Improved BP arithmetic in measuring tobacco moisture by resonant microwave sensor

Measurement accuracy of moisture is influenced by conventional linear regression mainly using microwave resonator. Based on improved BP arithmetic, a regression is presented to calibrate the measurement results. First, GA is used to pre-optimize the regressive neural network for its global search ability, parallel processing and strong robust. Then the gradient descent method of BP arithmetic is integrated to avoid getting into infinitesimal locally effectively and keep the merits of high prediction precision and rapid convergence. The experiment shows there is a great improvement between the predicted value and the real one.