power prediction model based on the multiple verification indicator values ​​of each power prediction model and the weight value of each verification indicator; and determining the power prediction model whose evaluation score is greater than a preset threshold as the target power prediction model.

[0008] The method provided by this optional implementation method calculates the evaluation score of each power prediction model through a pre-set verification indicator, determines the power prediction model with an evaluation score greater than a preset threshold as the target power prediction model, and eliminates incorrect models from power prediction models from different sources, which helps to give full play to the advantages of strong models and improve power prediction accuracy.

[0009] In an optional embodiment, the first predicted power generation data of the new energy station in the first time period output by each target power prediction model is determined by the following steps: obtaining the third meteorological characteristic data and the third actual power generation data of the new energy station in the third time period, the third time period being earlier than the first time period; inputting the third meteorological characteristic data and the third actual power generation data into each target power prediction model, so that the corresponding target power prediction model outputs the first predicted power generation data of the new energy station in the first time period.

[0010] In an optional embodiment, the method also includes: obtaining second measured power generation data of the new energy station in the second time period; calculating an evaluation value based on the second measured power generation data and the target predicted power generation data; and performing accuracy evaluation on the power generation prediction result of the new energy station based on the evaluation value.

[0011] In an optional embodiment, a power prediction model is constructed by the following steps: obtaining a historical meteorological feature sequence data set, a historical actual power generation sequence data set, and an initial model provided by a target manufacturer of a new energy site; dividing the historical meteorological feature sequence data set and the historical actual power generation sequence data set using a preset time step as a sliding window to obtain first training data; dividing the historical actual power generation sequence data set using a preset time step as a sliding window to obtain second training data; associating the first training data and the second training data according to prediction requirements to construct an associated data set; and using the associated data set to train the initial model until the model accuracy meets the preset requirements to obtain a power prediction model.

[0012] In the second aspect, the present invention provides a new energy power generation prediction device, which includes: a first acquisition module for acquiring target power prediction models from different sources, first predicted power generation data of the new energy station in the first time period output by each target power prediction model, first actual power generation data of the new energy station in the first time period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model; a first determination module for inputting the first meteorological characteristic data and the first actual power generation data into each target power prediction model so that the corresponding target power prediction model outputs the second predicted power generation data of the new energy station in the second time period, and the second time period is later than the first time period; solving A module is used to solve a pre-constructed weight coefficient optimization model using the first actual power generation data and the first predicted power generation data output by each target power prediction model to obtain the weight coefficients corresponding to different target power prediction models. The weight coefficient optimization model is constructed with the goal of minimizing the overall prediction error value. The overall prediction error value is calculated using the first actual power generation data, the weight coefficients of each target power prediction model, and the first predicted power generation data output by each target power prediction model; a first calculation module is used to calculate the target predicted power generation data of the new energy station in the second time period based on the weight coefficients and predicted power generation data corresponding to different target power prediction models.

[0013] In an optional embodiment, the target power prediction model is obtained through the following steps: obtaining power prediction models from multiple different sources and actual power generation data for a second preset time period, preset weight values ​​of multiple verification indicators, meteorological characteristic data for a first preset time period, and actual power generation data; inputting the meteorological characteristic data and actual power generation data for the first preset time period into each power prediction model so that the corresponding power prediction model outputs predicted power generation data for the second preset time period; calculating multiple verification indicator values ​​of the corresponding power prediction model based on the actual power generation data for the second preset time period and the predicted power generation data for the second preset time period output by each power prediction model; calculating an evaluation score of the corresponding power prediction model based on the multiple verification indicator values ​​of each power prediction model and the weight value of each verification indicator; and determining the power prediction model whose evaluation score is greater than a preset threshold as the target power prediction model.

[0014] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to execute the new energy power generation power prediction method of the first aspect or any corresponding embodiment thereof.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the new energy power generation power prediction method of the above-mentioned first aspect or any corresponding embodiment thereof.

