Derivative Control Guidance for Model-Based Controller Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Model-based advanced controllers face performance deterioration when control periods are below the modeling threshold, leading to increased sensitivity to identification errors and poor predictive capabilities, especially in applications requiring shorter control periods.
Innovation Solution
A guidance system that generates a desired trajectory for a process output variable and computes a set point for its derivative variable, which is applied to a model-based advanced controller, allowing for improved control performance even at shorter control periods by guiding the process output variable towards the desired trajectory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the control period is reduced below the modeling threshold to improve response speed, then the control frequency increases, but the predictive model parameters become overly sensitive to identification errors causing performance deterioration
Solution Approach 1:
The patent introduces a derivative variable as an intermediary between the control system and the process output variable. Instead of directly controlling the process output variable y(t), the system controls its derivative dy(t)/dt. This intermediary approach allows the use of shorter control periods without directly exposing the predictive model to high-frequency sensitivity issues, as the derivative control smooths the control actions and reduces the impact of parameter identification errors.
Solution Approach 2:
The patent changes the control parameter from the process output variable itself to its derivative. By controlling dy(t)/dt instead of y(t) directly, the system transforms the control problem into a form that is less sensitive to rapid parameter variations. This parameter transformation allows shorter control periods to be used while maintaining model reliability, as the derivative control inherently filters high-frequency noise and parameter sensitivity.
2Loss of time
If the control period is reduced below the modeling threshold, then the control response becomes faster, but the sensitivity to identification errors increases sharply
Solution Approach 1:
The derivative variable serves as a mediator that decouples the direct relationship between control period and parameter sensitivity. By introducing dy(t)/dt as the controlled variable, the system can use shorter control periods without the exponential increase in sensitivity to identification errors that would occur with direct control of y(t). The derivative operation inherently provides smoothing that reduces the impact of identification errors.
Solution Approach 2:
The patent applies derivative control as a form of beforehand cushioning against the harmful effects of parameter sensitivity. By pre-processing the control signal through differentiation, the system cushions against the amplification of identification errors that would otherwise occur at high control frequencies. This prior processing prevents the sensitivity issue from manifesting in the final control performance.
3Manufacturing precision
If model-based advanced controllers are used to achieve precise control, then the control accuracy improves, but the performance deteriorates when control periods are below the modeling threshold
Solution Approach 1:
The patent introduces derivative control as an intermediary layer between the model-based advanced controller and the process. This intermediary approach allows the controller to maintain high accuracy through model-based prediction while avoiding the performance deterioration that occurs at short control periods. The derivative variable acts as a buffer that preserves the benefits of model-based control without exposing the system to high-frequency sensitivity issues.
Solution Approach 2:
The patent makes the control approach dynamic by switching from direct control of y(t) to control of dy(t)/dt. This dynamic change in control strategy allows the system to adapt to short control periods while maintaining reliability. The derivative control provides a dynamic filtering effect that maintains controller performance across a wider range of control periods, including those below the traditional modeling threshold.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Guidance method and system applied to the control of monovariable or multivariable process variables, with parameters that are known or unknown and variable with time. The guidance system of the present invention generates a desired trajectory for a process output variable and, on the basis of said desired trajectory, calculates a reference of the derived variable of said process output variable. Said reference is then applied to a model based advance controller based on a model of said derived and the control action generated by said model-based advance controller is applied to the process and guides the evolution thereof such that said process output variable converges towards said desired trajectory