Aiming at the problem that back propagation（ BP） neural network is easy to fall into local minimum

the network structure is not easy to be determined

and the original spectral wavelength variable experimentally measured is used as input to make the prediction model highly complex and can’t meet the requirements of realtime rapid measurement of blood glucose concentration

a new modeling prediction method based on OV-GA-BP neural network is presented

that is

preprocessing spectral data combining with orthogonal signal correction（ OSC）and variable importance in the projection（ VIP）

that can reduce dimension and number of input variables to reduce the complexity of the model

genetic algorithm（ GA） is used to optimize the BP neural network. The simulation results show that the correlation coefficient between the true value of the validation sets and the predicted value can reach 99. 84 %

the coefficient of variation（ CV） of the root mean square error reaches0. 079 412 and mean absolute percent error（ MAPE） reaches 0. 102 69. Compared with the traditional BP neural network

prediction precision of glucose solution has been greatly improved.