Vehicle Trip Anomaly Detection Using Neural Network Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting harsh driving events in vehicles consume excessive computing resources and fail to accurately identify all events, leading to undetected alerts, incorrect classifications, and potential traffic accidents.
Innovation Solution
An anomaly detection system using a neural network model processes historical data to generate clusters, allowing it to automatically detect anomalous driving patterns by comparing new trips to trained clusters, thereby conserving resources and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed threshold methods are used for harsh driving event detection, then the detection process is simple, but the accuracy of identifying all harsh driving events is poor
Solution Approach 1:
The system transforms fixed threshold parameters into dynamic, adaptive parameters by training neural network models on historical data. The detection thresholds are no longer static values but are derived from learned patterns in the data, allowing the system to adapt to different driving conditions and vehicle types, thereby improving detection accuracy without requiring overly complex manual configuration
Solution Approach 2:
The patent replaces the mechanical threshold-based detection system with an intelligent neural network-based system. Instead of using predetermined fixed thresholds that require manual adjustment, the system employs machine learning models that automatically learn optimal detection criteria from historical data, substituting rigid mechanical logic with adaptive intelligent processing
2Reliability
If manual threshold adjustment is performed to capture majority of harsh driving events, then coverage is improved, but computing resources and networking resources are excessively consumed
Solution Approach 1:
The system performs preliminary action by pre-training neural network models offline using historical harsh driving event data. This preprocessing step creates ready-to-use detection models that can be deployed to edge devices, eliminating the need for continuous heavy computing during actual operation. The models are prepared in advance to enable efficient real-time detection with minimal resource consumption
Solution Approach 2:
The patent creates simplified copies of complex detection logic by training neural network models on historical data and deploying them to edge devices. Instead of centralizing all processing and requiring constant networking resources, the system replicates intelligent detection capabilities across multiple distributed devices, allowing local processing that reduces networking overhead while maintaining comprehensive coverage
3Use of energy by moving object
If fixed threshold detection is used, then resource consumption is reduced, but false classifications and undetected events occur leading to potential accidents
Solution Approach 1:
The system implements self-service by enabling neural network models to automatically learn and optimize detection parameters from historical data without requiring continuous manual intervention or centralized reconfiguration. The models autonomously adapt to new patterns and conditions, improving classification accuracy while maintaining efficient resource utilization through decentralized intelligent processing
Data Source
AI summary
A device may receive historical input data associated with trips traversed by a plurality of vehicles with vehicle tracking units (VTUs), and may process the historical input data to generate training data. The device may train a neural network model, with the training data, to generate a trained neural network model that provides a latent space representation of vectors, and may cluster the latent space representation of vectors to generate clusters. The device may receive input data associated with a trip traversed by a vehicle of the plurality of vehicles with the VTUs, and may process the input data to generate time series data. The device may compare the time series data and the clusters to determine whether the trip is anomalous or not anomalous, and may perform one or more actions based on the determination of whether the trip is anomalous or not anomalous.