摘要
在考虑工程风险及保险实际理赔情况的基础上,形成了含自然灾害、项目环境等7个指标维度的风险评价体系,利用粒子群(PSO)算法优化BP神经网络的初始阈值及权值,建立了公路工程保险费率厘定模型。将该模型应用于34个公路工程保险实际案例,通过PSO-BP神经网络拟合保险样本中风险指标因素与费率之间的关系,实现费率预测。对比分析PSO-BP神经网络与BP神经网络的仿真效果,结果表明,PSO-BP神经网络模型能较好地反映公路工程实际风险水平,预测准确度高,收敛速度快,适用于保险费率厘定。
A model for determining the insurance premium rate of highway engineering is established in this paper.The model is based on the consideration of engineering risk and the actual claim,using the PSO method to optimize the initial biases and weights of BP neural network.Applied to the 34 actual cases,the model determine the relationship between risk index factor in insurance sample and rate by PSO-BP neural network and realize the rate prediction.And the simulation results of PSO-BP neural network and BP neural network are compared and analyzed.The results show that the PSO-BP neural network model can reflect the actual risk level of highway engineering better,and is suitable for the determination of cost.Compared with BP neural network,PSO-BP Neural Network has higher prediction accuracy and faster convergence.
作者
仇一颗
周翔
王家
QIU Yike;ZHOU Xiang;WANG Jia(College of Civil Engineering,Hunan University,Changsha,Hunan, 410082,China)
出处
《公路工程》
北大核心
2020年第2期80-85,共6页
Highway Engineering
基金
国家自然科学基金资助项目(51308204)。
