New energy station power prediction technology evaluation and scheduling method, system and device

Abstract

The present application relates to a method, system and device for evaluating and scheduling power prediction technology for new energy stations. The method includes: based on the regional information of the new energy station, generating model evaluation index information of the new energy station, and based on each measured operation information and measured environmental information, predicting the operation result corresponding to each measured operation information predicted by each power prediction model; by comparing and evaluating each operation result, identifying the evaluation value of each power prediction model, thereby screening each initial target power prediction model adapted to the new energy station; based on the target model function information corresponding to the advantageous evaluation index information in each initial target power prediction model, through the model combination strategy, generating a fusion power prediction model containing all target model function information, that is, the target power prediction model of the new energy station. The present invention realizes the objective evaluation of the prediction performance of various prediction technologies and the comparison between them, and provides support for the selection of the optimal prediction technology for new energy stations.

Description

Technical Field

[0001] The present application relates to the technical field of power generation prediction of new energy sites, and in particular to a method, system and device for evaluating and scheduling power prediction technology of new energy sites. Background Art

[0002] The power generation of new energy stations has unstable characteristics such as randomness, volatility, and intermittency. With the large-scale grid connection of new energy, the difficulty of power grid dispatching agencies to formulate power generation plans has greatly increased. Therefore, predicting the power generation of new energy stations is helpful to reasonably arrange power generation plans and improve the economy, stability, and reliability of the power grid. There are various existing power prediction models, which have different adaptability to different climates and terrain environments. Although they can all predict the power of new energy stations, there is no technology to objectively evaluate and compare the prediction performance of each power prediction model, resulting in the inability of new energy stations to select the power prediction model suitable for the new energy station. Therefore, how to evaluate and compare the prediction performance of each power prediction model is the current research focus.

[0003] The prediction performance evaluation and comparison method of traditional power prediction models evaluates the prediction performance of each power prediction model by identifying the accuracy of the predicted power generation of different power prediction models. However, it only analyzes and evaluates through the predicted power generation, and is unable to select the target power prediction model for different new energy sites. As a result, the selected target power prediction model has poor compatibility with the new energy site. Summary of the invention

[0004] Based on this, it is necessary to provide a new energy station power prediction technology evaluation and scheduling method, system, device, computer equipment, computer-readable storage medium and computer program product to address the above-mentioned technical problems.

[0005] In a first aspect, the present application provides a new energy station power prediction technology evaluation and scheduling method.

[0006] The method comprises:

[0007] Acquire each historical operation information of the new energy station, the environmental information corresponding to each historical operation information, and a plurality of power prediction models, and identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs based on each historical operation information and each environmental information;

[0008] Based on the regional information, model evaluation index information of the new energy station is generated, and each operation status information and each meteorological information are respectively input into each power prediction model to obtain a prediction result corresponding to each historical operation information predicted by each power prediction model;

[0009] Based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction results corresponding to each historical operation information predicted by each power prediction model, the evaluation value of each power prediction model is identified respectively, and the power prediction model corresponding to the evaluation value with the highest evaluation value is selected as each initial target power prediction model of the new energy station;

[0010] Based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model, screening the dominant evaluation index information in each evaluation index information of each initial target power prediction model;

[0011] In each initial target power prediction model, the model function information corresponding to each advantageous evaluation index information is selected as the target model function information corresponding to each evaluation index information;

[0012] Based on the sub-model algorithm of the target model function information corresponding to each evaluation index information, and the weight value of each evaluation index information, a fusion power prediction model including all sub-model algorithms is generated through a model combination strategy, and the fusion power prediction model is used as the target power prediction model of the new energy station.

[0013] Optionally, the generating the model evaluation index information of the new energy station based on the regional information includes:

[0014] Identifying regional characteristic information of the regional information, and based on the regional characteristic information, determining a target power generation mode of the new energy station, and obtaining normal evaluation index information;

[0015] In the database, the power generation characteristic information of the target power generation mode and the characteristic evaluation index information corresponding to each power generation characteristic information are queried, and the normal evaluation index information and the characteristic evaluation index information are used as the model evaluation index information of the new energy station.

[0016] Optionally, the step of inputting each operating state information and each meteorological information into each power prediction model to obtain a prediction result corresponding to each historical operating information predicted by each power prediction model includes:

[0017] For each piece of operating state information, identify the state parameters of the operating state information and the meteorological parameters of the meteorological information corresponding to the operating state information, and input each power prediction model based on the state parameters and the meteorological parameters, adjust the initial operating parameters of each power prediction model, and obtain a new power prediction model;

[0018] Based on each new power prediction model, the predicted power generation power of the operating status information in the future time period is predicted, and the prediction log information of each new power prediction model is obtained, and the predicted power generation power predicted by each new power prediction model and the prediction log information of each new power prediction model are used as the prediction result of the historical operating information predicted by the power prediction model corresponding to each new power prediction model.

[0019] Optionally, the historical power generation based on each historical operation information, the model evaluation index information of the new energy station, and the prediction result corresponding to each historical operation information predicted by each power prediction model, respectively identifying the evaluation value of each power prediction model, includes:

[0020] For each power prediction model, based on the prediction log information of each historical operation information by the power prediction model, identifying the operation characteristic information in each prediction log information;

[0021] Based on each sub-evaluation index information of the model evaluation index information, each evaluation feature information in each operation feature information is identified respectively, and for each operation feature information, a first evaluation value corresponding to each evaluation feature information of the operation feature information is identified;

[0022] Calculate the predicted power generation of each historical operation information predicted by the power prediction model and the deviation value between the historical power generation of each historical operation information to obtain the second evaluation value of the power prediction model, and determine the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model.

[0023] Optionally, determining the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model includes:

[0024] Obtaining a weight value for each evaluation index information, and respectively calculating an average value of the first evaluation value of each historical operation information of the power prediction model for each evaluation index information, to obtain a target evaluation value for each evaluation index information corresponding to the power prediction model;

[0025] Based on the weight value of each evaluation index information, the target evaluation value of each evaluation index information is weighted and summed to obtain a first target evaluation value, and the first target evaluation value and the second evaluation value are averaged to obtain the evaluation value of the power prediction model.

[0026] Optionally, after using the fusion power prediction model as the target power prediction model of the new energy site, the method further includes:

[0027] Obtaining an evaluation threshold for each evaluation index information, and respectively calculating a deviation value between a target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold for each evaluation index information, and selecting, from among the deviation values, evaluation index information corresponding to a deviation value greater than the deviation threshold as abnormal evaluation index information of the target power prediction model;

[0028] Based on the operation interference information in each of the historical operation information, the first evaluation value of each of the historical operation information corresponding to the abnormal evaluation index information of the target power prediction model is identified, in each of the operation interference information, the target interference information is screened, and based on the meteorological information of each of the historical operation information and the deviation value corresponding to the predicted power generation of each of the historical operation information predicted by the target power prediction model, the target interference meteorological information of the target power prediction model is identified;

[0029] The target interference information and the target interference meteorological information are used as prediction deviation causes of the target power prediction model, and an error diagnosis report of the target power prediction model is generated based on the prediction deviation causes.

