摘要
在组合预报模型研究基础上,提出基于神经网络的非线性组合预报模型,由神经网络给出常规预报方法的最佳组合。首先从函数逼近角度研究了这种模型的理论基础,在此基础上给出了实现策略和神经网络的有效训练算法,将该模型应用于发电机组状态检修的振动参数的趋势分析和故障预报,仿真结果表明该模型有更高的预报精度。
On the basis of study of the combining predic-tion model, a new combining prediction model based on neural network is submitted. The best combination of traditional prediction method is given by neural network. Firstly, theoreti-cal foundation of this model is studied in sense of function approximation. Then, the implement strategy of the model and effective training algorithm for neural network are given. The model is applied to vibration trend prediction in condition based maintenance of turbo-generator set, and the result shows that the prediction precision of this model is better.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2003年第9期204-206,211,共4页
Proceedings of the CSEE
