Improved adaptive neural-network-based fuzzy inference system angular sensor error compensation method

The control of the measurement accuracy of angular sensors is very important in engineering applications; and has significant effects on the operation performances of the applications. Traditional methods cannot provide satisfactory results when the input-output relationship of angular sensors is complex and nonlinear. To deal with this problem, we propose the error compensation method based on the improved adaptive neural-network-based fuzzy inference system (ANFIS). The modeling procedures are demonstrated step by step in this paper. This method has been applied to calibrate a 16-bit absolute-type photoelectric encoder based on the accuracy test. The results show that, compared with the polynomial fitting and BP neural network, the improved ANFIS enhances the measurement accuracy markedly. The measurement accuracy of the optical encoder is raised up to at least 7.5 times higher than that of the original value.