论文部分内容阅读
针对大坝变形系统的非线性、复杂性以及不确定等特点,提出一种优化多核相关向量机的大坝变形预测模型方法。通过对实验数据进行归一化处理,核函数的加权组合以及遗传算法对模型参数的优化,建立遗传算法优化多核相关向量机的大坝变形预测模型。实验结果表明:数据归一化能归纳统一样本的统计分布性,加快梯度下降求解最优解速度和提高预测精度;优化的加权核函数能有效提高模型预测精度;各项精度指标值均优于BP神经网络方法、多项式核相关向量机方法预测精度,证实优化的多核相关向量机模型是一种精度较高的大坝变形预测方法。
According to the nonlinear, complex and uncertain characteristics of dam deformation system, a method of dam deformation prediction based on multi-core correlation vector machine is proposed. Through the normalization of the experimental data, the weighted combination of kernel function and the optimization of the model parameters by the genetic algorithm, the dam deformation prediction model based on genetic algorithm to optimize the multi-core correlation vector machine is established. The experimental results show that the data normalization can be used to summarize the statistical distribution of uniform samples, accelerate the gradient descent to find the optimal solution speed and improve the prediction accuracy. The optimized weighted kernel function can effectively improve the prediction accuracy of the model. BP neural network method and polynomial kernel-based correlation vector machine method to predict the accuracy. It is confirmed that the optimized multi-core correlation vector machine model is a high accuracy dam deformation prediction method.