【摘要】 基坑变形是造成施工事故的主要原因之一，其变形特性具有很大不确定性及随机性。为了防止基坑变形事故的发生，实现基坑变形的准确预测，提出一种将相空间重构、遗传算法(GA)、极限学习机(ELM)相结合的基坑变形时序混沌预测模型。首先，对基坑变形量时序进行相空间重构，通过混沌理论中的G-P算法及自相关法获得重构相空间参数；然后，采用相空间重构后的变形时间序列，建立基坑变形量预测的ELM模型，同时采用GA算法对ELM模型参数进行优化。最后，训练得到最优GA-ELM模型。以某基坑为例，采用GA-ELM模型进行变形量的混沌预测。研究结果表明：GA-ELM模型的预测结果优于单一ELM模型的预测结果，预测精度较高，有利于基坑变形的长期预测。更多还原

【Abstract】 Foundation pit deformation is one of the main causes of construction accidents, and its deformation characteristics is provided with great uncertainty and randomness.In order to prevent the occurrence of foundation pit deformation accidents, realize accurate prediction of foundation pit deformation.A chaotic prediction model of foundation pit deformation time series combining phase space reconstruction, genetic algorithm(GA) and extreme learning machine(ELM) is proposed.First, the phase space reconstruction is performed on the time series of the deformation of the foundation pit, and the reconstructed phase space parameters are obtained through the G-P algorithm and autocorrelation method in the chaos theory.Then, the deformation time series after phase space reconstruction is used to establish the ELM model for predicting the deformation of the foundation pit.At the same time, the GA algorithm is used to optimize the ELM model parameters.Finally, the optimal GA-ELM model is trained.Taking a certain foundation pit as an example, the GA-ELM model is used to predict the amount of deformation in a chaotic manner.The research results show that the prediction result of the GA-ELM model is better than that of a single ELM model, and the prediction accuracy is higher, which is conducive to the long-term prediction of foundation pit deformation.更多还原