[0016] In a fifth aspect, the present invention provides a computer program product comprising computer instructions for enabling a computer to execute the new energy power generation prediction method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 is a flow chart of a method for predicting renewable energy power generation according to an embodiment of the present invention;

[0019] Figure 2 is a flow chart of another method for predicting power generation from renewable energy according to an embodiment of the present invention;

[0020] Figure 3This is a structural block diagram of a specific example in the embodiments of the present application;

[0021] Figure 4 This is a flow chart of a specific example in the embodiments of the present application;

[0022] Figure 5 is a structural block diagram of a new energy power generation prediction device according to an embodiment of the present invention;

[0023] Figure 6 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0025] In related technologies, the fusion of power prediction-related algorithm models should be concentrated at the underlying algorithm level, that is, the prediction results of multiple basic algorithms such as support vector machines, random forests or neural networks are integrated to determine the final prediction results. When integrating the prediction results of different algorithm models, this fusion method generally performs linear or nonlinear combinations of the prediction results of different models according to preset rules to obtain the fusion results. The preset rules are generally determined based on past experience and have a certain degree of randomness, which leads to the final prediction results being not accurate enough.

[0026] In view of this, a new energy power generation prediction method provided in an embodiment of the present application can be applied to a server to realize the power generation prediction of a new energy station. The method provided by the present invention uses the first actual power generation data of the new energy station in the first time period and the first predicted power generation data of the new energy station in the first time period output by each target power prediction model to solve the weight coefficient optimization model, and obtains the optimal weight coefficients corresponding to different target power prediction models. The optimal weight coefficient can minimize the overall prediction error value, and determines the target predicted power generation data of the new energy station in the second time period according to the weight coefficients of each target power prediction model and the predicted power generation data of the second time period, effectively improving the power prediction accuracy of the new energy station, and solving the problem in the related art that the prediction results of different models are linearly or nonlinearly combined through past experience to obtain the fused prediction results, resulting in inaccurate prediction results.

[0027] According to an embodiment of the present invention, an embodiment of a method for predicting power generation from renewable energy is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] In this embodiment, a new energy power generation power prediction method is provided, which can be used for the above-mentioned server. Figure 1 FIG. 1 is a flow chart of a method for predicting new energy power generation according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0029] Step S101: Obtain target power prediction models from different sources, first predicted power generation data of new energy sites in the first time period output by each target power prediction model, first actual power generation data of new energy sites in the first time period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model.

[0030] Exemplarily, the target power prediction models referred to by different sources may be power prediction models provided by different manufacturers that meet the prediction requirements. The target power prediction models may include but are not limited to short-term prediction models, ultra-short-term prediction models, etc. The underlying algorithm adopted by the model is not restricted by the module, and the power prediction manufacturer can choose it at will. The new energy station can be any new energy station that needs to perform power generation prediction. The first meteorological characteristic data can be the meteorological characteristic time series data of the new energy station in the first time period. The first time period can be any time period that needs to perform power generation prediction. The first actual power generation data can be the power time series data measured by the new energy station in the first time period. The first actual power generation data can be obtained by collecting the power acquisition module in advance. The first predicted power generation data is the power time series data of the new energy station in the first time period predicted by the target power prediction model.

[0031] Step S102: input the first meteorological characteristic data and the first actual power generation data into each target power prediction model, so that the corresponding target power prediction model outputs second predicted power generation data of the new energy station in a second period, which is later than the first period.

[0032] For example, in an embodiment of the present application, the first meteorological characteristic data and the first actual power generation data of the new energy station in the first time period are input into each target power prediction model, and the corresponding target power prediction model will output the second predicted power generation data of the new energy station in the second time period.

[0033] Step S103 : solving a pre-built weight coefficient optimization model using the first actual generated power data and the first predicted generated power data output by each target power prediction model to obtain weight coefficients corresponding to different target power prediction models.

[0034] Exemplarily, the weight coefficient optimization model is constructed with the goal of minimizing the overall prediction error value, and the overall prediction error value is calculated using the first actual generated power data, the weight coefficients of each target power prediction model, and the first predicted generated power data output by each target power prediction model. In an embodiment of the present application, the weight coefficient optimization model is solved using the first actual generated power data and the first predicted generated power data output by each target power prediction model to obtain the optimal solution of the model. The optimal solution minimizes the overall prediction error value, and the optimal solution is used to characterize the weight coefficients corresponding to different target power prediction models. In an embodiment of the present application, the weight coefficient optimization model can be expressed by the following formula:

[0035]

[0036] In the formula, r represents the overall prediction error value, P at represents the actual power generation of the new energy station at time t, P it represents the predicted generated power output by the i-th target power prediction model at time t, α i Represents the weight coefficient of the i-th target power prediction model.

[0037] Step S104 , calculating the target predicted power generation data of the new energy station in the second time period based on the weight coefficients and predicted power generation data corresponding to different target power prediction models.

