基于空间加权的LS-SVM城市轨道交通车站客流量预测

Direct Ridership Forecast Model of Urban Rail Transit Stations Based on Spatial Weighted LS-SVM

为提高城市轨道交通车站客流预测模型精度,简化模型数据需求规模,提出基于空间加权的LS-SVM城市轨道交通车站客流预测模型。基于交通网络距离重新划分车站的影响范围,提出分距离影响带的线型和指数型空间权重系数方程,结合空间权重系数,输入区域特征变量和车站属性变量构建城市轨道交通车站客流LSSVM预测模型,运用动态改变惯性权重自适应粒子群优化算法(DCW-APSO)对模型参数进行优化选取。应用模型预测2011年成都市地铁1号线部分车站客流,并与其他模型进行比较,结果表明:模型明显提高客流预测精度,简化数据需求量,作为城市轨道交通客流预测的补充模型可以进一步提高系统的可靠性。

In order to improve the accuracy of the urban rail transit （URRT） station ridership forecast model and to simplify the model data scale, the direct ridership forecast model of urban rail transit stations based on the spatial weighted LS-SVM was proposed. The station influencing area was calculated by transport network distances, and the linear and exponential spatial weight equations associated with different distance bands were put forward. By inputting area characteristics variables and station attribute variables based on spatial weight coefficients, The LS-SVM forecast model of URRT station ridership was built. The dynamic change inertia weight adaptive particle swarm optimization algorithm （DCW-APSO） was applied to optimize selection of mod- el parameters. The proposed model was applied to forecast the ridership of some sations of Chengdu Metro Line in 2011, and the results were compared with other types of models. The comparison show that the pro-posed model improves forecast accuracy, simplifies model data demand and upgrades the model system reliabili-ty as a supplementary model of URRT passenger flow forecast.