Polynomial Filter for Valve Stiction Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In offshore oil and gas platforms, real-time control systems face challenges in detecting equipment deterioration and malfunction conditions, such as valve stiction, leading to poor control performance and potential system failures due to inadequate monitoring and tuning.
Innovation Solution
A method involving data-driven approaches, including polynomial filtering and pattern recognition algorithms, is used to monitor control systems and detect valve stiction by smoothing process data and applying a stiction index calculation, enabling early detection and prevention of equipment failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional control monitoring methods are used, then system simplicity is maintained, but detection precision of equipment deterioration and valve stiction is insufficient
Solution Approach 1:
A polynomial filter is introduced as an intermediary component between raw process data and the detection algorithm. The filter smooths the data while preserving important features, enabling more accurate detection of valve stiction and equipment deterioration without requiring complex measurement systems. This intermediary processing layer resolves the contradiction by enhancing detection precision through mathematical transformation rather than hardware complexity.
Solution Approach 2:
The invention transforms the raw process data parameters through polynomial filtering, changing the mathematical representation of the data to reveal underlying patterns. By applying a polynomial fit to the smoothed data and analyzing the residuals, the system detects equipment issues with higher precision. This parameter transformation approach allows accurate detection without increasing physical system complexity.
2Measurement precision
If polynomial filtering and pattern recognition algorithms are applied, then detection precision is improved, but computational complexity increases
Solution Approach 1:
The polynomial filter applies a predetermined order of filtering (e.g., 3rd or 5th order) to the process data, which is sufficient to smooth noise while preserving critical features. This partial application of filtering—using a specific order rather than excessive smoothing—achieves the necessary detection precision without unnecessary computational overhead. The pattern recognition algorithm then applies targeted analysis only where needed, avoiding exhaustive computation across all data points.
Solution Approach 2:
The invention extracts only the essential features from the process data by applying polynomial filtering and analyzing the residuals. Instead of processing the entire raw data set with complex algorithms, the system extracts the relevant information needed for detection (the deviation from the polynomial fit), significantly reducing computational complexity while maintaining high detection accuracy for valve stiction and equipment deterioration.
3Reliability
If real-time monitoring is implemented, then reliability is improved, but loss of time for data processing increases
Solution Approach 1:
The polynomial filter is applied continuously to the process data in real-time, maintaining a running smoothed data set that is ready for immediate analysis. This preliminary filtering action ensures that when equipment deterioration or valve stiction occurs, the system can detect it immediately using the pre-computed smoothed data and polynomial residuals, without requiring time-consuming post-processing. This enables real-time monitoring with minimal processing delay.
Data Source
AI summary
A method for monitoring a control of a parameter of one or more devices or systems in an oil or gas production site includes receiving process data, the process data being a result of the control of the parameter of the one or more devices or systems in the production site; smoothing the process data using a polynomial filter while preserving features of the process data to obtain smoothed data; and applying a pattern recognition algorithm to the smoothed data to determine whether there is a malfunction condition in the one or more devices or systems.