CN115564115A - Wind power plant power prediction method and related equipment - Google Patents
Wind power plant power prediction method and related equipment Download PDFInfo
- Publication number
- CN115564115A CN115564115A CN202211239008.4A CN202211239008A CN115564115A CN 115564115 A CN115564115 A CN 115564115A CN 202211239008 A CN202211239008 A CN 202211239008A CN 115564115 A CN115564115 A CN 115564115A
- Authority
- CN
- China
- Prior art keywords
- data
- power
- target
- wind
- wind turbine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Power Engineering (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The embodiment of the invention discloses a method for predicting power of a wind power plant and related equipment, wherein the method comprises the following steps: acquiring target historical meteorological data and target historical power data of each wind turbine generator in a wind power plant; training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, and carrying out dynamic time establishment on a multivariable time sequence through a lightGBM model by the preset model; respectively substituting the weather forecast data into the target model of each wind turbine to obtain the predicted power of each wind turbine; and adding the predicted power of each wind turbine generator to obtain the predicted power of the wind power plant. And establishing a target model of each wind turbine generator according to the difference of the single wind turbine generator in the wind power plant, and improving the prediction accuracy of the single wind turbine generator.
Description
Technical Field
The invention relates to the technical field of wind power plant power prediction, in particular to a wind power plant power prediction method and related equipment.
Background
Wind power has an important position in renewable energy sources in China, wind power plants in China are mainly and intensively built in regions rich in wind energy resources, namely northeast, northwest and north China, and the biggest difficulty in large-scale centralized power development is the problem of power delivery. The distributed wind power has the advantages that the distributed wind power has strong adaptability to the surrounding environment, almost all landforms can be constructed and used, and the utilization rate of wind energy resources in remote areas can be effectively improved. With the gradual expansion of the scale of the distributed wind power, the prediction of the power of the distributed wind power becomes an important means for ensuring the stable operation of the distributed wind power and the safety and reliability of the power grid connected with the distributed wind power.
For the traditional wind power prediction problem, a large number of scholars have done research work at home and abroad, and the research work mainly comprises a physical method and a statistical method. The physical method needs complex wind turbine related information and weather forecast information, and is difficult to apply. The statistical method only needs weather conditions such as wind speed and wind direction and the time sequence related to the wind power. With the gradual development of artificial intelligence, a large number of statistical methods are developed for wind power prediction, and common methods include an artificial neural network method, a support vector machine method, xgboost and the like.
The existing prediction technology is mainly applied to a centralized large wind power plant, and the prediction adaptability of a single unit of a distributed wind power plant is weak; the difference of a single unit cannot be considered by a model for the whole wind power plant prediction; for the characteristics of unobvious periodic changes such as wind speed and wind direction, the traditional network model cannot extract effective front and back related information for prediction, and further influences the prediction result.
Disclosure of Invention
In view of the above, the invention provides a wind power plant power prediction method and related equipment, which are used for solving the problems that the difference of a single unit cannot be considered and the prediction result is inaccurate in the prior art. To achieve one or a part of or all of the above or other objects, the present invention provides a method for predicting power of a wind farm, comprising: acquiring target historical meteorological data and target historical power data of each wind turbine generator in a wind power plant;
training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, and carrying out dynamic time establishment on a multivariable time sequence through a lightGBM model by the preset model;
respectively substituting the meteorological forecast data into the target model of each wind turbine to obtain the predicted power of each wind turbine;
and adding the predicted power of each wind turbine generator to obtain the predicted power of the wind power plant.
Optionally, the step of obtaining target historical meteorological data and target historical power data of each wind turbine generator in the wind farm includes:
acquiring initial historical meteorological data and initial historical power data of each wind turbine generator in a wind power plant;
carrying out outlier removal processing on the initial historical meteorological data and the initial historical power data by using a local outlier algorithm;
performing linear interpolation on the processed initial historical meteorological data and initial historical power data to obtain complete historical meteorological data and complete historical power data;
and carrying out normalization processing on the complete historical meteorological data and the complete historical power data to obtain target historical meteorological data and target historical power data of each wind turbine generator in the wind power plant.
