Model Predictive HVAC Control with disturbance variable forecasting for city buses

For city buses one of the key characteristics is their operating range, which among other things can be heavily influenced by the power consumption of the on board heating, ventilation and air conditioning (HVAC) system. Especially in the case of battery electric buses (BEBs) the energy consumption of the HVAC system can add up to a similar energy expenditure needed for powering the electric drive train. For addressing this challenge, this work proposes to create a new controller for the outer control loop of the cascaded HVAC control system in city buses, which is able to factor in external influences. The objective of such a controller is constituted of ensuring adherence to thermal requirements within the bus cabin while simultaneously consuming less energy than currently employed solutions. To achieve this goal, this work embeds the predictive capabilities of artificial neural networks (ANNs) into a model predictive control (MPC) architecture, for forming a neural network model predictive control (NNMPC). The prototypical NNMPC implementation is evaluated in a model-in-the-Ioop simulation and its control performance is investigated and compared to the control behaviour of a currently employed nroportlonal-Intearal-derivative (PID) controller.