Research On Gasoline Mass Spectrometry Analysis And Application Based On Machine Learning

Mass spectrometry is a chemical analysis technique used for compound detection,which can provide information such as the precise mass,molecular formula,and molecular structure of a molecule.Therefore,it is widely used in the field of gasoline quality assessment.In order to further improve the identification ability of mass spectrometry for gasoline components,this article proposes to combine machine learning with gasoline mass spectrometry analysis to improve the accuracy and efficiency of gasoline quality detection.The main research content is as follows:(1)Data normalization.This article analyzes the original gasoline mass spectrometry data and finds that there are differences in the mass-to-charge ratio of different gasoline mass spectrometry data,which makes these data unsuitable for subsequent data analysis.To solve this problem,this article uses fixed attribute differences to convert the original data into standardized data.(2)Feature selection.This article focuses on the research of the gasoline model identification model.To address the problem of high dimensionality of gasoline mass spectrometry data,a feature selection algorithm called RAP,which combines Relief-F algorithm and Pearson correlation coefficient method,is proposed to realize the feature selection of gasoline mass spectrometry data.The experimental results show that after the feature selection algorithm proposed in this article is used,the accuracy and training time of the final gasoline model identification model have been improved.(3)Gasoline model identification model.This article uses the XGBoost classification algorithm for gasoline model identification.To address the problem that XGBoost has many parameters that are difficult to adjust and may result in low gasoline model identification rates,an improved marine predator algorithm called IMPA is proposed.It can quickly find the optimal parameter combination of XGBoost and propose a gasoline model identification model based on IMPA-XGBoost.The experimental results show that compared with other common intelligent optimization algorithms for optimizing XGBoost,the IMPA-XGBoost gasoline model identification model proposed in this article has higher accuracy.(4)Design and implementation of the gasoline model identification system.Based on the gasoline mass spectrometry data,the feature selection algorithm RAP,and the IMPA-XGBoost gasoline model identification model,this article uses the Python language to design and implement a gasoline model identification system,and applies this system to the identification of gasoline models in real-life commercial gasoline.The results show that this system also has good identification effects on commercial gasoline in the field.This study utilizes machine learning techniques to analyze and apply gasoline mass spectrometry data,achieving rapid evaluation of gasoline quality,which has certain application prospects and promotion value.