[0030] In a second aspect, the present application also provides a new energy station power prediction technology evaluation and scheduling device. The device includes:

[0031] An acquisition module, used to acquire each historical operation information of the new energy station, environmental information corresponding to each historical operation information, and multiple power prediction models, and based on each historical operation information and each environmental information, identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs;

[0032] A prediction module, used to generate model evaluation index information of the new energy station based on the regional information, and input each operation status information and each meteorological information into each power prediction model respectively, to obtain a prediction result corresponding to each historical operation information predicted by each power prediction model;

[0033] A first screening module is used to identify the evaluation value of each power prediction model based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction result corresponding to each historical operation information predicted by each power prediction model, and screen the power prediction model corresponding to the evaluation value with the highest evaluation value as each initial target power prediction model of the new energy station;

[0034] A second screening module is used to screen the dominant evaluation index information in each evaluation index information of each initial target power prediction model based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model;

[0035] The third screening module is used to screen the model function information corresponding to each advantage evaluation index information in each initial target power prediction model as the target model function information corresponding to each evaluation index information;

[0036] A fusion module is used to generate a fusion power prediction model containing all sub-model algorithms based on the sub-model algorithms of the target model function information corresponding to each evaluation index information and the weight value of each evaluation index information through a model combination strategy, and use the fusion power prediction model as the target power prediction model of the new energy station.

[0037] Optionally, the prediction module is specifically used to:

[0038] Identifying regional characteristic information of the regional information, and based on the regional characteristic information, determining a target power generation mode of the new energy station, and obtaining normal evaluation index information;

[0039] In the database, the power generation characteristic information of the target power generation mode and the characteristic evaluation index information corresponding to each power generation characteristic information are queried, and the normal evaluation index information and the characteristic evaluation index information are used as the model evaluation index information of the new energy station.

[0040] Optionally, the prediction module is specifically used to:

[0041] For each piece of operating state information, identify the state parameters of the operating state information and the meteorological parameters of the meteorological information corresponding to the operating state information, and input each power prediction model based on the state parameters and the meteorological parameters, adjust the initial operating parameters of each power prediction model, and obtain a new power prediction model;

[0042] Based on each new power prediction model, the predicted power generation power of the operating status information in the future time period is predicted, and the prediction log information of each new power prediction model is obtained, and the predicted power generation power predicted by each new power prediction model and the prediction log information of each new power prediction model are used as the prediction result of the historical operating information predicted by the power prediction model corresponding to each new power prediction model.

[0043] Optionally, the first screening module is specifically used to:

[0044] For each power prediction model, based on the prediction log information of each historical operation information by the power prediction model, identifying the operation characteristic information in each prediction log information;

[0045] Based on each sub-evaluation index information of the model evaluation index information, each evaluation feature information in each operation feature information is identified respectively, and for each operation feature information, a first evaluation value corresponding to each evaluation feature information of the operation feature information is identified;

[0046] Calculate the predicted power generation of each historical operation information predicted by the power prediction model and the deviation value between the historical power generation of each historical operation information to obtain the second evaluation value of the power prediction model, and determine the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model.

[0047] Optionally, the first screening module is specifically used to:

[0048] Obtaining a weight value for each evaluation index information, and respectively calculating an average value of the first evaluation value of each historical operation information of the power prediction model for each evaluation index information, to obtain a target evaluation value for each evaluation index information corresponding to the power prediction model;

[0049] Based on the weight value of each evaluation index information, the target evaluation value of each evaluation index information is weighted and summed to obtain a first target evaluation value, and the first target evaluation value and the second evaluation value are averaged to obtain the evaluation value of the power prediction model.

[0050] Optionally, the device further comprises:

[0051] A calculation module, used to obtain an evaluation threshold of each evaluation index information, and respectively calculate the deviation value between the target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold of each evaluation index information, and select the evaluation index information corresponding to the deviation value greater than the deviation threshold from among the deviation values ​​as the abnormal evaluation index information of the target power prediction model;

[0052] an identification module, for screening target interference information from each of the operation interference information based on the operation interference information in each of the historical operation information and the first evaluation value of each of the historical operation information corresponding to the abnormal evaluation index information of the target power prediction model, and identifying target interference meteorological information of the target power prediction model based on the meteorological information of each of the historical operation information and the deviation value corresponding to the predicted power generation of each of the historical operation information predicted by the target power prediction model;

[0053] A generating module is used to use the target interference information and the target interference meteorological information as the prediction deviation cause of the target power prediction model, and generate an error diagnosis report of the target power prediction model based on the prediction deviation cause.

[0054] In a third aspect, the present application provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0055] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0056] In a fifth aspect, the present application provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0057] The above-mentioned new energy station power prediction technology evaluation and scheduling method, system and device obtain each historical operation information of the new energy station, the environmental information corresponding to each historical operation information, and multiple power prediction models, and based on each of the historical operation information and each of the environmental information, identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs; based on the regional information, generate model evaluation index information of the new energy station, and input each operation status information and each meteorological information into each of the power prediction models, respectively, to obtain the prediction result corresponding to each of the historical operation information predicted by each of the power prediction models; based on the historical power generation of each of the historical operation information, and the The model evaluation index information of the new energy station and the prediction results corresponding to each historical operation information predicted by each power prediction model are used to identify the evaluation value of each power prediction model respectively, and the power prediction model corresponding to the evaluation value with the highest evaluation value is selected as the initial target power prediction model of the new energy station; based on the prediction results of each historical operation information predicted by the initial target power prediction model, the evaluation value of the initial target power prediction model, and the operation interference information in each historical operation information, the cause of the prediction deviation of the initial target power prediction model is analyzed, and based on the cause of the prediction deviation, a model adjustment strategy of the initial target power prediction model is generated; based on the model adjustment strategy, the initial target power prediction model is adjusted to obtain the target power prediction model of the new energy station. In this scheme, different historical operation information of the new energy station is predicted by different power prediction models to obtain the prediction results corresponding to different power prediction models, and then, each prediction result is evaluated and processed by the model evaluation index information corresponding to the regional information of the new energy station, so as to select the initial target power prediction model of the new energy station. Then, the model function information corresponding to the advantageous evaluation index system in each initial target power prediction model is screened and fused and spliced ​​to obtain the target power prediction model of the new energy station, so that the target power prediction model contains the advantageous prediction function of each initial target power prediction model, thereby improving the accuracy of the target prediction model in predicting the power generation power of the new energy station, thereby comprehensively improving the adaptability between the screened target power prediction model and the new energy station. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of a flow chart of a power prediction technology evaluation and scheduling method for a new energy station in one embodiment;

[0059] Figure 2 A schematic diagram of the structure of a new energy station power prediction technology evaluation and dispatching system in one embodiment;

[0060] Figure 3 It is a structural block diagram of a new energy station power prediction technology evaluation and scheduling device in one embodiment;

[0061] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0063] The power prediction technology evaluation and scheduling method for new energy stations provided in the embodiment of the present application can be applied to the application environment of the power generation prediction model screening of new energy stations. The method can be applied to a terminal, a server, or a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, etc. Among them, the terminal predicts different historical operation information of the new energy station through different power prediction models, obtains the prediction results corresponding to different power prediction models, and then, through the model evaluation index information corresponding to the regional information of the new energy station, evaluates and processes each prediction result, thereby screening out the initial target power prediction model of the new energy station. Then, the model function information corresponding to the advantage evaluation index system in each initial target power prediction model is screened and fused and spliced ​​to obtain the target power prediction model of the new energy station, so that the target power prediction model contains the advantage prediction function of each initial target power prediction model, and improves the accuracy of the target prediction model in predicting the power generation of the new energy station, thereby comprehensively improving the adaptability between the screening target power prediction model and the new energy station.

[0064] In one embodiment, Figure 1 As shown, a new energy station power prediction technology evaluation and scheduling method is provided, which is described by taking the application of the method to a terminal as an example, and includes the following steps:

[0065] Step S101, obtain each historical operation information of the new energy station, the environmental information corresponding to each historical operation information, and multiple power prediction models, and based on each historical operation information and each environmental information, identify the historical power generation power of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs.

