【摘要】 结合汶川8.0级地震资料,利用神经网络原理和粒子群优化算法,提出了基于PSO-BP神经网络的地震地质灾害综合评价模型。该模型选取地震灾害、斜坡灾害、地面变形、斜坡分布特征4个指标作为输入,选用地质灾害危险度和分级2个指标为输出,引入粒子群算法对BP网络的权值和阈值进行优化,获得了BP网络模型参数。研究结果表明,PSO-BP网络模型不但能克服BP算法收敛速度慢和易陷于局部极小的缺陷,而且计算精度高,泛化能力强;对地质灾害的评价、防范和灾后重建具有一定的参考作用。更多还原

【Abstract】 Using the data of Wenchuan MS8.0 earthquake,we establish a comprehensive evaluation model for earthquake geological disasters based on the principle of neural networks and particle swarm optimization algorithm.Four parameters including the earthquake hazards,slope hazards,ground deformation and distribution characteristics of slope are taken as input,and risk of geological disasters and hazard class as output in the model.So we obtain the parameters of BP network model through weights and thresholds of BP network which is optimized by particle swarm algorithm.The results show that PSO-BP network model can overcome disadvantage of slow convergence speed of BP algorithm and falling into local minimum easily,which has higher calculation accuracy convenient and stronger generalization ability.Moreover,it can provide the reference for the evaluation and prevention of geological hazards,and reconstruction after earthquake.更多还原