Reliability-Based Adaptive Control In Manufacturing Decision Systems
Cheng Hsu · 2005
General system control can be considered as an adaptive process that alters the system by monitoring its actual behavior. In this process, confidence intervals serve to provide reliability statements regarding the system's performance under the condition of uncertainty, and lead to reliability - based constraints or standards required of the system. Whenever the system's actual performance violates the confidence interval-based requirements, a potential decision point is reached and corrective measures my be triggered. A prime example illustrating the above process is the control of production systems. In the classical literature, production systems including its configuration, operation mode, and demand (job type and resources requirements) - are generally considered as stationary over the system's planning horizons. In Flexible Manufacturing System (FMS), however, the basic conditions of systems are highly dynamic, thus, the adaptive control may require confidence intervals for complex stochastic processes which are beyond the classical results of probability. In this paper, we survey such situations and propose a framework to model the problem of reliability-based adaptive control in no decision systems. Current research on this method include: (a) combining reliability constraints with such control-theoretic models as stochastic Petri Nets and (b) incorporating the adaptive control capabilities into the design of a metadatabase for a computer-integrated manufacturing system, both are undertaken at RPI.