Research On Driver Identification Mechanism Based On Historical Driving Characteristics

With the increasing diversification of driver-related services,the personal identity of drivers has become increasingly important.Accurate driver identification technology can bring many conveniences to both the individual and service providers.Consequently,there has been a continuous emergence of various driver identification methods.Among them,mainstream identification methods include facial recognition,fingerprint identification,and so on.Although these identification methods are simple and efficient,they are not fully applicable to the driver identification scenario,especially when vehicles do not have additional camera sensors.These methods appear to be more passive in such cases.Therefore,a driver identification mechanism focusing on driving behavior is of significant importance.This paper explores a driver identification mechanism based on sensor data generated from driving behavior to address the driver identification problem in scenarios where only vehicle speed and acceleration sensors are available.The research content of this paper is subdivided as follows:(1)In response to the data mining problem in the driver identification scenario,this paper presents a comprehensive data processing workflow,including operations such as data preprocessing,feature extraction,data augmentation,and data standardization.It incorporates data denoising algorithms and data augmentation algorithms,resulting in the generation of statistical features that capture rich driver behavior characteristics in both the time domain and frequency domain.This achieves the generation of high-quality samples.(2)To address the post-processing of the probability distribution output by the softmax layer,a driver identification algorithm is designed.This algorithm takes the probability distribution as input and identifies the target driver through a process of filtering and iterative updating of the distribution.It accomplishes the secondary processing of the probability distribution.(3)Based on the proposed driver identification mechanism,a driver identification system is developed.It incorporates functionalities such as data collection,data upload and download,and data analysis.The algorithms proposed in this paper are validated in the experimental section.The experiments demonstrate that the model performance is stable,and the recognition rate based on the identification algorithm is approximately2 percentage points higher than traditional methods in terms of performance.This proves the effectiveness and feasibility of the proposed identification mechanism.