[0066] In this embodiment, the terminal obtains historical operation information of different time periods in the historical operation database of the new energy station, wherein the historical operation information is the power generation record information of the new energy station in different time periods, and the power generation record information includes the historical power generation of the time period, and the operation status information. The operation status information is the initial operation parameters of each power generation equipment of the new energy station before performing the power generation task. Among them, the initial operation parameters characterize the power generation state data related to the power generation task of the new energy station, and the power generation point state data, for example, the current unit power generation, the current power generation fluctuation data, the equipment wear and aging data, and the variable correspondence data between the power generation and the environmental influencing factors. Then, the terminal obtains the environmental information of each historical operation information, and the environmental information includes the regional information to which the new energy station belongs and the meteorological information corresponding to the historical operation information, and the meteorological information includes wind speed, wind direction, air pressure, temperature, and relative humidity. Among them, the new energy station is a power generation station that generates electricity through new energy, and the new energy includes wind power, potential energy electricity (such as the gravitational potential energy of water used in dam power stations for power generation), tidal energy electricity, light energy, etc. Among them, wind power can be divided into offshore wind power and onshore wind power according to the geographical information. Onshore wind power can be divided into plateau wind power and plain wind power according to the geographical information. Then, the terminal responds to the user's power prediction model upload operation and obtains different power prediction models, wherein the prediction algorithms of each power prediction model include regression analysis, time series analysis, neural network, genetic algorithm, etc.

[0067] Step S102, based on the regional information, generate model evaluation index information of the new energy station, and input each operating status information and each meteorological information into each power prediction model respectively to obtain the prediction result corresponding to each historical operating information predicted by each power prediction model.

[0068] In this embodiment, the terminal generates model evaluation index information of the new energy station based on regional information, and inputs each operating state information and each meteorological information into each power prediction model, and obtains the prediction result corresponding to each historical operating information predicted by each power prediction model. The regional characteristics of different regional information are different. For example, the regional characteristics of offshore wind power are variable wind direction sources, the regional characteristics of plateau wind power are large wind force fluctuations and rapid wind force changes in a short time, and the regional characteristics of plain wind power are fixed wind direction sources and stable wind force. The terminal presets the characteristic evaluation index information corresponding to different regional characteristics, and determines the model evaluation index information of the new energy station based on the conventional characteristic evaluation index information and the characteristic evaluation index information corresponding to the regional information. The conventional evaluation index includes prediction accuracy, prediction time, prediction process complexity, prediction influence corresponding to meteorological information, prediction influence corresponding to station equipment information, etc.; the characteristic evaluation index information is the prediction influence index information of different regional characteristics. The prediction results include the predicted power generation and the prediction process log information, which is used to record the prediction duration, the prediction algorithm calculation process, the parameter association information of different prediction influencing parameters, etc.

[0069] Step S103, based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction results corresponding to each historical operation information predicted by each power prediction model, respectively identify the evaluation value of each power prediction model, and select the power prediction model corresponding to the evaluation value with the highest evaluation value as the initial target power prediction model of each new energy station.

[0070] In this embodiment, the terminal identifies the evaluation value of each power prediction model based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction results corresponding to each historical operation information predicted by each power prediction model, and selects the power prediction model corresponding to the evaluation value with the highest evaluation value as the initial target power prediction model of the new energy station. Among them, the data of the initial target power prediction model can be one or more. The specific evaluation process and the process of identifying the evaluation values ​​of each power prediction model will be described in detail later.

[0071] Step S104 , based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model, select the dominant evaluation index information in each evaluation index information of each initial target power prediction model.

[0072] In this embodiment, the terminal screens the dominant evaluation index information in each evaluation index information of each initial target power prediction model based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model. The dominant evaluation index information is the evaluation index information with the best evaluation value among the evaluation index information of each initial target power prediction model.

[0073] Step S105 , in each initial target power prediction model, the model function information corresponding to each advantageous evaluation index information is selected as the target model function information corresponding to each evaluation index information.

[0074] In this embodiment, in each initial target power prediction model, the terminal selects the model function information corresponding to each advantageous evaluation index information as the target model function information corresponding to each evaluation index information.

[0075] Step S106, based on the sub-model algorithm of the target model function information corresponding to each evaluation index information, and the weight value of each evaluation index information, a fusion power prediction model including all sub-model algorithms is generated through a model combination strategy, and the fusion power prediction model is used as the target power prediction model of the new energy station.

[0076] In this embodiment, the terminal generates a fused power prediction model including all sub-model algorithms based on the sub-model algorithm of the target model function information corresponding to each evaluation index information and the weight value of each evaluation index information through a model combination strategy, and uses the fused power prediction model as the target power prediction model of the new energy station.

[0077] Specifically, for example, when the new energy site is a wind farm, the modeling method of the multi-model fusion method is:

[0078] 1) Modeling of multi-source model fusion problem. Specifically, k single power prediction models are selected to model the wind power series prediction problem of a certain station, and each model is denoted as M i (i=1, 2, ..., k). Suppose that at time t (t=1, 2, ..., n) in the time series T, the model M i The power prediction value is M i (t), the actual power observation value taken from the trusted data space is P actual (t). Let the weight vector λ = (λ1, λ2, ..., λ k ) is the weighted weight of each model combination prediction, and satisfies Then, the wind power combination prediction model expression for a specific station can be written as:

[0079]

[0080] How to select the weights in the combined model is the key to improving the accuracy of wind power prediction. By deploying a hybrid weight coefficient calculation method, the weights of the combined prediction model can be optimized.

[0081] 2) Calculate the sMAPE weight coefficient of the model. Symmetric mean absolute percentage error (sMAPE) is an effective model accuracy evaluation indicator that overcomes the asymmetry of the conventional MAPE indicator. Its expression is:

[0082]

[0083] The smaller the sMAPE, the higher the prediction accuracy of the model on the specified sequence. First, calculate the M of each model according to formula (2): i sMAPE value i Then the sMAPE weight coefficient v of the i-th model is determined according to the following formula: i :

[0084]

[0085] 3) Calculate the least squares weight coefficient of the model. Considering the overall fitting of the combined model at a single moment, there is a combined prediction model M ω (t)=ω1M1(t)+ω2M2(t)+…+ω m M m (t), the model M at time t ω The prediction error of (t) is recorded as e(t), that is, In order to obtain the combined model M over the entire time series ω The optimal weight vector ω=(ω1,ω2,…ω n ), construct the following least squares problem:

[0086]

[0087] Solving the above optimization problem, we can get the least squares weight coefficient ω of each model i .

[0088] 4) Calculate the combined weight coefficient. Based on the sMAPE weight coefficient and the least squares weight coefficient, the combined weight coefficient is constructed as follows:

[0089]

[0090] In the formula, μ i is the artificial correction coefficient, usually set to 1; when the combined model deviates, μ can be manually corrected according to the real-time operation situation. i , to adjust the participation level of some models for flexible management; iFor model M i The final determined weight coefficient can be substituted into formula (1) to generate the fusion power prediction model.

[0091] The above steps combine the sMAPE weight coefficient with the least squares weight coefficient to limit the contribution of each individual model (i.e., the initial target power prediction model) to the combined model (i.e., the fusion power prediction model). Through dual condition screening, the dominant model will be given a greater weight, thereby improving the effect of wind power prediction.

