Design of an Improved Single Neuron-Based PI Controller for an HVAC System in a Test Room

The paper presents an improved single neuron-based adaptive PI controller (SNPI) designed for regulating a heating, ventilation and air-conditioning (HVAC) system in a test room, which can be represented by a First-order-plus-dead-time (FOPDT) model. The characteristics including nonlinear and time-varying dynamics in HVAC systems have been considered. By using an improved learning algorithm, which adopts a modified least mean square (LMS) method, the proportional and integral gains of PI controller in the control scheme can be adjusted in real-time based on a single neuron. Furthermore, the proportional coefficient of the neuron can be adjusted according to the error of the system output. To evaluate the control performance of the improved SNPI controller in the HVAC system, it is compared with a conventional PI controller and a classical SNPI controller based on a common LMS algorithm. Simulation results demonstrate the effectiveness and superior performance of the proposed control method.