Model Predictive Control With Soft Dynamics Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current model predictive control (MPC) methods face challenges in accurately controlling machine operations due to uncertainties in model parameters, measurement inaccuracies, and model reduction errors, often leading to constraint violations or reduced performance.
Innovation Solution
The approach involves treating the machine's dynamic model as a soft constraint rather than a hard constraint, incorporating it into the cost function to balance performance optimization with parameter uncertainty, and using data assimilation methods to improve accuracy under uncertainties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the machine's dynamic model is treated as a hard constraint in MPC, then control performance is optimized, but constraint violations occur due to parameter uncertainties and measurement inaccuracies
Solution Approach 1:
The patent changes the parameter of the dynamic model constraint from hard (strict equality) to soft (inequality with tolerance bounds). This is achieved by reformulating the dynamic model constraint as an inequality constraint that allows deviations within acceptable bounds, thereby accommodating parameter uncertainties and measurement inaccuracies while maintaining control performance.
2Object-affected harmful factors
If the dynamic model is treated as a soft constraint, then constraint violations are reduced, but control accuracy decreases due to model parameter uncertainties
Solution Approach 1:
The patent incorporates feedback mechanisms through the receding horizon control structure, where the soft constraint formulation allows the controller to continuously adjust control actions based on actual system behavior. The tolerance bounds in the soft constraints are updated iteratively, enabling the system to learn from deviations and improve control accuracy over time while maintaining feasibility.
3Measurement precision
If unknown model parameters are estimated during operation, then model accuracy is improved, but constraint enforcement becomes more difficult
Solution Approach 1:
The patent applies preliminary action by pre-defining tolerance bounds for the soft constraints based on expected parameter uncertainties and measurement inaccuracies. This allows the constraint enforcement mechanism to be prepared in advance, simplifying real-time implementation while accommodating model parameter variations without increasing computational complexity during operation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A model predictive control (MPC) system for controlling an operation of a machine according to a model of the machine dynamics optimizes a cost function over a time-horizon subject to constraints to produce a sequence of control inputs to control the state of the machine over the time horizon. The machine is control using the first control input in the sequence. The cost function includes a first term defined by an objective of the MPC and a second term penalizing deviation of a state of the machine from a value satisfying an equation of dynamics of the machine.