Driving behavior classification method and device based on noise reduction diffusion probability model and product

Through the method based on the noise reduction diffusion probability model, the problem of insufficient data volume of EEG signal is solved, the accuracy of driving behavior classification is improved, early detection of driver abnormal behavior is achieved, and the risk of traffic accidents is reduced.

CN119004264AActive Publication Date: 2024-11-22TSINGHUA UNIVERSITY

Patent Information

Application Number
CN202411441083.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-16
Publication Date
2024-11-22
Estimated Expiration
2044-10-16

AI Technical Summary

Technical Problem

In the prior art, the use of electroencephalogram signals (EEG) for driving behavior analysis has the problem of insufficient data, which limits the accuracy of the driver's behavior model and makes it difficult to effectively predict abnormal behaviors of the driver, such as fatigue or distraction, resulting in The risk of traffic accidents increases.

Method used

The method based on the noise reduction diffusion probability model is adopted to enhance EEG characteristics by generating new samples and learning reconstruction errors, improve the accuracy of driving behavior classification, and solve the problem of insufficient sample capacity. The specific steps include: obtaining the sample EEG signal, coding and diffusion processing, inputting the noise reduction diffusion probability model for denoising and reconstruction processing, capturing the representation using the decoder, input classifier for classification, and updating the model parameters.

Benefits of technology

By enhancing the EEG characteristics, the accuracy of the identification of driving behavior models is improved, and the abnormal reactions of drivers can be detected earlier, reducing the risk of traffic accidents.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004264A_ABST
    Figure CN119004264A_ABST
Patent Text Reader

Abstract

The invention provides a driving behavior classification method, device and product based on a noise reduction diffusion probability model, and relates to the technical field of data processing. The method comprises the following steps: acquiring a sample electroencephalogram signal carrying a driving behavior label; inputting the sample electroencephalogram signal into an encoder for encoding to obtain a sample encoding result; performing diffusion processing on the sample electroencephalogram signal to obtain a noisy sample electroencephalogram signal; inputting the noisy sample electroencephalogram signal into a noise reduction diffusion probability model to obtain a de-noised sample electroencephalogram signal; inputting the de-noised sample electroencephalogram signal and the sample coding result into a decoder to obtain a sample decoding result; inputting the sample decoding result into a classifier to obtain a corresponding driving behavior classification result; and according to the corresponding driving behavior classification result and the corresponding driving behavior label, training the driving behavior classification model to obtain a trained driving behavior classification model for performing driving behavior prediction on the to-be-classified electroencephalogram signal, thereby improving the accuracy of driving behavior classification.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Method for detecting vigilance state of driver

    CN105726046A

  • Method for recognizing road rage states of drivers on basis of electroencephalography and pulse information

    CN109498041A

  • Driving ability determination method, electronic equipment and storage medium

    CN116350190A

  • Driving intention recognition method based on function connection and graph neural network

    CN117076980A

  • Model training method and device, electronic equipment, storage medium and program product

    CN117216546A

Cited By

  • Automatic driving emergency processing method and system based on brain-computer interface

    CN121388816A