Driver character recognition method

Abstract

The invention discloses a driver character recognition method. The method includes the steps that firstly, experimental data are acquired through multiple sensors and filtered; secondly, the filtereddata are normalized, then cluster analysis is conducted on the data, and time series segmented data are obtained; thirdly, a double-layer HMM model base is established and comprises a lower-layer behavior recognition layer and an upper-layer character recognition layer, and the upper layer recognizes the characters of a driver according to a lower-layer behavior recognition result; and fourthly, model verification is conducted. An established recognition model can be closely related to the reality, the characters of various driving conditions of the driver are recognized in combination with the practical condition, the model is further close to the reality, and universality and popularization are high. Based on the hidden Markov model (HMM) theory, the credibility of the obtained factor weight is high.

Description

technical field [0001] The invention relates to a driver characteristic identification method, which belongs to the field of vehicle control. Background technique [0002] In today's society, cars have become an indispensable means of transportation in people's lives. With the development of science and technology and the improvement of people's living standards, people have higher and higher requirements for cars. Cars are no longer just ordinary means of transportation, but have risen to become a culture and a symbol. The importance of cars is self-evident. Metaphor. [0003] Driver characteristics classification, identification and reference model modeling are the basis for vehicle dynamics control considering driver characteristics. Due to the use of wire control technology, four-wheel independent drive and steering electric vehicles do not have the limitations of mechanical or hydraulic systems of traditional vehicles. The driver's manipulation and actuators are conne...

AI Technical Summary

Problems solved by technology

[0004] The existing driver characteristic identification method is not precise enough in data processing, and the interference data filtering is not accurate enough, which affects the overall data analysis and affects the accuracy of driver characteristic analysis. Paper: Based on double-layer implicit Mark Driving characteristics identification and thesis of the Cove model: In the research of the driver characteristics identification method based on the invisible Markov model, there are hidden dangers in the filtering of interference data

[0005] The inventor's prior application for Chinese patent authorization announcement number: CN105034986A discloses a method and device for online identification of driver's steering characteristics, using BP neural network to establish an offline identification model, extracting weights and thresholds, and then establishing an online identification model. The ability of neural networks to process time series is not strong, and they are mostly used for static identification. At the same time, most of the current research on driver characteristic identification is limited to a single working condition level

Images