Multiobjective Optimization including Safety of Operation Applied to a Linear Drive System

In this contribution, we introduce a multiobjective optimization used to calculate safe optimal working points for a mechatronic system by including stochastic safety-critical signals in an objective function. Our application example consists of a linear drive for a rail-bound vehicle and an actuation unit. The linear drive's secondary part is fixed; the primary part is vehicle-mounted and can be adjusted vertically to account for deviations of the height of the secondary part. A small air gap between both parts improves efficiency, but increases the risk of a collision between the two parts. Using height data of the secondary part, a trajectory for the vertical adjustment of the primary part is calculated. However, unexpected deviations necessitate a readjustment of the air gap. The probability of such unexpected height deviations can be calculated from the readjustment data. The system is equipped with sensors to measure the air gap. Assuming that the sensor noise is normally distributed, noise characteristics are determined. Using this information and the probability distribution of unexpected height deviations, the probability of a collision is determined.T he sensor noise and the probability of a collision between both parts of the linear drive are included in the dynamical model of the system. Using multiobjective optimization, pareto-optimal working points for the controller of the air gap are obtained. By selecting an appropriate working point, safe operation can be ensured.