摘要
为提高短时交通流量预测的准确性,提出一种基于改进LSSVM的短时交通流量预测模型。针对传统混合蛙跳算法(SFLA)容易陷入局部最优的问题,提出基于新局部更新策略的改进混合蛙跳算法(ISFLA),在此基础上将其与最小二乘支持向量机(LSSVM)相结合,通过采用该算法优化LSSVM的关键参数,从而提高LSSVM的预测能力。结合实例,对模型和算法进行仿真分析,证明模型的可行性和算法的有效性。
To improve the predicting accuracy, proposes a short-term traffic flow forecasting model which is based on the improved least square support vector machine. Proposes an improved algorithm based on the traditional shuffled frog leaping algorithm, which can let shuffled frog leaping algorithm be not easy to fall into local optima. Uses the ISFLA algorithm to determine the important parameters in LSSVM, which significantly influences the predicting performance in the model. Simulation analysis demonstrates the feasibility of the model and the effectiveness of the algorithm.
出处
《现代计算机》
2015年第3期3-8,共6页
Modern Computer
基金
国家自然科学基金(No.11102124)
关键词
短时交通流量预测
最小二乘支持向量机
改进蛙跳算法
参数选择
智能优化算法
Short-Term Traffic Flow Forecasting
Least Square Support Vector Machine
Improved Shuffled Flog Leaping Algorithm
Parameter Selecting
Intelligent Optimization Algorithm
