Flow Datasets Research Articles

정책자문 ODA가 개발도상국 정부 역량에 미치는 영향에 관한 연구

Purpose: The purpose of this paper is to empirically analyze the impact of policy advisory ODA distributed by recipient countries on government capabilities from a sustainable development perspective.
 Originality: Sustainable development requires the capacity of the government to allocate resources efficiently and to design and operate the optimal policies to support it. In this regard, policy advisory ODA aims to contribute to the sustainable development by providing advice to developing country governments so that they can effectively formulate and implement policies. However, previous studies have focused on which factors ODA is distributed and how ODA contributes to development in terms of the total amount of ODA. By quantitatively measuring the policy advisory ODA, and focusing on its impact on government capacity in developing countries, this study draws more concrete academic and practical implications for development effectiveness.
 Methodology: This paper constructs a panel dataset of policy advisory ODA flows and government capacity for 152 developing countries from 2000 to 2021, and conducts a random effects panel regression analysis.
 Result: It was found that the relationship between the amount of policy advisory ODA a country receives and its government capacity is statistically significant: on average, the more policy advisory ODA a country receives, the better its government capacity is. The empirical analysis also shows that the impact of the policy advisory ODA on government capacity is greater in Fragile and Conflict-Affected Countries, where government capacity is relatively weak, and the Least Developed Countries and Lower-Middle Income Countries.
 Conclusion and Implication: This study has both academic and policy implications in that it empirically analyzes the impact of policy advisory ODA on recipient countries’ government capacity, an area that has been under-explored in the literature despite its importance. The empirical results draws attention to the importance and effectiveness of the policy advisory ODA.

Flow-models 2.0: Elephant flows modeling and detection with machine learning

This article presents the new version of the flow-models IP network flow modeling framework. The improved features include flow skipping and counting, flow filtering, IP address anonymization, and time series data calculation. The new version also enables simulation of the first packet mirroring feature and provides tools for modeling the detection of elephant flows. It includes examples of using the scikit-learn library to build machine learning models for elephant flow detection based on the first packet. Furthermore, it provides an anonymized flow dataset, enabling researchers to train and validate machine learning models for traffic analysis in a reproducible and comparable manner.

A tensorial weighted Schatten-[formula omitted] norm model with neighbor regularization for traffic data completion and traffic system correlation exploration

Traffic data exhibit the spatiotemporal heterogeneity. Revealing traffic system correlations (TSC) from traffic data, which refers to discovering inter-station correlations incurred from movements of passengers, is a meaningful task for releasing traffic congestion and sustainable urban operations. Since traffic data involve a great loss or corruption due to equipment and transmission failures, traffic data completion (TDC) serves as a necessary task for TSC exploration. We observe that: for the TDC task, traffic data exhibit anisotropic periodicity and repetition along different temporal dimensions, which cannot be captured by traditional low-rank TDC methods; for the TSC exploration task, a traffic network obeys graphical structure and exhibits a strong spatial correlation between stations or road segments. Based on these observations, we propose a Tensorial Weighted Schatten-pNorm model with Neighbor Regularization (TWSN2R) for accomplishing the two tasks simultaneously. TWSN2R consists of a tensorial weighted Schatten-p norm term and a neighbor regularization term. The first term mines the anisotropic periodicity with a tensor form of weighted Schatten-p norm which preserves suitable rank components of traffic data by allocating different weights in both tensorial dimension and decomposed mode dimension. The second term mines a spatial correlation of a traffic network by enforcing a neighbor contribution of passenger flows incurred from the network, and extrinsically reveals an inter-station (or inter-road-segment) TSC of the network. Experiments on the subway flow dataset MetroBJ and the road network vehicle speed dataset RoadBJ demonstrate that, TWSN2R not only outperforms several state-of-the-art TDC methods, but also reveals two pieces of regulations of TSC. (a) For subway passenger flows, positive correlated stations and negative correlated stations contribute to passenger flows differently according to each station’s attributes. (b) For road networks, urban loops, expressways and transportation hubs contribute to the vehicle flows at a majority, and serve as the critical roads for mitigating the congestion of whole road networks.

