Variable frequency wind speed trend prediction system based on combined neural network and improved multi-objective optimization algorithm

Trend prediction data with low measurement frequency has always been needed in wind power station, but the traditional multi-step prediction method has caused error accumulation and led to poor prediction accuracy, in order to solve this problem, a new wind speed trend prediction system is proposed which includes data preprocessing (Fuzzy Information Granulation), combined neural network prediction and an improved multi-objective manta rays foraging optimization based on Tent chaotic map and T -distribution perturbation operator ( IMOMRFO ). The algorithm not only has a good ability to escape from the local optimal solution, but also proves theoretically that the Pareto optimal solution is obtained. Through the simulation of four groups of experiments, it is obvious that the stability, generalization and accuracy of the model are satisfactory. It is confirmed that the model greatly improves the accuracy of trend prediction and makes a certain contribution to solve the problem of wind speed prediction, through the test of the ability of point prediction and interval prediction of the model. • A new wind speed trend prediction system is proposed. • FIG method can effectively realize the frequency conversion of wind speed. • An improved optimizer based on chaotic map and perturbation operator is proposed. • The combined neural network can effectively improve the prediction accuracy. • More information can be provided by point prediction and interval prediction.