Robust Control Optimization with Metaheuristics
Publication Date: January 2017 Hardback 448 pp.
In industry, the control engineer must design a unique control law that is valid on a single prototype, with a sufficient degree of robustness that satisfies complex scope statements on many systems. For this, the development methodology that the control engineer actually uses is an iterative experimental process (trial and error phase), which relies heavily on the experience of the engineer.
In this book, we try to make this servo controller synthesis methodology more efficient as it is more direct and therefore has a less costly development time because it calculates a final structured controller by a direct attack on the high-level specification system.
The complexity of the high-level system specification drives us to use metaheuristics: these optimization techniques do not require the formulation of a gradient, the only constraint being the possibility of evaluating the specification. Thus we propose in this work to reformulate the robust control problems for stochastic optimization, showing how to synthesize structured controllers from H-type issues, u-synthesis or LPV synthesis and that the interest of the formulated approach lies in its flexibility and the consideration of complex “exotic” constraints.
Evolutionary algorithms are proving to be very effective. We develop on this basis an original synthesis method of structured and robust controllers against high-level requirements of any form. This methodology results in the implementation of a Matlab numerical tool for controller synthesis included within book. The validation of this work was carried out on industrial problems such as the line of sight stabilization problem.
1. Metaheuristics for Controller Optimization .
2. Reformulation of Robust Control Problems for Stochastic Optimization.
3. Optimal Tuning of Structured and Robust H Controllers Against High-level Requirements
4. HinfStoch: A Toolbox for Structured and Robust Controller Computation Based on Stochastic Optimization
About the Authors
Philippe Feyel works as a R&D engineer for the high-tech company Safran Electronics & Defense (Safran Group). An expert in automation applied to the line of sight stabilization problem and with a PhD in Automation sciences, he works in partnership with the academic world on the industrial implementation of robust control through using modern optimization techniques (stochastic optimization by metaheuristics, non-smooth optimization, etc.).