Abstract
Parallel computing is used to evaluate compensator robustness numerically and to automate the design of control systems. Robustness is quantified by Monte Carlo evaluation of the effects of plant-parameter uncertainty, and it is maximized by searching the control-design-parameter space with a genetic algorithm. This design approach requires considerable computation, and parallel computing can reduce execution times. Since evaluation times vary among trials, computation time could be prolonged if some nodes remain idle while others finish their tasks. A dynamic scheduler is proposed as a solution. Theoretical and experimental results for a shared-virtual-memory computer illustrate that speed-up of controller design computations is nearly linear in the number of computing nodes.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
