基于GA-BP-AdaBoost强预测模型的大坝变形应用

Application of dam deformation based on GA-BP-AdaBoost strong prediction model

基于大坝变形量与多影响因素复杂非线性关系问题,提出了基于遗传算法优化BP神经网络的AdaBoost强预测模型(GA-BP-AdaBoost)。算例分析表明,该强预测模型融合了遗传算法全局优化和BP神经网络的局部寻优的特点,同时AdaBoost强预测器通过给弱预测器的预测序列赋予不同的权重,综合不同预测序列的精度优势,实现了AdaBoost强预测器“优中选优”的目的,最大限度地提高了预测精度,验证了本文基于遗传算法优化BP神经网络的AdaBoost强预测模型在大坝变形监测中的可行性和实用性。

From the complex non-linear relationship between dam deformation and multi-influence factors, an AdaBoost strong prediction model (GA-BP-AdaBoost) based on genetic algorithm for BP neural network optimization is proposed. Research analysis shows that the model combines the global optimization of genetic algorithm and the local optimization of BP neural network.AdaBoost strong predictor gives different weight to the prediction sequence of weak predictor. The prediction accuracy is maximized.The feasibility and practicability of the AdaBoost strong forecasting model based on BP neural network optimized by genetic algorithm in dam deformation monitoring are verified.