A fitting method of unlabeled interference for spectra based on multi-stage cross optimization

[Display omitted] •An impact model of single interference on the spectrum was constructed.•The difficulty of spectral analysis without interference label was pointed out.•A concept of virtual interference was proposed.•A method that fit the impact of unlabeled interference on spectra was proposed.•This method was successfully applied to cobalt ion spectral detection. When constructing a spectral detection model for a substance, it will be very difficult to ensure the accuracy of the model if there is an unlabeled interference in the spectrum. To overcome the issue, we used a mathematical model to describe the impact mode of single interference on spectra, and pointed out the difficulties of spectral analysis in the case of unknown interference label. Then, a fitting method based on multi-stage cross optimization was proposed, which can fit the impact of single interference on spectra without interference label. Finally, we applied the proposed method on cobalt ion spectral detection with temperature interference. The application results indicate that although the influence modes of temperature interference and cobalt ion concentration on spectra are similar, this method can still accurately fit their influence modes, and the effectiveness of the fitting model was verified through spectral detection for blank sample.