Engine Fault Isolation via Log-Likelihood Ratio Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current engine diagnostic systems struggle to accurately identify the specific component causing a fault or error, often requiring extensive shutdowns and multiple repairs before the correct issue is addressed, due to their inability to detect small degradations or component-level failures.
Innovation Solution
A method utilizing an engine control module and diagnostics module to determine engine parameters, apply log-likelihood ratio analysis, and provide likelihood-based fault determinations, filtering and ranking potential failure modes to guide repair efforts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If system level monitoring is used to detect faults, then system level faults can be detected, but component level failures cannot be accurately identified
Solution Approach 1:
The patent segments the fault detection problem by dividing system-level faults into component-level failures. It applies log-likelihood ratio analysis to evaluate multiple potential failure modes individually, transforming a single system-level detection task into multiple component-level diagnostic evaluations, thereby achieving precise fault localization.
Solution Approach 2:
The patent changes the diagnostic parameters from binary pass/fail system-level monitoring to continuous probability-based component-level assessment. By calculating likelihood ratios for different failure modes and comparing them against thresholds, the system transforms coarse system-level fault detection into fine-grained component-level diagnosis.
2Ease of operation
If general fault codes are provided, then fault detection is simple, but specific component identification is lost
Solution Approach 1:
The patent implements feedback by continuously monitoring multiple parameters and updating the likelihood assessments of different failure modes. The system provides feedback in the form of ranked probability lists that guide technicians through iterative diagnostic refinement, maintaining simplicity while progressively revealing component-level details.
Solution Approach 2:
The patent adds a probabilistic dimension to traditional fault codes. Instead of providing single discrete fault codes, the system generates multi-dimensional diagnostic information including likelihood ratios, probability rankings, and confidence levels, enabling technicians to prioritize investigations while maintaining diagnostic simplicity.
3Measurement precision
If extensive troubleshooting is performed to identify exact faults, then accurate repair can be achieved, but repair time increases significantly
Solution Approach 1:
The patent performs preliminary action by pre-calculating likelihood ratios and failure mode probabilities based on historical data and system models. When a fault occurs, the system immediately provides a ranked list of probable causes, allowing technicians to skip extensive troubleshooting and directly address the most likely failures, thereby reducing repair time while maintaining high identification accuracy.
Data Source
AI summary
A method (50) for isolating a fault within an engine system (10) including an engine control module (40) and a diagnostics module (36) includes determining, by the engine control module (40), a plurality of parameters of the engine system (10), calculating a likelihood of the fault using a log-likelihood ratio analysis of data within the engine control module (40), and providing a plurality of fault determinations and the likelihood of each of the plurality of fault determinations using plurality of parameters and the diagnostics module (36) of the engine system (10).