A randomised algorithm is proposed for computing globally optimal static-output-feedbacks for large-scale systems. The algorithm is based on the Ray-Shooting Method and involves some heuristics to accelerate the searc...
详细信息
A randomised algorithm is proposed for computing globally optimal static-output-feedbacks for large-scale systems. The algorithm is based on the Ray-Shooting Method and involves some heuristics to accelerate the search. We also improve the basic Ray-Shooting Algorithm and make the search in the controller parameter-space (which generally, is much more smaller than the certificate parameter-space), thus enabling its efficient use for large-scale systems.
A randomised optimisation method, called the ray-shooting method that efficiently solves the minimisation problem of continuous functions over compact non-convex unconnected regions, was introduced recently. Based on ...
详细信息
A randomised optimisation method, called the ray-shooting method that efficiently solves the minimisation problem of continuous functions over compact non-convex unconnected regions, was introduced recently. Based on the ray-shooting method, an algorithm is suggested for solving the problems of structured and structured-sparse stabilising static-output-feedbacks. The algorithm presented here is a randomised algorithm, with a proof of convergence in probability. Its practical implementation has a good performance, in terms of quality of controllers, percent of success and run-time.
For a given system with time-invariant affine uncertainties, ranging in a unit hypercube or, equivalently, in a hyper-rectangle, new LMI sufficient conditions for the verification of a given simultaneous static-output...
详细信息
For a given system with time-invariant affine uncertainties, ranging in a unit hypercube or, equivalently, in a hyper-rectangle, new LMI sufficient conditions for the verification of a given simultaneous static-output-feedback for hypercube nodes, being robust static-output-feedback for the whole system, are proved. The conditions generalise previous results in the sense that we do not assume that the state-to-measurement or the input-to-state matrix is constant. Based on the LMI conditions, a randomised algorithm is suggested. The effectiveness of the method is reinforced through experiments on real-life systems.
暂无评论