作者: Zoltán Petres , Szabolcs Nagy , Péter Zoltán Baranyi
DOI:
关键词: Soft computing 、 Transformation (function) 、 Fuzzy logic 、 Canonical form 、 Context (language use) 、 Algorithm 、 Model transformation 、 Computer science 、 Tensor product model transformation 、 Control engineering 、 Linear matrix inequality
摘要: The Tensor Product model transformation is a numerical method that capable of uniformly transforming LPV (linear parameter-varying) dynamic models into polytopic forms, both in theoretical and algorithmic context. Using the TP transformation, different optimization convexity constraints can be considered, transformations executed without any analytical interactions, within reasonable amount time (irrespective whether given form equations resulting from physical considerations, as an outcome soft computing based identification techniques such neural networks or fuzzy logic methods, result blackbox identification). Thus, replaces usual oftentimes complex conversions with numerically tractable straightforward series operations. generates two kinds models. Firstly, it reconstructs HOSVD (Higher Order Singular Value) canonical This new unique representation. extracts structure various important properties same sense does for matrices tensors. Secondly, convex upon which LMI (Linear Matrix Inequality) multi-objective control design immediately order to satisfy performance requirements. tool MATLAB Toolbox implements Model Transformation Control Design framework. It available at http://tptool.sztaki.hu.