基于BP神经网络的容错控制技术在通风机监控系统中的应用

Application of Fault Tolerant Control Based on BP Neural Network in Monitor and Control System of Mine Ventilator

矿井通风机监控系统有效地保障通风机的正常运行,针对监控系统中众多的传感器和执行器故障,BP神经网络对正常状态和故障状态分别进行训练,将得到的网络结构、权值和阈值进行存储,再通过容错控制中的重构,对故障点进行诊断和处理。利用风量传感器进行了仿真实验,证明基于BP神经网络的容错控制可以有效地修正传感器故障,从而推广至监控系统的所有传感器。

The monitor and control system of mine ventilator is established to guarantee the normal operation of the ventilator effectively. For the purpose of solving a large number of faults of the sensors and actuators, the normal and fault conditions were trained by the BP neural network. The network structures,weights and thresholds obtained were storaged in a computer. Through the reconstruction of fault-tolerant control,the failure points were got to be diagnosed and treated. This paper used air flow sensor to do the simulation experiments for proving that fault-tolerant control system based on BP neural network can correct sensor failure effectively and it can extend to all sensors of monitor system.