Research On Identification And Detection Methods Of Trace Components In Lubricant Additives

As a common method to improve the performance of lubricating oil,lubricating oil additives generally exist in various lubricating oils.Physical and chemical methods are often used to determine the additives contained in lubricating oils.Physical and chemical methods include Raman spectroscopy,physical and chemical performance testing,and gas chromatography etc.these methods take a long time,the measurement results are large in error,and the consumption cost is high.In order to quickly identify the types and content of additives in lubricating oils,the additives are sulfurized isobutylene(T321),alkyl diphenylamine(T534),and phosphorothioate(T307)in different proportions in the base oil,using an Extreme Learning Machine trains and tests the infrared spectral data model of the oil sample,and uses Greedy Algorithm and Genetic Algorithm to optimize the input band,and selects the optimal band interval combination to eliminate the highly correlated bands so we can improve the computing efficiency.Research indicates:1.The Extreme Learning Machine model can effectively identify the types and content of lubricant additives,verifying the feasibility of using the model to analyze the infrared spectrum as a rapid determination method for lubricant additives,which is a new and economical method compared to traditional physical and chemical detection methods.Lubricant additive detection method;and the model output is optimized after Genetic Algorithm band screening and optimization.The accuracy rate of the three types of additives is increased from 75%,85%,and 75%to 100%,and the content prediction determination coefficient R2 is respectively Increased from 0.6278,0.6436,and 0.6853 to 0.9028,0.8945,and 0.8524,an increase of 43.8%,39.0%,and 24.4%,respectively2.The method of machine learning combined with infrared spectroscopy is applicable to the detection of trace components of lubricating oil additives.It is not difficult to foresee that after a large amount of collection work,a relatively complete database can be constructed,and the unknown can be quickly identified based on this database.The additives contained in the lubricating oil are combined with various experimentally determined performance indicators of the lubricating oil.It is hoped that the performance of the additive-containing lubricating oil can be inferred only by analyzing the infrared spectrum,saving the cost of detection and analysis and improving Detect and analyze speed,and provide simulation models for scientific research on lubricant performance to guide research directions.