基于改进极限学习机的公交站点 短时客流预测方法

以公交车IC 卡和GPS数据为基础，提出了一种基于改进粒子群算法优化极限学习机(IPSO-ELM)的公交站点短时客流预测模型.依托IC 卡和GPS 数据在站点的特征表现和内在联系，定义了站点间距，并分析了站间距和车辆到总站距离间的联系；提出了公交乘客上车站点确定方法，进而得到公交站点上车客流量；通过分析公交客流数据特征，确定ELM输入参数维度，并采用IPSO 算法找到ELM的最优隐含层节点参数；最后依托广州市19 路公交车客流数据仓库进行了方法验证.结果表明：所用优化后的ELM方法预测误差在10%以内，并与应用广泛的SVM、ARIMA和传统ELM模型进行对比分析，发现改进的ELM方法拥有更高的可靠性和泛化性能.

Short-term Passenger Flow Prediction Method on Bus Stop Based on Improved Extreme Learning Machine

A short-term passenger flow prediction model on bus stop of optimized extreme learning machine and improved particle swarm optimization is proposed based on the bus IC card and GPS data. According to the characteristic and connection of IC card and GPS data at the bus station, station spacing is defined. Through analyzing the relationship between station spacing and the distance from the traffic to central station, a method to determining the bus passenger boarding station is proposed, and then get the number of boarding passenger at each stop. By analyzing the passenger flow data features of the bus, the dimension of input parameters of ELM is determined. Besides, the improved particle swarm optimization algorithm is used to find the optimal hidden layer node parameters of the extreme learning machine. Finally, the automated fare collection data of the 19 bus in Guangzhou city are taken to method verification. The results show that the prediction error of the optimized ELM method is less than 10%. Compared with SVM, ARIMA and traditional ELM which are widely used, the improved ELM methods has better reliability and generalization performance.