Adding a small amount of additives to the lubricating oil can make the lubricating oil obtain some new characteristics or improve the properties of some existing characteristics in the lubricating oil.Aiming at the problem of identifying various kinds of tiny additives in lubricating oil of mechanical equipment.Based on Python,eight different samples were prepared with Base Oil PAO-10 and three Commercial lubricating oil additives, T321,T534,and T307, in different proportions.Thermo Scientific Nicolet IS5 Fourier collected the mid-infrared spectra of the samples transform infrared spectrometer in the range of 4 000～400 cm-1,and the infrared spectra of the samples were normalized by Min-Max.For nearly category mechanical equipment of tiny amount of additive in lubricating oil variety identification,four classical classification algorithms are studied,including the Support Vector Classifier (OVR SVMs),Random Forests Classifier (RF),embedded in the Genetic Algorithm (GA),and Local search Genetic Algorithm (LGA) optimization technologies,infrared spectrum characteristic band many category classification model building methods are established.Example test results show that the accuracy of the new model improves the original classical algorithms OVR SVMs (91.67%)and RF (79.17%) to OVR SVMs (100%)and RF (100%).With the new models embedded in LGA,the length of the characteristic band was shortened to 36.7% of the length of the original band.The new model applies to the case with only one additive and has a high recognition rate of 100% for the simultaneous identification of two or more additives.The results show that the model can effectively realize the rapid,accurate,and multi-type synchronous recognition of small amounts of lubricanting oil additives.