[0092] Based on the above scheme, different historical operation information of new energy stations is predicted through different power prediction models to obtain prediction results corresponding to different power prediction models. Then, each prediction result is evaluated and processed through the model evaluation index information corresponding to the regional information of the new energy station, so as to screen out the initial target power prediction model of the new energy station. Then, the model function information corresponding to the advantageous evaluation index system in each initial target power prediction model is screened and fused and spliced ​​to obtain the target power prediction model of the new energy station, so that the target power prediction model contains the advantageous prediction function of each initial target power prediction model, which improves the accuracy of the target prediction model in predicting the power generation of the new energy station, thereby comprehensively improving the adaptability between the selected target power prediction model and the new energy station.

[0093] Optionally, based on the regional information, model evaluation index information of the new energy station is generated, including: identifying the regional characteristic information of the regional information, and based on the regional characteristic information, determining the target power generation mode of the new energy station, and obtaining normal evaluation index information; in the database, querying the power generation characteristic information of the target power generation mode, and the characteristic evaluation index information corresponding to each power generation characteristic information, and using the normal evaluation index information and the characteristic evaluation index information as the model evaluation index information of the new energy station.

[0094] In this embodiment, the terminal identifies the regional characteristic information of the regional information, and based on the regional characteristic information, determines the target power generation mode of the new energy station, and obtains the normal evaluation index information. Then, the terminal queries the power generation characteristic information of the target power generation mode (i.e., the regional characteristic information mentioned above) and the characteristic evaluation index information corresponding to each power generation characteristic information in the database, and uses the normal evaluation index information and the characteristic evaluation index information as the model evaluation index information of the new energy station.

[0095] Based on the above scheme, model evaluation index information of different new energy stations is constructed through different regional information, which improves the evaluation accuracy of the power prediction model evaluated by the model evaluation index information of the new energy station.

[0096] Optionally, each operating status information and each meteorological information are respectively input into each power prediction model to obtain a prediction result corresponding to each historical operating information predicted by each power prediction model, including: for each operating status information, identifying the state parameters of the operating status information and the meteorological parameters of the meteorological information corresponding to the operating status information, and based on the state parameters and the meteorological parameters, inputting each power prediction model, adjusting the initial operating parameters of each power prediction model, and obtaining a new power prediction model; based on each new power prediction model, predicting the predicted power generation of the operating status information in a future time period, and obtaining the prediction log information of each new power prediction model, and using the predicted power generation predicted by each new power prediction model and the prediction log information of each new power prediction model as the prediction result of the historical operating information predicted by the power prediction model corresponding to each new power prediction model.

[0097] In this embodiment, the terminal identifies the state parameters of the operating state information and the meteorological parameters of the meteorological information corresponding to the operating state information for each operating state information, and inputs each power prediction model based on the state parameters and the meteorological parameters, adjusts the initial operating parameters of each power prediction model, and obtains a new power prediction model. Then, based on each new power prediction model, the terminal predicts the predicted power generation of the operating state information in the future period, obtains the prediction log information of each new power prediction model, and uses the predicted power generation predicted by each new power prediction model and the prediction log information of each new power prediction model as the prediction result of the historical operating information predicted by the power prediction model corresponding to each new power prediction model. Among them, the corresponding relationship between each meteorological information and the meteorological parameter, and the corresponding relationship between the state parameters corresponding to each state information are stored in the parameter database. The terminal obtains the state parameters of the operating state information and the meteorological parameters corresponding to the operating state information by querying the corresponding relationship between each meteorological information and the meteorological parameter, and the corresponding relationship between the state parameters corresponding to each state information in the parameter database.

[0098] Based on the above scheme, the accuracy of the power prediction model prediction is improved by adjusting the initial operating parameters of the power prediction model and then performing power prediction processing.

[0099] Optionally, based on the historical power generation of each historical operation information, the model evaluation index information of the new energy site, and the prediction results corresponding to each historical operation information predicted by each power prediction model, the evaluation value of each power prediction model is identified respectively, including: for each power prediction model, based on the prediction log information of each historical operation information by the power prediction model, the operation characteristic information in each prediction log information is identified; based on each sub-evaluation index information of the model evaluation index information, each evaluation characteristic information in each operation characteristic information is identified respectively, and for each operation characteristic information, the first evaluation value corresponding to each evaluation characteristic information of the operation characteristic information is identified; the predicted power generation of each historical operation information predicted by the power prediction model and the deviation value between the historical power generation of each historical operation information are calculated to obtain the second evaluation value of the power prediction model, and the evaluation value of the power prediction model is determined based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model.

[0100] In this embodiment, the terminal identifies the operation characteristic information in each prediction log information based on the prediction log information of each historical operation information by the power prediction model for each power prediction model. Then, the terminal identifies each evaluation characteristic information in each operation characteristic information based on each sub-evaluation index information of the model evaluation index information, and identifies the first evaluation value corresponding to each evaluation characteristic information of the operation characteristic information for each operation characteristic information. The evaluation characteristic information is the characteristic information in the operation characteristic information used for evaluation index evaluation, such as the duration value in the prediction duration, the number of process nodes in the prediction algorithm calculation process, and the number of iterative operations.

[0101] Then, the terminal calculates the predicted power generation of each historical operation information predicted by the power prediction model and the deviation between the historical power generation of each historical operation information to obtain the second evaluation value of the power prediction model. Specifically, the terminal presets the deviation value range corresponding to different second evaluation values, and then determines the second bottle value corresponding to each deviation value by identifying the deviation value range to which each deviation value belongs.

[0102] Finally, the terminal determines the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model. The specific determination process will be described in detail later.

[0103] Based on the above scheme, the evaluation value of the power prediction model is determined by combining the operating characteristic information and the deviation value corresponding to the expected power generation, thereby improving the accuracy of determining the deviation value.

[0104] Optionally, based on the first evaluation value of each historical operation information predicted by the power prediction model, and the second evaluation value of each historical operation information predicted by the power prediction model, the evaluation value of the power prediction model is determined, including: obtaining the weight value of each evaluation index information, and respectively calculating the average value of the first evaluation value of the power prediction model in each historical operation information of each evaluation index information, to obtain the target evaluation value of each evaluation index information corresponding to the power prediction model; based on the weight value of each evaluation index information, performing weighted sum processing on the target evaluation value of each evaluation index information to obtain a first target evaluation value, and averaging the first target evaluation value and the second evaluation value to obtain the evaluation value of the power prediction model.

[0105] In this embodiment, the terminal obtains the weight value of each evaluation index information, and respectively calculates the average value of the first evaluation value of each historical operation information of the power prediction model in each evaluation index information, and obtains the target evaluation value of each evaluation index information corresponding to the power prediction model. Then, based on the weight value of each evaluation index information, the terminal performs weighted summation processing on the target evaluation value of each evaluation index information to obtain the first target evaluation value, and then the terminal averages the evaluation value of the deviation value corresponding to the power generation power of each historical operation information to obtain the target second evaluation value. After that, the terminal averages the first target evaluation value and the target second evaluation value to obtain the evaluation value of the power prediction model. The weight value of each evaluation index information is the weight value of each evaluation index information uploaded by the user obtained by the terminal in response to the user's weight value upload operation.

[0106] Based on the above scheme, the target evaluation value is determined by performing weighted summation on the information of each evaluation index, thereby improving the accuracy of the determined target evaluation value.

