Analysis And Study Of Gasoline Mass Spectrometric Data

In this paper,based on gasoline mass spectrum data,the classification algorithm based on data analysis and machine learning is proposed for visual display of gasoline mass spectrum data and model prediction.The main research contents of this paper are as follows:(1)By analyzing the specific situation of the original data of gasoline,it is found that the original data is composed of multiple single groups of data,and these single groups of data cannot reflect the overall trend of the original data.In this paper,by fixing the difference value of mass charge ratio,the mass charge ratio representing different substances is defined as the attribute of the data to analyze the data.After the attributes of each group are fixed,the normalized data can be obtained by transposing the data.(2)Analyze gasoline models from two independent technical forecasts.The first is to use the principal component analysis algorithm and the t-distributed stochastic neighbor embedding algorithm to reduce the dimensionality of the preprocessed data.By comparing the results of the two dimensionality reduction algorithms,The various gasoline data of t-SNE obtained by the algorithm’s dimensionality reduction show good discrimination and aggregation,while the PCA algorithm has certain advantages in running time;the second is to build a classification algorithm model for prediction.During the model establishment process,In this paper,three algorithms of stochastic gradient descent,random forest and XGBoost are used for model training.After ten iterations,the average prediction accuracy of the three models is over 73%.Finally,compare the prediction results of the three models to reduce the one-sided impact of a single model and improve the fairness and accuracy of the prediction.(3)On this basis,this paper based on the Python language developed a gas mass spectrometry data analysis software,the field of gasoline fuel types classification and identification,respectively determine the same brand and gasoline to distinguish between different brands and the results of the system output can consumption choices for consumers and businesses in developing marketing strategies and provide efficient and reliable reference information.This article enriches the related research on gasoline model classification,provides solutions to the gasoline model classification problem,and provides guidance and reference data for consumers or enterprises,which has certain practical significance.