Adaptive MPC Model Switching for Changing Process States
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial process control systems face challenges in maintaining accurate model predictive control due to changing process characteristics over time, leading to performance degradation when using outdated process models.
Innovation Solution
Implementing a model predictive control (MPC) device with multiple MPC models that can switch between different operating states based on current process parameters, using gain scheduling and online model estimation to maintain model accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single MPC process model is used for process control, then the device complexity is reduced, but the control performance degrades when process characteristics change over time
Solution Approach 1:
The patent divides the single MPC model into multiple segmented models, each representing different operating states or regions of the process. The system switches between these segmented models based on current process conditions, allowing accurate control across varying operating ranges while maintaining manageable complexity through modular model structures.
Solution Approach 2:
The patent implements dynamic model selection where the MPC model is not fixed but changes adaptively based on real-time process state parameters. The system dynamically identifies the current operating state and switches to the appropriate pre-trained model, enabling the control system to adapt to changing process characteristics without requiring a completely new model.
2Reliability
If multiple MPC process models are implemented to handle different operating states, then the control performance is maintained under varying conditions, but the device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training multiple MPC models offline for different operating states before deployment. During real-time control, the system only needs to identify the current state and switch to the corresponding pre-prepared model, avoiding the complexity of real-time model training while maintaining adaptive performance. The computationally intensive model development is performed in advance.
Solution Approach 2:
The patent manages complexity by parameterizing the model selection process based on key state parameters. Instead of managing completely independent complex models, the system uses parameter-based switching where models are distinguished by their applicable operating range parameters, simplifying the overall system architecture while maintaining the benefits of multiple specialized models.
3Adaptability or versatility
If the MPC model is updated frequently to track process changes, then the adaptability to process variations is improved, but the loss of time for model regeneration increases
Solution Approach 1:
The patent performs model training and adaptation offline in advance for various operating conditions. During real-time operation, the system switches between these pre-adapted models rather than regenerating models continuously, eliminating real-time model regeneration delays while maintaining adaptability through model switching based on current operating state identification.
Data Source
AI summary
A model predictive control (MPC) device includes an input interface configured to receive an industrial process input associated with at least one component of a process automation plant, an output interface configured to transmit a control instruction to control the component, memory configured to store first and second MPC process models corresponding to different states, and a processor configured to identify a current state parameter of an industrial process, and predict a future industrial process output using the first or second MPC process model, based on the current state parameter being associated with the first or second MPC process model. The processor is configured to calculate a target operating point according to the predicted future industrial process output, determine a control signal to drive the industrial process to the calculated target operating point, and output the determined control signal to control operation of the component of the industrial process plant.