Vehicle Control Anomaly Detection Using Bayesian Event Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Implementing multiple monitoring devices for anomaly detection in vehicle on-board systems increases system costs.
Innovation Solution
A vehicle control device utilizing a command unit, state quantity acquiring unit, event storage unit, event acquiring unit, and anomaly determining unit, which employs a Bayesian network to determine anomalies in vehicle on-board devices based on vehicle state quantities and event probabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple monitoring devices are implemented for anomaly detection, then detection reliability is improved, but system cost increases
Solution Approach 1:
The patent merges the anomaly detection function into the existing travel control device, combining multiple monitoring functions (actuator monitoring, sensor monitoring, detection system monitoring) into a single integrated system. This eliminates the need for separate main and auxiliary monitoring devices while achieving comprehensive anomaly detection through unified processing of command values, state quantities, and event correlations.
Solution Approach 2:
The travel control device is designed to perform multiple functions: it controls vehicle travel, monitors actuator operations, detects anomalies, and correlates events. By making the monitoring system universal and multi-functional, the patent avoids the need for dedicated separate monitoring devices, thereby reducing system cost while maintaining detection reliability.
2Measurement precision
If multiple monitoring devices execute the same determining process, then anomaly detection accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the anomaly detection process into distinct functional units within a single device: a command value generation unit, a state quantity acquisition unit, an event correlation unit, and a determination unit. Each unit performs a specific function, and their coordinated operation achieves high detection accuracy without requiring multiple complete monitoring devices.
Solution Approach 2:
The patent introduces event correlation as an intermediary mechanism that connects command values, state quantities, and anomaly determinations. By using event correlation tables and probabilistic reasoning, the system achieves accurate anomaly detection through indirect inference rather than direct redundant monitoring, reducing system complexity.
3Device complexity
If a single monitoring system is used, then system cost is reduced, but anomaly detection reliability may deteriorate
Solution Approach 1:
The patent implements feedback mechanisms where the determination result feeds back into the event correlation process, and state quantities are continuously monitored and compared against expected values. This closed-loop feedback ensures high detection reliability by continuously validating system state and adjusting anomaly probability assessments.
Solution Approach 2:
The patent performs preliminary actions by pre-establishing event correlation tables and expected state quantity relationships before anomaly detection begins. This preparation allows the single monitoring system to quickly and reliably determine anomalies by comparing real-time data against pre-computed expectations, enhancing detection reliability without additional hardware.
Data Source
AI summary
A vehicle control device includes a command unit that generates a command value for an actuator and causes a vehicle to travel by outputting the command value to a control unit, a state quantity acquiring unit that acquires at least two values of a vehicle state quantity among a vehicle state quantity ideal value, a vehicle state quantity detection value, and a vehicle state quantity operation value, an event storage unit that stores multiple events that can occur when the vehicle on-board device is not functioning normally, an event acquiring unit that compares at least two values of the vehicle state quantity and acquires an event corresponding to a result of the comparison, and an anomaly determining unit that determines whether there is an anomaly in the vehicle on-board device by using a Bayesian network that includes, as a node, an occurrence probability of the event.