Boiler Combustion Model Using Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Boilers in coal-fired power plants face inefficiencies in combustion, leading to increased pollutant generation and higher operational costs due to incomplete combustion and excessive nitrogen oxide emissions, necessitating a method to optimize combustion processes.
Innovation Solution
A method and apparatus for deriving a boiler combustion model using artificial neural networks (ANN) that sets input and output data, including flow rates, pressures, temperatures, and fuel composition, to optimize combustion efficiency and reduce exhaust gas production, by learning from computational fluid dynamics (CFD) and actual grid data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional combustion methods are used in coal-fired boilers, then power generation is achieved, but combustion efficiency is reduced and pollutant generation increases
Solution Approach 1:
The patent replaces traditional mechanical combustion control systems with an artificial neural network-based intelligent control system. The ANN model learns optimal combustion parameters from CFD simulation data and actual grid data, substituting conventional mechanical control mechanisms with data-driven intelligent decision-making to improve combustion efficiency and reduce pollutants
Solution Approach 2:
The patent changes combustion parameters (air-fuel ratio, temperature distribution, flow rates) based on ANN predictions. By dynamically adjusting these parameters according to learned patterns from training data, the system optimizes combustion efficiency while minimizing harmful emissions of nitrogen oxide and other pollutants
2Loss of energy
If incomplete combustion occurs to reduce fuel consumption, then operational costs decrease, but combustion efficiency is reduced and pollutant generation increases
Solution Approach 1:
The patent implements a feedback mechanism where the ANN model continuously receives input data from the combustion system, processes it through learned patterns, and outputs optimized control parameters. This closed-loop feedback system ensures complete combustion by constantly adjusting parameters based on actual system state, preventing both fuel waste and inefficient combustion
3Productivity
If excessive air is supplied to improve combustion completeness, then combustion efficiency increases, but nitrogen oxide emissions increase
Solution Approach 1:
The patent applies local quality optimization by controlling air distribution in different zones of the combustion chamber. The ANN model determines optimal air supply to specific regions rather than uniform distribution, ensuring complete combustion in fuel-rich zones while limiting excess air in other areas, thereby reducing nitrogen oxide formation while maintaining combustion completeness
Data Source
AI summary
A method of and an apparatus for deriving a boiler combustion model by setting input and output data and using a combination of mathematical models including an artificial neural network (ANN) are provided. The method of deriving a boiler combustion model may include setting input and output data required for derivation of a boiler combustion model, deriving a boiler combustion model including an input layer configured to receive the input data, a hidden layer including multiple nodes configured to perform computation on the input data, and an output layer configured to output a computation result transferred from the hidden layer as the output data, and determining a precision of the derived boiler combustion model.