通过对开采煤层自燃机理的阐释，选取其主要影响因素作为判别指标，建立了开采煤层自燃危险性的自组织神经网络预测模型，并选用典型的样本对该神经网络进行了训练，尔后用测试样本进行验证，结果表明该方法具有很高的准确性。 Based on the interpretation of spontaneous combustion mechanism of coal seam and its main influence factors as the discriminant index, a self-organizing neural network prediction model for mining spontaneous combustion danger of coal seam was established. The typical samples were used to train the neural network. Then, The results show that the method has high accuracy.