Application of optimal IA-BP algorithm in moisture content detecting with microwave resonator

The linear regression that founds the function between the moisture content and the detecting parameters brings significant errors which can be reduced by BP algorithm. There is one fatal disadvantage that conventional BP algorithm tends to get into infinitesimal locally, which worsens the stability of the measurement precision. An evolutionary neural network model based on optimal IA-BP algorithm is presented in this paper. In the model, IA algorithm is first used for global search and then BP algorithm for local search. The optimal IA-BP algorithm has the merits of high prediction precision, rapid convergence, global superiority and accuracy for optimization compared with conventional BP algorithm. Experiments shows the remarkable improvements in the measurement with the mean squared error 0.0125, the mean absolute error 0.0715, the mean relative error 0.1186 and the certain coefficient 0.9965 between the predicted moisture content and the real value.