Optionally, before the step of training the preset model based on the target historical meteorological data and the target historical power data, the method further includes:
and performing correlation analysis on the target historical meteorological data and the target historical power data by adopting an MIC algorithm, and eliminating meteorological features of which the correlation degree with the target historical power data is lower than a threshold value in the target historical meteorological data to obtain optimized target historical meteorological data.
Optionally, the step of training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator includes:
dividing the optimized target historical meteorological data and the target historical power data into a training set, a verification set and a test set;
and training the preset model based on the training set, the verification set and the test set to obtain a target model of each wind turbine.
Optionally, the step of training the preset model based on the training set, the verification set, and the test set to obtain a target model of each wind turbine generator includes:
optimizing the hyper-parameters of the preset model according to the verification set and a firefly lifting algorithm introducing inertial weight;
and training the optimized preset model based on the training set and the test set to obtain a target model of each wind turbine.
Optionally, the inertial weight is a nonlinear inertial weight, a nonlinear function is used for adjustment, and a tanh function is used as a basis function for constructing the nonlinear inertial weight.
Optionally, the step of training a preset model based on the target historical meteorological data and the target historical power data includes:
and training a preset model based on the target historical meteorological data and the target historical power data by adopting a gradient unilateral sampling method and a mutual exclusion feature binding method, wherein the gradient unilateral sampling method is a method for sampling according to the gradient size, the mutual exclusion feature binding method is a method for combining mutual exclusion features into a single feature, and the mutual exclusion feature is a feature which does not adopt a non-zero value in a sparse feature space at the same time.
In another aspect, the present application provides a wind farm power prediction apparatus, where the maintenance apparatus includes:
the data acquisition module is used for acquiring target historical meteorological data and target historical power data of each wind turbine generator in the wind power plant;
the model training module is used for training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, and the preset model carries out dynamic time establishment on a multivariable time sequence through a lightGBM model;
the prediction module is used for substituting the meteorological forecast data into the target model of each wind turbine generator respectively to obtain the predicted power of each wind turbine generator;
and the calculation module is used for adding the predicted power of each wind turbine to obtain the predicted power of the wind power plant.
In a third aspect, an embodiment of the present application provides an electronic device, including: a processor, a memory and a bus, the memory storing machine readable instructions executable by the processor, the processor and the memory communicating over the bus when an electronic device is running, the machine readable instructions when executed by the processor performing the steps of the wind farm power prediction method as described above.
In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, performs the steps of the wind farm power prediction method as described above.
The embodiment of the invention has the following beneficial effects:
acquiring target historical meteorological data and target historical power data of each wind turbine in a wind power plant; training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, wherein the preset model carries out dynamic time establishment on a multivariable time sequence through a lightGBM model; respectively substituting the weather forecast data into the target model of each wind turbine to obtain the predicted power of each wind turbine; and adding the predicted power of each wind turbine generator to obtain the predicted power of the wind power plant. And a target model of each wind turbine is established according to the difference of the single turbine in the wind power plant, so that the prediction accuracy of the single turbine is improved, and the power prediction accuracy of the wind power plant is further improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Wherein:
FIG. 1 is a flow chart of a method for predicting power of a wind farm provided by an embodiment of the present application;
FIG. 2 is a flow chart of yet another wind farm power prediction method provided by an embodiment of the present application;
FIG. 3 is a schematic structural diagram of a wind farm power prediction device provided in an embodiment of the present application;
fig. 4 is a schematic structural diagram of an electronic device provided in an embodiment of the present application;
fig. 5 is a schematic structural diagram of a storage medium according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, an embodiment of the present application provides a method for predicting power of a wind farm, including:
s101, obtaining target historical meteorological data and target historical power data of each wind turbine generator in a wind power plant;
illustratively, target historical power data and target historical meteorological data of each wind turbine in a wind power plant are obtained, the target historical power data comprise a wind power sequence, the wind power sequence can be wind power sampled according to sampling time, and the sampling time interval is 15 minutes. The target historical meteorological data comprise meteorological information such as wind speed, wind direction, air temperature, air pressure and humidity which are related to the wind power value.
