摘要
介绍了电力系统由结构稳定到倍周期分岔,直至混沌的自然规律,提出在先于混沌的分岔区域内仍可进行系统预测,其预测方法一定要注意接近混沌区域这一特性,应注重开发对初始条件敏感的预测方法;进行了工程实例验证,说明采用灰色模型(GM(1,1)、GM(1,1,ω*))、神经网络(BP)等预测方法可在小扰动临界状态得到较好的预测结果,优化后的GM模型、与BP网络相融合的预测模型的模拟精度更高。
The law from structural stability to period doubling bifurcation and chaos in power systems was analyzed. Bifurcation region before chaos could be predicated. The feature that is close to chaos must be noticed to get a reasonable result, and predication methods should be sensitive to original conditions. Predication methods, such as Gray Model (GM(1,1),GM(1,1, ω *)) and Back-propagation (BP) neural networks, are tested by engineering examples on small signal critical state and its more accurate predicated results can be gained by optimized GM model and BP neural networks.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2004年第12期52-57,共6页
Proceedings of the CSEE
基金
山东省自然科学基金项目(Y2002F19)山东省科学技术发展计划项目(012050107)山东省技术创新项目(鲁经贸技字200311821)。
