基于神经网络模型辨识的串级控制系统传感器容错技术研究  

Research on Fault Tolerant Technique of Sensor of Cascade Control System Based on Neural Network Model Identification

作　　者：彭桂芳[1] 宋彤[1] 

机构地区：[1]大连理工大学电信工程学院,大连116024

出　　处：《自动化信息》2007年第10期61-62,67,共3页Automation Information

摘　　要：本文根据控制律重构设计的思想，提出了串级控制系统主动容错方案。在工业过程中串级控制系统发生传感器失效故障的情况下，利用人工神经网络在线辨识被控对象模型，在故障诊断系统输出的故障信息驱动下，实现控制律在线重构设计，使工业过程在重构的控制系统控制下稳定工作，保证性能指标。最后，以一个串级控制回路为对象进行了仿真验证，结果表明上述方案达到比较满蔷的容错控制效果。This paper proposes an active fault tolerance scheme of cascade control system according to the thought of reconstruction of control law. When the malfunction of sensors occurs in cascade control system in industrial process, driven by output information of fault diagnosis system, the on-line reconstructed design of control law is implemented by using artificial neural network identifying model of controlled object on-line, It leads to the stabilized work of the industrial process controlled by reconstructed control system and guarantees to meet the requirements of performance. Finally, it proves that the fauh tolerant control performance of the above said scheme is of satisfactory through the simulation of a cascaded loop object.

关 键 词：串级控制系统 传感器故障 主动容错控制 模型辨识 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]