基于PSO-VMD-LSTM模型的城市轨道交通短期客流预测

Short-term passenger flow forecast of urban rail transit based on PSO-VMD-LSTM model

为减少噪声对客流预测模型的干扰,采用粒子群优化(particle swarm optimization,PSO)算法确定变分模态分解(variational mode decomposition,VMD)的参数,通过PSO算法优化的VMD对原始客流序列进行降噪处理,将客流数据分解为不同时间尺度下的本征模态函数(intrinsic mode function,IMF)和余量;采用贝叶斯优化(Bayesian optimization,BO)算法确定长短期记忆(long short term memory,LSTM)神经网络的超参数,构建PSO-VMD-LSTM客流预测模型。以重庆轨道交通1号线沙坪坝站客流数据为例,验证模型的预测准确度。结果表明:PSO-VMD-LSTM模型的均方根误差比反向传播(back propagation,BP)神经网络、径向基函数(radial basis function,RBF)神经网络、LSTM神经网络分别降低268.03、204.41、221.66,平均绝对百分误差分别降低13.16%、10.21%、11.06%。PSO-VMD-LSTM模型对轨道交通短期客流预测具有较高的适用性和预测准确度。

In order to reduce the interference of noise on the passenger flow prediction model,the particle swarm optimization(PSO) algorithm is used to determine the parameters of variational mode decomposition(VMD).The original passenger flow sequence is denoised using VMD,and the passenger flow data is decomposed into intrinsic mode functions(IMF) and residuals at different time scales.The Bayesian optimization(BO) algorithm is used to determine the hyperparameters of the long short term memory(LSTM) neural network,and the PSO-VMD-LSTM passenger flow prediction model is constructed.Taking the passenger flow data of Shapingba Station of Chongqing Metro Line 1 as an example,the prediction accuracy of the model is verified.The results show that compared with back propagation(BP) neural network,radial basis function(RBF) neural network,and LSTM neural network,the PSO-VMD-LSTM model reduces the root mean square error by 268.03,204.41,and 221.66,and reduces the mean absolute percentage error by 13.16%,10.21%,and 11.06% respectively.The PSO-VMD-LSTM model has high applicability and prediction accuracy for short-term passenger flow prediction in urban rail transit.