基于PCA-SSA-ELM的混凝土坝变形预测模型

Prediction Model of Concrete Dam Deformation Based on PCA-SSA-ELM

混凝土坝变形与环境量之间有着复杂的函数关系,传统统计模型泛化能力较弱,难以处理高维非线性问题。为此,提出了一种基于主成分分析法和麻雀搜索算法优化极限学习机的混凝土坝变形预测模型,该模型通过主成分分析法(PCA)提取环境量中的关键因子作为模型输入变量,采用寻优能力强的麻雀搜索算法(SSA)选取极限学习机(ELM)中的初始输入权重和偏置的最优解。将该PCA-SSA-ELM模型应用到某高拱坝拱冠梁坝段测点径向位移的预测中,并与ELM、BP神经网络模型的计算结果进行对比分析,验证了新模型的有效性。

Complex functional relationship exists between the deformation of concrete dam and the environmental variables,while traditional statistical models are too weak in generalization capability to deal with high-dimensional nonlinear problems.A concrete dam deformation prediction model based on principal component analysis,sparrow search algorithm and extreme learning machine(PCA-SSA-ELM)is proposed.The model extracts the key factors in the environmental variables as the model input variables by PCA,and selects the optimal solution of initial input weights and biases in ELM by using SSA with strong optimization ability.The model is applied to predict the radial displacement of the measured points in the arch crown beam section of a high arch dam,and the predicted results are compared with that of the ELM and the BP neural network.The effectiveness of the proposed model is verified.