论文部分内容阅读
理论研究及大量实践表明:径向基函数神经网络具有较强的函数逼近能力,学习速度优于常用的BP网络。本文利用径向基神经网络构成传感器输出预测器实现了多传感器故障在线检测和信号恢复。文中阐述了预测器的构成及其在线学习算法。通过仿真研究证明:该预测器对传感器输出具有很好的在线预测、跟踪能力。当某传感器发生故障时,在及时准确地发出报警信号的同时,对瞬时故障,能很好地恢复故障期间传感器正常的输出,即消除瞬时故障对系统正常运行的影响;对长期故障,能在故障发生后一定的时间范围内,正确估计出传感器正常输出,以保证系统的正常运行。
Theoretical research and extensive practice show that radial basis function neural network has strong function approximation ability and learning speed is superior to the commonly used BP network. In this paper, the radial basis neural network is used to form the sensor output predictor to realize the multi-sensor fault online detection and signal recovery. The article describes the structure of the predictor and its online learning algorithm. Simulation results show that this predictor has good on-line predictive and tracking capability for sensor output. When a sensor fails, timely and accurate alarm signal at the same time, the instantaneous fault, the sensor can recover the normal fault during the normal output, that is, to eliminate the instantaneous fault on the normal operation of the system; for long-term fault, in the Within a certain period of time after the fault occurs, the correct sensor output is correctly estimated to ensure the normal operation of the system.