Based on the error theory, “M+N” theory illustrates the effect of “N” factor such as external disturbance in the spectral analysis and provides a method to improve measurement accuracy. Multi-component ideal simulation model was established using the method that combines source spectra measured under different voltage with multi-component linear absorption spectral model. Meanwhile, source voltage was chosen as the “N” factor influencing the accuracy of measured component concentration. “N” factor distribution mode of samples was set to two groups with three different modes under different signal-to-noise ratio. It is concluded that prediction accuracy of component concentration is certainly influenced by “N” factor distribution range of calibration set and prediction set, which is system error. It was verified by the experimental results that “N” factor can be used to improve measurement accuracy based on “M+N” theory. Moreover, this research supplies a theoretic guidance for reducing interference and improving measurement accuracy of complex measurement system.