Wind power forecast based on broad learning system and simplified long short term memory network

Due to its strong randomness and volatility, the modes of wind power are complex. After decomposing time series into three subseries, complex deconstructed each subseries maintain unique characteristics. Aiming at characteristics a forecast method based on Broad Learning System (BLS) Simplified Long Short Term Memory (SLSTM) is proposed. Firstly, decomposed analysed. The first layer reflects main amplitude changes caused by ramp events second represents local fluctuations accompanying events. events, new appear frequently in two subseries. third consists continuous irregular jitter. Then, BLS introduced By adjusting structure dynamically, various can be learned from change fluctuation. Additionally, according jitter needn't transmitted long-term memory. Setting forget gate 0, SLSTM proposed Finally, data Belgian ELIA website used verify superiority method.