1999 Volume 11 Issue 2 Pages 320-327
Sensor fusion systems for estimating the states from sensor information have received much attention in various fields. In these days, various kinds of sensors have been developed. The measuring time and accuracy of each sensor are different because each sensor has particular characteristic for measuring. We have proposed the Recurrent Fuzzy Inference(RFI) with incremental learning for fusing multiple sensor information with different measuring time and accuracy. However, our method and traditional methods cannot evaluate the inferred result. In this paper, we propose the evalution method for the inferred result. In this method, we calculate the reliability of it by the degree of fuzziness which expresses the fuzziness of the learned membership function. For showing the effeciveness of our method, we apply our method to the grinding process. Then, we infer the surface roughness and calculate the reliability of it.
