Multivariable MPC Linearization for Multistep Plant Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Model predictive control in processing systems, such as chemical plants, often relies on assumptions and is computationally intensive, leading to poor control due to faulty system identification and inefficiencies in handling nonlinear models.
Innovation Solution
A multivariable model predictive control system that linearizes multistep plant processes, allowing for independent control of cycle steps based on output signals to optimize plant performance, using a model predictive controller (MPC) that generates a linear model of the plant and applies control signals to optimize outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model predictive control uses assumed models obtained through system identification, then control capability is provided, but control accuracy deteriorates due to faulty system identification and assumptions
Solution Approach 1:
The patent transforms the nonlinear process model into a linearized representation by changing the mathematical parameters and form of the model. This linearized model uses simplified parameters that can be accurately identified from data, avoiding the accuracy issues of complex nonlinear models while maintaining control capability.
Solution Approach 2:
The patent replaces the traditional system identification approach with a data-driven linearization method. Instead of relying on complex mechanical/physical models that require difficult identification, the system uses direct linearization of process data to create an accurate, simple model for control.
2Manufacturing precision
If model predictive control uses nonlinear models, then modeling accuracy is improved, but computational complexity increases making it impractical
Solution Approach 1:
The patent changes the mathematical parameters of the model from nonlinear to linear form. This parameter transformation maintains the essential process dynamics while dramatically reducing computational complexity, making the model predictive control practical for real-time implementation.
Solution Approach 2:
The patent extracts the essential linear behavior from the nonlinear process by applying linearization techniques. This extraction removes the computationally intensive nonlinear components while retaining the core process characteristics needed for effective control.
3Adaptability or versatility
If model predictive control uses complex assumed models, then control coverage is improved, but computational time increases
Solution Approach 1:
The patent changes the model parameters from complex nonlinear forms to simple linear forms, enabling fast computation while maintaining broad control coverage through the multivariable nature of the linearized model.
Solution Approach 2:
The patent segments the complex control problem into independent linearized process models for each controlled variable. This segmentation allows parallel computation and reduces overall computational time while maintaining comprehensive control coverage through the multivariable framework.
Data Source
AI summary
Systems and methods presented herein provide for multivariable model predictive control of a multistep plant. In one embodiment, a model predictive controller (MPC) includes a model of the multistep plant. The MPC is operable to linearize at least two steps of the multistep plant into cycle steps based on the model, to process an output signal from the multistep plant, and to independently control the cycle steps based on the output signal to optimize an output of the multistep plant.