基于混沌时间序列的Elman神经网络工业用电预测

Elman neural network for forecasting industrial electricity consumption based on chaotic time series

针对电力负荷数据在多重因素相互影响下呈现非线性特性甚至是混沌性的问题,采用基于相空间重构的Elman神经网络方法进行全社会工业月用电量预测.利用小数据量法计算最大Lyapunov指数,判别负荷时间序列的混沌性,进而确定最优延迟时间及最佳嵌入维数进行相空间重构,以此确定Elman神经网络的拓扑结构,并将实测数据带入模型进行训练.通过对实测数据进行预测仿真,表明该模型达到了较好的预测效果,验证了提出的时间序列相空间重构与Elman神经网络结合的正确性与有效性.

In order to solve the problem that the electric power load data displays a non-linear feature and even chaos characteristics under the mutual influence of multiple factors, the industrial month electricity consumption in the whole society was forecasted with Elman neural network method based on phase space reconstruction. The largest Lyapunov exponent was calculated with small-data method, and the chaos characteristics of load time series were judged. In addition, the optimal delay time and the best embedding dimension were determined to perform the phase space reconstruction, and thus the topology structure of Elman neural network was determined. Moreover, the measured data were put into the model for training. The forecasting and simulation for the measured data indicate that the model achieves better prediction effect, and the correctness and effectiveness in the combination of time series phase space reconstruction and Elman neural network is verified.