FUZZY LOGIC FOR BUILDING REAL-TIME INTELLIGENT SYSTEMS

Abstract

Fuzzy logic (FL) is attracting a huge amount of attention and research effort from across different fields these days. There are many successful applications of FL in these areas: control applications, scheduling and optimisation and signal analysis, etc. A fuzzy logic system (FLS) is actually a nonlinear mapping of input data vector into a scalar output. So it can be utilised in many nonlinear-related problems. The richness of FLS is that there are enormous numbers of possibilities that lead to lots of different mappings. This research focusses on the structure of FLS. Because single-scale ( conventional) fuzzy model can not appropriately capture the features activated if the structure is not set well, a new multi-scale fuzzy model is proposed, which has the structure-adaptive scheme. Firstly, the author approaches the multi-scale fuzzy model with the product-sum composition form. Secondly, the multi-scale fuzzy model with the min-max composition form is also addressed. Both model structures are used for nonlinear dynamical system modelling, and simulations show the effectiveness of the both schemes. The multi-scale fuzzy scheme is applied in a practical optimisation method. The global optimal searched shows the robustness of the result in noisy environment. More importantly, optimisation with fuzzy scheme has the potential to combine the expert knowledge with the numerical data.