S102, training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, wherein the preset model carries out dynamic time establishment on a multivariable time sequence through a lightGBM model;
illustratively, a python library LightGBM model is called as a preset model, the target historical meteorological data is used as input data, the corresponding target historical power data is used as corresponding output data, and the input data and the corresponding output data are substituted into the LightGBM model to be trained to obtain a target model of each unit. The LightGBM model is an improved gradient boosting decision tree framework.
S103, substituting the meteorological forecast data into the target model of each wind turbine generator to obtain the predicted power of each wind turbine generator;
exemplarily, the meteorological forecast data of the area where the wind power plant is located are respectively substituted into the target model of each wind power unit to obtain the predicted power of each wind power unit, and the meteorological features in the meteorological forecast data are extracted and substituted into the target model of each wind power unit to obtain the predicted power of each wind power unit.
And S104, adding the predicted power of each wind turbine to obtain the predicted power of the wind power plant.
The problem that when the existing prediction technology is applied to a centralized large wind farm, the prediction adaptability of a single unit of a distributed wind farm is weak is solved; the difference of single units cannot be considered; for the characteristics of unobvious periodic changes of wind speed, wind direction and the like, the problem that the traditional network model cannot extract effective front and back related information for prediction so as to influence the prediction result is solved.
In a possible embodiment, the step of obtaining target historical meteorological data and target historical power data of each wind turbine in the wind power plant includes:
acquiring initial historical meteorological data and initial historical power data of each wind turbine in a wind power plant;
carrying out outlier removal processing on the initial historical meteorological data and the initial historical power data by using a local outlier algorithm;
performing linear interpolation on the processed initial historical meteorological data and initial historical power data to obtain complete historical meteorological data and complete historical power data;
and carrying out normalization processing on the complete historical meteorological data and the complete historical power data to obtain target historical meteorological data and target historical power data of each wind turbine generator in the wind power plant.
Exemplarily, abnormal value detection is carried out on initial historical meteorological data and initial historical power data of each wind turbine generator in a wind power plant by using a local abnormal factor algorithm, abnormal data in the initial historical meteorological data and the initial historical power data are detected and removed in a self-adaptive mode to obtain a feature data set containing a missing value, and linear interpolation operation is carried out on the obtained feature data set containing the missing data to obtain a complete data set.
Illustratively, because there are differences in dimension between different variables, which will cause differences in quantity, in order to consider the influence of each factor variable more comprehensively and fairly, each variable and the wind power time series are normalized, due to the aperiodic trend of weather data, the signal is transformed from non-stationary state to stationary state by the method of first order difference, the wind speed, wind direction, air temperature, air pressure and wind power are normalized by the limit value of sample data, and the value is assigned to [ -1,1], and the formula is as follows:
in the formula, x min And x max Respectively a minimum and a maximum of the variable.
In a possible embodiment, before the step of training the preset model based on the target historical meteorological data and the target historical power data, the method further comprises:
and performing correlation analysis on the target historical meteorological data and the target historical power data by adopting an MIC algorithm, and eliminating meteorological features of which the correlation degree with the target historical power data is lower than a threshold value in the target historical meteorological data to obtain optimized target historical meteorological data.
Illustratively, a MIC algorithm is used for performing correlation analysis on the target historical meteorological data and the target historical power data of each unit, a correlation coefficient is calculated, meteorological features with high correlation with the target historical power data in the target historical meteorological data are reserved, meteorological features with low correlation with the target historical power data in the target historical meteorological data are removed, and therefore prediction accuracy of a model is improved while overfitting of the model is reduced.
And (4) performing a Maximum Information Coefficient (MIC) algorithm to select an effective input variable set. The MIC algorithm assumes that a grid is drawn on a data scatter diagram of related variables to measure the correlation between the two, and defines the mutual information between x and y as:
selecting the maximum value of mutual information as a final MIC under the condition of adopting grid division of different standards, wherein the calculation formula is as follows:
in the formula: the number of the division grids in the x and y directions is a and B, respectively, and the maximum value of the division grid is B.
