Vehicle Fault Sign Detection Using Normal and Fault Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fault detection systems in vehicles face challenges in accurately predicting faults without specialized sensors and often result in false detections due to environmental factors affecting tire pressure fluctuations.
Innovation Solution
A fault sign detection device that compares observation data against predetermined normal and fault pattern files, using threshold values to determine the presence of a fault, thereby improving detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If specialized sensors are installed to detect fault signs such as noise and odor, then detection capability is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates virtual sensor data by copying and processing existing vehicle operation data through machine learning models. Instead of installing physical sensors to detect noise and odor, the system generates synthetic sensor readings from patterns in existing data, thereby achieving fault detection capability without adding physical sensing devices.
Solution Approach 2:
The patent replaces the mechanical/physical sensor system with an information processing system. Machine learning models process existing vehicle operation data to infer fault conditions that would traditionally require specialized sensors, substituting computational analysis for physical measurement devices.
2Device complexity
If fault detection is performed by comparing with normal conditions only, then simplicity is maintained, but false detections increase due to environmental factors
Solution Approach 1:
The patent segments the detection process into multiple stages: first comparing with normal condition patterns, then evaluating against multiple environmental and operational factors. This multi-stage segmentation allows the system to maintain simplicity in individual comparison steps while achieving high reliability through cumulative evaluation of multiple conditions.
Solution Approach 2:
The patent dynamically adjusts detection parameters and thresholds based on environmental conditions and vehicle operation states. By changing the parameters used for comparison based on contextual factors, the system maintains simple comparison logic while adapting to different conditions to reduce false detections and improve accuracy.
3Measurement precision
If multiple environmental factors are considered in fault detection, then detection accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary processing of environmental and operational data by pre-training machine learning models with historical data that encompasses multiple environmental factors. This preliminary action during the training phase allows the system to automatically learn and integrate multiple factors without requiring complex real-time processing during actual fault detection.
Solution Approach 2:
The patent introduces machine learning models as intermediary components that automatically process and integrate multiple environmental and operational factors. These models serve as mediators between raw multi-factor data and fault detection decisions, handling the complexity of factor integration while presenting simplified outputs for fault determination.
4Reliability
If fault patterns are predicted based on elapsed time only, then age-related faults are detected, but daily life faults are missed
Solution Approach 1:
The patent merges time-based degradation patterns with operational condition patterns in a unified detection framework. By combining elapsed time analysis with real-time operational data analysis, the system simultaneously detects both age-related faults (from time patterns) and daily life faults (from operational patterns), achieving comprehensive fault coverage.
Solution Approach 2:
The patent creates a universal detection system that handles multiple fault types through a single integrated approach. The machine learning models are trained to recognize both time-dependent degradation patterns and operation-dependent fault patterns, enabling the system to perform multiple detection functions (age-related and daily faults) with a unified methodology.
Data Source
AI summary
An input unit 81 receives input of a normal pattern file including information indicating a normal condition of a vehicle, and a fault pattern file including information indicating a sign that a vehicle fault is about to occur. A collection unit 82 collects observation data observed by each device in the target vehicle. A comparison unit 83 compares the content of the normal pattern file with the content of the observation data. when the difference between the content of the normal pattern file and the content of the observation data is greater than a predetermined first threshold value, the comparison unit 83 further compares the content of the fault pattern file and the observation data, and when the difference between the content of the fault pattern file and the observation data is within a predetermined second threshold value, determines that there is the sign of the fault in the target vehicle.