A new Takagi-Sugeno fuzzy approach of process modeling and fault detection

Process modeling and fault detection using a new Takagi-Sugeno (T-S) fuzzy method which is obtained by decomposing structural risk minimization has been proposed in this paper. The proposed method makes use of some ideas on Least square support vector regression (LSSVR). First of all, a novel cost function consisted of R terms is constructed and used for solving consequent parameters of T-S fuzzy model based on process data. Then, the normal model corresponding to no fault is used to obtain nonlinear process models for the process running in normal operation. When a fault occurs, fault detection is performed using the residuals. Finally, several types of faults have been considered on the benchmark Tennessee Eastman Process data. The effectiveness of the obtained simulation results indicate that the developed method is capable to correctly detect various faults by simulations.