Deep LiDAR-Radar-Visual Fusion for Object Detection in Urban Environments

Robust environmental sensing and accurate object detection are crucial in enabling autonomous driving in urban environments. To achieve this goal, autonomous mobile systems commonly integrate multiple sensor modalities onboard, aiming to enhance accuracy and robustness. In this article, we focus on achieving accurate 2D object detection in urban autonomous driving scenarios. Considering the occlusion issues of using a single sensor from a single viewpoint, as well as the limitations of current vision-based approaches in bad weather conditions, we propose a novel multi-modal sensor fusion network called LRVFNet. This network effectively combines data from LiDAR, mmWave radar, and visual sensors through a deep multi-scale attention-based architecture. LRVFNet comprises three modules: a backbone responsible for generating distinct features from various sensor modalities, a feature fusion module utilizing the attention mechanism to fuse multi-modal features, and a pyramid module for object reasoning at different scales. By effectively fusing complementary information from multi-modal sensory data, LRVFNet enhances accuracy and robustness in 2D object detection. Extensive evaluations have been conducted on the public VOD dataset and the Flow dataset. The experimental results demonstrate the superior performance of our proposed LRVFNet compared to state-of-the-art baseline methods.

A Spatial-Temporal Graph Convolutional Recurrent Network for Transportation Flow Estimation.

Accurate estimation of transportation flow is a challenging task in Intelligent Transportation Systems (ITS). Transporting data with dynamic spatial-temporal dependencies elevates transportation flow forecasting to a significant issue for operational planning, managing passenger flow, and arranging for individual travel in a smart city. The task is challenging due to the composite spatial dependency on transportation networks and the non-linear temporal dynamics with mobility conditions changing over time. To address these challenges, we propose a Spatial-Temporal Graph Convolutional Recurrent Network (ST-GCRN) that learns from both the spatial stations network data and time series of historical mobility changes in order to estimate transportation flow at a future time. The model is based on Graph Convolutional Networks (GCN) and Long Short-Term Memory (LSTM) in order to further improve the accuracy of transportation flow estimation. Extensive experiments on two real-world datasets of transportation flow, New York bike-sharing system and Hangzhou metro system, prove the effectiveness of the proposed model. Compared to the current state-of-the-art baselines, it decreases the estimation error by 98% in the metro system and 63% in the bike-sharing system.

Performance Evaluation of Machine Learning Nave Bayes Algorithms for Network Traffic Classification

Network Traffic Classification (NTC) is an important field for different network statistics like management, malware detection and other paramount constraints. Artificial Intelligence (AI) including Machine Learning (ML) and Deep Learning (DL), on the other hand, plays a very important field nowadays due to its significant capabilities with an extremely different fields and for complex problems. ML, specifically, provides tools in the most important network fields like traffic management, security, etc. Performance evaluation is a very important aspect of any system. This research paper provides a method for NTC using ML Nave Bayes (NB) algorithm in terms of Bernoulli, Multinominal and Gaussian for classifying captured network traffic in two different datasets and perform a performance evaluation and a comparison among these algorithms. The first dataset is a VPN-nonVPN (ISCXVPN2016) dataset while the second is a packet-captured regular Wi-Fi traffic flow dataset for video browsing on the web. Results were comparable in terms of f1-score, accuracy and processing time. Bernoulli NB provides average 93.05% accuracy with 742 ms, Multinominal NB provides average 98.78% accuracy with 78.3 ms processing time and finally, Gaussian NB provides average 69.14% with 46.85 ms processing time.

Flow evaluation software for four-dimensional flow MRI: a reliability and validation study

PurposeFour-dimensional time-resolved phase-contrast cardiovascular magnetic resonance imaging (4D flow MRI) enables blood flow quantification in multiple vessels, which is crucial for patients with congenital heart disease (CHD). We investigated net flow volumes in the ascending aorta and pulmonary arteries by four different postprocessing software packages for 4D flow MRI in comparison with 2D cine phase-contrast measurements (2D PC).Material and methods4D flow and 2D PC datasets of 47 patients with biventricular CHD (median age 16, range 0.6–52 years) were acquired at 1.5 T. Net flow volumes in the ascending aorta, the main, right, and left pulmonary arteries were measured using four different postprocessing software applications and compared to offset-corrected 2D PC data. Reliability of 4D flow postprocessing software was assessed by Bland–Altman analysis and intraclass correlation coefficient (ICC). Linear regression of internal flow controls was calculated. Interobserver reproducibility was evaluated in 25 patients.ResultsCorrelation and agreement of flow volumes were very good for all software compared to 2D PC (ICC ≥ 0.94; bias ≤ 5%). Internal controls were excellent for 2D PC (r ≥ 0.95, p < 0.001) and 4D flow (r ≥ 0.94, p < 0.001) without significant difference of correlation coefficients between methods. Interobserver reliability was good for all vendors (ICC ≥ 0.94, agreement bias < 8%).ConclusionHaemodynamic information from 4D flow in the large thoracic arteries assessed by four commercially available postprocessing applications matches routinely performed 2D PC values. Therefore, we consider 4D flow MRI-derived data ready for clinical use in patients with CHD.