In a possible implementation manner, the step of training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator includes:
dividing the optimized target historical meteorological data and the target historical power data into a training set, a verification set and a test set;
and training the preset model based on the training set, the verification set and the test set to obtain a target model of each wind turbine.
Illustratively, the training set, the validation set, and the test set are proportionally arranged in the normalized data set, i.e., the optimized target historical meteorological data and the target historical power data, and 80% of the training set, 10% of the validation set, and 10% of the test set are set.
In a possible implementation manner, the step of training the preset model based on the training set, the verification set, and the test set to obtain the target model of each wind turbine includes:
optimizing the hyper-parameters of the preset model according to the verification set and a firefly lifting algorithm introducing inertial weight;
and training the optimized preset model based on the training set and the test set to obtain a target model of each wind turbine.
Illustratively, the training set trains the model, the verification set is used for improving the firefly algorithm to optimize the hyper-parameters of the preset model, and the test set tests the generalization ability of the optimized preset model to avoid overfitting of the optimized preset model.
In one possible embodiment, the inertial weight is a nonlinear inertial weight, and the adjustment is performed by using a nonlinear function, and the tanh function is used as a basis function for constructing the nonlinear inertial weight.
Illustratively, the model is tested using a validation set, and the lightGBM model, i.e., the hyper-parameters of the pre-set model, is tuned using an improved firefly algorithm. The tuning standard comprises the following steps: root mean square error:
in the formula, n is the number of the predicted verification data; y is i And
actual values and predicted values of the data are respectively; i is the predicted point sequence number.
The firefly algorithm is mainly optimized according to the principle that fireflies with large brightness attract fireflies with small brightness in the whole search space, and comprises the following steps:
initializing parameters such as population scale, space dimension, iteration times and the like, and randomly initializing the position x of the firefly t ＝[x 1 ,x 2 ,L,x n ]The maximum fluorescence brightness of firefly is set as the objective function value.
Calculating the relative brightness I and the attraction degree beta of the fireflies in the population, and determining the moving direction of the fireflies according to the relative brightness I, wherein the calculation formulas of I and beta are as follows:
in the formula I 0 Maximum fluorescence brightness; gamma is the light intensity absorption coefficient; r is a radical of hydrogen ij Is the distance between fireflies i and j; beta is a 0 Is the maximum attraction factor.
Aiming at the condition that the standard firefly algorithm is easy to be premature and the oscillation is easy to be generated near the global optimal solution in the later period, an inertia weight w is introduced, and the position iteration is disclosed as follows:
x i (t+1)＝w i (t)·x i +β×(x j -x i )+α×(rand-0.5)
when w is larger, the algorithm has stronger global search capability, and when w is smaller, the algorithm has stronger local search capability. In order to search a global optimal solution, a nonlinear function is adopted to dynamically adjust the inertia weight. And selecting the tanh function as a basic function for constructing the nonlinear inertia weight. The functional formula is:
selecting a =5 to perform w function construction.
In order to make the tanh function conform to the variation law of w, the formula of w is defined as:
in the formula: t is the current iteration number; t is the maximum iteration number; w is a max ，w min The maximum and minimum values of w.
And recalculating the brightness of the firefly according to the updated position of the firefly.
And when the stopping condition is met, outputting a global extreme point and an optimal individual value, otherwise, searching for the next time.
In one possible embodiment, the step of training a preset model based on the target historical meteorological data and the target historical power data includes:
and training a preset model based on the target historical meteorological data and the target historical power data by adopting a gradient unilateral sampling method and a mutual exclusion feature binding method, wherein the gradient unilateral sampling method is a method for sampling according to the gradient size, the mutual exclusion feature binding method is a method for combining mutual exclusion features into a single feature, and the mutual exclusion feature is a feature which does not adopt a non-zero value in a sparse feature space at the same time.