[0107] Optionally, after using the fused power prediction model as the target power prediction model of the new energy station, it also includes: obtaining an evaluation threshold for each evaluation index information, and respectively calculating the deviation value between the target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold of each evaluation index information, and in each deviation value, screening the evaluation index information corresponding to the deviation value whose deviation value is greater than the deviation threshold as the abnormal evaluation index information of the target power prediction model; based on the operation interference information in each historical operation information, identifying the first evaluation value of each historical operation information corresponding to the abnormal evaluation index information of the target power prediction model, screening the target interference information in each operation interference information, and based on the meteorological information of each historical operation information and the deviation value corresponding to the predicted power generation of each historical operation information predicted by the target power prediction model, identifying the target interference meteorological information of the target power prediction model; using the target interference information and the target interference meteorological information as the cause of the prediction deviation of the target power prediction model, and generating an error diagnosis report of the target power prediction model based on the cause of the prediction deviation.

[0108] In this embodiment, the terminal obtains the evaluation threshold of each evaluation index information, and respectively calculates the deviation value between the target evaluation value of each evaluation index information of the initial target power prediction model and the evaluation threshold of each evaluation index information, and selects the evaluation index information corresponding to the deviation value greater than the deviation threshold in each deviation value as the abnormal evaluation index information of the initial target power prediction model. Then, based on the operation interference information in each historical operation information, the terminal identifies the first evaluation value of each historical operation information corresponding to the abnormal evaluation index information of the initial target power prediction model, and selects the target interference information in each operation interference information, and identifies the target interference meteorological information of the initial target power prediction model based on the meteorological information of each historical operation information and the deviation value corresponding to the predicted power generation of each historical operation information predicted by the initial target power prediction model. After that, the terminal uses the target interference information and the target interference meteorological information as the cause of the prediction deviation of the initial target power prediction model.

[0109] Specifically, (1) the starter sends an enable signal fenable to start the power prediction error diagnosis and quantitative evaluation.

[0110] (2) After receiving the enable signal, the power prediction trusted data space sends the meteorological data x(v, d, T, h, Va) at the corresponding time to the meteorological-power conversion model server.

[0111] (3) According to the value of the model controller, the meteorological-power conversion model server dispatches the corresponding meteorological-power conversion model to calculate the predicted power of the wind farm under the correct meteorological conditions.

[0112] Specifically: When the model controller S m= 1, the meteorological-power conversion model server uses the model f1(·) deployed by the manufacturer on the meteorological-power conversion model server to calculate the wind farm predicted power P under correct meteorological conditions. wind1 ; When the model controller S m = 2, the meteorological-power conversion model server uses the model f2(·) deployed on the meteorological-power conversion model server by the centralized control center to calculate the wind farm predicted power P under correct meteorological conditions. wind2 .

[0113] (4) Calculate the corresponding predicted power error and contribution based on the multiple parameters sent by the weather-power prediction conversion model server.

[0114] Specifically, the prediction error quantitative analysis and calculation module calculates the prediction power error En and error contribution Rn caused by the NWP link, the prediction power error Em and error contribution Rm caused by the Model link, and the prediction power error Er and error contribution Rr caused by the correction link according to Pwindi (i=1 or 2) sent by the meteorological-power prediction conversion model server, the manufacturer's predicted power Ppredict sent by the power prediction trusted data space, the actual output power Pactual of the station, the actual starting capacity Co of the station and the rated capacity Crated.

[0115] Among them, the weather-power conversion model server is mainly used to deploy the weather-power conversion model, and calculate the predicted power under the correct weather data at the specified time according to the actual weather data at the input time. The specific model expression is shown in formula (9):

[0116] P windi =f i (v,d,T,H,r h ) i=1,2 (9)

[0117] P windi is the station output power, i.e. the output of the model; [v,d,T,H,r h ] is the input vector of the prediction model, which are the observed wind speed, wind direction, temperature, air pressure, and relative humidity. In addition, other parameters related to wind power output can also be added to the model input vector; f i It is a weather-power conversion model, that is, the functional relationship between input and output.

[0118] The meteorological-power conversion model is the core of the technical diagnosis and quantitative evaluation module of the power prediction error of new energy stations. Its functions are realized in two ways: the new energy power prediction manufacturer directly deploys the actual prediction model of the station to the server and directly calculates the value of Pwind1 by calling the model; the platform develops a meteorological-power conversion model and calls the model to calculate the value of Pwind2.

[0119] Among them, for the construction of the above-mentioned meteorological-power conversion model, the present invention proposes a model update data set selection mechanism based on threshold judgment, the specific contents of which are as follows:

[0120] At present, there are many mature meteorological-power conversion modeling algorithms, including statistical models, physical models and other algorithms. The basic method of these modeling algorithms is to obtain the meteorological-power conversion model through training with historical meteorological data (v, d, T, H, rh) and historical power data of the station over a period of time. Taking the deep learning algorithm as an example, the training steps of the meteorological-power conversion model include:

[0121] 1) Preprocess the training data set, including missing data filling, data normalization, etc. The data set is divided into a training set and a test set, where the former is used to train the model and the latter is used to test the accuracy of the model and optimize the model parameters (tuning parameters).

[0122] 2) In view of the timing characteristics of wind power forecasting, it is planned to use Long-Short Term Memory (LSTM) as a deep learning algorithm.

[0123] 3) In order to improve the accuracy of the model, a feature learning link is added to the front end of the learning model, so as to effectively extract the features that are most conducive to the accurate conversion of wind power from multi-source data input. This patent uses stacked denoising autoencoders (SADE) for feature learning.

[0124] Due to changes in meteorological conditions, equipment status, and the environment, the weather-power conversion model needs to be updated in real time based on new operating data to maintain model accuracy. Model updates rely on the input of historical meteorological data and historical power data of the station. As the operating time of new energy stations increases, the amount of accumulated data also gradually increases. When updating the training model, if historical data is used as a training set indiscriminately, it may cause a series of problems such as increased demand for computing resources, longer model training time, and increased risk of overfitting.

[0125] In order to reduce the blindness of data set selection and improve the model prediction performance, the workflow of the threshold-based model update data set selection mechanism includes:

[0126] 1) Select the power prediction decision threshold ∈;

[0127] 2) Obtain the meteorological data at time t in the historical meteorological data [v, d, T, H, r h ] t , the meteorological-power conversion model predicts the power P predict and the actual output power P of the new energy stationactual ;

[0128] 3) Calculate the difference percentage ΔP between the predicted power and the actual power:

[0129] ΔP=|P predict -P actual | / P actual (10)

[0130] 4) If ΔP>∈, then the meteorological data at time t [v, d, T, H, r h ] t And the actual output power P of the station at time t actual The model update database is included; otherwise, the historical data at that moment is not included.

[0131] 5) Repeat steps 2) to 4) to obtain a weather-power conversion model training data set [(v, d, T, H, r h ),P actual ].

[0132] It is understandable that the above steps can more intelligently select the training data required for model update, thereby effectively reducing the amount of model training data and avoiding the model update training problem caused by too much data. Selecting the data required for training through model prediction error can more accurately capture the changes in station operation mode and external environment, and reduce training costs and time while improving model prediction accuracy.

[0133] Based on the above scheme, by identifying abnormal evaluation index information, target interference information and target interference meteorological information are screened, thereby improving the accuracy of screening target interference information and target interference meteorological information.

[0134] In one embodiment, Figure 2 As shown, a new energy station power prediction technology evaluation and scheduling system is provided, the system includes a data support layer, a model deployment layer, a control middle platform layer, and an advanced application layer, wherein:

[0135] The data support layer is used to obtain the initial historical operation information of the new energy station and the initial environmental information corresponding to each initial historical operation information, and pre-process the historical operation information and the initial environmental information to obtain the historical operation information of the new energy station and the environmental information corresponding to each historical operation information.

