Research On Evaluation And Suggestion Method Of Bus Driving Status Based On Deep Learning

Drivers' decision is an important aspects that can affect the driving safety,at the same time,it also affects vehicle exhaust emissions and energy consumption level.Keeping good driving status can not only improve ride safety and comfort,but also reduce fuel consumption.Keeping the driver in good driving status firstly requires a large amount of driving data to model driving data.According to the results of modeling processing,the driver is provided with real-time evaluation and suggestions of driving status.This kind of research can help drivers improve driving skill,improve driving comfort and reduce energy consumption.Based on optimizing the state of bus driving status,this paper proposes a new concept-the evaluation zone,which are special driving areas that has great influence on energy consumption and ride comfort.Then,an evaluation zone oriented bus driving status model is proposed to get good driving status dataset,which is then used to provide training samples for driving status evaluation and recommended deep learning neural networks.The trained network can provide drivers with driving status evaluation and recommendations,the main contents of the thesis are as follows:First of all,the background and significance of bus driving status evaluation and suggestion are described,and the research status of driving data modeling,driving status evaluation and suggestion and deep learning in driving status are introduced.After that,the system framework of driving status analysis is proposed,and the key technologies used about driving status analysis in this paper are introduced and determined.Then,based on the energy-saving and comfort of eco-driving,the evaluation zone oriented bus driving status model and the data processing flow in the model are introduced.The driving data is segmented according to the evaluation zone,time period and airconditioning usage.Cluster analysis of each data segment is performed to obtain a good bus driving status' s dataset,which is used as a training sample for deep learning neural networks.Finally,the neural network of bus driving status evaluation and bus driving status suggestion is designed by using Long Short-Term Memory network.The experimental results show that the bus driving status evaluation and suggested neural networks which based on the evaluation zone oriented bus driving status model can give reasonable bus driving status evaluation and recommendations.The application scenarios of the model are introduced,and the test results of the real-time feedback system developed in this paper are given.