对掘进巷道瓦斯涌出时间序列(Q序列)的多重分形特征的研究表明,在掘进工作面有突出危险时Q序列的多重分维Dq图谱与无突出危险时相比,Dq图谱中负q与正q之间张口变大,在Dq-q曲线上表现为负q部分上升,正q部分下降,q=0附近区域曲线明显变陡。而无突出危险时段中Dq-q曲线过渡平缓,无明显异常。最后据此规律,建立了基于Q序列多重分形特征的煤巷突出危险性人工神经网络(ANN)预测模型。 The research on multifractal characteristics of gas emission time series (Q sequence) in tunneling roadway shows that the multifractal DQ spectrum of Q series is significantly different from that of non-prominent hazard when there is outstanding danger in driving face, Positive q between the mouth becomes larger, in the Dq-q curve showed a negative q-part rise, positive q part of the decline, q = 0 in the vicinity of the curve became steep. The DQ-q curve in the period of no prominent danger transitions smoothly with no obvious abnormality. Finally, based on this rule, a prediction model of outburst dangerous artificial neural network (ANN) based on multi-fractal features of Q-sequence is established.