Predicting gas diffusion layer flow information in proton exchange membrane fuel cells from cross-sectional data using deep learning methods

Obtaining transient flow field information of gas diffusion layers (GDLs) is a crucial issue for improving the performance of proton exchange membrane fuel cells (PEMFCs). While physically-based methods, such as effective medium theory and solving partial differential equations, can be utilized to calculate fluid flow in GDL, the computational cost associated with such methods remains substantial. In this study, a 3D multiphase flow dataset of GDLs is obtained using fluid of volume (VOF) calculation, and the 3D data is sliced. The resulting cross-sectional data are used to establish a convolutional neural network (CNN) model based on two-dimensional (2D) data for predicting fluid flow in GDLs. The reliability of predicting the saturation information of 3D GDLs using 2D cross-sectional data is verified and discussed. Subsequently, a comparison is made between the accuracy and computational cost of predicting the saturation of 3D GDLs using 2D and 3D CNN models. The results indicate that using 25 cross-sectional images provide accurate predictions of GDL saturation in 3D. When using cross-sectional images as inputs, 100 × 100 images are found to be more representative than 50 × 50 images, resulting in an average increase of 1.86% and 13.36% in R2 and RMSE, respectively. While the 3D CNN model outperforms the 2D CNN model in predicting GDL saturation in 3D by only 0.62% in terms of R2 score, its computational cost is two orders of magnitude higher. These findings suggest that 2D GDL cross-sectional images can also be used for predicting 3D GDL flow information and effectively reducing computational costs. The findings of this study provide a profound insight into the GDL flow phenomena and contribute to the development of more efficient and accurate fuel cell models.

Multi-scale spatial-temporal aware transformer for traffic prediction

Traffic prediction is an important part of smart city management. Accurate traffic prediction can be deployed in urban applications such as congestion alerting and route planning, thus providing sustainable services to the public or relevant departments. Although some improvements have been made in existing traffic prediction methods, there are challenges due to the following: (1) Time series has multi-scale nature, that is, from different scale time ranges, traffic flow changes show different trends; (2) Spatial heterogeneity, meaning that traffic conditions in similar functional areas are usually similar. This task remains difficult. To address the above challenges, we propose a new spatial-temporal prediction method, namely Multi-Scale Spatial-Temporal Aware Transformer (MSSTAT), which is a Transformer architecture with multi-scale characteristics. Specifically, compared to the input of encoder, the input of different decoder layers has different scale information, MSSTAT synchronizes model the connection between time steps and scale information by a kind of Parallel Cross Multi-Head Attention, which gives each time step several times the perceived field while also being able to weaken the impact brought by anomaly point. In addition, to add connections between regions with similar functions, we map the traffic data of each node as a probability distribution and then measure the similarity between the nodes by the Wasserstein Distance, which leads to our proposed spatial-temporal aware adjacency matrix. Experimental results on four traffic flow datasets show that MSSTAT outperforms the state-of-the-art baseline.

What determines Tanzania's trade? A gravity approach

Based on the gravity model, this study examines the determinants of trade flows between Tanzania and its trading partners and attempts to predict the country's trade potential. Using a panel dataset of Tanzania's trade flows of 100 countries from 2000 to 2016, this study finds that the results are consistent with the predictions of the gravity model, with positive and negative coefficients for gross domestic product and distance, respectively. The predicted trade potential reveals that neighboring countries and some middle and superpowers have significant trade potential. Considering the important contribution of trade to economic growth and notwithstanding Tanzania's active membership in regional economic communities, such as the East African Cooperation, Southern African Development Cooperation, and the anticipated engagement in the African Continental Free Trade Agreement, this study provides policy recommendations to reduce trade barriers and diversify production to promote intra-African trade and enhance trade with the rest of the world.

