CNN-LSTM Mining Parameter Optimization for Outburst Coal Seams
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for selecting mining parameters in coal mines rely heavily on experience and do not comprehensively consider multiple factors, leading to safety hazards and resource wastage, while existing numerical simulation methods are complex and not user-friendly.
Innovation Solution
A method involving the construction of an automatic numerical simulation model using a CNN-LSTM predicting model, which includes convolutional neural layers and a long-short term memory network, to optimize key mining parameters by analyzing stress, displacement, and gas field changes, and utilizing grey relational analysis and Lorenz's chaotic system for predicting coal and gas outbursts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If experience-based methods are used for selecting mining parameters, then the operation is simple, but the prediction accuracy and reliability are insufficient
Solution Approach 1:
The patent introduces an intelligent prediction system as an intermediary between experience-based methods and accurate prediction. This system uses machine learning models (CNN-LSTM), grey relational analysis, and Lorenz chaotic system as mediators to process multiple factors (stress field, displacement field, gas pressure field, fissure field) and provide optimized mining parameter recommendations, thereby maintaining operational simplicity while achieving high prediction accuracy.
2Measurement precision
If comprehensive numerical simulation is used to analyze multiple factors, then the prediction accuracy is improved, but the operation complexity increases
Solution Approach 1:
The patent segments the complex numerical simulation process into distinct modular components: stress field analysis module, displacement field analysis module, gas pressure field analysis module, and fissure field analysis module. Each module independently processes specific aspects of coal mine behavior. The system then integrates these segmented results through the intelligent prediction system to provide comprehensive analysis, thereby reducing operation complexity while maintaining prediction accuracy.
Solution Approach 2:
The intelligent prediction system acts as an intermediary that automatically integrates results from multiple complex simulation modules. By using machine learning models and chaotic system analysis as mediators, the system synthesizes data from stress, displacement, gas pressure, and fissure fields without requiring users to manually complex operations, thus reducing operational complexity while preserving comprehensive analysis capabilities.
3Adaptability or versatility
If existing numerical simulation software is used, then the analysis comprehensiveness is improved, but the ease of operation deteriorates
Solution Approach 1:
The patent creates a universal intelligent prediction system that performs multiple functions: it analyzes stress fields, displacement fields, gas pressure fields, and fissure fields simultaneously; it optimizes mining parameters; and it provides predictions for coal and gas outbursts. This multi-functional system replaces the need for users to operate multiple separate simulation software programs, thereby improving ease of operation while maintaining comprehensive analysis capabilities.
Solution Approach 2:
The system implements self-service by automatically performing comprehensive analysis of multiple fields and providing optimized mining parameter recommendations without requiring extensive user intervention. The machine learning models and chaotic system analysis automatically process simulation data and generate predictions, making the comprehensive analysis accessible to users regardless of their expertise level.
Data Source
AI summary
A method for quickly optimizing key mining parameters of an outburst coal seam as provided includes steps of constructing a graphic basic information model of the coal mine, giving coal mine characteristic information, performing mining simulation, constructing a CNN-LSTM predicating model, obtaining changes under different mining conditions, constructing a Lorenz chaotic primer, and the like. The model can be improved with continuous breakthroughs in theory, so that the model has a strong learning ability and can adapt to the constantly changing complex geological environment. The method has very good predictability for the determination of coal seam group parameters, and can efficiently select and output a set of candidate parameters.