Boiler Combustion Control Using Separate Interference And Process Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for controlling thermodynamic processes, such as combustion in boilers or furnaces, are inefficient due to the costly and time-consuming process of predicting and analyzing process variables using neural networks, which fail to accurately distinguish between short-term and long-term interference-based system changes.
Innovation Solution
The method involves creating separate interference and process models that describe the effects of system changes and setting actions on state variables, with continuous adaptation and consideration of past data using neural networks, allowing for more accurate control by distinguishing between short-term and long-term changes and predicting future effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single process model is used to describe both setting actions and interference-based system changes, then the model structure is simpler, but the accuracy of process control deteriorates
Solution Approach 1:
The patent divides the single process model into two separate models: a process model for setting actions and an interference model for interference-based system changes. This segmentation allows each model to specialize in its specific function, improving the accuracy of process control while maintaining manageable complexity through dedicated model structures.
2Measurement precision
If neural networks are continuously trained for predictions, then the prediction accuracy improves, but the time and personnel costs increase
Solution Approach 1:
The patent segments the neural network functionality into separate process and interference models, allowing independent training and optimization. This enables more efficient training strategies where each specialized model can be trained on relevant data without the overhead of training a single comprehensive model, reducing both time and computational resources required.
Solution Approach 2:
The patent performs preliminary training of the process and interference models using historical data before actual process control operations. This preliminary action ensures the models are pre-trained and ready for deployment, reducing the need for continuous training during operation and minimizing time losses during actual control processes.
3Reliability
If three cyclical steps (process analysis, training, and applying) are performed to consider process changes, then the control comprehensiveness improves, but the operational efficiency deteriorates
Solution Approach 1:
The patent merges the process analysis, training, and application steps into an integrated control system where the process model and interference model work together continuously. This integration eliminates the need for separate cyclical execution of three distinct steps, improving operational efficiency while maintaining comprehensive control through the coordinated operation of both models.
Solution Approach 2:
The patent implements continuous monitoring and control using the trained process and interference models, eliminating the need to repeatedly perform discrete training and analysis cycles. The models continuously process incoming data and generate control actions, maintaining comprehensive control while significantly improving operational efficiency through uninterrupted useful action.
Data Source
AI summary
The invention relates to an apparatus and a method for controlling 4 a process within a system, particularly a combustion process in a boiler or furnace, comprising the following steps: capturing 1 of state variables (st) of the system; creating 2 an interference model, which describes the effects of interference-based system changes (vt) on the state variables (st) of the system; creating 3 a process model, which describes the effects of setting actions (at) on the state variables (st) of the system; and controlling 4 the process within the system by performing setting actions (at) by considering the process model, the interference model and predetermined controlling goals.