[0038] For example, in the embodiment of the present application, the target predicted power generation data of the new energy station in the second period is calculated by the following formula:

[0039]

[0040] Where, P p Represents the target predicted power generation data, P i represents the predicted power generation output by the i-th target power prediction model, α i Represents the weight coefficient of the i-th target power prediction model.

[0041] The new energy power generation prediction method provided in this embodiment uses the first actual power generation data of the new energy station in the first time period and the first predicted power generation data of the new energy station in the first time period output by each target power prediction model to solve the weight coefficient optimization model, and obtains the optimal weight coefficients corresponding to different target power prediction models. The optimal weight coefficient can minimize the overall prediction error value, and determines the target predicted power generation data of the new energy station in the second time period according to the weight coefficients of each target power prediction model and the predicted power generation data in the second time period, effectively improving the power prediction accuracy of the new energy station, and solving the problem in related technologies of obtaining a fused prediction result by linearly or nonlinearly combining the prediction results of different models through past experience, resulting in inaccurate prediction results.

[0042] In this embodiment, a new energy power generation power prediction method is provided, which can be used for the above-mentioned server. Figure 2 FIG. 1 is a flow chart of a method for predicting new energy power generation according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0043] Step S201: Obtain target power prediction models from different sources, first predicted power generation data of new energy stations in the first period output by each target power prediction model, first actual power generation data of new energy stations in the first period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0044] In some optional implementations, the target power prediction model is obtained by the following steps:

[0045] Step a1: obtain power prediction models from multiple different sources, actual power generation data for a second preset period, weight values ​​of multiple preset verification indicators, meteorological characteristic data for a first preset period, and actual power generation data.

[0046] For example, in the embodiments of the present application, multiple power prediction models from different sources can be trained power prediction models uploaded and deployed by power prediction manufacturers in accordance with system specifications, including ultra-short-term prediction models, short-term prediction models, and other models. The underlying algorithms used by the models are not restricted by the module and can be freely selected by the power prediction manufacturer. In addition, the manufacturer can choose to upload and deploy both the ultra-short-term prediction model and the short-term prediction model simultaneously, or deploy only one of them.

[0047] Step a2: inputting meteorological characteristic data and actual power generation data of the first preset period into each power prediction model, so that the corresponding power prediction model outputs predicted power generation data of the second preset period.

[0048] For example, in the embodiment of the present application, the first preset time period can be any time period. The embodiment of the present application does not limit the specific content of the first preset time period, and those skilled in the art can determine it according to needs.

[0049] Step a3: calculating a plurality of verification index values ​​corresponding to the power prediction model based on the actual generated power data of the second preset period and the predicted generated power data of the second preset period output by each power prediction model.

[0050] For example, in this embodiment of the present application, the multiple verification indicators may include, but are not limited to, a primary verification indicator and at least one secondary verification indicator. In this embodiment of the present application, the evaluation indicator system integrates a variety of prediction deviation evaluation indicators, which can adaptively complete the accuracy assessment of the multi-source power prediction model based on the business needs of the new energy station. Specifically, the disclosed pre-verification evaluation indicator system includes Mode M1 and Mode M2.

[0051] Mode M1 is a dual-rule assessment mode. In this mode, the system automatically uses the dual-rule prediction deviation assessment formula for the region where the new energy station F is located as the pre-verification main indicator. The default deviation formula is:

[0052]

[0053] Where, f m1 is the double-rule prediction deviation value, n is the total number of power prediction points in time period T; P p To predict the power generation data; P a P is the actual power generation data of the station; pi 、P ai P p ,P a The i-th value, C F To increase the operational capacity of new energy stations.

[0054] Mode M2 ​​is the power trading mode. In this mode, the system mainly uses the closeness between the predicted power curve and the actual power curve as the measurement standard. Specifically, the average relative error of the predicted value is used as the pre-verification main indicator. The calculation formula is:

[0055]

[0056] Where, f m2 represents the average relative error, and the meanings of other variables are not repeated here.

[0057] In addition to the main indicators, the disclosed system evaluation system also includes a set of secondary verification indicators f s , including but not limited to: Pearson correlation coefficient, R 2 Coefficient of determination and root mean square error.

[0058] The calculation formula of Pearson correlation coefficient is as follows:

[0059]

[0060] Where, f s1 represents the Pearson correlation coefficient of the secondary verification indicator, Represents i predicted power generation data P pi The mean of Represents i actual power generation data P pi The meanings of other variables are not described in detail. 2 The formula for calculating the coefficient of determination is as follows:

[0061]

[0062] Among them, f s2 Represents the secondary verification index R 2 The meanings of the other variables will not be elaborated.

