Abstract:We propose a combined prediction model of building deformation based on fractional Fourier transform(FrFT) and support vector machine(SVM). First, FrFT is used for multi-scale analysis of deformation time series, and the complex time series is decomposed into a series of subsequences with relatively simple structure. Then, we use SVM to establish prediction models for each subsequence, and the final prediction results are obtained by comprehensively superimposing the prediction results of each subsequence. At the same time, considering the difficult problem of parameter selection of SVM model, we propose an improved fruit fly optimization algorithm(IFOA) to globally optimize it and improve the prediction performance. Taking the measured deformation data of a concrete dam in Southwest China as an example, the results show that the proposed combined prediction model can fully mine the trend and regularity information hidden in the data and obtain high prediction accuracy.