MPC Model Adaptation via Fuzzy Logic Submodel Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Model Predictive Control (MPC) performance degrades over time due to process changes, making it challenging to identify and update problematic models in large-scale applications, requiring significant resource-intensive efforts and disrupting normal operations.
Innovation Solution
An automated method and apparatus for monitoring, auditing, and updating MPC models using closed-loop step testing techniques, data validation, and fuzzy logic to identify and re-identify problematic submodels, minimizing operational impact and reducing data requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated closed-loop step testing and fuzzy logic assessment are used to identify problematic submodels, then model maintenance efficiency is improved and resource consumption is reduced, but the complexity of the model assessment system increases
Solution Approach 1:
The patent segments the overall model into multiple submodels, each representing specific process relationships. The fuzzy logic assessment system evaluates each submodel independently, allowing targeted identification of problematic areas without requiring complete model re-testing, thus improving efficiency while managing system complexity through modular assessment
Solution Approach 2:
The patent introduces a fuzzy logic assessment module as an intermediary between data collection and model evaluation. This intermediary layer processes raw process data and controller output through fuzzy inference rules to generate quality scores, simplifying the overall assessment architecture while enabling efficient automated model maintenance
2Reliability
If comprehensive model re-identification is performed to maintain control performance, then model predictive quality is improved, but operational disruption and time loss increase
Solution Approach 1:
The patent extracts and isolates only the problematic submodels identified through fuzzy logic assessment, rather than performing comprehensive re-identification of the entire model. This selective approach maintains control performance by updating only the necessary model portions, significantly reducing the time and operational disruption associated with model maintenance
Solution Approach 2:
The patent performs preliminary fuzzy logic assessment of model quality using existing process data before initiating any re-identification activities. This preliminary evaluation identifies which submodels require updating, allowing planners to prepare targeted re-identification strategies in advance and minimize operational disruption
3Measurement precision
If frequent model auditing is conducted to detect performance degradation, then model quality monitoring is improved, but resource consumption and operational impact increase
Solution Approach 1:
The fuzzy logic assessment module serves multiple functions: it evaluates model quality, identifies problematic submodels, and provides prioritization for re-identification activities. This multi-functional approach enables comprehensive model quality monitoring using a single assessment system, improving detection accuracy while avoiding the need for multiple separate analysis tools that would increase resource consumption
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
Apparatuses and methods for model quality estimation and model adaptation in multivariable process control are disclosed. A method for updating a multiple input multiple output (MIMO) dynamical model of a process includes perturbing the process, auditing the controller model, identifying poor performing submodels and re-testing the relevant process variables, re-identifying submodels and adapting the model online while the process continues to operate within normal operating parameters. An apparatus comprises an online multivariable controller, a tester, a database to store data corresponding to manipulated variables and controlled variables, and a performance diagnosis module configured to identify problematic submodels and adapt a model used by the controller.