Research On Wind Power Ramp Detection And Prediction Based On Data Driven

Wind power ramp events refers to the large fluctuation of wind power power in a short time interval,which will seriously affect the safe and stable operation of the grid system.In order to maintain the stable operation of the power grid system,wind power ramp detection and wind power ramp prediction are extremely necessary.Therefore,how to improve the accuracy of wind power power climbing detection and climbing prediction is a problem worthy of study.In the field of ramp events detection and ramp events prediction,there are their own problems.In the existing wind power ramp events detection algorithm,the accuracy of the ramp endpoint is not considered.In the existing wind power ramp events prediction algorithm based on indirect prediction,there are problems of prediction lag and accuracy when the prediction step is large.Aiming at the problem of the accuracy of the end point of the ramp section,this paper proposes a wind power ramp detection algorithm RPCRD(Ramp Point Correct Ramping Detection)based on the correction of the ramp point.The algorithm proposes a merging method of ramp sections to solve the fracture problem,and proposes a scoring mechanism for ramp point selection to find the two extreme points that best match the ramp characteristics,and correct the separate ramp points of the wind power ramp section.The experimental results show that the RPCRD algorithm can effectively improve the accuracy of the ramp events detection by modifying the ramp point.Aiming at the problem of prediction lag and accuracy when the prediction step is large,this paper proposes a wind power ramp events prediction algorithm RCRP(Residual Correction Ramping Prediction)based on an end-to-end model and residual correction.In order to solve the problem of the lag of multi-step prediction results and the problem of prediction accuracy when the prediction step size is large,RCRP innovatively proposed a wind power prediction model based on Bi LSTM-Seq2Seq-AM and a residual prediction model based on TCN.Experimental results show that the wind power prediction model in the RCRP algorithm can effectively solve the problem of prediction lag,and the residual correction model can improve the prediction accuracy when the prediction step is large.In order to verify the availability of the RPCRD algorithm and the RCRP algorithm,this paper implements the ramp events prediction system Ramp Pred based on the above algorithm.The system test verifies the usability and real-time performance of Ramp Pred.