Development of on-line neural network based adaptive fractional-order sliding mode robust controller on electromechanically actuated engine cooling system

The reduction of temperature fluctuations around desired reference and control input energy for cooling actuators are among the most important aims of engine thermal management system. However, maintaining of engine/radiator outlet temperature within certain limits is a challenging process in different engine operating conditions due to unknown in-cylinder variable heat transfer rate in terms of the control theory. For this purpose, the thermodynamic model is obtained by dividing into two subsystems, such as the engine and the radiator subsystem in the presence of unknown heat transfer rate, external disturbance, and uncertainties in this study. To increase system robustness, tracking engine inlet/outlet temperature control of engine cooling components is performed using the adaptive fractional-order sliding mode control structure. Moreover, the disturbances acting on the system and unknown heat transfer dynamics of the system are estimated with a radial basis function on-line neural network estimator. The efficiency of proposed controller strategy for electromechanically actuated engine cooling system is compared by proportional–integral–derivative and classical integral sliding mode control under the NEDC (New European Driving Cycle) and WLTP (Worldwide Harmonized Light Vehicles Procedure) transient operating conditions. The temperature error tracking performance is tested by paying attention integral square error, integral absolute error, integral time-weighted absolute error along to the driving cycles. The obtained results showed that the adaptive fractional-order sliding mode control–based engine cooling system outperforms compared to the optimally gain tuned proportional–integral–derivative and integral sliding mode control schemes in terms of reducing of tracking error and engine cooling actuators energy consumption for all engine operating cycles.