摘要
为了有效预测交通事故,提出一种基于改进粒子群算法优化支持向量回归机的预测模型。改进粒子群算法利用网格搜索对全局最优粒子的邻域进行精细搜索,结合粒子群算法较快的收敛速度和网格搜索较强局部搜索能力的优点,提高了支持向量回归机相关参数的优化精度,进而改善了交通事故预测模型的预测性能。仿真结果表明,基于改进粒子群算法优化支持向量回归机的交通事故预测模型达到了较快的学习速度和较高的预测精度,具有良好的工程应用性。
In order to predict traffic accidents effectively, this article introduced a traffic accident forecast model based on SVR optimized by improved PSO. The improved particle swarm algorithm searches the neighborhoods of the global optimal particle by grid search which combines the faster convergence speed of the particle swarm algorithm with the stronger local search ability of the grid search. The improved particle swarm algorithm improves the optimization precision of SVR parameters and improves the performance of the traffic accident prediction model. The simulation results show that the traffic accident forecast model based on SVR optimized by improved PSO gets faster learning speed and higher prediction precision, it is of good engineering applicability.
出处
《计算机与现代化》
2014年第5期28-32,共5页
Computer and Modernization
基金
轻工过程先进控制教育部重点实验室开放课题资助项目(APCLI1004)
国家青年科学基金资助项目(F030204)
现代农业产业技术体系专项资金资助项目(CARS-49)
江苏高校人文社会科学校外研究基地:中国物联网发展战略研究基地资助项目
关键词
粒子群优化
网格搜索
支持向量回归机
参数优化
交通事故预测
particle swarm optimization
grid search
support vector regression
parameter optimization
traffic accident forecast
