-
摘要: 研究了一类具有未建模动态或扰动的非线性系统的鲁棒故障检测与诊断问题,利用神 经网络、模糊系统或小波网络等对非线性故障模式进行在线逼近的方法进行故障诊断.第一步， 对用于鲁棒故障检测的观测器,建立了保证观测器稳定的增益阵的选择条件;第二步,若检测出 发生故障,则用神经网络、模糊系统或小波网络进行故障的在线估计,建立了估计误差界,结果 显示输出估计误差将收敛到由扰动上界或建模误差上界线性确定的范围内.Abstract: The fault detection and diagnosis for a class of nonlinear systems with unmodeled dynamics or noise are considered. Nonlinear online approximator, such as neural network, fuzzy system and wavelet network, is used to monitor the system for any abnormal behaviour due to faults. First, robust fault detection observer is designed, and the condition to choose observer-gain matrix for a stable observer is given. After the fault is detected, online approximator will give an estimation of the fault. The bound of output estimation error is established, which is linearly dependent on the bound of unmodeled dynamics or disturbances.
-
Key words:
- Fault detection and diagnosis /
- nonlinear /
- robust /
- neural networks /
- fuzzy systems
-
点击查看大图
计量
- 文章访问数: 2727
- HTML全文浏览量: 75
- PDF下载量: 1065
- 被引次数: 0
下载:
