Optimization-based reliability control of mechatronic systems

Reliability-adaptive systems allow an adaptation of system behavior based on current system reliability. They can extend their lifetime at the cost of lowered performance or vice versa. This can be used to adapt failure behavior according to a maintenance plan, thus increasing availability while using up system capability fully. To facilitate setup, a control algorithm independent of a degradation model is desired. A closed loop control technique for reliability based on a health index, a measure for system degradation, is introduced. It uses self-optimization as means to implement behavior adaptation. This is based on selecting the priorities of objectives that the system pursues. Possible working points are computed beforehand using model-based multiobjective optimization techniques. The controller selects the priorities of objectives and this way balances reliability and performance. As exemplary application, an automatically actuated single plate dry clutch is introduced. The entire reliability control is setup and lifetime experiments are conducted. Results show that the variance of time to failure is reduced greatly, making the failure behavior more predictable. At the same time, the desired usable lifetime can be extended at the cost of system performance to allow for changed maintenance intervals. Together, these possibilities allow for greater system usage and better planning of maintenance.