Wind Power Interval Prediction Based on Improved Whale Optimization Algorithm and Fast Learning Network

The uncertainty of wind power brings great challenges to large-scale wind power integration. The conventional point prediction of wind power is difficult to meet the demand of power grid planning and operation. In contrast, interval prediction is gaining increasing attention as an effective approach as it can describe the uncertainty of wind power. An interval prediction model based on an improved whale optimization algorithm (IWOA) and fast learning network (FLN) was developed in this study. First the convergence speed and accuracy of the IWOA was enhanced by adjusting the nonlinear convergence factor, and by adding adaptive inertia weights and a chaos search strategy. Second, a novel evaluation index was proposed according to the lower and upper bound estimation method. The proposed evaluation index was considered as a fitness function, and the FLN parameters were optimized by the IWOA to output the final prediction interval. The examples considered in this study revealed that the proposed method can be employed to reduce the prediction interval normalized root-mean-square bandwidth (PINRW) and the prediction interval average deviation (PIAD) more than 3% and 8% at the 90% and 80% confidence level respectively, which has a high practical significance.