为提高风速序列的预测性能，提出一种改进的遗传混沌算子网络预测方法。混沌算子网络由输入层、中间层和输出层3层组成，网络的输入层与中间层的连接权值采用线性衰减的方式设计，中间层混沌算子单元的激励函数为混沌映射函数，采用遗传算法优化网络的权值和混沌算子控制参数。利用差分方法对被预测序列进行平稳化预处理，结合相空间重构理论利用平稳化后的数据构造网络的训练样本。仿真实验结果表明：该方法能够实现风速序列的多步预测分析，其预测性能优于传统预测方法，尤其随着预测步长的增加，该方法具有相对稳定的预测性能。%In order to enhance the prediction performance of the wind speed series, an improved chaotic operator network based on genetic algorithm is proposed. The chaotic operator network contains input layer , middle layer and output layer. The connective weights between the input layer and the middle layer are designed by the linear attenuation. The activation functions of the chaotic operators in the middle layer are chaotic map functions. Genetic algorithm is used to optimize the weights and the control parameters in the chaotic operators. Difference method is used to preprocess the predicted wind speed series as the stationary time series. Combing the phase space reconstruction theory, the training samples are constructed by the stationary time series. The simulation results show that the method proposed in the paper can complete the multi-step-ahead prediction analysis of the wind speed series, and it has better prediction performance than the conventional method. Especially, the prediction performance of the method is relatively stable as the prediction step increases.
展开▼
