To address the current problem that the printability of water-based inks is difficult to detect and control accurately, a method based on visible/near-infrared spectroscopy combined with chemometrics is proposed to accurately detect the type and concentration of water-based ink additives. First, a micro-spectrometer (380-980 nm) was used to obtain spectral data of water-based ink samples containing different concentrations of alcohol, toning red, and toning yellow additives, and by extracting the characteristic bands that can completely characterize the spectral information of the samples through principal component analysis, redundant information is reduced and accurate identification of the auxiliary types is achieved. Then, the prediction models constructed by six different pretreatment methods combined with partial least squares (PLS) and interval partial least squares (iPLS) for each additive concentration were compared and discussed. The experimental results showed that the cumulative contribution rate of the first two principal components in the spectral band of 617-726 nm was as high as 99.909% by principal component analysis, and the accurate identification of water-based ink additives types was achieved. Among them, the determination coefficient and root mean square error of alcohol additive prediction model were 0.9798 and 0.0223, while the determination coefficient and root mean square error of toning yellow additive were 0.9870 and 0.0075, and the determination coefficient and root mean square error of the toning red additive were 0.9948 and 0.0038. Experimental results prove that the model established by the optimal pretreatment method combined with the iPLS method is generally better than the single PLS model, which can accurately detect the type and concentration of water-based ink additives, meet the application needs of water-based ink detection and control in the printing production process, and provide a technical basis for the later realization of online detection of water-based ink.