[0063] The formula for calculating the root mean square error is as follows:

[0064]

[0065] Among them, f s3 represents the root mean square error, and the meanings of other variables are not repeated here.

[0066] Step a4: Calculate the evaluation score of the corresponding power prediction model based on the multiple verification index values ​​of each power prediction model and the weight value of each verification index.

[0067] For example, in the embodiment of the present application, the evaluation score of the power prediction model can be calculated by the following formula:

[0068]

[0069] Where S x Indicates the evaluation score, represents the normalized main verification indicator, represents the first secondary verification indicator after normalization, represents the normalized k-th secondary verification indicator; ω m Represents the weight value of the main verification indicator, ω s1 Represents the weight value of the first secondary verification indicator, ω sk Represents the weight value of the kth secondary verification indicator.

[0070] Step a5: Determine the power prediction model with an evaluation score greater than a preset threshold as the target power prediction model.

[0071] Exemplarily, the preset threshold can be set based on experience, and the embodiment of the present application does not limit the specific value of the preset threshold.

[0072] In some optional implementations, the power prediction model is constructed by the following steps:

[0073] Step b1: Obtain the historical meteorological characteristic sequence data set, the historical actual power generation sequence data set of the new energy station, and the initial model provided by the target manufacturer.

[0074] For example, the historical meteorological characteristic sequence dataset includes historical meteorological characteristic time series data for new energy stations, and the historical actual power generation sequence dataset includes historical actual power generation time series data for new energy stations. The initial model can be a trained underlying algorithm model. This embodiment of the application does not limit the specific content of the initial model, and those skilled in the art can determine it based on their needs.

[0075] Step b2: using a preset time step as a sliding window to divide the historical meteorological feature sequence data set and the historical actual power generation sequence data set to obtain first training data.

[0076] Exemplarily, the preset time step is determined based on the predicted duration. The present embodiment does not limit the specific content of the preset time step, and those skilled in the art can determine it as needed. In the present embodiment, the historical meteorological feature sequence data set is partitioned using a sliding window to obtain multiple segments of meteorological feature sequence data and multiple segments of actual power generation sequence data as the first training data. The preset time step may include, but is not limited to, one day.

[0077] Step b3: using the preset time step as a sliding window to divide the historical actual power generation sequence data set to obtain second training data.

[0078] For example, in an embodiment of the present application, the historical actual power generation sequence data set is divided by a sliding window to obtain multiple segments of actual power generation sequence data as the second training data.

[0079] Step b4: Associating the first training data and the second training data according to the prediction requirements to construct an associated data set.

[0080] Exemplarily, in an embodiment of the present application, the first training data and the second training data are associated based on prediction requirements. For example, if it is necessary to use the power data and meteorological data of the previous day to predict the power data of the next day, the meteorological characteristic sequence data and the actual power generation power sequence data of the previous day are associated with the actual power generation power sequence data of the next day to obtain associated data, and associated data sets are obtained based on different associated data.

[0081] Step b5: Use the associated data set to train the initial model until the model accuracy meets the preset requirements to obtain a power prediction model.

[0082] For example, in the embodiment of the present application, the preset requirements can be determined according to the needs, and the embodiment of the present application does not limit the specific content of the preset requirements.

[0083] In an optional embodiment, the first predicted power generation data of the new energy station in the first time period output by each target power prediction model is determined by the following steps:

[0084] Step c1: Acquire third meteorological characteristic data and third actual power generation data of the new energy station in a third time period.

[0085] Exemplarily, the third time period is earlier than the first time period.

[0086] Step c2: input the third meteorological characteristic data and the third actual power generation data into each target power prediction model, so that the corresponding target power prediction model outputs the first predicted power generation data of the new energy station in the first time period.

[0087] Step S202: Input the first meteorological characteristic data and the first actual power generation data into each target power prediction model, so that the corresponding target power prediction model outputs the second predicted power generation data of the new energy station in the second period, which is later than the first period. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0088] Step S203: Use the first actual generated power data and the first predicted generated power data output by each target power prediction model to solve the pre-built weight coefficient optimization model to obtain the weight coefficients corresponding to the different target power prediction models. The weight coefficient optimization model is constructed with the goal of minimizing the overall prediction error value. The overall prediction error value is calculated using the first actual generated power data, the weight coefficients of each target power prediction model, and the first predicted generated power data output by each target power prediction model. For details, please refer to Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0089] Step S204: Calculate the target predicted power generation data of the new energy station in the second period based on the weight coefficients and predicted power generation data corresponding to the different target power prediction models. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0090] Step S205: Acquire the second measured power generation data of the new energy station in the second time period.

