Optimal Parameter Calibration for Physics Based Multi-Mass Engine Model

Designing an efficient transient thermal system model has become a very important task in improving fuel economy. As opposed to steady-state thermal models, part of the difficulty in designing a transient model is optimizing a set of input parameters. The first objective in this work is to develop an engine compatible physics-based 1D thermal model for fuel economy and robust control. In order to capture and study the intrinsic thermo-physical nature, both generic “Three Mass” and “Eight Mass” engine models are developed. The models have been correlated heuristically using Simulink. This correlation and calibration process is challenging and time consuming, especially in the case of the 8-mass model. Hence, in this work a Particle Swarm Optimizer (PSO) method has been introduced and implemented on a simple 3-mass and more complex 8-mass engine thermal model in order to optimize the input parameters. PSO has been proven to be effective in handling a large set of parameters which need calibration (i.e. optimization). These parameters were optimized and validated over different transient drive cycles: both fuel economy and extreme driving conditions. Results demonstrate that the use of the PSO guarantees a better correlation of the transient models to vehicle level test data, while providing a systematic way to find an optimal set of parameters for the transient model. The generic framework can be extended to different vehicle thermo-mechanical components such as transmissions or heat-exchangers.