摘要
提出了一种基于粒子群优化算法的邻域粗糙集-神经网络的发动机智能故障诊断方法,首先利用基于邻域粗糙集模型的属性约简方法对样本数据进行属性约简,然后采用粒子群优化算法替代传统BP算法来训练神经网络的权值和阈值,再用训练好的神经网络对航空发动机气路故障进行诊断.仿真结果表明:该方法降低了神经网络结构的复杂性,减少了网络训练时间,提高了诊断精度.
A new method based on neighborhood rough set model and neural networks （NN）integrated with particle swarm optimization （PSO） algorithm was presented in this paper for fault diagnosis of aeroengine. Firstly, using the algorithm of attribute reducing based on neighborhood rough set model, we deleted the unnecessary attributes from the decision table. Secondly, the PSO was used to train the weights and the thresholds of NN instead of BP algorithm. Therefore, The NN trained by PSO was applied to aeroengine fault diagnosis. The simulation results indicate that the method has shortened the training time and increased the diagnosis accuracy.
出处
《航空动力学报》
EI
CAS
CSCD
北大核心
2009年第2期458-464,共7页
Journal of Aerospace Power
基金
国家自然科学基金(50576033)
关键词
粗糙集
粒子群优化算法
神经网络
航空发动机
故障诊断
rough set
particle swarm optimization algorithm
neural networks
aeroengine
fault diagnosis