[0091] Exemplarily, the second measured power generation data is collected by a power generation acquisition module pre-set at the new energy station. The embodiment of the present application does not limit the specific content of the power generation acquisition module, as long as it is reasonable.

[0092] Step S206: Calculate an evaluation value based on the second measured power generation data and the target predicted power generation data.

[0093] For example, in an embodiment of the present application, the evaluation value is determined based on the deviation between the second measured power generation data and the target predicted power generation data. The embodiment of the present application does not limit the calculation method of the evaluation value, and those skilled in the art can determine it according to needs.

[0094] Step S207: performing an accuracy evaluation on the power generation prediction result of the new energy station based on the evaluation value.

[0095] Exemplarily, the accuracy of the power generation prediction result is evaluated by the evaluation value. The embodiment of the present application does not limit the evaluation method, and those skilled in the art can determine it according to needs.

[0096] The new energy power generation power prediction method provided by the present invention is described below through a specific embodiment.

[0097] Example:

[0098] The new energy power generation power prediction method provided in the embodiment of the present application is applied to the new energy multi-source time-sharing adaptive fusion power prediction system, such as Figure 3 As shown, the multi-source time-sharing adaptive fusion prediction system for renewable energy power generation provided by the embodiment of the present invention includes five parts: a renewable energy power prediction manufacturer model deployment module, a renewable energy site data organization module, a model pre-verification module, a time-sharing adaptive fusion power prediction module, and a fusion model post-evaluation module. The composition and functions of each module are as follows:

[0099] (1) New energy power prediction manufacturer model deployment module ①. The system disclosed in the present invention is characterized in that it simultaneously accepts power prediction services provided by multiple manufacturers for specific new energy sites. In module ①, each power prediction manufacturer uploads and deploys trained power prediction models in accordance with system specifications, including ultra-short-term prediction models, short-term prediction models, etc. The underlying algorithm used by the model is not restricted by the module, and the power prediction manufacturer can choose it at will. In addition, the manufacturer can choose to upload and deploy the ultra-short-term prediction model and the short-term prediction model at the same time, or deploy only one of them.

[0100] (2) New energy station data organization module ②. Module ② obtains historical and real-time power generation data from designated new energy stations through data communication protocols and stores them. It also provides historical power generation data and real-time power generation data for training and verification as needed by other modules.

[0101] (3) Model pre-verification module ③. Module ③ presets a set of power forecast accuracy evaluation indicators. Its function is to call and organize the historical power generation data within a specified period provided by module ② based on the main indicators and sub-indicators selected by the system administrator, verify and score the power forecast models deployed by various manufacturers, and select several models with the best forecast accuracy to form a short-term forecast pre-selected model set and an ultra-short-term forecast pre-selected model set.

[0102] (4) Time-based adaptive fusion power prediction module ④. Based on the ultra-short-term and short-term pre-selected model sets, module ④ first constructs a power prediction fusion model, then calls and organizes the historical power generation data within the specified time period provided by module ②, determines the weights of each sub-model in the fusion model through a weight optimization algorithm, and uses the optimized fusion model to perform the power generation prediction task.

[0103] (5) Post-fusion model evaluation module ⑤. Based on the power prediction accuracy evaluation index set, module ⑤ evaluates the performance of the fusion model so as to issue an alarm when the accuracy does not meet the requirements, reminding the system administrator to adjust and update the fusion model in time.

[0104] The disclosed multi-source, time-sharing, adaptive fusion prediction system for renewable energy power generation is characterized by using automatic or manual modes to trigger model fusion instructions, complete fusion model construction or update, and perform multi-source fusion power prediction. Trigger modes include but are not limited to:

[0105] Mode M1: Scheduled automatic update, that is, the fusion power prediction system triggers the model fusion instruction at fixed time intervals, where the time intervals include annual, quarterly, monthly, ten-day, weekly, daily and intra-day time periods; or any two or more time scales are combined to form scheduled fusion instructions with irregular time intervals; for short-term predictions, the default time interval for the system's scheduled automatic update is monthly; for ultra-short-term fusion predictions, the default time interval for the system's scheduled automatic update is weekly.

[0106] Mode M2: Manual update, that is, the system administrator manually triggers the model fusion instruction when needed based on the running status of the fusion model.

