中文摘要：

结合灰色模型和神经网络的数据处理特点,提出串联、并联和混联式3种结构的灰色神经网络滑坡变形预测模型。串联式将滑坡变形位移时序分解为趋势项和随机项,采用灰色模型提取滑坡位移时序趋势,利用神经网络逼近随机波动;并联式以灰色模型和神经网络分别对滑坡预测,采用智能非线性组合,按照预测目标精度动态调整权重,从而获取最终组合预测结果;混联式通过增加灰白化层及灰模型群,对神经网络拓扑结构进行优化,达到弱化滑坡原始监测数据随机性、提高预测模型稳健性的目的。将3种模型应用于古树屋滑坡变形预测,并对其适用性进行讨论。结果表明,3种结构的灰色神经网络耦合模型均提高了预测精度,适用于复杂状况下滑坡体的变形预测。

英文摘要：

This paper proposes a new model of landslide deformation prediction based on the grey artificial neural network,combining the data processing characteristics of the grey model and the artificial neural network,respectively.There are three kinds of forecasting model structure：series grey artificial neural network（SGANN）,parallel grey artificial neural network（PGANN）,and inlaid grey artificial neural network（IGANN）.The landslide deformation time series is decomposed into trend item and random item in SGANN.The trend item of the deformation time series can be extracted by the grey model,using the artificial neural network to construct the nonlinear relationship between the deformation time series and the trend item.PGANN uses the grey model and the artificial neural network to predict separately,while the weight value of this model is subject to the required accuracy of the experiment.IGANN optimizes the topological structure of the neural network by adding agrey layer and grey model group,in order to reduce the randomness of the original monitoring data and to enhance the model robustness ability.The above three new models are employed to forecast the deformation time series data monitored at the Gushuwu landslide.The cases show that the grey artificial neural network model is valid and feasible in prediction of landslide deformation under complicated conditions.