Economic Analysis Of Pure Electric Bus Based On Driving Style

New energy vehicles are developing rapidly and expanding their market share under the dual promotion of "dual carbon" policy and technological innovation.As the main component of public transport,the trend of pure electric bus is becoming more and more clear.Improving the economy of electric buses can not only help energy conservation and emission reduction,but also effectively reduce the operating costs of bus companies.In addition to vehicle design and environmental impact,driving style has great correlation with vehicle energy consumption.Therefore,it is of great significance to analyze the potential relationship between driving style and vehicle energy consumption to make improvement and optimization from the perspective of driver operation for improving vehicle economy.At the same time,improving driving style can also effectively improve vehicle safety.This paper combines driving style analysis and energy consumption prediction system to analyze the law.Specifically,the main contents include the following aspects:1.Conduct a series of data preprocessing to improve data accuracy.Delete outliers and invalid data.Compare the accuracy of KNN interpolation and EM interpolation,and select the more accurate method to interpolate the vacancy data.Aiming at the noise problem of data acquisition,filter the interpolated data to eliminate the interference.Finally,according to Pearson correlation coefficient,determine the degree of correlation and complete data dimension reduction,and divide the trip to change the data saving form.2.To solve the problem that the classification of driving style is separate from specific driving behavior,propose a scoring system based on driving behavior to classify driving style.Firstly,in view of the characteristics of large volume and multi-dimension of data,select K-means clustering for cluster analysis.Use contour coefficient method to determine the optimal number of categories of data and ANOVA analysis to determine whether the clustering was complete.Ten pure electric buses are divided into five groups according to the driving characteristics obtained by clustering.Secondly,nine indexes of driving behavior,driving speed and driving time are selected as references.Analyze and verify the data distribution form,so as to design the targeted recognition algorithm according to the actual situation.Propose an integrated weighting method by comparing the results of subjective and objective weighting methods.In this fusion weighting method,the overall trend is controlled by subjective weighting,and the final weight of the refined index is adjusted by objective weighting.Build a scoring system for driving behavior to score the five groups based on the weights,and finally complete the driving style classification according to the scores.3.Propose a KF-LSTM prediction model combining Kalman filter and LSTM aiming at the misalignment of energy consumption forecast caused by complex and variable traffic conditions.KF-LSTM model,which has higher accuracy than the conventional LSTM model,combines the selective memory prediction process in LSTM with the observation update process in Kalman filter.Complete the setting of the forecast set by comprehensively considering the difference of temperature and working conditions during the week and weekends.The final results show that driving style has obvious influence on vehicle energy consumption under the same working conditions.The difference in energy consumption between radical and conservative styles can reach more than 10% between the mean and the minimum value,and the maximum difference has reached more than 5%.