Optimization of the control parameters of a pneumatic servo cylinder driveusing genetic algorithms

Genetic algorithms are applied here to optimize the control gains of a thre e-loop controller for a pneumatic servo cylinder drive. The three-loop cont roller has position, velocity, and acceleration feedback gain, all of which have formerly been manually tuned. A general procedure for optimizing the control parameters of fluid power systems by using genetic algorithms is pr esented. It is shown that, by using minimum information specific to the sys tem, near-optimal values of the control gains can be obtained within 10 gen erations. It is also shown that the numerical values of both the optimal ve locity and the acceleration gain tend to increase as either the load mass o r reference position increases, when the optimal position gain is fixed at a constant value. The optimized gains are confirmed by a plot of the fitnes s distribution defined in this study, which represents the control performa nce of the system in a given gain space. (C) 1998 Published by Elsevier Sci ence Ltd. All rights reserved.