Driving behavior mode real-time classification method and system based on short-term observation

Abstract

The invention discloses a driving behavior mode real-time classification method and system based on short-term observation. The system comprises a data preprocessing module, a data dimension reduction module, a driving behavior mode classification module and a driving behavior mode classification module. The method comprises a training stage and a reasoning stage. Carrying out preprocessing and data dimension reduction on the vehicle driving data, and extracting vehicle driving behavior data characteristics of different sources; and classifying the driving behavior data to obtain a driving behavior mode. The technical scheme of the invention can be applied to the intelligent vehicle to classify the driving behavior mode of the driver by using the short-term observation data, and the method is high in effectiveness, good in interpretability of the classification result and high in applicability.

Description

technical field [0001] The invention belongs to the technical field of intelligent traffic data analysis, and in particular relates to a real-time classification method and system of driving behavior patterns based on short-term observation, which is a technology for classifying drivers' driving behavior patterns using short-term observation data applied to intelligent vehicles , through the observation data in a short time window, using the algorithm to obtain the classification results of driving patterns, and the intuitive interpretation of different driving behavior patterns, which can be further used for vehicle behavior prediction. Background technique [0002] In recent years, humans have made great progress in transportation systems, especially as advanced sensors are installed on road infrastructure and vehicles, which can collect a variety of perception data to realize situational awareness and machine intelligence, thereby realizing intelligent transportation syste...

AI Technical Summary

Problems solved by technology

However, this kind of work often requires a long observation time to obtain sufficient vehicle and driver data, and is not suitable for online analysis systems that require high real-time performance.

In another recent work, the researchers used semi-supervised SVM to reduce the burden of data labeling without reducing performance, but only focused on the behavior of the forward direction, only considering the vehicle speed and throttle status

[0005] All in all, the above existing work is classified based on some artificially predefined driving behavior patterns, and for different traffic conditions, the rules for judging different styles are very different, such as whether the traffic is crowded or not, night and day, etc., generally Poor adaptability; or a large amount of data is required to achieve the expected accuracy, but it is difficult to collect all the required data in practical applications

It is difficult for existing technologies to realize real-time and continuous classification of driving behavior patterns in any actual environment, only through observation data within a short period of time

Images