PSO优化算法飞机操纵面故障辨识研究

Fault Identification Research of Aircraft Control Surface by PSO Optimization Algorithm

针对飞机复杂的非线性操纵面故障系统,建立故障模型,提取各种故障数据,并将粒子群优化算法应用于BP网络系统,提出了一种基于粒子群优化神经网络的故障诊断方法;该方法分阶段实施神经网络的训练,有效地加强了算法的全局搜索能力,采用PSO算法优化了传播中的权值和阈值,弥补了BP算法收敛速度慢和易陷入局部极小点的不足,提高了故障模式识别的能力;实验结果表明该方法加快了神经网络的学习收敛速度,提高了故障模式的识别正确率,对飞机操纵面的各种典型故障都能有效加以辨识。

Aimed at complex nonlinear fault system of aircraft control surface, the fault models are set up. Various fault data are acquired. By using PSO optimization algorithm in BP neural networks system, a new fault diagnosis method is proposed. The neural networks are trained by steps. The global search ability is enhanced. The connecting weighs and thresholds of neural networks are optimized. It makes up the lacks of slow convergence speed and falling into local minimum point easily in BP algorithm, and improves the ability of fault pattern recognition. The experimental results show that the method speeds up the convergence speed of neural networks, enhances the correctness in fault pattern recognition, and identifies all typical faults of aircraft control surface effectively.