Data-Driven Deployment of Cargo Drones: A U.S. Case Study Identifying Key Markets and Routes

Electric and autonomous aircraft (EAA) are set to disrupt current cargo-shipping models. To maximize the benefits of this technology, investors and logistics managers need information on target commodities, service location establishment, and the distribution of origin–destination pairs within EAA’s range limitations. This research introduces a three-phase data-mining and geographic information system (GIS) algorithm to support data-driven decision-making under uncertainty. Analysts can modify and expand this workflow to scrutinize origin–destination commodity flow datasets representing various locations. The algorithm identifies four commodity categories contributing to more than one-third of the value transported by aircraft across the contiguous United States, yet only 5% of the weight. The workflow highlights 8 out of 129 regional locations that moved more than 20% of the weight of those four commodity categories. A distance band of 400 miles among these eight locations accounts for more than 80% of the transported weight. This study addresses a literature gap, identifying opportunities for supply chain redesign using EAA. The presented methodology can guide planners and investors in identifying prime target markets for emerging EAA technologies using regional datasets.

PDC-Net+: Enhanced Probabilistic Dense Correspondence Network

Establishing robust and accurate correspondences between a pair of images is a long-standing computer vision problem with numerous applications. While classically dominated by sparse methods, emerging dense approaches offer a compelling alternative paradigm that avoids the keypoint detection step. However, dense flow estimation is often inaccurate in the case of large displacements, occlusions, or homogeneous regions. In order to apply dense methods to real-world applications, such as pose estimation, image manipulation, or 3D reconstruction, it is therefore crucial to estimate the confidence of the predicted matches. We propose the Enhanced Probabilistic Dense Correspondence Network, PDC-Net+, capable of estimating accurate dense correspondences along with a reliable confidence map. We develop a flexible probabilistic approach that jointly learns the flow prediction and its uncertainty. In particular, we parametrize the predictive distribution as a constrained mixture model, ensuring better modelling of both accurate flow predictions and outliers. Moreover, we develop an architecture and an enhanced training strategy tailored for robust and generalizable uncertainty prediction in the context of self-supervised training. Our approach obtains state-of-the-art results on multiple challenging geometric matching and optical flow datasets. We further validate the usefulness of our probabilistic confidence estimation for the tasks of pose estimation, 3D reconstruction, image-based localization, and image retrieval. Code and models are available at https://github.com/PruneTruong/DenseMatching.

Competition in Multi-Airport Regions: Measuring airport catchments through spatial interaction models

The paper makes use of a novel dataset of the surface access flows of passengers departing from the four main airports surrounding London to construct a spatial interaction model for the region. This model explains the spatial variability in the flows through four separate components being [1] spatial separation between the origin of the flow and the destination airport, [2] the demand characteristics at the origin, [3] the attractiveness of the service offered by the airport, and [4] the presence of intervening opportunities. A spatial econometric approach is taken in the modelling to account for the presence of spatial dependence in the data.The output of the model reveals a strong distance decay effect, where the level of interaction between origins and airports displays a negative spatial gradient. Each of the four airports dominate passenger flows in their immediate vicinity, with the market in the region being hotly contested in central London. All four components of the model are useful in explaining spatial variation in passenger flows, demonstrating the efficacy of this approach in considering how airports source their demand in a Multi-Airport Region. The performance of the model is superior when considering passenger surface access flows for scheduled flights, while the explanatory power is reduced for flows associated with chartered flights. Having verified the applicability of the model, it is possible for policy makers to utilise the approach to consider such issues of expanding, limiting or reducing capacity at existing airports in the region, establishing new airport facilities, and the effect of population growth on the geographic form of airport demand.

Passenger flow anomaly detection in urban rail transit networks with graph convolution network-informer and Gaussian Bayes models.

Passenger flow anomaly detection in urban rail transit networks (URTNs) is critical in managing surging demand and informing effective operations planning and controls in the network. Existing studies have primarilyfocused on identifying the source of anomalies at a single station byanalysing the time-series characteristics of passenger flow. However, they ignoredthe high-dimensional and complex spatial features of passenger flow and thedynamic behaviours of passengers in URTNs during anomaly detection. This article proposes a novel anomaly detection methodology based on a deeplearning framework consisting of a graph convolution network (GCN)-informer model and a Gaussian naive Bayes model. The GCN-informer model is used to capture the spatial and temporal features of inbound and outbound passenger flows, and it is trained on normal datasets. The Gaussian naive Bayes model is used to construct a binary classifier for anomaly detection, and its parameters are estimated by feeding the normal and abnormal test data into the trained GCN-informer model. Experiments are conducted on a real-world URTN passenger flow dataset from Beijing. The results show that the proposed framework has superior performancecompared to existing anomaly detection algorithms in detecting network-level passenger flow anomalies. This article is part of the theme issue 'Artificial intelligence in failure analysis of transportation infrastructure and materials'.

