摘要
文中提出了一种基于小波神经网络非线性观测器的故障检测方法。它是一种把信号分析和数学模型相结合的故障检测方法 ,通过小波对信号的去噪和神经网络的自学习功能 ,来获取系统输入输出的非线性动力学特性 ,进而实时计算出残差并进行逻辑判决 ,可提高故障检测的速度和准确率。对同步交流电机的结构损伤故障进行了仿真 ,结果表明了该方法是可行的。
In this paper, a method of fault detection based on nonlinear observer using wavelet neural networks is presented. This approach is a combination of signal analysis and mathematical model .By the denoising function of wavelet and the learning itself function of neural network , the input and output nonlinear dynamic characteristic of system is obtained . The output prediction error, generated from the real output and wavelet neural networks estimated output, is used as a residual error to execute logical judge. and this approach can improve the speed and accuracy rate of fault detection. Simulation for structural damage faults of nonlinear synchronous motors show that the fault detection approach is feasible.
出处
《计算机仿真》
CSCD
2000年第1期24-27,共4页
Computer Simulation
