Auto-Associative Neural Network for Multivariate Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current condition-based maintenance approaches for industrial assets rely on univariate techniques that fail to effectively detect anomalies in complex systems, as they do not account for interactions between sensor measurements, leading to delayed fault detection and increased operational costs.
Innovation Solution
The implementation of an auto-associative neural network (AANN) for advanced condition monitoring, which senses actual values, estimates them, determines residual vectors, and performs fault diagnostics to provide timely alerts for asset system changes, utilizing techniques like Hotelling's T2 statistic for multivariate change detection and sensor validation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If univariate techniques are used to detect changes in individual sensor measurements, then the measurement process is simple, but the anomaly detection accuracy deteriorates because interactions between sensor measurements are not captured
Solution Approach 1:
The patent combines multiple sensor measurements into a unified multivariate monitoring framework. By integrating data from multiple sensors and analyzing their interactions simultaneously, the system captures the complex relationships between different measurement parameters, thereby improving anomaly detection accuracy while managing system complexity through structured integration approaches.
Solution Approach 2:
The patent transitions from univariate (single-dimension) analysis to multivariate (multi-dimension) analysis by incorporating multiple sensor measurements and their interactions. This dimensional expansion allows the system to detect anomalies that involve relationships between multiple parameters, significantly improving detection accuracy while the structured multivariate framework manages the increased complexity.
2Reliability
If multivariate techniques are used to analyze interactions between sensor measurements, then anomaly detection accuracy improves, but the device complexity increases
Solution Approach 1:
The patent segments the complex multivariate analysis into manageable components by dividing the monitoring system into modular elements that handle different aspects of multivariate analysis. This segmentation allows the system to process complex interactions between sensor measurements in a structured manner, improving fault detection reliability while keeping system complexity manageable through modular architecture.
Solution Approach 2:
The patent introduces intermediary computational layers and processing mechanisms that facilitate multivariate analysis. These intermediaries handle the complex interactions between sensor measurements, transforming raw multivariate data into meaningful fault indicators, thereby improving detection reliability while managing system complexity through layered processing architectures.
3Productivity
If traditional monitoring approaches are used, then the system is easier to operate, but maintenance responsiveness deteriorates due to delayed fault detection
Solution Approach 1:
The patent implements preliminary action by establishing comprehensive multivariate monitoring frameworks and analytical models in advance. This preparatory work enables the system to quickly detect and respond to faults when they occur, improving maintenance responsiveness. The pre-configured multivariate analysis capabilities allow for rapid fault identification without requiring complex real-time decision-making, thus maintaining ease of operation.
Data Source
AI summary
A method for advanced condition monitoring of an asset system includes sensing actual values of an operating condition for an operating regime of the asset system using at least one sensor; estimating sensed values of the operating condition by using an auto-associative neural network; determining a residual vector between the estimated sensed values and the actual values; and performing a fault diagnostic on the residual vector. In another method, an operating space of the asset system is segmented into operating regimes; the auto-associative neural network determines estimates of actual measured values; a residual vector is determined from the auto-associative neural network; a fault diagnostic is performed on the residual vector; and a change of the operation of the asset system is determined by analysis of the residual vector. An alert is provided if necessary. A smart sensor system includes an on-board processing unit for performing the method of the invention.