Greenhouse Climate Control Enhancement by Using Genetic Algorithms

Abstract

This paper develops a control system for greenhouse climate with PID controllers tuned based on genetic algorithms. The greenhouse climate nonlinear model with dead-time and measurable disturbances is decoupled by feedback-feedforward linearization method, tacking the feedback without delay from an internal model. The equivalent system consists in two integrator plus dead-time channels for temperature and humidity suitable for PID control. The genetic algorithms (GA) with fitting objective cost functions are developed for PID tuning, including in initial population the PID controllers from four conventional tuning methods. Simulation results in real scenarios show improved performances for PID tuning by GA in comparison with conventional methods, where Ziegler-Nichols method is the best.