Process Equipment State Monitoring Using Correlated Reference Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In process industries, it is challenging to determine the health condition of equipment online due to complex and sensitive correlations between numerous variables, leading to difficulties in monitoring equipment states and predicting faults in real-time.
Innovation Solution
A method and apparatus for monitoring the state of a device in a process industry, which involves collecting multi-dimensional state data, performing correlation analysis using algorithms like Pearson or Kendall, establishing a device state reference model using the Dirichlet process algorithm, and determining real-time device states to identify normal or abnormal conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If planned routine maintenance is performed based on scheduled time, then maintenance can be performed regularly, but it causes excess maintenance or lack of maintenance due to inability to predict actual fault timing
Solution Approach 1:
The system performs preliminary analysis of historical operation data to establish equipment state reference models before actual faults occur. By pre-processing data and identifying normal state patterns, the system enables early detection of deviations from normal operation, allowing maintenance to be scheduled based on actual equipment condition rather than fixed time intervals.
Solution Approach 2:
The system continuously monitors equipment state dimensions in real-time and compares them against the established reference models. When deviations indicate potential faults, the system provides feedback alerts that trigger maintenance actions. This closed-loop feedback mechanism replaces open-loop scheduled maintenance with condition-based maintenance timing.
2Reliability
If online monitoring of equipment state is implemented, then unplanned stops can be reduced by early fault detection, but it requires accurate definition of boundary between normal and abnormal states which is difficult due to complex variable correlations
Solution Approach 1:
The system segments the complex equipment state into multiple independent state dimensions (vibration, temperature, pressure, etc.). Each dimension is monitored and analyzed separately against its own reference model, rather than attempting to define a single complex boundary for the entire equipment state. This segmentation simplifies the boundary definition problem while maintaining comprehensive monitoring capability.
Solution Approach 2:
The system transforms the abstract concept of equipment health into concrete measurable parameters and state dimensions. By changing from a holistic state assessment to specific parameter monitoring with statistically-defined boundaries (reference models), the system makes the boundary between normal and abnormal states objectively definable and measurable.
3Productivity
If equipment runs until scheduled maintenance time, then production continuity is maintained, but potential faults may go undetected between maintenance intervals
Solution Approach 1:
The system enables continuous monitoring of equipment state dimensions throughout operation, rather than periodic inspections. This continuous observation maintains production continuity while simultaneously providing ongoing fault detection capability, eliminating the trade-off between running until maintenance and detecting potential faults.
Solution Approach 2:
The system introduces state reference models as intermediaries between raw equipment data and fault detection decisions. These models continuously compare actual state dimensions against normal operation patterns, enabling real-time fault detection without interrupting production. The intermediary model translates continuous data streams into actionable maintenance insights.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method and an apparatus for monitoring the state of a device in process industries and a medium. The method comprises: collecting and saving data of the multi-dimensional state of a device in the process industry; performing correlation analysis on the collected historical data of the multi-dimensional state of the device to obtain state dimensions having correlation; acquiring, according to the obtained state dimensions having correlation, historical data of the dimensional states having correlation, and modeling the acquired historical data of the dimensional states to obtain a device state reference model; acquiring real-time data of the dimensional states having correlation corresponding to the device state reference model in real time, and determining whether the current state of the device is normal according to a preset device state determination condition and the acquired real-time data of the dimensional states having correlation, and in combination with the device state reference model. The invention realizes online monitoring of the state of a device in the process industry.