Model Predictive Control With Filtered Input for Vibration Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing control systems face challenges in suppressing vibration during high-speed positioning of control objects, leading to increased completion time and deteriorated followability when steady-state deviations occur, especially when target commands change in real-time.
Innovation Solution
A control device employing model predictive control with a filter unit for attenuation and a model predictive control unit that defines correlations between extended state variables and control inputs, optimizing control inputs in real-time to follow target commands while suppressing vibration, using a prediction model that includes a filter state variable and a state variable related to the control object.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If terminal state control is used to suppress vibration during positioning, then vibration is reduced, but the control input must be determined in advance and cannot adapt to real-time target command changes
Solution Approach 1:
The invention transforms the static terminal state control into a dynamic model predictive control system that continuously recalculates control inputs based on current target commands. The prediction model and evaluation function are updated in real-time, allowing the control input to adapt dynamically to changing targets while maintaining vibration suppression through the terminal state constraints in the prediction model.
Solution Approach 2:
The invention incorporates terminal state constraints into the prediction model in advance, which preliminarily defines the desired end-state conditions. This preliminary action ensures that vibration suppression is built into the control strategy from the beginning, while the model predictive control framework allows these preliminary constraints to be satisfied even as target commands change in real-time.
2Speed
If model predictive control is used to follow target commands, then followability is improved, but steady-state deviation occurs when target constantly changes
Solution Approach 1:
The invention incorporates feedback mechanisms into the model predictive control framework by using the evaluation function to continuously assess the deviation between predicted and target states. The terminal state constraints in the prediction model provide feedback guidance to eliminate steady-state deviation, ensuring that the control object accurately follows changing target commands without sacrificing followability.
3Speed
If high-speed positioning is performed according to target command, then response speed is improved, but vibration occurs immediately before positioning and completion time increases
Solution Approach 1:
The invention uses dynamic model predictive control to continuously optimize the control input trajectory. The prediction model incorporates terminal state constraints that anticipate the positioning endpoint, allowing the system to dynamically adjust the control input to reduce vibration before positioning completion. This dynamic optimization enables high-speed response while minimizing vibration-induced delays in positioning completion.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
The present invention is a control device which includes a filter unit for performing an attenuation process at a predetermined frequency on a control input based on a predetermined target command, generates the control input through model predictive control executed by a model predictive control unit and causes an output of a predetermined control object to follow the predetermined target command. A prediction model defines a correlation between the control input and predetermined extended state variables including a state variable related to a predetermined control object and a predetermined filter state variable related to the filter unit, and a predetermined evaluation function for model predictive controls configured to calculate a state quantity cost that is a stage cost with respect to state variables except the predetermined filter state variable among the predetermined extended state variables, and a control input cost that is a stage cost related to the control input.