In this paper a temperature control system for an automated educational
classroom is optimized with several advanced computationally intelligent
methods. Controller development and optimization has been based on developed
and extensively tested mathematical and simulation model of the observed
object. For the observed object cascade P-PI temperature controller has been
designed and conventionally tuned. To improve performance and energy
efficiency of the system, several metaheuristic optimizations of the
controller have been attempted, namely genetic algorithm optimization,
simulated annealing optimization, particle swarm optimization and ant colony
optimization. Efficiency of the best results obtained with proposed
computationally intelligent optimization methods has been compared with
conventional controller tuning. Results presented in this paper demonstrate
that heuristic optimization of advanced temperature controller can provide
improved energy efficiency along with other performance improvements and
improvements regarding equipment wear. Not only that presented methodology
provides for determination and tuning of the core controller, but it also
allows that advanced control concepts such as anti-windup controller gain are
optimized simultaneously, which is of significant importance since
interrelation of all control system parameters has important influence on the
stability and performance of the system as a whole. Based on the results
obtained, general conclusions are presented indicating that meta-heuristic
computationally intelligent optimization of heating, ventilation, and air
conditioning control systems is a feasible concept with strong potential in
providing improved performance, comfort and energy efficiency.