摘要
提出了基于集成信息融合技术的实时智能故障诊断系统结构。在对传统BP神经网络分析的基础上引入了聚类分析方法与遗传优化算法,有效降低了BP神经网络的训练难度并提高其训练精度。将遗传神经网络与自适应神经-模糊推理系统相集成用于特征级信息融合,既提高了诊断可靠性又充分利用了诊断知识;引入D-S证据理论进行决策级融合,有效地利用了各诊断单元的诊断结果。仿真测试结果表明,该故障诊断系统能迅速、准确、可靠的诊断出各种故障。
A structure of real-time intelligent fault diagnosis system based on compositive information fusion is designed.Method of clustering analysis and algorithm of genetic optimization is introduced to traditional BP neural-network,in order to reduce training difficulty of BP neural-network,and in order to increase training precision.Method of genetic NN and method of adaptive-NN-fuzzy-reasoning-system are integrated to information fusion diagnostic level in order to improve fault diagnosis reliability and to use diagnosis knowledge in training data fully.D-S evidence theory is used to analyze conclusions in the level of decision in order to utilize diagnosis result of each diagnosis cell effectively.It’s validated from emulational test that this fault diagnosis system can diagnose fault rapidly,exactly and reliably.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第20期4476-4479,共4页
Computer Engineering and Design
关键词
信息融合
故障诊断
神经网络
遗传算法
自适应神经-模糊推理系统
information fusion fault diagnosis neural network genetic algorithm adaptive neuro-fuzzy reasoning diagnosis system
