Adaptive Denoising Spatio-Temporal Attention Network Fused With External Factors for Passenger Flow Prediction

Traditional station passenger flow prediction can no longer meet the application needs of urban rail transit vehicle scheduling. Station passenger flow can only predict station distribution, and the passenger flow distribution in general sections is unknown. Accurate short-term travel origin and destination (OD) passenger flow prediction is the main basis for formulating urban rail transit operation organization plans. To simultaneously consider the spatiotemporal characteristics of passenger flow distribution and achieve high precision estimation of origin and destination (OD) passenger flow quickly, a predictive model based on a temporal convolutional network and a long short-term memory network (TCN–LSTM) combined with an attention mechanism was established to process passenger flow data in urban rail transit. Firstly, according to the passenger flow data of the urban rail transit section, the existing data characteristics were summarized, and the impact of external factors on section passenger flow was studied. Then, a temporal convolutional network and long short-term memory (TCN–LSTM) deep learning model based on an attention mechanism was constructed to predict interval passenger flow. The model combines some external factors such as time, date attributes, weather conditions, and air quality that affect passenger flow in the interval to improve the shortcomings of the original model in predicting origin and destination (OD) passenger flow. Taking Chongqing Rail Transit as an example, the model was validated, and the results showed that the deep learning model had significantly better prediction results than other baseline models. The applicability analysis in scenarios such as high/medium/low passenger flow could achieve stable prediction results.

Accurate short-term passenger flow prediction can help metro managers optimize train scheduling and formulate useful operation plan, so as to reduce metro transportation pressure and improve passenger comfort. However, existing research can’t fully explore the dynamic spatial-temporal correlation of passenger flow in metro network, and don’t fully consider the external characteristics such as weather and geographical, which affect the accuracy of metro passenger flow prediction. This paper proposes an attention based metro spatial-temporal convolution model (AMSTCN) for short-term passenger flow prediction. Firstly, aiming at the problem that the existing methods don’t fully consider the dynamic correlation of passenger flow between stations, the model uses the spatial-temporal convolution block with attention mechanism to capture the spatiotemporal dynamic transferability of passenger flow between stations; Secondly, in view of the shortcomings of the existing graph convolutional network that does not fully consider the topology of the metro network, the adjacency matrix is constructed by using passenger flow transfer information; Finally, the model captures the characteristics of external factors through the external module, and the learning ability is enhanced. Experimental results show that not only the dynamic spatial-temporal dependence of metro network passenger flow is captured, but also the prediction accuracy is further improved after considering the external factors. The model is effective and stable in predict short-term and mid-term outbound passenger flow.

Multi-point short-term prediction of station passenger flow based on temporal multi-graph convolutional network

This paper studies accurate short-term prediction of urban rail transit passenger flow in external passenger transport hub. Based on the conventional features that affect external rail transit passenger flows, we propose an innovative method of statistical features construction to make accurate prediction of frequently fluctuating passenger flow under the framework of machine learning. Firstly, by classifying the statistical features in time-series, we improve the previous long-short-term memory(LSTM) network model and establish a novel long-short-term memory(LSTM) network model that can deal with long-term dependent time-series data so as to accurately fit real-time passenger flow. Secondly, by introducing the lightweight implementation algorithm LightGBM based on the gradient decision boosting tree(GBDT), we build a short-term prediction model, which can reduce operation cost and improve accuracy so that accurate fitting of passenger flow peaks can be achieved. Thirdly, based on K-nearest neighbor algorithm, we select DRS as the dynamic regression device that can find a local optimal fusion method, and establish a LGB-LSTM-DRS local optimal fusion prediction model. Taking the AFC passenger flow data of Chengdudong Station as an example, we predict the short-term passenger flow. Compare with other commonly used models such as LR, RF, and GBDT, it's proved that the LGB-LSTM-DRS model has the smallest error (MAPE and RMSE) and the best prediction performance.

The accurate passenger flow prediction is the base of bus scheduling and bus dispatching. Many factors, including internal factors and external factors, have great impact on the fluctuation of passenger flow. In the modern informationized bus system, many influencing factors became available by multi-source data. Current passenger flow prediction methods are mainly based on statistical predicting methods and machine learning methods. The implication of interpolating prediction method on passenger flow prediction is preliminary. Interpolating prediction method makes use of historical data; the prediction result is generally accurate, and the method is robust. Interpolating prediction method shows good performance and has mature application in other research areas. This paper makes use of historical passenger data and multi-source data; apply Shepard model to predict public transportation passenger flow. The result shows that Shepard prediction model has better performance than that of neural network (NN) model and support vector machine (SVM) model. The mean absolute percentage error (MAPE) has increased 7.5 and 3.43%; the MSP has increased 16 and 10.51% compared with NN and SVM and has lower dependency of parameters.