[0107] When the model fusion instruction is triggered, the fusion power prediction system workflow is as follows:

[0108] 1. The new energy power forecasting service provider obtains the historical power generation data of the designated new energy site from Module ②, trains a dedicated short-term power forecasting model and / or ultra-short-term power forecasting model for the site, and uploads and deploys it in Module ①.

[0109] 2. Based on the deviation assessment rules for the location of the new energy station and other preset rules, the system automatically selects the main verification indicator and several secondary indicators in the indicator evaluation system of module ③; obtains the historical power generation data of the site within the specified time period from module ②, and evaluates the power prediction models deployed by each manufacturer in module ① based on the selected verification indicator set, calculating the model pre-verification score; and includes models that exceed the score threshold into the short-term prediction pre-selected model set or the ultra-short-term prediction pre-selected model set;

[0110] 3. After the system receives the model fusion instruction, module ④ constructs the short-term or ultra-short-term power forecast fusion original model according to the instruction requirements; obtains the historical power generation data within the specified time period from module ②, and determines the weight of each sub-model in the fusion model through the weight optimization algorithm; after the weight optimization is completed, the fusion model is used to perform the power generation forecast task, generate short-term and ultra-short-term power forecast files, and upload them to the power grid dispatch center;

[0111] 4. After the fusion model is started, module ⑤ regularly evaluates its indicators based on the selected evaluation indicator system (the default interval is days); when the fusion model evaluation score is lower than the threshold, the system issues an alarm message.

[0112] In the embodiment of this application, the adaptability of the power generation prediction model to different accuracy evaluation indicators varies. Therefore, before multi-source integration is performed, it is necessary to verify and optimize the models of each manufacturer based on the actual indicator requirements of a specific new energy site. This embodiment of the application provides a pre-verification method for a power prediction model that integrates multiple indicators. Taking the pre-verification of a short-term prediction model as an example, the process is as follows:

[0113] (1) The operator specifies hyperparameters such as the new energy station F, the time period T for pre-verification, and the pre-verification score threshold δ in the fusion power prediction system;

[0114] (2) Module ③ obtains the actual power generation P of station F during the specified time period T through module ②. a ;

[0115] (3) Module ③ loads and runs the short-term power prediction model M deployed by manufacturer X through module ①. x , obtain the power prediction data P of the model in time period T p ;

[0116] (4) According to the policy requirements of the region where the new energy station is located or the preset needs of the system administrator, module ③ comprehensively selects the main verification index of the model fm and the secondary verification indicators f1~f k ;

[0117] (5) Module ③ predicts power data P p , actual power generation P a Calculation model M x The index values ​​f m 、f1~f k ,Further, the pre-verification comprehensive score of the model is calculated based on the values ​​of each indicator;

[0118] (6) If the model M x If the pre-verification comprehensive score exceeds the threshold δ, it will be included in the short-term prediction pre-selected model set; otherwise, the model will be discarded or retrained.

[0119] In the embodiment of the present application, the multi-source model fusion power prediction process is as follows Figure 4 As shown in the figure, the statutory power of the forecast period is predicted by using the pre-selected power forecast model provided by the model pre-verification module ③, and the forecast result is obtained by optimizing the weights of each pre-selected model to obtain the optimal weight coefficient combination α i ~α k , the optimal fusion model can be built and used for power generation prediction in the next time period.

[0120] The system provided by the embodiments of the present application comprehensively prioritizes short-term or ultra-short-term power forecasting services provided by multiple vendors, avoiding the shortcomings of using a single model and enhancing system adaptability. Based on diverse evaluation indicators, different power forecasting models are dynamically selected for fusion forecasting at different time periods, promptly eliminating weak models rather than rigidly selecting the same or several basic models. This helps further leverage the advantages of strong models and improve power forecast accuracy throughout the year and all weather conditions. Dynamically updating the fusion model weights based on historical forecasts allows for timely model corrections to adapt to the needs of different scenarios.

