Researching And Realization Of Wind Speed Forecast Model For Wind Farms

Wind speed prediction is an important part in planning and designing in the wind electric field. In the process of prediction the model must be setup rightly. Wind speed in itself has a timing sequence and auto-correlation, therefore, the model can be set up by using the time-series prediction. At present, the model has been widely used. But,it exists the problem of delay and low prediction accuracy.To address the problem, scholars have put forward a time-series analysis and Kalman filter hybrid algorithm combining. To some extent, it achieved good results. However, its still not ideal,many scholars are still in exploring and studding to find a better mathod.In this paper, by analyzing the structure of interpolation filter (DDF), the time series and interpolation filtering algorithm are combined.By using the wind speed data measured at JiShan, And using time series method, time series and Kalman filter hybrid algorithm, time series with the interpolation filter (DDF) hybrid algorithm made predictions respectively, and then comparaed the results of these three prediction methods. that the mixed filter based on the time series and interpolation filter have the obvious advantages. Examples of analysis shows that using the hybrid algorithm has a higher prediction accuracy than a single time series, as well as by the time series prediction and Kalman filter prediction of the hybrid,Also it improves the delay problem to some extent.First of all,By using the wind speed data measured in Jishan,I setup the model based on time series theory. In this paper, It's the ARIMA (1,1,1),the information criteria AIC(An Information Criterion) function is used to test the validity of the model, By analysis the predicted results,it shows the existence of its low prediction accuracy and the latency problem. Then derived the state equation and observation equation, and than predicted the wind speed mixed filter based on time series and Kalman filter and the mixed filter based on time series and DDF respectively.Finally, to illustrate the practical feasibility,Here I used the TI`s DSP chip- TMS320F2812 as the controller to control the speed of motor based on the forecasted wind speed. Through the physical simulation illustrated its practical significance. By analysing the results of the simulation and experimental, shows that: the mixed way Based on the time series and DDF is feasible in the wind speed prediction, and the delay problem can be reduced to some extent, it can improve the accuracy of prediction also.Also here analysed the problem still exists.