Predictive Feedforward Control for Dedicated Variable Coordination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing model predictive control (MPC) technologies do not support the control scheme where certain outputs are controlled without compromising others, requiring a time-consuming trial and error approach for tuning the cost function weights.
Innovation Solution
A predictive feedforward control method that uses past and planned movements of certain process input variables to control selected output variables without impacting those input variables, integrated into an industrial control package.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If model predictive control coordinates all input movements to control all outputs, then all outputs can be controlled according to their importance weights, but it requires time-consuming trial and error tuning of cost function weights and cannot achieve control without trade-off
Solution Approach 1:
The patent segments the control variables into two distinct groups: primary controlled variables (those that must be controlled without compromise) and secondary controlled variables (those that should be controlled as well as possible given the input movements). This segmentation is achieved through the feedforward control structure where certain input variable movements are predetermined and used as fixed inputs to the MPC, rather than being optimized alongside other inputs. This eliminates the need for trial-and-error weight tuning while achieving the desired control hierarchy.
2Adaptability or versatility
If traditional MPC optimizes all input movements simultaneously, then overall system performance is optimized, but it cannot dedicate control to specific variables without impacting others
Solution Approach 1:
The patent introduces dynamics into the control structure by allowing the MPC to operate in different modes depending on the application requirements. The feedforward control structure enables the system to dynamically adjust which input variables are predetermined and which are optimized, providing flexibility in control strategy. This dynamic capability allows the system to switch between different control philosophies (dedicated control vs. overall optimization) without requiring multiple separate control systems, thereby achieving both adaptability and precise dedicated control.
Data Source
AI summary
Predictive feedforward control whereby past and the planned movements of certain process input variables are used in planning the control of some selected process output variables without allowing for the impact of the certain process input variables on those selected process output variables to impact planned movements of the certain process input variables. Existing model-based predictive control systems can be modified by incorporating a control package that is encoded with the predictive feedforward technique to control industrial multivariable process systems. A method includes (a) identifying a dynamic process model for the multivariable process system; and (b) implementing a dedicated feedforward control whereby within the processing unit certain process input variable signals are used by the processing unit for planning control of selected process signal output variable signals by adjustment of other process input variable signals to one or more manipulated actuators without impact to the certain process input variable signals.