[0121] In this embodiment, a new energy power generation power prediction device is also provided. The device is used to implement the above-mentioned embodiments and preferred embodiments. The details that have been described will not be repeated here. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0122] This embodiment provides a new energy power generation power prediction device, such as Figure 5 Shown, including:

[0123] A first acquisition module 501 acquires target power prediction models from different sources, first predicted power generation data of a new energy station in a first period output by each target power prediction model, first actual power generation data of the new energy station in the first period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model;

[0124] A first determining module 502 is configured to input the first meteorological characteristic data and the first actual power generation data into each target power prediction model, so that the corresponding target power prediction model outputs second predicted power generation data of the new energy station in a second time period, where the second time period is later than the first time period;

[0125] A solution module 503 is configured to solve a pre-constructed weight coefficient optimization model using the first actual generated power data and the first predicted generated power data output by each target power prediction model to obtain weight coefficients corresponding to different target power prediction models. The weight coefficient optimization model is constructed with the goal of minimizing an overall prediction error value, which is calculated using the first actual generated power data, the weight coefficients of each target power prediction model, and the first predicted generated power data output by each target power prediction model.

[0126] The first calculation module 504 is configured to calculate the target predicted power generation data of the new energy station in the second period based on the weight coefficients and predicted power generation data corresponding to different target power prediction models.

[0127] In some optional implementations, the target power prediction model is obtained by the following steps:

[0128] Obtaining power prediction models from multiple different sources, actual power generation data for a second preset period, weight values ​​of multiple preset verification indicators, meteorological characteristic data for a first preset period, and actual power generation data;

[0129] Inputting meteorological characteristic data and actual power generation data of a first preset period into each power prediction model so that the corresponding power prediction model outputs predicted power generation data of a second preset period;

[0130] Calculating a plurality of verification index values ​​corresponding to the power prediction model based on the actual generated power data of the second preset period and the predicted generated power data of the second preset period output by each power prediction model;

[0131] Calculating an evaluation score of the corresponding power prediction model based on multiple verification indicator values ​​of each power prediction model and a weight value of each verification indicator;

[0132] The power prediction model with an evaluation score greater than a preset threshold is determined as the target power prediction model.

[0133] In some optional implementations, the first predicted power generation data of the new energy station in the first time period output by each target power prediction model is determined by the following steps:

[0134] Obtaining third meteorological characteristic data and third actual power generation data of the new energy station in a third time period, where the third time period is earlier than the first time period;

[0135] The third meteorological characteristic data and the third actual power generation data are input into each target power prediction model, so that the corresponding target power prediction model outputs the first predicted power generation data of the new energy station in the first time period.

[0136] In an optional embodiment, the device further comprises:

[0137] The second acquisition module is used to obtain the second measured power generation data of the new energy station in the second time period;

[0138] A second calculation module is used to calculate an evaluation value based on the second measured power generation data and the target predicted power generation data;

[0139] The evaluation module is used to evaluate the accuracy of the power generation prediction results of the new energy station based on the evaluation value.

[0140] In an optional implementation, the power prediction model is constructed by the following steps:

[0141] Obtain historical meteorological characteristic sequence data sets, historical actual power generation sequence data sets, and initial models provided by target manufacturers for new energy stations;

[0142] The historical meteorological feature sequence data set and the historical actual power generation sequence data set are divided using a preset time step as a sliding window to obtain first training data;

[0143] The historical actual power generation sequence data set is divided using a preset time step as a sliding window to obtain second training data;

[0144] Associating the first training data and the second training data according to the prediction requirements to construct an associated data set;

[0145] The initial model is trained using the associated data set until the model accuracy meets the preset requirements, and a power prediction model is obtained.

[0146] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0147] The new energy power generation prediction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0148] The embodiment of the present invention also provides a computer device having the above Figure 5 The new energy power generation power prediction device shown.

[0149] See also Figure 6 , Figure 6 is a structural diagram of a computer device provided by an optional embodiment of the present invention, such as Figure 6 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of a GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0150] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0151] The memory 20 stores instructions that can be executed by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0152] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0153] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0154] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0155] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0156] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0157] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for predicting new energy power generation, characterized in that: The method comprises: Obtain target power prediction models from different sources, first predicted power generation data of the new energy station in the first period output by each target power prediction model, first actual power generation data of the new energy station in the first period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model; inputting the first meteorological characteristic data and the first actual power generation data into each target power prediction model so that the corresponding target power prediction model outputs second predicted power generation data of the new energy station in a second time period, where the second time period is later than the first time period; using the first actual generated power data and the first predicted generated power data output by each target power prediction model to solve a pre-constructed weight coefficient optimization model to obtain weight coefficients corresponding to different target power prediction models, wherein the weight coefficient optimization model is constructed with the goal of minimizing an overall prediction error value, and the overall prediction error value is calculated using the first actual generated power data, the weight coefficients of each target power prediction model, and the first predicted generated power data output by each target power prediction model; The target predicted power generation data of the new energy station in the second time period is calculated based on the weight coefficients corresponding to the different target power prediction models and the predicted power generation data.

