ILC Fault Detection for Time-Varying Casting Roll Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing iterative learning control (ILC) algorithms face challenges in maintaining stability and performance specifications, particularly when plant models include uncertainty and iteration-varying dynamics, leading to conservative learning algorithms and potential instability if the assumed uncertainty bounds are violated.
Innovation Solution
An ILC controller with a fault detection algorithm that monitors control signal saturation and system stability, using infinity and p-norms to determine fault conditions and adjust the algorithm accordingly, ensuring asymptotic stability and performance specifications are met.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If an optimal design procedure for ILC is used to balance performance against robust stability criterion, then system performance is improved, but the algorithm becomes overly conservative and may become unstable if the assumed uncertainty bound is violated
Solution Approach 1:
The patent implements dynamic adjustment of the ILC algorithm by introducing a fault detection mechanism that monitors system behavior in real-time and adapts the learning gain based on detected fault conditions. This allows the system to transition from a static, conservative design to a dynamic one that can achieve high performance during normal operation while maintaining stability when faults are detected.
Solution Approach 2:
The patent employs feedback through a fault detection algorithm that continuously monitors system outputs and compares them against expected behavior. When deviations indicating potential instability or fault conditions are detected, the feedback mechanism adjusts the ILC algorithm parameters to prevent instability, thus resolving the contradiction between performance and reliability.
2Reliability
If a conservative ILC algorithm is used to ensure robust stability, then algorithm stability is improved, but system performance deteriorates
Solution Approach 1:
The patent applies partial action by implementing the fault detection mechanism selectively - the system operates with aggressive, high-performance ILC parameters during normal operation, and only activates conservative adjustments when fault conditions are partially or fully detected. This allows the system to achieve high performance most of the time while maintaining stability safeguards.
Solution Approach 2:
The patent changes algorithm parameters dynamically based on system state. The learning gain and other ILC parameters are adjusted according to fault detection results, allowing the system to switch between conservative and aggressive parameter settings rather than being locked into a single conservative configuration.
3Manufacturing precision
If an aggressive ILC algorithm is used to achieve high performance, then system performance is improved, but the system becomes unstable when uncertainty bounds are violated
Solution Approach 1:
The patent implements preliminary action through proactive fault detection that identifies potential instability conditions before they manifest as actual system failure. The fault detection algorithm monitors for early signs of uncertainty bound violations and triggers preventive adjustments to the ILC algorithm, preventing instability before it occurs.
Solution Approach 2:
The patent provides beforehand cushioning by preparing conservative backup parameter settings that are activated when fault conditions are detected. This cushioning mechanism acts as a safety buffer that prevents the aggressive ILC algorithm from causing instability when uncertainty bounds are violated, allowing high performance during normal operation while protecting against instability.
Data Source
AI summary
A twin roll casting system includes a pair of counter-rotating casting rolls having an adjustable nip therebetween, a casting roll controller configured to adjust the nip between the casting rolls in response to control signals; a cast strip sensor measuring a parameter of the cast strip and generating strip measurement signals; and an iterative learning control (ILC) controller receiving the strip measurement signals and providing control signals to the casting roll controller. The ILC controller includes a fault detection algorithm receiving the control signals and the strip measurement signals and generating a fault detection signal indicating when a fault condition is detected and an iterative learning control algorithm to generate the control signals. The fault detection algorithm indicates a fault condition when it detects the control signal exceeding an upper control saturation threshold or the ILC controller operating a state that is not guaranteed as stable.