Iterative learning fault-tolerant control for injection molding processes against actuator faults

Abstract

The injection molding process is a typical multi-phase batch process. As the filling and packing-holding phases share the same actuator, faults occurring in the actuators may cause serious impact on the performance and running time. Because these two phases are of crucial importance in relation to the final quality of the product, to solve this problem is essentially meaningful. This paper proposes iterative learning fault-tolerant control (ILTFC) in terms of common multi-phase batch processes and then applies it to the injection molding processes. To develop the ILFTC design, the multi-phase batch process is treated as a switched system composed of different dimensional subsystems and then converted to an equivalent two-dimensional (2D) switched fault-tolerant Rosser model. A hybrid fault-tolerant law is then designed based on an average dwell time method. Sufficient conditions and minimum running time guaranteeing the exponential stability under both normal and fault conditions are obtained. Under the proposed control law, the control performance and running time will restore to the previous level before actuator faults occur. The efficiency and merits of the proposed scheme is illustrated by an injection molding process, and results show that it can guarantee the stability and minimum running time whether the process is in normal operation or in case of actuator faults.