Illustratively, the LightGBM model, i.e., the predictive model, is an improved gradient boosting decision tree framework. Combining M weak regression trees into a strong regression tree in a linear mode, wherein the formula is as follows:
in the formula: f (x) is the final output value; f. of m (x) The output value of the m weak regression tree.
Aiming at the defects of efficiency and expandability of an algorithm under the traditional Boosting framework, the LightGBM model is improved, namely a gradient unilateral sampling method and a mutual exclusion characteristic binding method are adopted. Gradient unilateral sampling means that the LightGBM does not use all samples of a training set when training parameters of a model, but samples according to the gradient, only selects sample data with high gradient to calculate information gain, and mutually exclusive feature binding means that in a sparse feature space, most features cannot simultaneously take a non-0 value, and the mutually exclusive features are combined into a single feature so as to achieve the purpose of reducing feature dimensionality.
In order to improve the parallelism capacity of model training, the LightGBM model not only performs a gradient sampling strategy on data of a training sample, but also performs fusion binding on high-dimensional sparse mutual exclusion features, for example, the LightGBM binds the high-dimensional sparse and mutual exclusion features together in a histogram-based manner to improve the splitting efficiency of nodes, and first disperses data of input continuous features into k integer values to form k bound results, wherein each feature in each result is mutually exclusive, and then constructs a k-wide histogram, and only the duration accumulation of each discrete value on the histogram needs to be counted when training data passes through.
When the splitting gain is calculated, the discrete value of the sequencing histogram is traversed, so that only k times of calculation are needed, and when the feature is split, only the value after the feature discretization is needed to be stored.
In a possible implementation manner, as shown in fig. 2, in the data acquisition process, acquisition errors caused by hardware devices such as sensors may have a certain influence on the later-stage wind power prediction result. According to the method, meteorological data and power data collected by each wind turbine generator sensor are subjected to data preprocessing in advance, a local abnormal factor algorithm is used for removing abnormal values and then linear interpolation is carried out, finally, the characteristic set is subjected to normalization processing, in order to avoid prediction accuracy of a model to overfitting or reducing the model caused by excessive input data, correlation analysis is carried out on the meteorological data and the power data of each wind turbine generator by using an MIC algorithm, meteorological features with high correlation are reserved, meteorological features with low correlation are removed to improve the prediction accuracy of the model, GMB model is adopted to carry out dynamic time modeling on a multivariable time sequence, and improved firefly algorithm is used for optimizing the hyper-parameters of the model to improve the prediction performance of the model, wherein inertia weight is introduced to improve the global search capability of the firefly algorithm.
In another aspect, as shown in fig. 3, the present application provides a wind farm power prediction device, where the maintenance device includes:
the data acquisition module 201 is used for acquiring target historical meteorological data and target historical power data of each wind turbine generator in the wind power plant;
the model training module 202 is configured to train a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine, and the preset model performs dynamic time establishment on a multivariate time sequence through a lightGBM model;
the prediction module 203 is used for substituting the weather forecast data into the target model of each wind turbine generator respectively to obtain the predicted power of each wind turbine generator;
and the calculating module 204 is configured to add the predicted power of each wind turbine to obtain the predicted power of the wind farm.
In one possible implementation, as shown in fig. 4, an embodiment of the present application provides an electronic device 300, including: comprising a memory 310, a processor 320 and a computer program 311 stored on the memory 310 and executable on the processor 320, when executing the computer program 311, implements: acquiring target historical meteorological data and target historical power data of each wind turbine generator in a wind power plant; training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, wherein the preset model carries out dynamic time establishment on a multivariable time sequence through a lightGBM model; respectively substituting the meteorological forecast data into the target model of each wind turbine to obtain the predicted power of each wind turbine; and adding the predicted power of each wind turbine to obtain the predicted power of the wind power plant.
