Research And Application Of Driving Pattern Recognition Based On Hidden Markov Model

With the rapid development of our society and economy, increasing well-development of the logistics industry, the road transport as an efficient mode of logistics, presents an unprecedented prosperity. However, road transport has also brought serious problemssecurity issues, energy issue and environmental issues. Accidents often occur during transport, oil prices continue to rise, exhausting pollution is getting worse. All have increased pressures on logistics enterprises. Energy conservation has been the focus of logistics enterprises. In order to improve vehicle safety and fuel economy, the major car manufacturers invested heavily in research and development costs and introduced a variety of innovative technologies, so far, has been highly developed automotive technology, every step will need to invest huge cost.In recent years, people gradually realized that the driver’s driving behavior has an impact not only on safe, but also on vehicle fuel consumption, statistics show that traffic accidents caused by drivers of human factors as high as 80%, while the driving habits on fuel economic impact of up to 22%. Therefore, the study of energy-saving driving behavior and promote safe driving techniques to improve the traffic environment, reduce the cost of logistics enterprises has essential significance. Modeling is an effective method to research the driving behavior, through the investigation found that the current research lacks of easy-to-guide modeling technology for driving behavior, and relatively rare model by establishing relationships with energy-efficient driving behavior, mostly stay in the empirical description.The development of car networking technology has achieved easily to get the vehicles running parameters and location. The collection of large amount of driving data has made a hot spot to study traffic problem though big data. In this paper, with traffic data sets provided by TY company, proposed a driving behavior model based on a two-layer structure of HMM to identify safe, energy-efficient driving patterns. In the double-layer HMM structure, lower multidimensional Gaussian mixture HMM, in a variety of automotive sensor data observed variables, it is easy to understand the various driving operating results for the reasoning and enter into the upper characterization driving intention multidimensional discrete HMM. The double-layer HMM is used to characterize different driving patterns. In order to verify the double-layer HMM driving model, this paper takes energy-saving driving pattern recognition for example, after quantitative analysis of the impact of various driving behavior on fuel consumption, using the model to test real environment data, the results show that the model can effectively identify energy-saving and non-energy-saving driving patterns.Based on the objective needs of transport companies, with the aim to help improve the driver’s driving skills, this paper proposes a driving pattern recognition method, which has important reference to the research and promotion of safety and energy-saving driving technology.