基于粒子群—高斯过程回归的高拱坝变形预测模型及其应用

针对高拱坝变形问题,提出应用粒子群算法优化高斯过程回归参数的高拱坝变形预测模型,基于高斯过程回归可将低维非线性关系通过核函数投射到高维线性空间的特点,利用高斯过程回归模型来表征水压、温度、时效等因素与坝体变形之间的非线性关系;同时针对迭代求解高斯过程回归模型的超参数效率低的问题,采用粒子群优化算法全局搜索模型超参数,提高了求解效率。对某高拱坝径向位移的拟合预测结果表明,粒子群优化高斯过程回归模型能较好地表征输入因子与变形之间的关系,预测坝体变形,误差在工程允许范围内,可应用于坝体变形预测分析中。

Deformation Prediction Model of High Arch Dam Based on Particle Swarm Optimization and Gaussian Process Regression and Its Application

Aiming at the deformation problem of high arch dam, this paper presents a prediction model of high arch dam deformation by using particle swarm optimization to optimize the parameters of Gauss process regression. Based on the characteristics of Gauss process regression, the low-dimensional nonlinear relationship can be projected into the high-dimensional linear space through the kernel function, and the Gauss process regression model is used to characterize the nonlinearity between water pressure, temperature, time and other factors and the dam deformation. At the same time, aiming at the problem of low efficiency in solving the hyper-parameters of the Gaussian process regression model in iteration process, particle swarm optimization algorithm is adopted to search the hyper-parameters of the model, which improves the solution efficiency. The fitting prediction results of radial displacement of a high arch dam show that the Gaussian process regression model based particle swarm optimization can better represent the relationship between input factors and deformation, and predict the deformation of dam body as well as the error is within the allowable range of engineering. Thus, the method can be applied to the prediction and analysis of dam deformation.