Coal Mill Neural Network Prediction for Real-Time Coal Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Thermal coal power plants face challenges in safely operating coal mill equipment and managing boiler combustion due to uncertainty in coal quality changes, leading to issues like deflagration, over-temperature, and boiler coking, with existing coal quality analyzers being costly, inaccurate, and posing safety risks.
Innovation Solution
A neural network-based prediction method that analyzes coal mill data to accurately determine coal quality in real-time, using a multi-layer perceptron structure with specific activation functions, integrating with a distributed control system to adjust operating conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional online coal quality analyzers are installed on each coal feeder, then real-time coal quality detection is achieved, but the cost becomes prohibitively high (more than 1 million USD for six units)
Solution Approach 1:
The patent creates a virtual copy of coal quality detection by training a neural network model on historical data from conventional analyzers. This software-based copy replicates the detection capability without requiring physical hardware on each feeder, dramatically reducing cost while maintaining measurement precision
Solution Approach 2:
The patent develops a universal prediction model that can analyze coal quality data from multiple coal feeders using a single system. This multi-functional approach allows one analyzer to serve multiple purposes and locations, eliminating the need for dedicated expensive equipment at each feeder point
2Measurement precision
If conventional online coal quality analyzers are used, then coal quality detection is provided, but the measurement accuracy is low due to few data points and environmental factors
Solution Approach 1:
The patent performs preliminary data collection and cleaning before model training by gathering historical data from multiple sources including conventional analyzers, DCS systems, and fuel management systems. This pre-processing ensures high-quality input data that improves measurement accuracy and reliability
Solution Approach 2:
The patent implements a feedback mechanism where the neural network model continuously learns from new data points and adjusts its predictions. The model compares predicted values with actual measurements and refines its parameters over time, improving accuracy and stability through iterative optimization
3Measurement precision
If conventional online coal quality analyzers with neutron radioactive sources are installed, then coal quality detection is achieved, but radioactive safety risks are introduced
Solution Approach 1:
The patent replaces the mechanical/physical neutron-based detection system with a computational/software-based neural network model. This substitution eliminates the need for radioactive sources while maintaining coal quality detection capabilities through data-driven predictions
Solution Approach 2:
The patent introduces an intermediary computational layer that processes data from existing sensors and systems to infer coal quality. This intermediary approach allows accurate detection without direct contact with radioactive materials, using software as a mediator between available data and quality assessment
4Measurement precision
If conventional online coal quality analyzers are installed, then coal quality analysis is provided, but the devices are impractically bulky and slow to generate results
Solution Approach 1:
The patent extracts the essential detection function from bulky physical hardware and isolates it as a software model. By taking out only the necessary analytical capability and implementing it computationally, the system achieves rapid results generation without the physical constraints of conventional analyzers
Solution Approach 2:
The patent implements periodic data collection and model updating cycles that enable rapid, continuous coal quality assessment. The system processes data in efficient batches and generates results quickly, improving response speed while maintaining analysis capability
Data Source
AI summary
A computer-implemented method for predicting coal quality of coal mill based on neural network is provided. The method includes: establishing a prediction model based on a neural network structure; importing cleaned coal mill data for model training; connecting the trained prediction model with a distributed control system (DCS) in real time; calculate online the running coal type of the coal mill; and determining the real-time running coal quality of the coal mill according to the calculation results.