Predictive Reliability Mining Using DTC-Based Failure Forecasting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional predictive reliability mining techniques fail to detect potential systemic problems in machine populations early and often miss anomalies hidden within the overall population, as they rely on significant deviations in failure counts that may not be evident across the entire population.
Innovation Solution
The system employs on-board diagnostic sensors in connected machines to generate Diagnostic Trouble Codes (DTCs), which are analyzed to correlate with future failures, using a temporal conditional dependence model based on past failure data and association rule mining to identify discriminative DTCs and predict future failures, while performing root cause analysis to identify subsets with anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional reliability mining techniques are used based on historical warranty claims data, then the analysis covers the full population, but early detection of potential systemic problems is delayed until significant deviations in failure counts occur
Solution Approach 1:
The system performs preliminary actions by continuously monitoring diagnostic sensor data and generating early warnings before significant deviations in failure counts occur. The temporal conditional dependence model analyzes sensor data in real-time to detect potential systemic problems early, rather than waiting for accumulated failure data to show significant deviations from expected patterns.
Solution Approach 2:
The patent introduces an intermediary mechanism - the temporal conditional dependence model - that bridges the gap between traditional failure count analysis and early problem detection. This model uses sensor data as an intermediary signal to detect potential issues before they manifest as significant deviations in failure counts, enabling earlier intervention.
2Measurement precision
If traditional reliability analysis is applied to the full population, then statistical power is maintained, but anomalies in specific subsets (batches, manufacturing plants, years) remain hidden and unidentified
Solution Approach 1:
The system applies segmentation by dividing the full population into meaningful subsets based on manufacturing attributes such as batch, plant, and year. The temporal conditional dependence model can analyze these segmented groups separately, enabling detection of anomalies specific to particular subsets while maintaining the overall population context. This segmentation reveals hidden patterns that would be masked in aggregate analysis.
Solution Approach 2:
The patent implements local quality by allowing different analysis parameters and thresholds for different subsets of the population. Instead of applying a uniform analysis to all machines, the system can tailor the temporal conditional dependence model to specific manufacturing batches, plants, or time periods, detecting local anomalies with precision appropriate to each subset's characteristics.
3Reliability
If sensor data from connected machines is integrated into reliability models, then early warnings can be generated before failures occur, but system complexity increases
Solution Approach 1:
The temporal conditional dependence model serves multiple functions simultaneously: it analyzes sensor data, detects anomalies, predicts failures, and identifies subset-specific patterns. This multi-functionality reduces the need for separate specialized systems for each task, managing complexity through a unified framework that handles diverse reliability analysis requirements.
Solution Approach 2:
The system manages complexity by dynamically adjusting parameters within the temporal conditional dependence model based on the specific analysis context. Rather than using fixed complex rules, the model adapts its parameters to the characteristics of the data being analyzed, allowing flexible handling of different sensor types, time periods, and machine populations without requiring completely different models for each scenario.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for predictive reliability mining are provided that enable predicting of unexpected emerging failures in future without waiting for actual failures to start occurring in significant numbers. Sets of discriminative Diagnostic Trouble Codes (DTCs) from connected machines in a population are identified before failure of the associated parts. A temporal conditional dependence model based on the temporal dependence between the failure of the parts from past failure data and the identified sets of discriminative DTCs is generated. Future failures are predicted based on the generated temporal conditional dependence and root cause analysis of the predicted future failures is performed for predictive reliability mining. The probability of failure is computed based on both occurrence and non-occurrence of DTCs. The root cause analysis enables identifying a subset of the population when an early warning is generated and also when an early warning is not generated.