边坡变形时序分析的进化-自适应神经模糊推理模型

变形监测与预报是保证边坡工程施工安全与工程质量的重要措施，但由于位移时间序列的强非线性，边坡变形预报成为非常困难的问题．自适应模糊神经推理系统（ANFIS）有优越的学习和泛化性能，而遗传算法（GA）是优秀的全局优化工具．采用遗传算法优化ANFIS参数，并编制了相应的计算程序．结合三峡工程永久船闸施工变形监测和新滩滑坡变形监测，建立了边坡变形时序分析的GA-ANFIS智能模型．为了对比该模型的预测精度，采用GA优化支持向量回归（SVR）和BP神经网络的模型参数，编制了GA-SVR及GA-BP程序，对相同的算例进行了变形预测分析．按滚动预测法对三峡永久船闸高边坡和新滩滑坡的计算结果表明，文中提出的GA-ANFIS模型能够获得比GA-SVR和GA-BP模型更高的预测精度，可以应用于边坡工程变形监测预报分析，并为类似工程提供参考．

Analysis model of slope deformation time series based on the genetic-adaptive neuron-fuzzy inference system

Deformation monitoring and forecasting is the key measurement to ensure construction secu- rity and quality of slope engineering. However, the deformation prediction has become a very difficult problem due to the strong nonlinear features of displacement time series during slope construction period. Adaptive neuron-fuzzy inference system （ANFIS） has good learning and generalization ability. Meanwhile, the genetic algorithm （GA） is an excellent global optimization tool and can be used to optimize the parameters of ANFIS. The corresponding calculation procedure has been programmed in this paper. Considering the permanent ship lock deformation monitoring of Three Gorges project and Xintan Slope, the GA-ANFIS intelligent displacement time series model of these slopes were established. In order to judge the performance of the GA-ANFIS model, GA was also applied to optimize the parameters of support vector regression （SVR） and BP neural network. Then, the two algorithms were used to analyze the same time series of the two slopes according to the rolling prediction method. By comparing the calculation results, it can be concluded that the GA-ANFIS model in this paper can obtain a better forecasting result than GA-BP and the GA-SVR model, and can be applied to deformation prediction of slope engineering well. At the same time, it can offer reference for other similar engineering.