基于Fisher-逐步判别法的煤与瓦斯突出预测

Prediction of coal and gas outburst based on Fisher-stepwise discriminant analysis method

为准确、可靠地对煤与瓦斯突出进行预测,基于Fisher判别分析理论,建立了煤与瓦斯突出预测的Fisher判别模型。通过逐步判别分析法筛选出瓦斯放散初速度、瓦斯压力和埋深3个煤与瓦斯突出敏感指标作为突出判别因子,利用所构建的判别模型对20组煤与瓦斯突出实例数据进行训练学习得出相应的判别函数,用回代估计的方法进行逐一验证,误判率仅为5%。将建立的判别模型应用于10组突出实例进行判别预测,判别正确率达100%。结果表明:Fisher-逐步判别分析模型稳定性好,判别准确度高,是一种十分有效的煤与瓦斯突出预测方法。

Fisher discriminant model was established to predict the coal and gas outbursts accurately and reliably based on the principle of Fisher discriminant analysis theory.Four sensitive indexes of coal and gas outburst such as initial speed of methane diffusion,gas pressure,buried depth were regarded as the discriminant factors of the Fisher discriminant analysis model by the stepwise discriminant analysis method.By using the constructed discriminant model,discriminant function was obtained through training of twenty sets of coal and gas outburst data.Each of the twenty sets of samples was tested by using resubstitution method according to the Fisher discriminant method,and the misjudgment rate was only equal to zero.The established discriminant model was applied to eight sets of coal and gas outburst examples for discriminant prediction.The accuracy of the test were 100%.The results showed that the Fisher-stepwise discriminant analysis method is a very effective method for predicting coal and gas outburst with good stability and high discrimination accuracy.