Eddy current sensor mode building based on polynomial basis functions artificial neural network

This paper presents a method used to the eddy current sensor modeling based on a polynomial basis functions neural network to settle its non-linear problem. The principle and algorithms of weight values of neural network are introduced. In this method, the non-linear model is set up by polynomial basis functions neural network according to measurement data. The results show that the sensor modeling has the character of strong robustness and on-line modeling compared with the sensor modeling of least square method. The maximum nonlinearity error can be reduced to 0.036% by using polynomial basis functions neural network. However, the maximum nonlinearity error is 0.075% using the least square method.