Process Abnormality Detection Using Rate-of-Change Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In processes where the operation state varies significantly due to preceding processes, such as in a recovery plant, it is difficult to generically define whether the operation state is normal or abnormal.
Innovation Solution
An abnormality detection method that involves reading multiple types of process data, calculating the rate of change for each type, and detecting abnormalities when a combination of these rates matches a predetermined pattern, which can be defined for each assumed cause of the abnormality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If generic abnormality detection criteria are used, then the detection method is simple and easy to implement, but it cannot accurately detect abnormalities in processes where operation states vary significantly
Solution Approach 1:
The patent applies dynamics by making the abnormality detection criteria adaptive rather than static. The system dynamically adjusts detection thresholds and patterns based on the actual operation state of the process, allowing the detection method to evolve with changing conditions. This enables accurate abnormality detection in variable processes without requiring overly complex fixed rules for every possible scenario.
Solution Approach 2:
The patent changes the parameters used for abnormality detection from fixed generic thresholds to dynamic parameters that reflect actual process conditions. By monitoring operation state parameters and adjusting detection criteria accordingly, the system achieves high detection accuracy while maintaining reasonable complexity through parameter adaptation rather than structural complexity.
2Reliability
If dynamic detection criteria are created for each process condition, then abnormality detection accuracy improves, but the system complexity increases
Solution Approach 1:
The system applies self-service by automatically adapting detection criteria based on monitored process conditions without requiring manual intervention. The abnormality detection system serves itself by dynamically adjusting its own parameters based on observed operation states, maintaining high accuracy while keeping operation simple through automated adaptation rather than manual configuration.
Solution Approach 2:
The patent implements feedback mechanisms where the system continuously monitors operation states and uses this information to adjust detection criteria in real-time. This closed-loop approach allows the system to maintain high detection accuracy while keeping operation simple, as the feedback automatically drives adaptations without requiring complex manual control or intervention.
3Measurement precision
If multiple process data types are monitored with complex pattern matching, then detection precision improves, but processing time increases
Solution Approach 1:
The patent applies segmentation by dividing the abnormality detection process into distinct stages: data collection, pattern matching, and determination. By segmenting the analysis of multiple data types and using hierarchical pattern matching, the system achieves high detection precision while reducing processing time through structured, modular analysis rather than monolithic complex processing.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
To provide a technique for performing abnormality detection in a process in which a preferred operation state changes. An abnormality detecting method is executed by one or more computers. The method includes reading a plurality of types of process data output from a plant, calculating a rate of change for each of the plurality of types of process data, and determining that an abnormality or a sign of the abnormality is detected when a combination of degrees of the rates of change related to the plurality of types of process data matches a predetermined pattern.