Application of Artificial Neural Network in Mass Calibration of FTMS

The multiple regression is often used in mass calibration in Fourier transform mass spectrometry (FTMS). Because observed frequency can shift caused by collisional damping and ion space charge effect, it is difficult to express the relationship between ion mass and observed frequency in FTMS. The relationship was investigated by neural networks. The prediction performance of the calibration models constructed by multiple regression and neural networks were compared, and no hidden layer BP with 41 is superior on robustness of prediction ions mass. Artificial neural network approach can provide better prediction results. The relative error of ions mass are less than 2×10^-6.