Traffic Flow Prediction Based on Hybrid Deep Learning Models Considering Missing Data and Multiple Factors

In the case of missing data, traffic forecasting becomes challenging. Many existing studies on traffic flow forecasting with missing data often overlook the relationship between data imputation and external factors. To address this gap, this study proposes two hybrid models that incorporate multiple factors for predicting traffic flow in scenarios involving data loss. Temperature, rainfall intensity and whether it is a weekday will be introduced as multiple factors for data imputation and forecasting. Predictive mean matching (PMM) and K-nearest neighbor (KNN) can find the data that are most similar to the missing values as the interpolation value. In the forecasting module, bidirectional long short-term memory (BiLSTM) network can extract bidirectional time series features, which can improve forecasting accuracy. Therefore, PMM and KNN were combined with BiLSTM as P-BiLSTM and K-BiLSTM to forecast traffic flow, respectively. Experiments were conducted using a traffic flow dataset from the expressway S6 in Poland, considering various missing scenarios and missing rates. The experimental results showed that the proposed models outperform other traditional models in terms of prediction accuracy. Furthermore, the consideration of whether it is a working day further improves the predictive performance of the models.

Improving airport arrival flow prediction considering heterogeneous and dynamic network dependencies

Predicting airport arrival flow serves as a crucial technique in air traffic flow management. Given the unique operational characteristics of air traffic systems, airport arrival flow simultaneously presents complex dynamics in spatial–temporal dimensions, specific spatial heterogeneity, non-rigid periodicity, and robust plannability. These factors pose challenges to existing modeling methods in achieving optimal performance. To address these challenges, we propose a novel large-range air traffic flow prediction model to forecast airport arrival flow. More specifically, a dynamic multi-graph neural network is designed to automatically capture the time-evolving and heterogeneous spatial correlations using convolution and attention operations. In terms of the temporal dimension, a temporal-aware attention module is constructed to extract the temporal transitions of traffic data, considering both the local context and stationarity of the traffic representation sequence. Furthermore, a prior-guided recalibration fusion module is employed to explicitly incorporate the prior knowledge, including historical-periodic and future-scheduled arrival flow features, to recalibrate the temporal module prediction results, thereby enhancing the prediction accuracy. Experimental results on a real-world airport traffic flow dataset demonstrate that the proposed method outperforms the state-of-the-art baselines, and all proposed technical modules contribute to desired performance improvements.

Identification and restoration of hydrological processes alteration during the fish spawning period

The hydrological processes play an important role in stimulating fish spawning behavior. Changes in the natural hydrological processes will alter the populations and distribution of fish, which may have a negative impact on the native aquatic organisms. The aim of this study is to identify the alteration of the water rising process during the fish spawning period and to construct an ecological flow optimization model to restore the water rising conditions for fish reproduction. The Mann–Kendall test and the sliding t-test were used to detect the mutation year of the mean daily flow data sets in the fish spawning period in each monitoring year. Then the data sets can be divided into pre-altered and post-altered periods. The water rising process was characterized by the water rising processes count, the duration, the daily flow increase rate, the date of the water rising process, and the initial water rising flow. The changes in hydrological processes in the middle reaches of the Yangtze River were investigated by comparing the post-altered and pre-altered characteristic parameters. Furthermore, we integrated the statistical values of the five characteristic parameters in pre-altered into an ecological flow optimization model to simulate the natural water rising processes for the spawning of the Four Major Chinese Carps (FMCC) and Chinese Sturgeon (CS). The analysis showed that after the hydrological mutation year, the duration and the initial water rising flow in the FMCC spawning season were increased, with hydrological alteration degrees of 63.10% and 70.16%, respectively; however, the daily flow increase rate was significantly decreased, with hydrological alteration of 86.50%. During the CS spawning season, the water rising processes count and the initial water rising flow were dramatically altered parameters, with hydrological alteration degrees of 50.86% and 83.27%, respectively. The former parameter increased, but the latter decreased significantly in the post-altered period. To induce the spawning activity of FMCC and CS, appropriate ecological flows and hydrological parameters were proposed. These results showed that during the spawning seasons of FMCC and CS, the hydrological processes of the middle reaches of the Yangtze River changed significantly. Therefore, ecological flow must be ensured through ecological operation of upstream reservoirs to provide suitable spawning conditions in target fish spawning grounds.