2. The method according to claim 1, characterized in that The target power prediction model is obtained by the following steps: Obtaining power prediction models from multiple different sources, actual power generation data for a second preset period, weight values ​​of multiple preset verification indicators, meteorological characteristic data for a first preset period, and actual power generation data; Inputting the meteorological characteristic data and the actual power generation data of the first preset time period into each power prediction model, so that the corresponding power prediction model outputs the predicted power generation data of the second preset time period; Calculating a plurality of verification index values ​​corresponding to the power prediction model based on the actual generated power data of the second preset period and the predicted generated power data of the second preset period output by each power prediction model; Calculating an evaluation score of the corresponding power prediction model based on multiple verification indicator values ​​of each power prediction model and a weight value of each verification indicator; The power prediction model whose evaluation score is greater than a preset threshold is determined as the target power prediction model.

3. The method according to claim 1 or 2, characterized in that The first predicted power generation data of the new energy station in the first period output by each target power prediction model is determined by the following steps: Obtaining third meteorological characteristic data and third actual power generation data of the new energy station in a third time period, wherein the third time period is earlier than the first time period; The third meteorological characteristic data and the third actual power generation data are input into each target power prediction model, so that the corresponding target power prediction model outputs the first predicted power generation data of the new energy station in the first time period.

4. The method according to claim 1 or 2, characterized in that The method further comprises: Obtaining second measured power generation data of the new energy station in the second period; calculating an evaluation value based on the second measured power generation data and the target predicted power generation data; An accuracy evaluation is performed on the power generation prediction result of the new energy station based on the evaluation value.

5. The method according to claim 2, characterized in that The power prediction model is constructed by the following steps: Obtain historical meteorological characteristic sequence data sets, historical actual power generation sequence data sets, and initial models provided by target manufacturers for new energy stations; Using a preset time step as a sliding window, the historical meteorological feature sequence data set and the historical actual power generation sequence data set are divided to obtain first training data; Dividing the historical actual power generation sequence data set using a preset time step as a sliding window to obtain second training data; Associating the first training data and the second training data according to prediction requirements to construct an associated data set; The initial model is trained using the associated data set until the model accuracy meets the preset requirements, thereby obtaining the power prediction model.

6. A new energy power generation power prediction device, characterized in that: The device comprises: a first acquisition module, configured to acquire target power prediction models from different sources, first predicted power generation data of the new energy station in the first period output by each target power prediction model, first actual power generation data of the new energy station in the first period, first meteorological characteristic data, and first predicted power generation data output by each target power prediction model; a first determining module, configured to input the first meteorological characteristic data and the first actual power generation data into each target power prediction model, so that the corresponding target power prediction model outputs second predicted power generation data of the new energy station in a second time period, where the second time period is later than the first time period; a solving module, configured to solve a pre-constructed weight coefficient optimization model using the first actual generated power data and the first predicted generated power data output by each target power prediction model to obtain weight coefficients corresponding to different target power prediction models, wherein the weight coefficient optimization model is constructed with the goal of minimizing an overall prediction error value, and the overall prediction error value is calculated using the first actual generated power data, the weight coefficients of each target power prediction model, and the first predicted generated power data output by each target power prediction model; The first calculation module is used to calculate the target predicted power generation data of the new energy station in the second time period based on the weight coefficients corresponding to the different target power prediction models and the predicted power generation data.

7. The device according to claim 6, characterized in that The target power prediction model is obtained by the following steps: Obtaining power prediction models from multiple different sources, actual power generation data for a second preset period, weight values ​​of multiple preset verification indicators, meteorological characteristic data for a first preset period, and actual power generation data; Inputting the meteorological characteristic data and the actual power generation data of the first preset time period into each power prediction model, so that the corresponding power prediction model outputs the predicted power generation data of the second preset time period; Calculating a plurality of verification index values ​​corresponding to the power prediction model based on the actual generated power data of the second preset period and the predicted generated power data of the second preset period output by each power prediction model; Calculating an evaluation score of the corresponding power prediction model based on multiple verification indicator values ​​of each power prediction model and a weight value of each verification indicator; The power prediction model whose evaluation score is greater than a preset threshold is determined as the target power prediction model.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the new energy power generation power prediction method according to any one of claims 1 to 5 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the new energy power generation power prediction method according to any one of claims 1 to 5.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the new energy power generation prediction method according to any one of claims 1 to 5.

Images