Boiler Combustion Control Using Separate Disturbance and Process Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for controlling combustion and grinding processes are time-consuming and labor-intensive due to the need for continuous training of neural networks or use of multiple networks for prediction, and they lack accuracy in distinguishing between control actions and disturbance effects on system variables.
Innovation Solution
The method involves creating separate models for disturbance and process effects on system variables, distinguishing between short-term and longer-term changes, and using computer-aided neural networks to adapt these models continuously, ensuring the best representation of current conditions is used for control actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single process model is used to describe both control actions and disturbance effects, then the model structure is simple, but the accuracy of distinguishing between control and disturbance effects deteriorates
Solution Approach 1:
The patent divides the single process model into two separate independent models: a process model that describes the effects of control actions on state variables, and a disturbance model that describes the effects of disturbance-related system changes on state variables. This segmentation allows each model to specialize in its specific function, improving the accuracy of distinguishing between control and disturbance effects while maintaining manageable complexity through modular architecture.
2Measurement precision
If neural networks are continuously trained or multiple networks are used for predictions, then the prediction accuracy is improved, but the time consumption and labor intensity increase
Solution Approach 1:
The patent segments the neural network functionality into two independent networks: one dedicated to the process model and another to the disturbance model. Each network is trained separately on its specific data type, which improves prediction accuracy for each function while reducing the overall computational burden compared to training a single large network on all data, thereby reducing time consumption and labor intensity.
3Measurement precision
If separate disturbance models and process models are created, then the accuracy of process control is improved, but the device complexity increases
Solution Approach 1:
The patent implements segmentation by creating separate disturbance models and process models that operate independently but are integrated in the control system. This segmentation improves control accuracy by allowing each model to specialize in its specific function while the modular architecture enables manageable complexity through independent training and validation of each model component.
Solution Approach 2:
The control system is designed with multi-functionality to handle both process model predictions and disturbance model predictions through a unified control architecture. This universal framework integrates multiple independent models without proportionally increasing complexity, as the same neural network infrastructure and control logic serve both modeling functions.
Data Source
Figure 1
AI summary
The invention relates to a device and a method for controlling a process within a system, in particular a combustion process in a boiler or furnace, comprising the following steps: - Acquiring state variables (st) of the system; - Creating a disturbance model (SM) that describes the effects of disturbance-induced system changes (vt) on the state variables (st) of the system; - Creating a process model (PM) that describes the effects of control actions (at) on the state variables (st) of the system; and - Controlling the process within the system by executing control actions (at) taking into account the process model (PM), the disturbance model (SM), and predetermined control objectives.