Search

For mass resonant system, a long-term operation causes the system degrading which ends up with system vibration and affects system control accuracy. A number of model-based feedforward control methods were developed to compensate for such systematic control error. However, the parameters of the complex and nonlinear control systems need to be identified during the degrading process so that the precise speed control can be achieved as well as the health condition of the mechanical system can be extracted. This study proposes an equipment health maintenance framework in a three-mass resonant servo control system with the LuGre friction model, which conducts model-based parameter estimation and feedforward compensation. Since the traditional methods require off-line frequency-sweep which causes the downtime in the production line, we suggest a constrained particle swarm optimization (CPSO) to estimate the mechanical parameters so that the machine can operate simultaneously. More specifically, with embedding the soft equality constraints of the anti-resonance frequency in the mass resonant system, the CPSO enables the feasible region shrank and the vibration suppressed. In particular, we address these constraints by obtaining the feasibility-preserving approach with the dynamic relaxing constraints. An experimental study of a leading electronics manufacturing company in Taiwan has been conducted to validate the proposed approach with a designed experiment of a belt drive system. The results show that the mass resonant system with the LuGre friction model via CPSO successfully reflects the main effects of current variation in the mechanical system. [ABSTRACT FROM AUTHOR]