An RBF neural network approach towards precision motion system with selective sensor fusion

• An RBF based multi-sensor fusion architecture in precision motion systems is proposed.
• An algorithm is developed to model nonlinear weights evolved with selector attributes.
• A case study on a linear motor with two sensors is demonstrated.
• Improved performance is observed compared with existing methods.

A radial basis function (RBF) neural network approach with a fusion of multiple signal candidates in precision motion control is studied in this paper. Sensor weightages are assigned to sensor measurements according to the selector attributes and approximated using RBF neural network in multi-sensor fusion. A specific application towards precision motion control of a linear motor system using a magnetic encoder and a soft position sensor in conjunction with an analog velocity sensor is demonstrated. Motion velocity and noise level in the sensor are chosen as the selector attributes, and the optimal sensor weightages under different attributes are approximated using RBF neural network with the reference data from laser interferometer. The experiment results illustrate that the proposed method can provide more accurate results than both single encoder measurement and existing sensor fusion methods including ordinary RBF neural network and Kalman filter based multi-sensor approach.