Multi-Dimensional Nonlinear Control for Complex Plants
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current advanced control algorithms for nonlinear processes are complex, difficult to implement, and fail to effectively handle wide-ranging nonlinearities, especially in highly nonlinear systems, due to their reliance on specific models and inability to accurately adapt to changing plant dynamics.
Innovation Solution
The development of Multi-Dimensional Nonlinear Control (MDNC) methodology, which formulates a continuous nonlinear function to recalculate the plant system matrix and control law at each sampling instant, allowing for online open-loop testing and reformulation of controllers, enabling improved control performance across various nonlinearities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model predictive control (MPC) algorithms are used to control nonlinear processes, then control performance is improved, but device complexity increases due to the large number of parameters affecting tuning
Solution Approach 1:
The patent segments the complex nonlinear control problem into multiple linear control problems by using multiple linear models to represent different operating regions of the nonlinear plant. Each linear model is valid in a specific region, and the controller switches between models based on the current operating point, thereby simplifying the overall control structure while maintaining effectiveness for highly nonlinear systems.
Solution Approach 2:
The patent implements dynamic model switching by continuously monitoring the operating point and selecting the appropriate linear model based on real-time plant conditions. This dynamic adaptation allows the controller to maintain optimal performance as the plant transitions between different operating regions, effectively handling the nonlinearities without requiring a single complex nonlinear controller.
2Reliability
If advanced predictive control schemes are used for highly nonlinear plants, then control performance is improved, but ease of operation deteriorates due to difficulty in understanding and implementing the complex formulations
Solution Approach 1:
The patent divides the complex nonlinear control task into simpler linear control tasks by creating multiple linear models for different operating regions. This segmentation makes the controller easier to understand and implement, as each linear model can be designed and tuned using standard linear control techniques, while the overall nonlinear performance is achieved through model switching.
Solution Approach 2:
The patent introduces an intermediary mechanism (the bank of linear models and switching logic) that mediates between the simple linear control techniques and the complex nonlinear plant behavior. This intermediary allows the use of straightforward linear control design methods while still achieving effective nonlinear control through the coordinated action of multiple linear models.
3Ease of operation
If conventional control schemes such as PID are used, then ease of operation is maintained, but reliability deteriorates when controlling systems with high degree of nonlinearity
Solution Approach 1:
The patent creates a universal controller that can handle both linear and nonlinear systems effectively. By using multiple linear models to represent different operating regions of a nonlinear plant, the controller achieves multi-functionality, capable of controlling highly nonlinear systems while maintaining the simplicity and ease of operation characteristic of linear control schemes.
Solution Approach 2:
The patent enhances the static PID controller by introducing dynamic model switching capability. The controller dynamically selects among multiple linear models based on the current operating point, allowing it to adapt to nonlinear plant behavior while maintaining the simple and familiar PID control structure for ease of operation.
4Reliability
If extended predictive control with unique weighting matrix is used, then control performance is improved for SISO and MIMO plants, but adaptability deteriorates for highly nonlinear plants due to inability to accurately model varying dynamics
Solution Approach 1:
The patent segments the nonlinear plant into multiple linear subsystems, each represented by a linear model valid in a specific operating region. This segmentation enables the controller to adapt to highly nonlinear plants by selecting the appropriate linear model for the current operating point, thereby achieving both good control performance and high adaptability to varying nonlinear dynamics.
Solution Approach 2:
The patent changes the controller parameters (the selected linear model) based on the operating conditions. By monitoring the plant's operating point and switching between different linear models, the controller adapts its parameters to match the current nonlinear dynamics, achieving both improved control performance and enhanced adaptability to highly nonlinear plants.
Data Source
AI summary
A computer implemented method of conducting closed-loop control of a physical system comprising the steps of carrying out an initialization of the physical system to commencing closed-loop control, evaluating the optimal constrained control move using the system error and the initial normalized matrix using a control move solver; calculating a first control action by the sum of delta u(0) and the initial control action; and implementing the result to the physical system by converting the control action to an output control signal to effect a change in at least one operating variable.