最大下沉角计算的神经网络模型研究

Determining angle of maximum subsidence based on neural network

最大下沉角是确定地表移动盆地走向主断面位置的关键性参数,关系到地表移动和变形预计的精度。系统分析了影响最大下沉角的地质采矿因素,根据国内大量的地表移动观测资料,建立了最大下沉角计算的神经网络模型。该模型与遗传算法相结合,克服了单一神经网络易陷入局部最优和收敛速度慢等缺点。研究表明,用基于遗传算法的神经网络来计算地表移动盆地的最大下沉角,其结果更加真实、可靠。

Angle of maximum subsidence, which is important to determine the position of the major cross-section of subsidence basin along strike, affects the prediction precision of the surface displacement and strain. The geological and mining factors that influence the angle of maximum subsidence are systematically analyzed; Based on the practical observational data from the ground movement monitoring stations of many mines in China the neural network model is developed to determine the angle of maximum subsidence. The combination of genetic algorithm and neural network can overcome the disadvantages of the artificial neural works such as limitation of local optimization and slow convergence rate. The validity and reliability of neural network method combined with genetic algorithm to determinate the angle of maximum subsidence are verified by the existing engineering instances.