Rapid Prediction of Pulverized Coal Particle Ignition Behavior Under Complex Atmospheric Conditions Based on BP Neural Networks

Objective The ignition characteristics of pulverized coal under complex atmospheres are a key focus of clean combustion research.

Method In this study, a one-dimensional transient single-particle ignition model was employed to systematically calculate the ignition delay characteristics and ignition mode criterion parameters of pulverized coal particles. By integrating the model calculation results with experimental observation data, a backpropagation (BP) neural network prediction model was constructed to achieve rapid estimation of key ignition indicators of pulverized coal.

Result The results indicate that, under an appropriate topology and hyperparameter configuration, the neural network can accurately predict the ignition time and ignition mode of pulverized coal particles under different operating conditions (including ambient temperature, oxygen concentration, pressure, volatile content, particle size, carbon dioxide concentration, water vapor concentration, turbulence intensity, etc.). Furthermore, a multi-factor weight analysis method is adopted to evaluate the influence degree of various factors on the ignition parameters. The results indicate that particle size, water vapor concentration, and volatile content of pulverized coal are the main factors affecting the ignition delay time, while particle size, ambient temperature, and ambient oxygen concentration are the three key factors affecting the ignition mode.

Conclusion The established BP neural network model can effectively predict the ignition time and ignition mode of pulverized coal particles under complex atmospheres, and identify the differences in the dominant effects of different operating parameters on the ignition delay time and ignition mode.