Passenger flow is an important benchmark for measuring tourism benefits, and accurate tourism passenger flow prediction is of great significance to the government and related tourism enterprises and can promote the sustainable development of China’s tourism industry. For daily passenger flow time series data, a passenger flow forecasting method based on convolutional neural network (CNN) and improved quantile regression long short-term memory network (QRLSTM), denoted as CNN-IQRLSTM, is proposed with reconstructed correlation features and in the form of sliding windows as inputs. First, four discrete variables such as whether the day is a weekend and holiday are created by time; then, a sliding window of width 42 is used to pass the passenger flow data into the network sequentially; finally, the loss function of the sparse Laplacian improved QRLSTM is introduced for passenger flow prediction, and the point prediction and interval prediction results under different quartiles are obtained. The application of quantile regression captures the overall picture of the data, enhances the robustness, fit, predictive power and nonlinear processing capability of neural networks, and fills the gap between quantile regression and neural network methods in the field of passenger flow prediction. CNN can effectively handle complex input data, and the improved nonlinear QR model can provide passenger flow quantile prediction information. The method is applied to the tourism traffic prediction of four 5A scenic spots in Jilin Province, and the effectiveness of the method is verified. The results show that the method proposed in this paper fits best in point prediction and has higher prediction accuracy. The MAPE of the Changbai Mountain dataset was 0.07, the MAPE of the puppet palace museum dataset was 0.05, the fit of the Sculpture Park dataset reached 93%, and the fit of the net moon lake dataset was as high as 99%. Meanwhile, the interval prediction results show that the method has a larger interval coverage as well as a smaller interval average width, which improves the prediction efficiency. In 95% of the interval predictions, the interval coverage of Changbai Mountain data is 99% and the interval average width is 0.49. It is a good reference value for the management of different scenic spots.

As residents are more and more inclined to choose rail transportation as their daily travel mode, the status of urban rail transportation is becoming more and more prominent. However, due to the level of passenger flow analysis and prediction accuracy, how to better meet passenger demand and improve operation efficiency has become a key issue to be solved by the operation department. In order to improve operational efficiency and service quality, this paper introduces weather-related data to analyze passenger flow at rail transit stations and develop short-term passenger flow prediction. Taking Beijing urban rail transit passenger flow as an example, cluster analysis and correlation analysis are conducted with corresponding weather data to verify the passenger flow characteristics in line with the actual travel. The long and short-term memory (LSTM) neural network model is selected for short-term passenger flow prediction, and weather data are added as features to improve the prediction accuracy. The results show that the passenger flow data with weather features are better fitted in the prediction, and the more accurate prediction results will play a more effective role for the relevant operation departments to schedule trains and improve the operation efficiency.

The travel planning system is an important reference tool for passengers to choose a travel route. An efficient and safe travel plan can improve the travel efficiency and ride safety of passengers, and improve the service level of the rail transit system. At the same time, travel planning information is of great value for more accurate prediction of short-term passenger flow, and more accurate short-term passenger flow prediction information can also provide support for higher-quality travel planning. Existing related studies have hardly considered the information interaction mechanism between passenger flow predicting and travel planning. To solve this problem, this paper proposes a new regional rail transit travel planning system design idea and implementation scheme based on passenger flow prediction. This system makes short-term prediction of passenger flow based on individual travel planning information. The passenger flow prediction results are used to provide support for subsequent passenger travel planning. Finally, the Chongqing rail transit network is taken as an example to analyze the prototype case, and the feasibility of the system scheme was preliminarily verified.

Passenger demand is the necessary foundation of urban rail transit operation management, and it’s very important to study the accuracy of short-term passenger flow prediction. However, the passenger flow is affected by many factors, so more comprehensive research is needed. This paper is based on multi-source data and the combined model. Firstly, the preliminary correlation between the passenger flow sequence and external factors is analyzed before the prediction, and the required features are extracted by using the convolutional neural network (CNN) and inputted into the prediction process. Also, the model is combined by Bi-directional Long Short-term memory network (Bi-LSTM) and attention mechanism (Attention), which can adapt to the nonlinearity and periodicity of short-term passenger flow prediction. Finally, the real example analysis shows that there is a certain correlation between subway passenger flow and multi-source data, and the CNN-BiLSTM-Attention model can improve the accuracy of passenger flow prediction and reduce errors, which is better than other traditional prediction methods.

