RVM-based nonlinear dynamic compensation of sensors

A technique based on the relevance vector machine (RVM) is proposed for nonlinear dynamic compensation of sensors in this paper. The relevance vector machine is a Bayesian extension of the support vector machine (SVM). The relevance vector machine has a simpler model structure, a fewer control parameter, and a faster computation speed in comparison with the support vector machine. Therefore, the compensator structure can be simpler by means of the relevance vector machine. Also, the rapid compensation of the sensor dynamic characteristics can be implemented with the proposed technique. The numerical experimental results show that the technique is still effective even if the sensor dynamic characteristics has a strong nonlinear.