A Framework for Utilizing Permutational Multiple Analysis of Variance as a Precursor for Nonparametric Statistical Learning with Cyber Network Data

INTRODUCTION: Although scientific hypothesis testing methodologies are well established, their application to falsifiable hypothesis testing for assessing causal relationships potentially identified by machine learning and artificial intelligence models is rare due to the primarily nonparametric statistical nature of these systems.OBJECTIVES: The primary objective of this study is to demonstrate the potential for applying nonparametric statistical tests to a mixed qualitative and quantitative cyber network dataset as a method to pre-assess the feasibility of applying forms of statistical hypothesis testing before a machine learning algorithm models the data.METHODS: A mixture of permuted analysis of variance models augmented by the use of transformed non-Euclidean multivariate distances between curated dependent variable classes produced this research data. Quasi-experimental data from an enclosed laboratory environment utilizing a monitored, locally unrestricted network that introduced known Internet of Things (IoT) malware software supplied network flow events.RESULTS: A PERMANOVA model was executed against 62,000 records of the network flow observations, using Euclidean distance measurements with variable-dependent relationship ordering, using terms added sequentially (first to last) in the order encountered in the raw network flow dataset, using 200 permutations. This precursor test resulted in a p-value for the PERMANOVA model that incorporated terms added sequentially of 0.02985, providing an F value of 0.00017 with which to determine the ratio of explained to unexplained variance. Utilizing an analysis of the F values for all of the residuals, we show 29,998 degrees of freedom with a residual F model score of 0.99983, indicating that there is a strong proportion of explained to unexplained variance across all of the independent variables contained in the model. The model is thus statistically significant with a p-value below the alpha test statistic of 0.05.CONCLUSION: This research has demonstrated that it is possible to apply tests of falsifiability that incorporate reproducible methods into the quasi-experiment design and apply this to the field of machine learning. Applied to AI/ML (artificial intelligence/machine learning) models, this pre-assessment methodology supports the appropriateness of cyber network flow datasets in which a final test of statistical significance would be required. The authors believe that this represents a substantially useful precursor assessment stage for the suitability and reliability of the utilization of any nonparametric statistical learning algorithms applied to cyber network data predictive analytics.

Comparative Assessment of Sap Flow Modeling Techniques in European Beech Trees: Can Linear Models Compete with Random Forest, Extreme Gradient Boosting, and Neural Networks?

Transpiration and sap flow are physiologically interconnected processes that regulate nutrient and water uptake, controlling major aspects of tree life. They hold special relevance during drought, where wrecked sap flow can undermine overall tree growth and development. The present study encompasses five-year (2012–2015 and 2017) sap flow datasets on European beech (Fagus sylvatica). Four different techniques were used for sap flow modeling, namely, a linear model (LM), random forest (RF), extreme gradient boosting machine (XGBM), and neural networks (NN). We used six variants (Variants 1–6) differing in the captured conditions and the dataset size. The ‘prediction power’ was the ratio of the predicted and observed sap flow. We found the LM had the maximum prediction power for the overall sap flow in beech trees with 1 h shift of global radiation. In the reaming variants, the LM provided comparable prediction power to RF and XGBM. At the same time, NN exhibited relatively poor prediction power over other machine learning models. The study supports an easier-to-apply and computationally simpler approach (LM) to assess sap flow over more sophisticated machine learning approaches (RF, XGBM, and NN).

Mul-DesLSTM: An integrative multi-time granularity deep learning prediction method for urban rail transit short-term passenger flow

It is critical for the management and control of urban rail transit (URT) to be able to predict passenger flow accurately and in real time. Considering that the high-resolution data aggregated by the automatic fare collection (AFC) system is wasted, this paper analyzes the problem of applying a multi-time granularity passenger flow data fusion forecasting process. First, we examine the challenge of constructing a dataset of passenger flow data with different time granularities. Thus, an algorithm is proposed for selecting passenger flow datasets with multi-time granularity. Furthermore, a multi-time granularity dense residual network (Mul-DesLSTM) with a dense residual structure and LSTM (long short-term memory) as the predictor is constructed, inspired by a residual network. Using Mul-DesLSTM, finer-grained passenger flow features can be fused layer by layer while maintaining the accuracy of traditional single-granularity passenger flow predictions. Lastly, Mul-DesLSTM is applied to the URT system of Shanghai, China, and compared with baselines. As a result, the proposed Mul-DesLSTM outperforms the baselines with LSTM as a predictor and state-of-the-art model. When the predicted time granularity is 30 min, compared to the single-time granularity LSTM network, the mean absolute error, root mean square error, and symmetric mean absolute percentage error can be reduced by 51%, 63%, and 15%, respectively. The results can serve as a reference and basis for the operation and management of URT systems.