Faulty Variable Identification for Process Plant Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multivariate statistics-based fault detection systems in process plants often fail to accurately identify faulty variables due to the propagation of estimation errors, leading to incorrect identification of normal variables as faulty, especially in complex systems with numerous interacting process variables.
Innovation Solution
The implementation of an improved faulty variable identification technique that constructs miniature system matrices for each pair of process variables and applies a modified binary search algorithm to divided system matrices, iteratively determining faulty variables based on estimation errors and reducing the number of processing resources consumed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional multivariate statistics-based fault detection systems are used to monitor process variables, then fault detection capability is provided, but estimation errors propagate causing incorrect identification of normal variables as faulty
Solution Approach 1:
The system segments the fault detection process into two distinct stages: (1) a multivariate statistics-based monitoring stage that detects abnormal conditions using control charts, and (2) a dedicated faulty variable identification stage that uses binary search algorithms to locate the specific faulty variable. This segmentation prevents estimation errors from the monitoring stage from directly causing misidentification, as the identification stage independently verifies which variable is actually faulty.
Solution Approach 2:
The patent introduces an intermediary binary search-based identification system that acts as a mediator between the multivariate monitoring system and the final fault diagnosis. This intermediary layer processes the alarm signals and systematically identifies the faulty variable by comparing process behavior against expected relationships, preventing direct propagation of estimation errors to the identification outcome.
2Measurement precision
If system matrices are constructed for all pairs of process variables to identify faulty variables, then identification accuracy improves, but processing resources and computational complexity increase significantly
Solution Approach 1:
The patent segments the set of all possible variable pairs into smaller subsets using a binary search approach. Instead of constructing system matrices for all n*(n-1)/2 pairs simultaneously, the algorithm divides variables into groups and systematically processes subsets, reducing memory requirements and computational complexity from O(n²) to O(n log n) while maintaining identification accuracy.
Solution Approach 2:
The system performs partial action by constructing system matrices only for relevant variable pairs identified through the binary search process, rather than pre-computing all possible pairs. This selective approach reduces unnecessary computational effort while ensuring that the specific faulty variable is accurately identified through targeted matrix construction.
3Reliability
If comprehensive fault detection monitoring is implemented across all process variables, then system safety improves, but operator confusion increases due to false alarms from incorrect faulty variable identification
Solution Approach 1:
The patent segments the alarm information presented to operators into two distinct components: (1) a clear indication that an abnormal condition exists (from the multivariate monitoring system), and (2) a separately derived, accurate identification of the specific faulty variable (from the binary search identification system). This segmentation eliminates confusion by ensuring operators receive precise fault location information rather than potentially misleading estimates from the monitoring system alone.
Data Source
AI summary
A real-time control system includes a faulty variable identification technique to implement a data-driven fault detection function that provides an operator with information that enables a higher level of situational awareness of the current and likely future operating conditions of the process plant. The faulty variable identification technique enables an operator to recognize when a process plant component is behaving abnormally to potentially take action, in a current time step, to alleviate the underlying cause of the problem, thus reducing the likelihood of or preventing a stall of the process control system or a failure of the process plant component.