论文部分内容阅读
为对井下落煤瓦斯涌出量进行预测,采用粗糙集与改进神经网络相结合的方法,在样本数据的筛选上吸取粗糙集数据约简的优点,使选择的数据样本简洁且更具代表性;充分利用BP神经网络的非线性拟合能力,将遗传算法与其相结合,避免BP网络陷入局部最优.利用编写的程序确定隐含层节点数,相比以往经验公式取值更具优势.最后在任家庄煤矿成功应用.研究结果表明:利用粗糙集与改进神经网络相结合模型进行预测,结果准确可靠,克服了以往BP模型的不足.该模型对井下落煤瓦斯涌出量预测具有一定参考价值.
In order to predict the amount of gas emission from downhole coal, a rough set and an improved neural network are combined to extract the advantages of the rough set data reduction in the selection of sample data, which makes the selected data sample concise and more representative Make full use of the nonlinear fitting ability of BP neural network and combine the genetic algorithm with it to avoid the BP network from getting into the local optimum.Using the programmed program to determine the number of hidden layer nodes is more advantageous than the previous empirical formula. Finally, it is successfully applied in Renjiazhuang Coal Mine.The results show that the prediction by using the combination of the rough set and the improved neural network is accurate and reliable, and overcomes the shortcomings of the previous BP model. Reference value.