Compressor Fault Diagnosis Using Fuzzy Anomaly Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional algorithms for diagnosing compressor faults are not always accurate and can be ambiguous, making it difficult to determine the relevance of symptoms and identify potential failure modes effectively.
Innovation Solution
A method and system that generate a feature vector from compressor measurement data, using fuzzy constraints to create an aggregated anomaly vector, and calculating a corroborating measure between this vector and predefined rules to identify potential faults or failure modes, incorporating fuzzy thresholds to assess parameter deviations and risk levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional algorithms are used for diagnosing compressor faults, then the diagnostic process is simple, but the accuracy and clarity of fault identification deteriorates
Solution Approach 1:
The diagnostic system segments the fault identification process into distinct modules: feature extraction from measurement data, anomaly detection using fuzzy constraints, rule-based fault mode matching, and corroborating measure calculation. This segmentation allows each module to specialize in specific tasks, improving overall diagnostic accuracy while maintaining manageable system complexity through modular architecture.
Solution Approach 2:
The patent introduces an intermediary aggregated anomaly vector that mediates between raw measurement data and fault diagnosis. This anomaly vector serves as an intermediate representation that captures system deviations without directly exposing the complexity of the underlying fuzzy logic operations, thereby improving measurement precision while shielding the user from excessive complexity.
2Reliability
If traditional anomaly detection methods are used, then the system is easy to operate, but the reliability of maintenance decisions deteriorates
Solution Approach 1:
The system implements feedback through the corroborating measure calculation, which continuously evaluates the consistency between observed anomalies and predefined fault rules. This feedback mechanism enhances maintenance decision reliability by providing quantitative validation, while the automated nature of the feedback loop maintains ease of operation without requiring manual intervention.
Solution Approach 2:
The patent transforms raw measurement parameters into an aggregated anomaly vector using fuzzy constraints, changing the parameter representation to better suit diagnostic purposes. This parameter transformation improves reliability by capturing subtle deviations that traditional methods might miss, while the automated transformation process preserves ease of operation.
3Measurement precision
If detailed symptom analysis is performed, then fault identification becomes more accurate, but the time required for diagnosis increases
Solution Approach 1:
The system performs preliminary action by pre-defining fault rules and constraints before actual diagnosis occurs. During operation, the aggregated anomaly vector is quickly compared against these pre-established rules, enabling accurate symptom analysis without requiring time-consuming real-time complex calculations, thus reducing diagnosis time while maintaining precision.
Solution Approach 2:
The patent applies partial action by focusing the detailed analysis only on the most significant anomalies identified through fuzzy constraints, rather than analyzing all possible symptoms equally. This selective approach maintains high accuracy for critical fault indicators while reducing overall diagnosis time by avoiding unnecessary analysis of less relevant parameters.
Data Source
AI summary
Method, system and computer software for diagnosing a compressor. The method includes generating a feature vector of the compressor, the feature vector of the compressor including components describing states of various parts of the compressor; determining, based on fuzzy constraints, an aggregated anomaly vector corresponding to the feature vector; defining rules for a preset list of possible faults/failure modes of the compressor; calculating a corroborating measure between the aggregated anomaly vector and the rules; and identifying a fault/failure mode of the compressor based on a result of the corroborating measure.