煤矿顶板矿压显现实时预报的自适应神经网络方法

AN ADAPTIVE NEURAL NETWORK FOR REAL-TIME PREDICTION OF ROCK BEHAVIOUR IN COAL MINES

应用神经网络系统理论，提出了煤矿顶板压力显现实时预报的自适应模式识别方法，它通过对井下实测压力曲线的记忆，可以预报出该顶板未来的矿压显现规律，包括来压步距和来压强度。实际应用表明，本方法可外推预报１０个工作面推进循环的来压规律，来压强度的预报准确率达到９３％，来压步距的预报准确率为１００％。

An adaptive recognition approach for real-time prediction of rock behaviour was proposed by application of artificial neural network theory. It can predict the rules of future rock behaviour including the weighting interval and strength based on output of underground observation data. Practical experience indicates that this method can be applied to prediction of weighting in advancing cycles in ten faces by extrapolation with an accuracy of 93% for weighting strength and 100% for weighting interval.