Mean time to failure (MTTF) is an important reliability index of fault‐tolerant control systems, which is chosen as a design objective in this paper. However, it is usually evaluated from stochastic reliability models, and no analytical expression is available to relate MTTF to controller parameters. To overcome this difficulty, a two‐stage design scheme is proposed in this paper: A gradient‐based search is firstly carried out on probabilistic H∞ performance characteristics for MTTF requirement; a sequential randomized algorithm with a weighted violation function is then developed for a controller design to satisfy the required H∞ performance, and its convergence is guaranteed with probability 1. Two iterative algorithms are carried out alternately to implement this scheme, and a controller can be designed for MTTF requirement. Copyright © 2008 John Wiley & Sons, Ltd.
平均故障时间(MTTF)是容错控制系统的一个重要可靠性指标,本文将其选作设计目标。然而,它通常是从随机可靠性模型中评估的,并且没有解析表达式可将MTTF与控制器参数相关联。为克服这一困难,本文提出了一种两阶段设计方案:首先针对MTTF要求对概率H∞性能特征进行基于梯度的搜索;然后开发一种带有加权违反函数的顺序随机算法用于控制器设计以满足所需的H∞性能,并且其收敛以概率1得到保证。两种迭代算法交替执行以实现该方案,并且可以针对MTTF要求设计一个控制器。版权所有©2008约翰威立父子有限公司