Process Dynamics Change Detection Using Response Time Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial process control systems face challenges in detecting changes in dynamic behavior due to noisy data and multiple parameters, making it difficult to determine when process behavior has shifted, which can lead to poor quality production if not addressed promptly.
Innovation Solution
The system maps process parameters to a lower dimensional space by combining parameters like time constant and time delay into a response time parameter or examining the frequency response, reducing the parameter space to two more stable parameters that are easier to monitor for changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple process parameters are monitored individually, then comprehensive process information is obtained, but noise and variability increase making change detection difficult
Solution Approach 1:
The patent combines multiple process parameters (time constant, time delay, and other dynamic characteristics) into a single composite parameter called 'response time'. This merging reduces the number of parameters from multiple individual measurements to one aggregated metric, thereby reducing noise and variability while maintaining comprehensive process information. The response time parameter integrates the effects of multiple underlying parameters, making change detection more reliable.
Solution Approach 2:
The patent segments the complex process dynamics into distinct components (time constant, time delay, gain) that can be individually analyzed and then recombined. By breaking down the overall process behavior into these fundamental elements, the system can identify which specific aspect is changing and combine them into a unified response time metric that highlights significant changes while filtering out noise.
2Reliability
If controller tuning parameters are adjusted to match process dynamic behavior, then control performance improves, but when process behavior changes, re-tuning is required which degrades productivity
Solution Approach 1:
The patent implements continuous monitoring of the response time parameter and compares it against reference values or historical data. When significant changes in response time are detected, the system generates alerts or automatically triggers re-tuning procedures. This feedback mechanism ensures that control performance is maintained by detecting process changes early, allowing for timely intervention before quality degradation occurs, thus balancing reliability with productivity.
Solution Approach 2:
The patent performs preliminary detection of process changes by continuously monitoring the response time parameter before actual quality degradation occurs. By detecting changes in dynamic behavior early, the system allows for proactive re-tuning or process adjustment, preventing the need for reactive measures that would cause production interruptions. This preliminary action maintains both control performance and productivity.
3Reliability
If frequent process monitoring is performed to detect behavior changes, then quality maintenance is improved, but unnecessary experiments and interventions increase
Solution Approach 1:
The patent transforms multiple process parameters into a single response time parameter that is more sensitive to meaningful changes and less sensitive to normal process variability. By monitoring this transformed parameter with appropriate thresholds and statistical methods, the system can distinguish between significant process changes requiring intervention and normal fluctuations that do not require action. This reduces unnecessary interventions while maintaining quality consistency.
Data Source
AI summary
A method includes acquiring process data collected in an industrial process control and automation system. The method also includes reducing a dimension space of the process data by combining two or more parameters of the process data or examining a frequency response of the process data. The method further includes determining a change in a process based on a change in the process data in the reduced dimension space. The method also includes outputting a result based on the determined change in the process.