As the intensity and depth of coal mining grow year by year, coal seam gas pressure increases and stope structures become more complex, which can easily cause coal and gas outburst. During the process of coal and gas outburst, a large amount of coal is broken and ejected, seriously threatening the safety of workers and coal mine production. Therefore, a multifunctional coal and gas outburst physical simulation test system was used to carry out three outburst tests under different gas pressures to study the particle size distributions and fragmentation characteristics of the ejected coal. The results showed that the relative intensity of outburst increased with gas pressure, but the increase rate decreased. Gas pressure also played a role in promoting the coal crushing. For the crushing product, the R–R (Rosin–Rammler) distribution model with high COD (coefficient of determination) was used to calculate the comminution energy at 0.35 MPa, while the fractal distribution model with high COD was used at 0.85 MPa and 2.00 MPa. When gas pressure increased, the basic shape of the R–R model curve remained unchanged, the probability density curve of fractal model changed from concave to nearly straight and then to convex and the basic shape of the cumulative distribution curve of fractal model remained constant. The values of α (uniformity coefficient) and x_e (characteristic particle size) impacted on the R–R model and the values of D_f (fractal dimension) and x_max (maximum particle size) impacted on the fractal model. Within a certain error range, the comminution energy could be approximated. The comminution energy increased with gas pressure, and the potential energy of crushing product decreased with the value of the n related to the crushing mechanism. There was a strong linear relationship between relative intensity of outburst and comminution coefficient. The combination of experiments and machine learning provided a new direction for outburst prediction and prevention at coal mine sites.

随着煤矿开采强度和深度的逐年增长，煤层气压增大，采场结构更加复杂，容易引起煤气外冲。在煤气突出过程中，大量煤炭破碎、喷出，严重威胁工人安全和煤矿生产。因此，采用多功能煤与瓦斯突出物理模拟试验系统，在不同瓦斯压力下进行了三次突出试验，研究了喷出煤的粒度分布和破碎特性。结果表明：突出的相对强度随瓦斯压力的增加而增加，但增加速率降低;瓦斯压力也起到了促进煤炭破碎的作用。对于破碎产品，采用高 COD（决定系数）的 R-R （松香-Rammler） 分布模型计算 0.35 MPa 时的粉碎能量，而高 COD 的分形分布模型在 0.85 MPa 和 2.00 MPa 时采用。当气压升高时，R-R 模型曲线的基本形状保持不变，分形模型的概率密度曲线由凹形变为近直再变为凸形，分形模型累积分布曲线的基本形状保持不变。α（均匀系数）和 x_e（特征粒径）的值影响了 R-R 模型，D _f （分形维数）和 x_max（最大粒径）的值影响了分形模型。在一定的误差范围内，可以近似计算出粉碎能量。 粉碎能量随气压的增加而增加，破碎产物的势能随与破碎机理相关的 n 值而降低。突增的相对强度与粉碎系数之间存在很强的线性关系。实验和机器学习的结合为煤矿现场的突出预测和预防提供了新的方向。