Model-Based Optimization for an AMT Clutch Control during the Vehicle Starting 2015-01-0161
With the continuous growth in the emission requirements and higher riding comfort demand, the shift quality becomes more and more an important evaluation index of the automated transmission control algorithms. Traditionally, the shift quality is assessed subjectively by the driver's feeling and then adjusted based on the calibration engineer experience. This classical calibration has disadvantages, such as low reproducibility of the shifting event and a high dependence on the driver's driving habit, so here a model-based multi-objective optimization method is proposed, and the optimization of the clutch control parameters during the vehicle starting is used as an example.
Firstly, a Modelica® based vehicle model is introduced. A second-order sliding mode control is applied to track the clutch position trajectory. The clutch engagement point and the clutch engagement speed in the slipping stage are taken as the optimization objects. The shift quality, clutch engagement duration and mean longitudinal jerk during the vehicle starting, are taken as the evaluation criteria. An evolutionary multi-objective optimization method (non-dominated neighbor immune algorithm, NNIA) is chosen to find out the optimal control parameters. Finally the optimal clutch control parameters are determined based on a proposed mathematical function in different accelerator positions, and the testing results show that the optimized clutch control meets the driver's requirements in different conditions.