Improving the adaptive neuro-fuzzy method to intellectualize multisensor signals processing

This research paper presents an improved method and software implementation of intellectualization processing for multisensor signals by integrating the modified method of identifying individual conversion functions of multisensor and adaptive neuro-fuzzy inference system. The results of the training and data approximation of adaptive neuro-fuzzy inference system with noise amplitude of 2.5% in training vector for an advanced set of rebuilt points on the surface of the individual conversion function are obtained. The proposed algorithm provides high accuracy of training and approximation data and a low error of artificial neural network training.