Abstract
Real-time condition acquisition and accurate time-to-failure (TTF) prognostic of machines are both crucial in the condition based maintenance (CBM) scheme for a manufacturing system. Most of previous researches considered the degradation process as a population-specific reliability characteristics and ignored the hidden differences among the degradation process of individual machines. Moreover, existing maintenance scheme are mostly focus on the manufacturing system with fixed structure. These proposed maintenance scheme could not be applied for the reconfigurable manufacturing system, which is quite adjustable to the various product order and customer demands in the current market. In this paper, we develop a systematic predictive maintenance (PM) framework including real-time prognostic and dynamic maintenance window (DMW) scheme for reconfigurable manufacturing systems to fill these gaps. We propose a real-time Bayesian updating prognostic model using sensor-based condition information for computing each individual machine’s TTFs, and a dynamic maintenance window scheme for the maintenance work scheduling of a reconfigurable manufacturing system. This enables the real-time prognosis updating, the rapid decision making for reconfigurable manufacturing systems, and the notable maintenance cost reduction.
Deadlock Control and Fault Detection and Treatment in Reconfigurable Manufacturing Systems Using Colored Resource-Oriented Petri Nets Based on Neural Network
