科研之友
微信
新浪微博
Facebook
分享链接
摘要
Artificially given amplitude and ensemble number of white noises and the randomness of white noises causes the uncertainty to the decomposed results of ensemble empirical mode decomposition （EEMD），leading to the imperfect decomposed results in application to the wind power prediction by EEMD. The effect mechanism of the parameters of white noises on decomposed results of EEMD is studied，and the method called adaptive EEMD based on fractal characteristics of modes is proposed in this paper. In the different white noises and different parameters of white noises，the modes decomposed by EEMD exhibit the different fractal characteristics. Particle swarm optimization algorithm is adopted to calculate the fractal dimensions of modes in different parameters，so as to achieve the precise decomposition for EEMD. Employing long short term memory（LSTM）algorithm which has great nonlinear modeling ability to predict decomposed components obtained by adaptive EEMD. Simulated signal and actual wind power data from two wind farms are analyzed. Adaptive EEMD could avoid the uncertainty brought by the randomness of white noises and artificially given parameters. Compared with three benchmark models，the RMSE is significantly reduced by adaptive EEMD combined with LSTM model on two groups of wind power data，which verifies the effectiveness of the proposed method. ? 2023 Science Press.
- 单位
