Research On Short-Term Wind Power Prediction Method Based On Combinatorial Model

In recent years,China’s wind power generation technology has gradually matured,and the installed capacity of wind power and the integration of wind power into the grid have been increasing,which has brought challenges to the balance of supply and demand of the power grid.Therefore,in the actual operation of wind farms,accurate wind power prediction is of great significance to grid dispatch and wind power consumption.Based on the combinatorial model,this paper studies short-term wind power prediction to improve the short-term wind power prediction accuracy of wind farms,and the main research contents are:(1)Aiming at the problems of local optimization and slow convergence in the late iteration of the sparrow search algorithm(SSA),the Tent chaotic mapping is used to initialize the sparrow population to improve the diversity and distribution uniformity of the sparrow population;Golden sine and linear decline strategies were introduced to improve the position update formula of finder and vigilant in sparrow populations.Gaussian variation and chaotic disturbance methods are used to improve the local search ability of the model and accelerate the convergence speed.Then,the ISSA algorithm is used to optimize the regularization parameters and kernel function width of the least squares support vector machine(LSSVM),and the ISSA-LSSVM prediction model is established.Through example simulation,the ISSA algorithm is used to optimize the hyperparameters,which has little impact on the overall operation time and improves the prediction accuracy by 1.16%.(2)After preprocessing the original dataset by using the isolated forest algorithm,the wind power sequence is decomposed into different sub-sequences according to frequency characteristics by variational mode decomposition(VMD),which weakens its non-stationarity and further mines the hidden information,and then enters each subseries into the ISSA-LSSVM prediction model respectively,establishes the VMD-ISSSA-LSSVM model for prediction,and finally adds the prediction results of each subseries to obtain the prediction results of wind power.Through the simulation of the example,the prediction accuracy reached 98.72%.(3)The subsequence obtained by VMD decomposition was classified into high-frequency components and low-frequency components by running discrimination method,and the high-frequency components with high complexity were predicted by the long short-term memory network(LSTM)model,and the low-frequency components were predicted by the ISSA-LSSVM model,and a short-term wind power prediction model based on the VMD-ISSA-LSSVM combined with VMD-LSTM combined model was established.Through the simulation of the example,the prediction accuracy of the proposed combined model reaches 99.082%,which can well integrate the advantages of the two prediction models and effectively improve the prediction effect of wind power.