地表移动预计参数选取的神经网络法

Neural network method used in adopting parameters for prediction of surface displacement

地表移动预计参数的选取是研究地表移动及其规律的重要内容,由于预计参数受多种复杂因素的影响,具有高度的不确定性和离散性,利用神经网络具有自组织、自学习和高度非线性映射的能力,并既能考虑定量因素又能考虑定性因素的优点,可建立地表移动预计参数选取的神经网络模型以及对BP神经网络进行改进。利用大量的地表移动实际观测数据样本对该网络模型进行训练和学习,并用该网络模型对地表移动参数进行预计,结果表明,该改进的BP神经网络具有收敛速度快、预计参数精度高的优点,从而为开采沉陷地表移动预计中参数的选取提供了新方法。

To predict parameters of surface displacement is an important field in researching on surface displacement and its rules, but the factors which control and affect the parameters are complicated and uncertain. As artificial neural network method not only possesses ability of self-taught, self-organized and high non-linear mapped, but also can consider both quantitative and quanlitative factors, based on improved BP neural network, a lot of neural network models are established to predict parameters of surface displacement. A lot of practical engineering data are collected to test and train the model,and application results show that the improved BP neural network model is of high convergent speed, good prediction precision. The model offers a new method to predict parameters of surface displacement in mining.