In one possible implementation, as shown in fig. 5, the present application provides a computer-readable storage medium 400, on which a computer program 411 is stored, where the computer program 411 implements, when executed by a processor: acquiring target historical meteorological data and target historical power data of each wind turbine generator in a wind power plant; training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, and carrying out dynamic time establishment on a multivariable time sequence through a lightGBM model by the preset model; respectively substituting the meteorological forecast data into the target model of each wind turbine to obtain the predicted power of each wind turbine; and adding the predicted power of each wind turbine to obtain the predicted power of the wind power plant.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It will be understood by those skilled in the art that the modules or steps of the present invention described above can be implemented by a general purpose computing device, they can be centralized in a single computing device or distributed over a network of multiple computing devices, and they can alternatively be implemented by program code executable by a computing device, so that they can be stored in a storage device and executed by a computing device, or they can be separately fabricated into various integrated circuit modules, or multiple modules or steps thereof can be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in some detail by the above embodiments, the invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the invention, and the scope of the invention is determined by the scope of the appended claims.
The above disclosure is only for the purpose of illustrating the preferred embodiments of the present invention, and it is therefore to be understood that the invention is not limited by the scope of the appended claims.
Claims (10)
1. A wind farm power prediction method is characterized by comprising the following steps:
acquiring target historical meteorological data and target historical power data of each wind turbine in a wind power plant;
training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, wherein the preset model carries out dynamic time establishment on a multivariable time sequence through a lightGBM model;
respectively substituting the meteorological forecast data into the target model of each wind turbine to obtain the predicted power of each wind turbine;
and adding the predicted power of each wind turbine generator to obtain the predicted power of the wind power plant.
2. The wind farm power prediction method of claim 1, wherein the step of obtaining target historical meteorological data and target historical power data for each wind turbine generator within the wind farm comprises:
acquiring initial historical meteorological data and initial historical power data of each wind turbine generator in a wind power plant;
carrying out outlier removal processing on the initial historical meteorological data and the initial historical power data by using a local outlier algorithm;
performing linear interpolation on the processed initial historical meteorological data and initial historical power data to obtain complete historical meteorological data and complete historical power data;
and carrying out normalization processing on the complete historical meteorological data and the complete historical power data to obtain target historical meteorological data and target historical power data of each wind turbine generator in the wind power plant.
3. The wind farm power prediction method of claim 1, further comprising, prior to the step of training a preset model based on the target historical meteorological data and the target historical power data:
and performing correlation analysis on the target historical meteorological data and the target historical power data by adopting an MIC algorithm, and eliminating meteorological features of which the correlation degree with the target historical power data is lower than a threshold value in the target historical meteorological data to obtain optimized target historical meteorological data.
4. The wind farm power prediction method according to claim 3, wherein the step of training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator comprises:
dividing the optimized target historical meteorological data and the target historical power data into a training set, a verification set and a test set;
and training the preset model based on the training set, the verification set and the test set to obtain a target model of each wind turbine.
5. The wind farm power prediction method according to claim 4, wherein the step of training the preset model based on the training set, the verification set and the test set to obtain the target model of each wind turbine includes:
optimizing the hyper-parameters of the preset model according to the verification set and a firefly lifting algorithm introducing inertial weight;
and training the optimized preset model based on the training set and the testing set to obtain a target model of each wind turbine.
6. A wind farm power prediction method according to claim 5, characterized in that the inertial weight is a non-linear inertial weight, which is adjusted using a non-linear function, and that a tanh function is used as a basis function for constructing the non-linear inertial weight.
7. The wind farm power prediction method of claim 6, wherein the step of training a predetermined model based on the target historical meteorological data and the target historical power data comprises:
training a preset model by adopting a gradient unilateral sampling method and a mutual exclusion feature binding method based on the target historical meteorological data and the target historical power data, wherein the gradient unilateral sampling method is a method for sampling according to the gradient size, the mutual exclusion feature binding method is a method for combining mutual exclusion features into a single feature, and the mutual exclusion feature is a feature which does not adopt a nonzero value in a sparse feature space at the same time.
8. A wind farm power prediction device, wherein the maintenance device comprises:
the data acquisition module is used for acquiring target historical meteorological data and target historical power data of each wind turbine generator in the wind power plant;
the model training module is used for training a preset model based on the target historical meteorological data and the target historical power data to obtain a target model of each wind turbine generator, and the preset model carries out dynamic time establishment on a multivariable time sequence through a lightGBM model;
the prediction module is used for substituting the meteorological forecast data into the target model of each wind turbine generator respectively to obtain the predicted power of each wind turbine generator;
and the calculation module is used for adding the predicted power of each wind turbine generator to obtain the predicted power of the wind power plant.
9. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine readable instructions executable by the processor, the processor and the memory communicating over the bus when an electronic device is running, the machine readable instructions when executed by the processor performing the steps of the wind farm power prediction method according to any of claims 1 to 7.
10. A computer-readable storage medium, characterized in that it has stored thereon a computer program which, when being executed by a processor, carries out the steps of the wind farm power prediction method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211239008.4A CN115564115A (en) | 2022-10-11 | 2022-10-11 | Wind power plant power prediction method and related equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211239008.4A CN115564115A (en) | 2022-10-11 | 2022-10-11 | Wind power plant power prediction method and related equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115564115A true CN115564115A (en) | 2023-01-03 |
Family
ID=84744833
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211239008.4A Pending CN115564115A (en) | 2022-10-11 | 2022-10-11 | Wind power plant power prediction method and related equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115564115A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117559519A (en) * | 2023-10-31 | 2024-02-13 | 北京瑞科同创能源科技有限公司 | Wind power plant power data prediction method and device and electronic equipment |
-
2022
- 2022-10-11 CN CN202211239008.4A patent/CN115564115A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117559519A (en) * | 2023-10-31 | 2024-02-13 | 北京瑞科同创能源科技有限公司 | Wind power plant power data prediction method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cui et al. | Wind power ramp event forecasting using a stochastic scenario generation method | |
| Li et al. | Applications of Bayesian methods in wind energy conversion systems | |
| CN111127246A (en) | Intelligent prediction method for transmission line engineering cost | |
| CN108053048A (en) | A kind of gradual photovoltaic plant ultra-short term power forecasting method of single step and system | |
| CN114006369B (en) | Regional wind and light station power joint prediction method and device, electronic equipment and storage medium | |
| Sogabe et al. | Optimization of decentralized renewable energy system by weather forecasting and deep machine learning techniques | |
| CN112100911B (en) | Solar radiation prediction method based on depth BILSTM | |
| CN112285807A (en) | Meteorological information prediction method and device | |
| CN113822418A (en) | Wind power plant power prediction method, system, device and storage medium | |
| CN115564115A (en) | Wind power plant power prediction method and related equipment | |
| CN116402203A (en) | Method, system and medium for predicting short-time photovoltaic power generation capacity considering weather conditions | |
| CN115239105A (en) | Method and device for evaluating wind resources of in-service wind power plant | |
| WO2020228568A1 (en) | Method for training power generation amount prediction model of photovoltaic power station, power generation amount prediction method and device of photovoltaic power station, training system, prediction system and storage medium | |
| Bianchini et al. | Estimation of photovoltaic generation forecasting models using limited information | |
| CN116703638A (en) | Greenhouse vegetable planting management system and method thereof | |
| Al-Hilfi et al. | Estimating generated power of photovoltaic systems during cloudy days using gene expression programming | |
| CN110019167A (en) | Long-term new forms of energy resource data base construction method and system in one kind | |
| CN113962432A (en) | Wind power prediction method and system integrating three-dimensional convolution and light-weight convolution threshold unit | |
| Guo et al. | Short‐term photovoltaic power forecasting with adaptive stochastic configuration network ensemble | |
| Nayak et al. | Short-term wind speed forecasting using multi-source multivariate rnn-lstms | |
| Zhao et al. | An interpretable ultra-short-term wind power prediction model based on the feature matrix reconstruction through regression trees | |
| Ma et al. | Crop Water Requirement Prediction Method Based on EEMD-Attention-LSTM Model | |
| CN116544931B (en) | Power load distribution prediction method based on integrated fragment transformation and time convolution network | |
| Su et al. | Growth height prediction for the trees under overhead lines based on deep learning algorithm | |
| Huang et al. | Air quality monitoring of landscape architecture based on multi-sensor fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |