改进哈里斯鹰优化算法与BP神经网络组合的滑坡位移高精度预测模型

开展滑坡位移高精度预测研究对于滑坡灾害的防灾预警具有重要意义。针对哈里斯鹰优化算法(HHO)搜索精度低且会陷入局部最优的问题,对其进行改进并进一步与BP神经网络组合,同时有效兼顾滑坡外部影响因子,发展了一种改进哈里斯鹰优化算法(IHHO)与BP神经网络组合(IHHO-BP)的滑坡位移高精度预测模型。结合我国典型黄土滑坡——甘肃黑方台党川滑坡HF08、HF05和HF09等3个监测点的北斗/GNSS实测数据,验证了IHHO-BP模型在3个实测数据集中的位移预测精度均优于单一BP神经网络模型,以及哈里斯鹰优化算法、麻雀搜索算法(SSA)、粒子群算法(PSO)、遗传算法(GA)与BP神经网络组合的预测模型。结果表明:引入Levy变异、局部增强和随机化Halton序列种群初始化策略的改进哈里斯鹰优化算法,可有效解决哈里斯鹰优化算法搜索精度低且会陷入局部最优的问题; IHHO-BP模型具有更好的泛化能力,可有效提升滑坡位移的预测精度,该组合预测模型具有更好的推广应用价值。

High-precision Landslide Displacement Prediction Model Based on IHHO Algorithm Combined with BP Neural Network

High-precision prediction of landslide displacement has important reference value for landslide disaster prevention and early warning. Aiming at the problem that the Harris Hawks optimization(HHO)algorithm has low search accuracy and falls into local optimum, while effectively taking into account the external influence factors of landslides, a high-precision landslide displacement prediction model was established based on improved Harris Hawks optimization(IHHO)algorithm combined with BP neural network(IHHO-BP). Combined with the measured Beidou/GNSS data(HF08, HF05 and HF09 monitoring points)of the Dangchuan landslide in Heifangtai area of Gansu, which is a typical loess landslide in China, the displacement prediction accuracy of IHHO-BP model is verified to be better than that of the single BP neural network, and the combination prediction models of HHO, sparrow search algorithm(SSA), particle swarm optimization(PSO), genetic algorithm(GA)and BP neural network. The results show that IHHO algorithm, which introduces Levy variation, local enhancement and randomized Halton sequence population initialization strategy, can effectively solve the problem that HHO algorithm has low search accuracy and falls into local optimization; the IHHO-BP model has better generalization ability and can effectively improve the prediction accuracy of landslide displacement, which is also has better popularization and application value.