Conference Year



transient model, optimization, automotive HVAC, time series


A novice method for calibrating the transient model of an automotive HVAC system is presented in this paper. Transient models can be of great importance in the development process of automotive HVAC control algorithms, especially model based ones, as it saves both time and effort. However, the calibration process is usually difficult and relies heavily on experience due to the complexity of the model. A set of customized measurement tools, which consists of several wireless temperature and humidity sensors and an OBD dongle, is used to capture time series data related to the HVAC system during normal driving. Parts of the time series data are then fed into an optimization algorithm to generate a cost function, which can be minimized when the measured data correspond to the simulation data generated by the transient model, while other parts of the data are remained for the validation step. A sensitivity analysis is then performed to find out which parameters in the HVAC transient model need to be optimized to calibrate the model. As the transient model is a physical network model which can be generally considered as a set of differential and algebraic equations, this presented method reduces the calibration process of a complex physical model into solving a common optimization problem. Therefore, various optimization algorithms and tools can be applied. The method is developed and tested during the modeling process of an automotive HVAC system. The efficiency of the modelling process is improved while the calibration results fit better with the measured data.Â