Accurate passenger flow prediction at bus stops is the basis of bus operations. Capturing the temporal, spatial and periodic correlations to improve the prediction accuracy of passenger flow efficiently is one of the main problems faced by bus-stop passenger flow prediction. In this study, a deep-learning framework that combines a graph-convolution network (GCN) and gated-recurrent unit (GRU) is proposed to predict citywide bus-stop passenger flow. First, to extract the spatial correlation of passenger flow, a method is proposed to construct a multi-semantic graph according to the topological attributes of the stop to capture deep spatial correlations. Second, the proposed framework considers the periodic correlation of passenger flow, which is usually neglected and differs from the temporal correlation. Finally, a bus-stop passenger flow prediction framework is constructed using a GCN to capture spatial correlations and a GRU to capture temporal and periodic correlations. The proposed framework was tested on real datasets. The results show that for time granularities of 30, 45, and 60 min, the root-mean-square errors (RMSE) yielded by this model were 2.2275, 2.7832, and 3.2902, respectively. Compared with the baseline models, the average RMSE was reduced by 21.98%, 27.31%, and 29.25%, respectively. The multi-semantic graph and periodic correlation proposed in this study improve the prediction accuracy of citywide passenger flow, and the fitting results of selected cases show that the model is robust. The results can provide data support for scheduling plan and departure frequency of urban public transport enterprises.

A novel passenger flow prediction model using deep learning methods

Rapid and accurate short-term passenger flow prediction plays an important and far-reaching role in passenger flow control and early warning. In fact, the short-term passenger flow presents the characteristics of non-linearity and randomness. Traditional machine learning algorithms can hardly meet current predictions. In this paper, the random forest(RF) is used to calculate the feature importance to filter the extracted features and remove the redundant features, and we apply the Long Short-Term Memory network(LSTM) algorithm model to predict the short-term passenger flow of the metro. First, we calculate the out-of-bag(OOB) error of the features by RF based on the characteristics of bootstrap sampling and regression tree, and calculate OOB error again after adding noise. According to the two OOB errors, the feature importance can be obtained through related formulas, and some features can be filtered. RF can effectively reduce redundant features to participate in calculation and improve operating efficiency. Second, we apply the LSTM model to predict passenger flow every 10 minutes for each station and use the important features selected by RF as the model inputs. LSTM has an excellent effect in dealing with problems that are highly related to time series, and it is very suitable for prediction on time series issues. The proposed model is evaluated with real metro card data, the prediction performance compares to single RF, LSTM, and other algorithm models. The experiment results show that the accuracy of the RF combined LSTM model algorithm is better than that of other existing models such as RF and LSTM model. It shows good prediction accuracy and has far-reaching significance in the field of passenger flow prediction.

Bus operation scheduling is closely related to passenger flow. Accurate bus passenger flow prediction can help improve urban bus planning and service quality and reduce the cost of bus operation. Using machine learning algorithms to find the rules of urban bus passenger flow has become one of the research hotspots in the field of public transportation, especially with the rise of big data technology. Bus IC card data are an important data resource and are more valuable to passenger flow prediction in comparison with manual survey data. Aiming at the balance between efficiency and accuracy of passenger flow prediction for multiple lines, we propose a novel passenger flow prediction model based on the point-of-interest (POI) data and extreme gradient boosting (XGBoost), called PFP-XPOI. Firstly, we collected POI data around bus stops based on the Amap Web service application interface. Secondly, three dimensions were considered for building the model. Finally, the XGBoost algorithm was chosen to train the model for each bus line. Results show that the model has higher prediction accuracy through comparison with other models, and thus this method can be used for short-term passenger flow forecasting using bus IC cards. It plays a very important role in providing decision basis for more refined bus operation management.

The rapid development of urban rail transit brings high efficiency and convenience. At the same time, the increasing passenger flow also remarkably increases the risk of emergencies such as passenger stampedes. The accurate and real-time prediction of dynamic passenger flow is of great significance to the daily operation safety management, emergency prevention, and dispatch of urban rail transit systems. Two deep learning neural networks, a long short-term memory neural network (LSTM NN) and a convolutional neural network (CNN), were used to predict an urban rail transit passenger flow time series and spatiotemporal series, respectively. The experiments were carried out through the passenger flow of Beijing metro stations and lines, and the prediction results of the deep learning methods were compared with several traditional linear models including autoregressive integrated moving average (ARIMA), seasonal autoregressive integrated moving average (SARIMA), and space–time autoregressive integrated moving average (STARIMA). It was shown that the LSTM NN and CNN could better capture the time or spatiotemporal features of the urban rail transit passenger flow and obtain accurate results for the long-term and short-term prediction of passenger flow. The deep learning methods also have strong data adaptability and robustness, and they are more ideal for predicting the passenger flow of stations during peaks and the passenger flow of lines during holidays.

It is a great significance for the accurate and real-time prediction of passenger flow in rail transit operation. In the process of passenger flow prediction, the new method of recurrent neural networks (RNNs) can well solve the problems of randomness and nonlinearity which can not be solved by the existed linear models. In this paper, the long short-term memory (LSTM) and the gated recurrent unit (GRU) networks, which are methods of RNNs, are employed to predict the dayparting passenger flow and the raw passenger flow data is denoised by the wavelet transform. Experimental results show that LSTM and GRU networks can well predict the passenger flow. And compared to LSTM, GRU is better for passenger flow prediction.