[0136] The model deployment layer is used to obtain multiple initial power prediction models, and perform model training on each of the power predictions to obtain each candidate power prediction model; through each of the historical operation information and the environmental information, each of the candidate power prediction models is evaluated and screened to obtain each power prediction model.

[0137] The control middle platform layer is used to identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs based on each historical operation information and each environmental information; based on the regional information, generate model evaluation index information of the new energy station, and input each operation status information and each meteorological information into each power prediction model respectively to obtain the prediction result corresponding to each historical operation information predicted by each power prediction model.

[0138] The advanced application layer includes a prediction comparison module, a model fusion module, a model screening module, and an error diagnosis module; the prediction comparison module is used to identify the evaluation value of each power prediction model based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction results corresponding to each historical operation information predicted by each power prediction model, and perform visual comparison and display; the model screening module is used to screen the power prediction model corresponding to the evaluation value with the highest evaluation value as each initial target power prediction model of the new energy station; the model fusion module is used to screen the dominant evaluation index information in each evaluation index information of each initial target power prediction model based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model; in each initial target power prediction model, the model function information corresponding to each dominant evaluation index information is screened as the target model function information corresponding to each evaluation index information; based on the sub-model algorithm of the target model function information corresponding to each evaluation index information and the weight value of each evaluation index information, a model combination strategy is used to generate a system including all sub-models. The invention relates to a fusion power prediction model of the algorithm, and uses the fusion power prediction model as the target power prediction model of the new energy station; the error diagnosis module is used to obtain the evaluation threshold of each evaluation index information, and respectively calculate the deviation value between the target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold of each evaluation index information, and select the evaluation index information corresponding to the deviation value greater than the deviation threshold in each deviation value as the abnormal evaluation index information of the target power prediction model; based on the operation interference information in each of the historical operation information, identify the first evaluation value of each of the historical operation information corresponding to the abnormal evaluation index information of the target power prediction model, select the target interference information in each of the operation interference information, and identify the target interference meteorological information of the target power prediction model based on the meteorological information of each of the historical operation information and the deviation value corresponding to the predicted power generation of each of the historical operation information predicted by the target power prediction model; use the target interference information and the target interference meteorological information as the prediction deviation cause of the target power prediction model, and generate an error diagnosis report of the target power prediction model based on the prediction deviation cause.

[0139] Preferably, the trusted data space layer is the infrastructure of the platform. Through hardware and software means such as servers, storage, networks and virtualization, it accesses, stores and analyzes the models and historical operation data of different new energy stations / units, measured / predicted meteorological data (wind speed, wind direction, temperature, humidity, etc.), maintenance plan data, wind abandonment, light abandonment data and other massive data; provides data preprocessing functions, including data cleaning, missing data completion, power limit data filtering and other operations; constructs a trusted execution environment through data encryption and setting access rights, and provides users with real, reliable and rich data resources while ensuring data security and reliability. The model deployment layer includes a power prediction model management module and a power prediction model operation module, which are responsible for the storage, management, operation and maintenance of the manufacturer's power prediction model. The control middle platform layer includes a real-scene verification module, which is responsible for organizing data, defining operation configurations, calling models, and providing offline and online verification services for other modules. The advanced application layer includes four main modules, namely, a power prediction module, an error diagnosis module, a model fusion module, and a model screening module. The main function is to provide top-level packaging services for end users (new energy stations, centralized control centers and power prediction manufacturers, etc.).

[0140] Preferably, (1) power prediction model management module ①. The function of the power prediction model management module ① is to access and manage the prediction models of various new energy power prediction manufacturers. This module is built based on containerization technology, and has built-in power prediction model image temporary warehouse, available image library and discarded image library. The image temporary warehouse is used to store the power prediction algorithm model images uploaded by the manufacturer; the available image library is used to store the power prediction algorithm model images that have been pre-verified by the real scene verification module ③ and can be used normally; the discarded image library is used to store the power prediction algorithm model images that have not passed the pre-verification. This module provides model access function for the power prediction model operation module ② in a timely manner according to demand.

[0141] (2) Power prediction model operation module ②. The power prediction model operation module function ② has a built-in container engine, which allows different power prediction manufacturers to establish independent algorithm execution environments based on the model operation environment and dependencies, and run all algorithm models such as ultra-short-term, short-term, medium-term, long-term deterministic and probabilistic forecasts in parallel in a containerized manner, and output the operation results. This module accepts the calls of modules ③, ④, ⑤, ⑥, and ⑦, thereby accessing the model operation capabilities of the above modules and providing the output results of the power prediction model based on the input given by the upper-level modules.

[0142] (3) Real-life verification module ③. Real-life verification module ③ provides power forecasting manufacturers with forecasting scenarios with diverse climate and terrain conditions, and creates an algorithm deployment and testing environment that is consistent with the actual dispatching environment, which can meet the verification needs of new energy power forecasting algorithms at multiple time scales, such as medium, short, and ultra-short term. Module ③ provides two types of forecasting algorithm verification services: offline real-life verification service and online real-life verification service. In the offline real-life verification service, module ③ organizes and provides meteorological data and historical operation data of typical stations at historical moments, and based on the capabilities of module ②, provides an algorithm deployment and testing environment that is consistent with the actual dispatching environment, and conducts offline real-life performance testing. In the online real-life verification service, module ③ takes actual operating new energy stations as the object, provides an algorithm deployment and testing environment that is consistent with the actual dispatching environment, and conducts online real-life performance testing. Module ③ accepts the calls of modules ①, ④, ⑤, ⑥, and ⑦, thereby connecting the above modules to the model verification capability for further analysis and processing.

[0143] (4) Power prediction system AI competition range module ④. The advanced application module ④ AI competition range provides unified and standardized power prediction multi-source data and deployment environment for different new energy power prediction systems, and conducts multi-dimensional evaluation of the prediction results of different new energy prediction systems on time scales such as long-term, medium-term, and ultra-short-term, as well as on spatial scales such as single machine, station, region / province, etc., to achieve fair and objective comparison of the prediction performance of different power prediction systems, and provide decision-making basis for the procurement and selection of new energy power prediction software and services.

[0144] (5) Power forecast error technical diagnosis and supervision module ⑤. Advanced application module ⑤ is aimed at new energy stations whose forecast accuracy does not meet the standard. Based on multi-source operation data such as meteorology and electricity, it uses data mining methods to decouple the meteorological-power forecast error in the operation results of module 5, quantitatively analyze the error causes of each key link in the forecast and the degree of influence on the overall error, realize quantitative evaluation of the timing error of each link, finely analyze the causes of forecast deviation events, accurately locate problem stations and forecast weak links, and provide support for new energy stations and forecast manufacturers to improve forecast algorithms.

[0145] (6) Power prediction algorithm multi-source fusion module ⑥. A single algorithm usually cannot achieve power prediction at all times. For specific new energy sites, the advanced application module ⑥ multi-source fusion platform freely combines the weather forecast algorithms and weather-power conversion algorithms of different power prediction manufacturers in module ①, assigns different weights, and compares the prediction performance of the original algorithm with the fused algorithm in multiple dimensions through a visual interface.

[0146] (7) Station and model adaptive adaptation module ⑦. Taking into account the characteristics of new energy stations in different regions, different installed capacities, and different weather conditions, and combining the historical operation data in module 1, the advanced application module ⑦ can establish a new energy station operation model. On this basis, the platform verifies the performance of various power prediction algorithms in module ① according to multi-dimensional indicators, establishes a mapping relationship between scene features and algorithm features, and adaptively selects the optimal power prediction model on different time scales (including but not limited to the past 7 days, the past 30 days, the past 2 months, the past six months, etc.) for specific stations.

