Extended Kalman filter training T-S fuzzy model for signal reconstruction of multifunctional sensor

Multifunctional sensor is an emerging sensor which can measure more than one physical or chemic parameters simultaneously. But how to establish the relationship between the outputs and inputs of multifunctional sensor, which called signal reconstruction, becomes a problem. A method based on T-S fuzzy model and extended Kalman filter (EKF) for multifunctional sensor signal reconstruction is proposed in this paper. The method firstly uses subtractive clustering to partition the sampled-data and confirm the structure and initial parameters of T-S fuzzy model. Then train T-S fuzzy model with extended Kalman filter and sampled-data continuously until reaching the expected criterion. The trained T-S fuzzy model is located behind the multifunctional sensor to convert the output of the sensor into the expected parameters in the practical application. The simulation results show that the method is of higher precision and accuracy than other methods, and is very suitable for practical use.