Operating Behavior Classification Interface for Machine Health Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Complex systems with multiple machines face inefficiencies due to degradation, failure, and sub-optimal use of components, which are difficult to detect and manage, especially in large and varied environments, leading to uncertainty and reliance on human expertise that is often scarce and inaccurate.
Innovation Solution
A computer-implemented method and system for generating and displaying interfaces that classify and predict machine operating behavior by storing sequences of values, filtering relevant data, and using graphical user interfaces and APIs to provide labeled information for maintenance engineers, enabling visualization and decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If sensors and customized triggers are used to monitor component operation, then detection capability is improved, but measurement precision deteriorates due to sensor degradation and uncertainty
Solution Approach 1:
The system implements feedback by continuously monitoring sensor measurements and comparing them against learned normal operating patterns. When deviations are detected, the system adjusts maintenance schedules and alerts operators, creating a closed-loop system that improves detection accuracy over time while compensating for sensor degradation through adaptive thresholding and pattern recognition.
Solution Approach 2:
The system performs self-service by automatically learning normal operating patterns from historical data and adapting to changing conditions without requiring manual recalibration. The machine learning models continuously refine their understanding of component behavior, enabling the system to maintain measurement precision despite sensor degradation or environmental changes.
2Reliability
If maintenance frequency is increased based on manual estimates, then reliability is improved, but productivity deteriorates due to unnecessary maintenance operations
Solution Approach 1:
The system transitions from static, schedule-based maintenance to dynamic, condition-based maintenance. Maintenance frequency and timing are continuously adjusted based on real-time sensor data and learned patterns of component degradation. This allows the system to perform maintenance only when actually needed, improving reliability while minimizing disruptions to productivity.
Solution Approach 2:
The system changes the parameter of maintenance timing from fixed calendar intervals to variable intervals based on actual component condition. By monitoring parameters such as vibration, temperature, and performance metrics, the system determines optimal maintenance moments, preventing both premature maintenance and failure-related downtime.
3Adaptability or versatility
If manual monitoring and judgment by maintenance engineers is used, then adaptability is improved, but loss of information increases due to scarcity and variability of human expertise
Solution Approach 1:
The system creates digital copies of maintenance engineer expertise through machine learning models trained on historical data and expert judgments. These models capture and reproduce the decision-making patterns and knowledge of experienced engineers, making this expertise widely available and consistent across the organization while preserving critical information that would otherwise be lost.
Solution Approach 2:
The system implements a universal monitoring platform that can analyze data from multiple sensor types and apply learned patterns across different machine types and operating conditions. This multi-functional system consolidates specialized knowledge into a generalizable framework that adapts to various scenarios, reducing information loss while maintaining adaptability.
Data Source
AI summary
A computer-implemented method of obtaining a label for graphically presented operating behavior is disclosed. The method comprises receiving a set of sequences of values that describe operating behavior of one or more machines that were observed over one or more windows of time, at least one sequence of values of the set of sequences of values corresponds to a time series of varying values; causing display of one or more graphical elements that change in appearance over time according to the at least one sequence of values, at least one of the one or more graphical elements representing a characteristic of a part of the one or more machines; detecting an upcoming occurrence of an event based on the one sequence of values, the event corresponding to a certain sequence of values of the set of sequences of values; causing, in response to detecting the upcoming occurrence of the event and prior to an occurrence of the event, display of the at least one graphical elements that change in appearance in a specific manner; receiving, after display of the certain sequence of values, a label for the event, storing, based at least in part on the input, the label in association with the certain sequence of values.