Predictive Component Modeling for Mechanical Fault Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing system monitoring technologies face challenges in isolating the causes of faults and predicting failures in complex mechanical systems, such as vehicles and HVAC systems, as they struggle to effectively utilize sensor data to declare fault conditions and anticipate impending failures.
Innovation Solution
A predictive system model is configured with component models to mimic the mechanical system's state, using sensor data to capture failure conditions and apply perturbation inputs to produce prediction results, which are then compared to a targeted mode, allowing for the capture of configuration parameters and perturbation inputs that reproduce observed conditions, enabling fault prediction and trend detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If general fault conditions are declared using sensor parameter monitoring, then fault detection capability is improved, but the ability to isolate likely causes and predict when faults are likely to occur deteriorates
Solution Approach 1:
The patent segments the monitoring system into multiple specialized modules: a fault detection module that identifies general fault conditions, a fault isolation module that determines likely causes, and a fault prediction module that anticipates future failures. Each module processes sensor data through specific algorithms tailored to its function, enabling the system to maintain fault detection capability while simultaneously improving cause isolation and prediction without information loss
Solution Approach 2:
The patent introduces an intermediary processing layer between raw sensor data and fault analysis results. This layer includes data fusion algorithms that integrate information from multiple sensors, feature extraction modules that identify relevant patterns, and knowledge bases that store fault signatures. These intermediaries transform raw sensor parameters into meaningful diagnostic information, enabling both fault detection and detailed fault analysis simultaneously
2Reliability
If complex engineered systems are systematically monitored with multiple sensors, then system health monitoring coverage is improved, but the complexity of isolating fault causes and predicting failures increases
Solution Approach 1:
The patent divides the complex monitoring system into modular functional components: sensor interface modules, data preprocessing modules, fault detection engines, isolation algorithms, and prediction systems. Each module handles specific aspects of the monitoring task independently, reducing the complexity burden on any single component while maintaining comprehensive system-wide monitoring coverage
Solution Approach 2:
The patent replaces complex manual fault analysis procedures with automated computational algorithms. Machine learning models and expert system algorithms automatically process sensor data to isolate fault causes and predict failures, substituting what would otherwise require complex human expert analysis with streamlined computational processes that reduce operational complexity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to an aspect, a method includes configuring a predictive system model (116) with a plurality of component models (204A-N) to correspond with a modeled state of a mechanical system (100). A set of perturbation inputs (222) is applied to one or more of the component models to produce a plurality of prediction results (118). The prediction results are compared to a targeted mode (212) of the mechanical system. A set of configuration parameters (220) of the mechanical system is captured in combination with one or more of the perturbation inputs that most closely results in the prediction results matching the targeted mode of the mechanical system.