Optimal tuning of proportional integral derivative controller for simplified heating ventilation and air conditioning system

Abstract

A Heating Ventilation and Air Conditioning system (HVAC) is an equipment that is designed to adapt and adjust the humidity as well as temperature in various places. To control the temperature and humidity of the HVAC system, various tuning methods such as Ziegler–Nichols (Z-N), Chien-Hrones-Reswick (CHR), trial and error, robust response time, particle swarm optimization (PSO) and radial basis function neural network (RBF-NN) were used. PID is the most commonly used controller due to its competitive pricing and ease of tuning and operation. However, to effectively control the HVAC system using the PID controller, the PID control parameters must be optimized. In this work, the epsilon constraint via radial basis function neural network method is proposed to optimize the PID controller parameters. The advantages of using this method include fast and accurate response and follow the target values compared to other tuning methods. This work also involves the estimation of the dynamic model of the HVAC system. The non-linear decoupling method is used to modify the model of HVAC system. The benefits of using the proposed simplification technique rather than other techniques such as the relative gain array techniques (RGA) is because of its simplification, accuracy, and reduced non-linear components and interconnection effect of the HVAC system. It is observed that the amount of integral absolute error (IAE) for temperature and humidity based on the simplified model are decreased by 18% and 20% respectively. Moreover, it is revealed that optimization of PID controller through multi objective epsilon constraint method via RBF NN of the simplified HVAC system based on non-linear decoupling method shows better transient response and reaches better dynamic performance with high precision than other PID control tuning techniques. The proposed optimum PID controller and estimation of dynamical model of the HVAC system are compared with the different tuning techniques such as RBF and ZN based on original system. It is observed that the energy cost function due to temperature (JT) and humidity (JRH) are lowered by 15.7% and 4.8% respectively; whereas the energy cost functions reflect the energy consumptions of temperature and humidity which are produced by the humidifier and heating coil. Therefore, based on the new optimization method the energy efficiency of the system is increased. The unique combination of epsilon constraint method and RBF NN has shown that this optimization method is promising method for the tuning of PID controller for non-linear systems.