Exact analytical inverse mapping of decomposable TS fuzzy systems with singleton and linear consequents

• A new method is proposed for the exact analytical inverse mapping of decomposable TS fuzzy systems with singleton and linear consequents.
• The proposed method is based on the decomposability property and the matrix multiplication form representation of decomposable TS fuzzy systems.
• The proposed method provides exact inverse solutions in a more simple and straightforward way due to its compact formulation.
• The proposed method simplifies the inversion of decomposable TS fuzzy systems having high number of input variables.
• The proposed inversion procedure can easily be implemented due to its compact formulation.

In this study, a new method is proposed for the exact analytical inverse mapping of Takagi–Sugeno fuzzy systems with singleton and linear consequents where the input variables are described by using strong triangular partitions. These fuzzy systems can be decomposed into several fuzzy subsystems. The output of the fuzzy subsystem results in multi-linear form in singleton consequent case or multi-variate second order polynomial form in linear consequent case. Since there exist explicit analytical formulas for the solutions of first and second order equations, the exact analytical inverse solutions can be obtained for decomposable Takagi–Sugeno fuzzy systems with singleton and linear consequents. In the proposed method, the output of the fuzzy subsystem is represented by using the matrix multiplication form. The parametric inverse definition of the fuzzy subsystem is obtained by using appropriate matrix partitioning with respect to the inversion variable. The inverse mapping of each fuzzy subsystem can then easily be calculated by substituting appropriate parameters of the fuzzy subsystem into this parametric inverse definition. So, it becomes very easy to find the analytical inverse mapping of the overall Takagi–Sugeno fuzzy system by composing inverse mappings of all fuzzy subsystems. The exactness and the effectiveness of the proposed inversion method are demonstrated on trajectory tracking problems by simulations.

Figure optionsDownload as PowerPoint slide