A comparison of neural networks, linear controllers, genetic algorithms and simulated annealing for real time control

1 This paper reports a performance comparison between traditional (linear PID controller) and cognitive methods (neural networks, genetic algorithms and simulated annealing) applied to the problem of real-time control. A one-axis magnetic bearing is considered as a case study and cognitive methods have been successfully applied to its control. Comparisons are made using a real test setup. It is shown that hybrid approaches provide the best performance among all the methods analysed.