Study On Short-term Wind Speed And Power Prediction Technology Of Wind Farm

With the advancement of society and the continuous development of science and technology,people’s consumption of energy is increasing,and the contradiction of energy crisis is becoming more and more serious.As a renewable energy source of cleaning and pollution-free,wind energy has been paid more and more attention to its development and utilization.Due to the random,intermittent and uncontrollable characteristics of wind,the prediction of wind speed and power is difficult,and the large-scale grid-connection of wind power has a serious impact on the scheduling and stability of power system.This is an effective way to solve the problem of wind power integration,reduce the impact on the power system and improve the security and economy of the system through accurate prediction of short-term wind speed and power of the wind farm.To this end,this paper study the prediction technology of short-term wind speed and power.The main research contents are as follows:(1)In order to improve the accuracy of wind power short-term wind speed prediction,this paper proposes a short-term wind speed combined forecasting model.Firstly,the complete integration based on adaptive noise empirical mode decomposition is used to decompose the original wind speed time series and reduce the influence between different characteristic scale sequences.Secondly,in order to reduce the scale of computation,the each component sequence of decomposition is calculated for PE,and the components with similar entropy value are superimposed to form a new sequence.Finally,BP neural network is used as the basic prediction model.according to the weights of BP neural network in the initialization and random problems existing in the selection of threshold,BP parameters are optimized by quantum genetic algorithm,Each new series is forecasted by the established model and the final wind speed forecasting value is obtained by superimposing the forecasting results.(2)For traditional deterministic point prediction of wind speed,there are generally different degrees of errors and uncertainty in the results,The interval prediction of Short-term wind speed is proposed.Firstly,the original time series of wind speed is decomposed by LMD into multiple sub-sequences to reduce the interaction between different feature scale sequences.Secondly,in order to improve the calculation efficiency,the fuzzy entropy is computed for each component sequence obtained by decomposition,and the components with similar entropy value are superimposed to form a new sequence.Finally,for the problem of randomness of ELM in the selection of input weight and hidden layer bias,the the hybrid grey Wolf algorithm algorithm is used to optimize the ELM parameters and each new sequence is predicted separately.The prediction results are superimposed and then the predicted value of the superposition is estimated by the T distribution to obtain the prediction interval.(3)Considering that the method of fitting power curve is used for short-term powerprediction,The key of the prediction is the accuracy of the power curve model.Therefore,a power curve model based on the fruit fly optimization algorithm is proposed.Firstly,for the problem of historical wind speed-power data have many bad points and the fitting curve can not reflect the real performance of the fan.The smoothing function is used to detect and filter the validity of the data to eliminate invalid bad points.Then,for the data of eliminating the bad points are fitted by curve fitting method separately.Finally,by comparing the fitting effects of several modeling methods,the fruit fly optimization algorithm is used to establish power curve model and combining short-term wind speed prediction results to achieve wind power prediction.