一种基于证据深度学习的车载异常综合检测方法及系统
By designing an evidence-based deep learning model for vehicle CAN networks and utilizing information entropy and dynamically optimized loss functions, the problems of asymmetric feature extraction and ambiguous uncertainty calculations in existing technologies are solved, enabling accurate classification and fault diagnosis of vehicle network anomalies.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HEFEI UNIV OF TECH
- Filing Date
- 2026-04-09
- Publication Date
- 2026-07-17
AI Technical Summary
Existing vehicle CAN network security detection models based on evidence theory and deep learning do not design input features specifically for the characteristics of vehicle CAN network communication, feature extraction lacks specificity, do not define a unified standard for uncertainty measurement calculation, and the loss function cannot be dynamically adjusted, resulting in low detection accuracy.
Design an evidence deep learning model that receives the information entropy of the vehicle CAN network message sequence through the feature input layer, combines the evidence reasoning layer and the uncertainty calculation layer, adopts a dynamically optimized loss function, uses the optimal annealing factor β for model training and detection, and combines local frequency domain features and dynamic time warping distance for fault judgment.
It achieves accurate classification of anomalies in vehicle networks, improves recognition sensitivity and robustness, reduces false alarm rate, and ensures high-precision detection in complex environments.
Smart Images
Figure CN121997183B_ABST