Combustion Process Control Using Multi-Time-Scale Process Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for controlling combustion processes in power plants, waste incineration plants, or cement plants do not effectively account for the impact of pre-processing raw materials on the combustion process or its outcomes, leading to suboptimal control and efficiency.
Innovation Solution
A method that captures input and output variables of raw material pre-processing and processing over specific periods, creating multiple process models to predict and adapt pre-processing and processing actions based on control objectives, using neural networks for self-learning and real-time control, considering short-term and long-term scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If existing control methods are used that do not account for pre-processing effects, then the control system is simpler, but the combustion process control precision and efficiency deteriorate
Solution Approach 1:
The control system is segmented into multiple independent process models: a pre-processing model that captures input variables from pre-processing steps, a processing model that captures variables during combustion, and an integration model that combines both. This segmentation allows each model to specialize in specific process stages, improving overall control precision without requiring a complete system redesign
Solution Approach 2:
The pre-processing model captures input variables from pre-processing steps (such as material preparation, heating, or treatment) before the main combustion process begins. By analyzing and predicting the effects of pre-processing actions in advance, the system can optimize subsequent combustion parameters, thereby improving control precision while keeping the overall system architecture manageable
2Measurement precision
If multiple process models are created to account for pre-processing effects, then the control accuracy improves, but the computational complexity and data processing requirements increase
Solution Approach 1:
The system merges the pre-processing model and processing model through an integration mechanism that combines their respective predictions. The pre-processing model outputs predicted effects that are fed into the processing model, creating a unified multi-stage prediction framework. This merging approach maintains high prediction accuracy across the entire process chain while avoiding the need for separate independent analysis of each stage
Solution Approach 2:
The pre-processing model acts as an intermediary between raw material inputs and the main combustion processing model. It transforms complex pre-processing effects into simplified prediction outputs that the processing model can efficiently utilize. This intermediary role reduces the computational burden on the main processing model while maintaining overall prediction accuracy
3Productivity
If real-time capture of input and output variables over multiple periods is implemented, then the process understanding and control capability improve, but the data processing time and computational load increase
Solution Approach 1:
The system implements variable capture over multiple time periods (first period for pre-processing, second period for processing) but focuses computational resources on capturing the most critical variables that have the greatest impact on combustion outcomes. By selectively monitoring key parameters rather than all possible variables continuously, the system maintains high control efficiency while reducing unnecessary data processing overhead
Data Source
AI summary
A method and apparatus for controlling a process in a system comprising pre-processing of a raw material, processing the pre-processed raw material and acquisition of the result of the processing of the pre-processed raw material, comprising the steps of: capturing input and output variables of the pre-processing; capturing output variables of the processing of the pre-processed raw material; creating a first, second and third process model for at least two different time scales, which describes the effects of adapting the pre-processing of raw material, the effects of adapting the processing of the pre-processed raw material, the effects of adapting the pre-processing of raw material and adapting the processing of pre-processed raw material on the output variables of the processing of pre-processed raw material; wherein the process in the system is controlled using the prediction of the process model which currently provides the best predictions for the process in the system.