Abstract:In the experimental process of separating coal seam gas using the CO2 displacement method, establishing a predictive model for key variables is essential to optimize displacement parameters, increase coal seam gas recovery, and improve CO2 sequestration efficiency. Traditional modeling methods often struggle with the complex nature of industrial data and are susceptible to overfitting due to multicollinearity caused by long-term datasets. This paper presents a hybrid predictive model based on variational mode decomposition (VMD), a spatiotemporal attention mechanism (STA), a bidirectional long short-term memory network (BiLSTM), and an extreme learning machine (ELM). During the offline phase, VMD is used to decompose raw data into intrinsic mode functions (IMFs). The hidden state from the last time step of the STA-BiLSTM is then added to the original data to enrich the features for ELM training. In the online prediction phase, the outputs from the VMD-STA-BiLSTM and VMD-STA-BiLSTM-ELM models are combined using an error reciprocal method to generate the final prediction. The proposed model is validated with experimental datasets from CO2 displacement for coal seam CH4 under various conditions, as well as with N2-ECBM and CO2-ECBM engineering datasets. The results show that the hybrid model surpasses VMD-ELM, STA-BiLSTM-ELM, BiLSTM, STA-BiLSTM, ELM, TCN, and Attention-TCN models in predictive accuracy. Even in multi-step and rolling predictions, the model exhibits minimal impact from cumulative errors, maintaining accurate forecasts with strong generalization and robustness. It effectively captures feature patterns across different datasets and accurately predicts unknown data. The model shows potential for application in diverse scenarios and complex environments, offering reliable support and decision-making for the field application of CO2 displacement in coal seam CH4 separation. It is an effective and promising predictive approach to enhance coal seam gas recovery and CO2 sequestration efficiency.

An optimized LSTM model for clean coal ash content prediction in dense medium separation scenarios on the basis of the dual decomposition method

Ash Accumulation Prediction in Coal-Fired Power Plant Boilers Based on GWO-VMD-BiLSTM

Predictive Combination Model for CH4 Separation and CO2 Sequestration with CO2 Injection into Coal Seams: VMD-STA-BiLSTM-ELM Hybrid Neural Network Modeling

Research on Dual GRU model driving by time series alignment for multi-step prediction of ash content of dense medium clean coal

Prediction Model of Suspension Density in the Dense Medium Separation System Based on LSTM

Integrated Estimation Model of Clean Coal Ash Content for Froth Flotation Based on Model Updating and Multiple LS-SVMs

Optimization of Soft-Sensing Model for Ash Content Prediction of Flotation Tailings by Image Features Tailings Based on GA-SVMR

Research on Prediction of Ash Content in Flotation-Recovered Clean Coal Based on NRBO-CNN-LSTM

Research on Gray Prediction of Heated Surface Combining Empirical Mode Decomposition and Long Short-term Memory Network

Time Series Prediction of Gas Emission in Coal Mining Face Based on Optimized Variational Mode Decomposition and SSA-LSTM

Detection Method for Flotation Tailings Ash Content Based on Regularized Stochastic Configuration Networks Dynamic Compensation

Prediction of Coal Ash Fusion Temperature by Least-Squares Support Vector Machine Model

Prediction of Clean Coal Ash Content in Coal Flotation through a Convergent Model Unifying Deep Learning and Likelihood Function, Incorporating Froth Velocity and Reagent Dosage Parameters

Product quality prediction in dense medium coal preparation process based on recurrent neural network

Real-time in situ monitoring of ash content in coal via dual-energy X-rays and an MLP neural network

An Intelligent prediction model for UCG state based on dual-source LSTM

Short-Term Prediction of the Intermediate Point Temperature of a Supercritical Unit Based on the EEMD–LSTM Method

A Hybrid Method for Prediction of Ash Fouling on Heat Transfer Surfaces

Dynamic Prediction of SO2 Emission Based on Hybrid Modeling Method for Coal-Fired Circulating Fluidized Bed

Short Term Coal Price Prediction Based on VMD-Informer-LSTM Model Considering Error Compensation

Real Time Prediction of Ash Content of Clean Coal Using Neural Network