Time Series Data Correlation for Predictive Maintenance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for predicting events in industrial systems, such as equipment failures, lack the capability to effectively identify new patterns in time series data and associate them with other types of data, like alarm and event data, to quantify interrelationships and predict future events.
Innovation Solution
A system and method that processes time series data to identify sequences of interest, extracts these sequences as events, and calculates a confidence level to quantify the relationship between the time series data and known events, using pattern matching and correlation approaches to predict future events by associating time series data with alarm and event data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If time series data is analyzed to detect equipment behavior changes, then anomaly detection capability is improved, but the ability to predict future events and quantify interrelationships with other data types remains insufficient
Solution Approach 1:
The patent combines time series data analysis with discrete event data and alarm data into a unified predictive system. The system merges multiple data types (continuous sensor data, discrete events, alarms) to comprehensively predict future equipment states and quantify interrelationships, resolving the limitation of analyzing time series data in isolation.
Solution Approach 2:
The patent introduces pattern recognition algorithms and correlation analysis as intermediary mechanisms that bridge time series data with discrete event data. These intermediaries enable the system to detect patterns in time series data, match them with known event patterns, and quantify relationships between different data types, thereby preventing information loss.
2Difficulty of detecting and measuring
If traditional anomaly detection methods are used, then equipment behavior changes can be detected, but new patterns in time series data cannot be effectively identified or associated with other data types
Solution Approach 1:
The patent employs dynamic pattern recognition that adapts to new equipment behavior patterns over time. The system continuously learns from incoming data, updating its understanding of normal and abnormal patterns, which enables it to identify new patterns and associate them with relevant events and alarms, thereby improving adaptability while maintaining detection capability.
Solution Approach 2:
The patent pre-establishes a framework for pattern association that links time series patterns with discrete events and alarms before actual prediction is needed. By pre-defining relationship structures and correlation methods, the system can quickly identify and associate new patterns with relevant data types when they occur, enhancing versatility without sacrificing detection accuracy.
3Productivity
If only time series data is monitored, then real-time equipment state can be tracked, but the system cannot predict future events or estimate remaining useful life
Solution Approach 1:
The patent performs preliminary pattern recognition and event association during real-time monitoring, preparing predictive models in advance. By continuously analyzing time series data against known patterns and associating them with discrete events, the system builds predictive capabilities on-the-fly, enabling both real-time tracking and future event prediction without sacrificing either function.
Solution Approach 2:
The patent implements feedback loops where predicted events and actual equipment states are continuously compared. This feedback mechanism refines the predictive models by learning from discrepancies between predictions and actual outcomes, thereby improving reliability of predictions while maintaining real-time monitoring efficiency through iterative optimization.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A system for predicting events by associating time series data with other types of non-time series data can include a processor configured to receive a data stream including time series data transmitted from a sensor configured to measure an operating parameter of a component being monitored. The processor identifies sequences of interest in the time series data having predictive value. The processor compares the real-time data stream to a set of pre-existing event data that act as effective leading indicators of different alarms and events. The processor extracts any identified sequences of interest from the time series data as an extracted event. The processor quantifies the relationship between the data of the extracted event and the known historical pattern by calculating a confidence level to denote a probability of occurrence of the event by comparing how closely the new time series data matches the data patterns associated with known events.