[0147] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0148] Based on the same inventive concept, the embodiment of the present application also provides a new energy station power prediction technology evaluation and scheduling device for implementing the new energy station power prediction technology evaluation and scheduling method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the new energy station power prediction technology evaluation and scheduling device provided below can refer to the limitations of the new energy station power prediction technology evaluation and scheduling method above, and will not be repeated here.

[0149] In one embodiment, Figure 3 As shown, a new energy station power prediction technology evaluation and scheduling device is provided, including: an acquisition module 310, a prediction module 320, a first screening module 330, a second screening module 340, a third screening module 350 and a fusion module 360, wherein:

[0150] The acquisition module 310 is used to acquire each historical operation information of the new energy station, the environmental information corresponding to each historical operation information, and a plurality of power prediction models, and identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs based on each historical operation information and each environmental information;

[0151] The prediction module 320 is used to generate model evaluation index information of the new energy station based on the regional information, and input each operation status information and each meteorological information into each power prediction model to obtain a prediction result corresponding to each historical operation information predicted by each power prediction model;

[0152] A first screening module 330 is used to identify the evaluation value of each power prediction model based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction result corresponding to each historical operation information predicted by each power prediction model, and screen the power prediction model corresponding to the evaluation value with the highest evaluation value as each initial target power prediction model of the new energy station;

[0153] A second screening module 340 is used to screen the dominant evaluation index information in each evaluation index information of each initial target power prediction model based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model;

[0154] The third screening module 350 is used to screen the model function information corresponding to each advantageous evaluation index information in each initial target power prediction model as the target model function information corresponding to each evaluation index information;

[0155] The fusion module 360 ​​is used to generate a fusion power prediction model including all sub-model algorithms based on the sub-model algorithm of the target model function information corresponding to each evaluation index information and the weight value of each evaluation index information through a model combination strategy, and use the fusion power prediction model as the target power prediction model of the new energy station.

[0156] Optionally, the prediction module 320 is specifically used for:

[0157] Identifying regional characteristic information of the regional information, and based on the regional characteristic information, determining a target power generation mode of the new energy station, and obtaining normal evaluation index information;

[0158] In the database, the power generation characteristic information of the target power generation mode and the characteristic evaluation index information corresponding to each power generation characteristic information are queried, and the normal evaluation index information and the characteristic evaluation index information are used as the model evaluation index information of the new energy station.

[0159] Optionally, the prediction module 320 is specifically used for:

[0160] For each piece of operating state information, identify the state parameters of the operating state information and the meteorological parameters of the meteorological information corresponding to the operating state information, and input each power prediction model based on the state parameters and the meteorological parameters, adjust the initial operating parameters of each power prediction model, and obtain a new power prediction model;

[0161] Based on each new power prediction model, the predicted power generation power of the operating status information in the future time period is predicted, and the prediction log information of each new power prediction model is obtained, and the predicted power generation power predicted by each new power prediction model and the prediction log information of each new power prediction model are used as the prediction result of the historical operating information predicted by the power prediction model corresponding to each new power prediction model.

[0162] Optionally, the first screening module 330 is specifically used to:

[0163] For each power prediction model, based on the prediction log information of each historical operation information by the power prediction model, identifying the operation characteristic information in each prediction log information;

[0164] Based on each sub-evaluation index information of the model evaluation index information, each evaluation feature information in each operation feature information is identified respectively, and for each operation feature information, a first evaluation value corresponding to each evaluation feature information of the operation feature information is identified;

[0165] Calculate the predicted power generation of each historical operation information predicted by the power prediction model and the deviation value between the historical power generation of each historical operation information to obtain the second evaluation value of the power prediction model, and determine the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model.

[0166] Optionally, the first screening module 330 is specifically used to:

[0167] Obtaining a weight value for each evaluation index information, and respectively calculating an average value of the first evaluation value of each historical operation information of the power prediction model for each evaluation index information, to obtain a target evaluation value for each evaluation index information corresponding to the power prediction model;

[0168] Based on the weight value of each evaluation index information, the target evaluation value of each evaluation index information is weighted and summed to obtain a first target evaluation value, and the first target evaluation value and the second evaluation value are averaged to obtain the evaluation value of the power prediction model.

[0169] Optionally, the device further comprises:

[0170] A calculation module, used to obtain an evaluation threshold of each evaluation index information, and respectively calculate the deviation value between the target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold of each evaluation index information, and select the evaluation index information corresponding to the deviation value greater than the deviation threshold from among the deviation values ​​as the abnormal evaluation index information of the target power prediction model;

[0171] an identification module, for screening target interference information from each of the operation interference information based on the operation interference information in each of the historical operation information and the first evaluation value of each of the historical operation information corresponding to the abnormal evaluation index information of the target power prediction model, and identifying target interference meteorological information of the target power prediction model based on the meteorological information of each of the historical operation information and the deviation value corresponding to the predicted power generation of each of the historical operation information predicted by the target power prediction model;

[0172] A generating module is used to use the target interference information and the target interference meteorological information as the prediction deviation cause of the target power prediction model, and generate an error diagnosis report of the target power prediction model based on the prediction deviation cause.

[0173] Each module in the above-mentioned new energy station power prediction technology evaluation and scheduling device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0174] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a new energy station power prediction technology evaluation and scheduling method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse.

[0175] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0176] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0177] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0178] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of any one of the methods in the first aspect.

[0179] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0180] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0181] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0182] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A new energy station power prediction technology evaluation and scheduling method, characterized in that: The method comprises: Acquire each historical operation information of the new energy station, the environmental information corresponding to each historical operation information, and a plurality of power prediction models, and identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs based on each historical operation information and each environmental information; Based on the regional information, model evaluation index information of the new energy station is generated, and each operation status information and each meteorological information are respectively input into each power prediction model to obtain a prediction result corresponding to each historical operation information predicted by each power prediction model; For each power prediction model, based on the prediction result of the power prediction model for each of the historical operation information, identifying the operation characteristic information in each of the prediction results; Based on each sub-evaluation index information of the model evaluation index information, each evaluation feature information in each operation feature information is identified respectively, and for each operation feature information, a first evaluation value corresponding to each evaluation feature information of the operation feature information is identified; Calculate the predicted power generation of each of the historical operation information predicted by the power prediction model and the deviation value between the historical power generation of each of the historical operation information to obtain a second evaluation value of the power prediction model, and determine the evaluation value of the power prediction model based on the first evaluation value of each of the historical operation information predicted by the power prediction model and the second evaluation value of each of the historical operation information predicted by the power prediction model; Selecting a power prediction model corresponding to the evaluation value with the highest evaluation value as each initial target power prediction model of the new energy station; Based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model, screening the dominant evaluation index information in each evaluation index information of each initial target power prediction model; In each initial target power prediction model, the model function information corresponding to each advantageous evaluation index information is selected as the target model function information corresponding to each evaluation index information; Based on the sub-model algorithm of the target model function information corresponding to each evaluation index information, and the weight value of each evaluation index information, a fusion power prediction model including all sub-model algorithms is generated through a model combination strategy, and the fusion power prediction model is used as the target power prediction model of the new energy station.

