Performance comparison of radial basis function with feed-forward neural network for sensor linearization

A comparative analysis of Linearization technique using two prominent neural network approaches is presented. Generally linearizing the non-linear input/output characteristics of most of the sensors available using hardware/software is indeed a formidable task. The ANN employs two different approaches of feed forward neural network using radial basis function and Levenberg-Marquardt algorithm, for automatic adjustments of weights and biases to arrive within a minimum mean square error and number of iterations. The performance comparison of the two algorithms for the linearization of NTC thermistor is analysed in this paper.