基于多ANN模型的复杂系统长时段预报

LONG TERM PREDICTION FOR COMPLEX SYSTEM BASED ON MULTIPLE ANN MODELS

提出一种多ＡＮＮ结构的极值聚类训练算法，并将这种方法应用于复杂系统长时段预报．采用这种方法，可以提高长时段预报精度、增强模型的可靠性．以这种模型为基础可以进一步建立基于多ＡＮＮ模型的复杂系统预测控制．

An extremum clustering approach for multiple artificial neural networks is developed in this paper. Then it is used for long term prediction of complex dynamic system. With this method the long term prediction precision and model robustness can be improved. This approach can be used for complex system predictive control using multiple ANNs.