2. The method according to claim 1, characterized in that The generating, based on the regional information, model evaluation index information of the new energy station comprises: Identifying regional characteristic information of the regional information, and based on the regional characteristic information, determining a target power generation mode of the new energy station, and obtaining normal evaluation index information; In the database, the power generation characteristic information of the target power generation mode and the characteristic evaluation index information corresponding to each power generation characteristic information are queried, and the normal evaluation index information and the characteristic evaluation index information are used as the model evaluation index information of the new energy station.

3. The method according to claim 2, characterized in that The step of inputting each operating state information and each meteorological information into each power prediction model to obtain a prediction result corresponding to each historical operating information predicted by each power prediction model includes: For each piece of operating state information, identify the state parameters of the operating state information and the meteorological parameters of the meteorological information corresponding to the operating state information, and input each power prediction model based on the state parameters and the meteorological parameters, adjust the initial operating parameters of each power prediction model, and obtain a new power prediction model; Based on each new power prediction model, the predicted power generation power of the operating status information in the future time period is predicted, and the prediction log information of each new power prediction model is obtained, and the predicted power generation power predicted by each new power prediction model and the prediction log information of each new power prediction model are used as the prediction result of the historical operating information predicted by the power prediction model corresponding to each new power prediction model.

4. The method according to claim 1, characterized in that: The determining the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model comprises: Obtaining a weight value for each evaluation index information, and respectively calculating an average value of the first evaluation value of each historical operation information of the power prediction model for each evaluation index information, to obtain a target evaluation value for each evaluation index information corresponding to the power prediction model; Based on the weight value of each evaluation index information, the target evaluation value of each evaluation index information is weighted and summed to obtain a first target evaluation value, and the first target evaluation value and the second evaluation value are averaged to obtain the evaluation value of the power prediction model.

5. The method according to claim 1, characterized in that After the fusion power prediction model is used as the target power prediction model of the new energy station, the method further includes: Obtaining an evaluation threshold for each evaluation index information, and respectively calculating a deviation value between a target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold for each evaluation index information, and selecting, from among the deviation values, evaluation index information corresponding to a deviation value greater than the deviation threshold as abnormal evaluation index information of the target power prediction model; Based on the operation interference information in each of the historical operation information, the first evaluation value of each of the historical operation information corresponding to the abnormal evaluation index information of the target power prediction model is identified, in each of the operation interference information, the target interference information is screened, and based on the meteorological information of each of the historical operation information and the deviation value corresponding to the predicted power generation of each of the historical operation information predicted by the target power prediction model, the target interference meteorological information of the target power prediction model is identified; The target interference information and the target interference meteorological information are used as prediction deviation causes of the target power prediction model, and an error diagnosis report of the target power prediction model is generated based on the prediction deviation causes.

6. A new energy station power prediction technology evaluation and dispatching system, characterized in that: The system includes a data support layer, a model deployment layer, a control platform layer, and an advanced application layer, wherein: The data support layer is used to obtain each initial historical operation information of the new energy station and the initial environmental information corresponding to each initial historical operation information, and pre-process each historical operation information and each initial environmental information to obtain each historical operation information of the new energy station and the environmental information corresponding to each historical operation information; The model deployment layer is used to obtain multiple initial power prediction models, and perform model training on each of the power predictions to obtain each candidate power prediction model; through each of the historical operation information and each of the environmental information, each of the candidate power prediction models is evaluated and screened to obtain each power prediction model; The control middle platform layer is used to identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs based on each of the historical operation information and each of the environmental information; based on the regional information, generate the model evaluation index information of the new energy station, and input each of the operation status information and each of the meteorological information into each of the power prediction models, respectively, to obtain the prediction result corresponding to each of the historical operation information predicted by each of the power prediction models; The advanced application layer includes a prediction comparison module, a model fusion module, a model screening module, and an error diagnosis module; The prediction comparison module is used to identify the evaluation value of each power prediction model based on the historical power generation of each historical operation information, the model evaluation index information of the new energy station, and the prediction result corresponding to each historical operation information predicted by each power prediction model, and perform visual comparison and display; The model screening module is used to screen the power prediction model corresponding to the evaluation value with the highest evaluation value as each initial target power prediction model of the new energy site; The model fusion module is used to screen the advantageous evaluation index information in each evaluation index information of each initial target power prediction model based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model; in each initial target power prediction model, screen the model function information corresponding to each advantageous evaluation index information as the target model function information corresponding to each evaluation index information; based on the sub-model algorithm of the target model function information corresponding to each evaluation index information and the weight value of each evaluation index information, generate a fusion power prediction model including all sub-model algorithms through a model combination strategy, and use the fusion power prediction model as the target power prediction model of the new energy station; The error diagnosis module is used to obtain the evaluation threshold of each evaluation index information, and respectively calculate the deviation value between the target evaluation value of each evaluation index information of the target power prediction model and the evaluation threshold of each evaluation index information, and select the evaluation index information corresponding to the deviation value greater than the deviation threshold in each deviation value as the abnormal evaluation index information of the target power prediction model; based on the operation interference information in each historical operation information, identify the first evaluation value of each historical operation information corresponding to the abnormal evaluation index information of the target power prediction model, select the target interference information in each operation interference information, and identify the target interference meteorological information of the target power prediction model based on the meteorological information of each historical operation information and the deviation value corresponding to the predicted power generation of each historical operation information predicted by the target power prediction model; use the target interference information and the target interference meteorological information as the prediction deviation cause of the target power prediction model, and generate an error diagnosis report of the target power prediction model based on the prediction deviation cause.

7. A new energy station power prediction technology evaluation and scheduling device, characterized in that: The device comprises: An acquisition module, used to acquire each historical operation information of the new energy station, environmental information corresponding to each historical operation information, and multiple power prediction models, and based on each historical operation information and each environmental information, identify the historical power generation of each historical operation information, the operation status information of each historical operation information, the meteorological information corresponding to each historical operation information, and the regional information to which the new energy station belongs; A prediction module, used to generate model evaluation index information of the new energy station based on the regional information, and input each operation status information and each meteorological information into each power prediction model respectively, to obtain a prediction result corresponding to each historical operation information predicted by each power prediction model; A first screening module is used to identify, for each power prediction model, the operation characteristic information in each prediction result based on the prediction result of each historical operation information by the power prediction model; to identify, based on each sub-evaluation index information of the model evaluation index information, each evaluation characteristic information in each operation characteristic information, and, for each operation characteristic information, to identify the first evaluation value corresponding to each evaluation characteristic information of the operation characteristic information; to calculate the deviation value between the predicted power generation of each historical operation information predicted by the power prediction model and the historical power generation of each historical operation information, to obtain the second evaluation value of the power prediction model, and to determine the evaluation value of the power prediction model based on the first evaluation value of each historical operation information predicted by the power prediction model and the second evaluation value of each historical operation information predicted by the power prediction model; to screen the power prediction model corresponding to the evaluation value with the highest evaluation value as each initial target power prediction model of the new energy station; A second screening module is used to screen the dominant evaluation index information in each evaluation index information of each initial target power prediction model based on the evaluation threshold of each evaluation index information and the target evaluation value of each evaluation index information of each initial target power prediction model; The third screening module is used to screen the model function information corresponding to each advantage evaluation index information in each initial target power prediction model as the target model function information corresponding to each evaluation index information; A fusion module is used to generate a fusion power prediction model containing all sub-model algorithms based on the sub-model algorithms of the target model function information corresponding to each evaluation index information and the weight value of each evaluation index information through a model combination strategy, and use the fusion power prediction model as the target power prediction model of the new energy station.

8. A computer 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 method according to any one of claims 1 to 5 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images