A graph deep learning method for short‐term traffic forecasting on large road networks

Short‐term traffic flow prediction on a large‐scale road network is challenging due to the complex spatial–temporal dependencies, the directed network topology, and the high computational cost. To address the challenges, this article develops a graph deep learning framework to predict large‐scale network traffic flow with high accuracy and efficiency. Specifically, we model the dynamics of the traffic flow on a road network as an irreducible and aperiodic Markov chain on a directed graph. Based on the representation, a novel spatial–temporal graph inception residual network (STGI‐ResNet) is developed for network‐based traffic prediction. This model integrates multiple spatial–temporal graph convolution (STGC) operators, residual learning, and the inception structure. The proposed STGC operators can adaptively extract spatial–temporal features from multiple traffic periodicities while preserving the topology information of the road network. The proposed STGI‐ResNet inherits the advantages of residual learning and inception structure to improve prediction accuracy, accelerate the model training process, and reduce difficult parameter tuning efforts. The computational complexity is linearly related to the number of road links, which enables citywide short‐term traffic prediction. Experiments using a car‐hailing traffic data set at 10‐, 30‐, and 60‐min intervals for a large road network in a Chinese city shows that the proposed model outperformed various state‐of‐the‐art baselines for short‐term network traffic flow prediction.     

Methodology for spatio‐temporal predictions of traffic counts across an urban road network and generation of an on‐road greenhouse gas emission inventory

On‐road emission inventories in urban areas have typically been developed using traffic data derived from travel demand models. These approaches tend to underestimate emissions because they often only incorporate data on household travel, not including commercial vehicle movements, taxis, ride hailing services, and other trips typically underreported within travel surveys. In contrast, traffic counts embed all types of on‐road vehicles; however, they are only conducted at selected locations in an urban area. Traffic counts are typically spatially correlated, which enables the development of methods that can interpolate traffic data at selected monitoring stations across an urban road network and in turn develop emission estimates. This paper presents a new and universal methodology designed to use traffic count data for the prediction of periodic and annual volumes as well as greenhouse gas emissions at the level of each individual roadway and for multiple years across a large road network. The methodology relies on the data collected and the spatio‐temporal relationships between traffic counts at various stations; it recognizes patterns in the data and identifies locations with similar trends. Traffic volumes and emissions prediction can be made even on roads where no count data exist. Data from the City of Toronto traffic count program were used to validate the output of various algorithms, indicating robust model performance, even in areas with limited data.     

Short‐Term Traffic Speed Prediction for an Urban Corridor

Short‐term traffic speed prediction is one of the most critical components of an intelligent transportation system (ITS). The accurate and real‐time prediction of traffic speeds can support travellers’ route choices and traffic guidance/control. In this article, a support vector machine model (single‐step prediction model) composed of spatial and temporal parameters is proposed. Furthermore, a short‐term traffic speed prediction model is developed based on the single‐step prediction model. To test the accuracy of the proposed short‐term traffic speed prediction model, its application is illustrated using GPS data from taxis in Foshan city, China. The results indicate that the error of the short‐term traffic speed prediction varies from 3.31% to 15.35%. The support vector machine model with spatial‐temporal parameters exhibits good performance compared with an artificial neural network, a k‐nearest neighbor model, a historical data‐based model, and a moving average data‐based model.     

Network‐wide traffic speed forecasting: 3D convolutional neural network with ensemble empirical mode decomposition

Accurate traffic speed forecasting is one of the most critical tasks in proactive traffic management and the deployment of advanced traveler information systems. This paper proposes a hybrid forecasting approach named DeepEnsemble by integrating the three‐dimensional convolutional neural network (3D CNN) with ensemble empirical mode decomposition (EEMD). There are four steps in this hybrid approach. First, EEMD is adopted to decompose the complex traffic speed time series data with noise into several intrinsic mode functions (IMFs) and a residue. Second, a three‐dimensional tensor is established and fed into 3D CNN for prediction. Third, the output of 3D CNN prediction is obtained by a linear combination of the results of all components. Finally, the 3D CNN prediction output, external features, and historical features are fused to predict the network‐wide traffic speed simultaneously. The proposed DeepEnsemble approach is tested on the three‐month traffic speed series data of a real‐world large‐scale urban expressway network with 308 traffic flow detectors in Beijing, China. The experimental results indicate that DeepEnsemble outperforms the state‐of‐the‐art network‐wide traffic speed forecasting models. 3D CNN learns temporal, spatial, and depth information better than 2D CNN. Moreover, forecasting accuracy can be improved by employing EEMD. DeepEnsemble is a promising model with scalability and portability for network‐wide traffic speed prediction and can be further extended to conduct traffic status monitoring and congestion mitigation strategies.     

Development of Recurrent Neural Network Considering Temporal‐Spatial Input Dynamics for Freeway Travel Time Modeling

Abstract: The artificial neural network (ANN) is one advance approach to freeway travel time prediction. Various studies using different inputs have come to no consensus on the effects of input selections. In addition, very little discussion has been made on the temporal–spatial aspect of the ANN travel time prediction process. In this study, we employ an ANN ensemble technique to analyze the effects of various input settings on the ANN prediction performances. Volume, occupancy, and speed are used as inputs to predict travel times. The predictions are then compared against the travel times collected from the toll collection system in Houston. The results show speed or occupancy measured at the segment of interest may be used as sole input to produce acceptable predictions, but all three variables together tend to yield the best prediction results. The inclusion of inputs from both upstream and downstream segments is statistically better than using only the inputs from current segment. It also appears that the magnitude of prevailing segment travel time can be used as a guideline to set up temporal input delays for better prediction accuracies. The evaluation of spatiotemporal input interactions reveals that past information on downstream and current segments is useful in improving prediction accuracy whereas past inputs from the upstream location do not provide as much constructive information. Finally, a variant of the state‐space model (SSNN), namely time‐delayed state‐space neural network (TDSSNN), is proposed and compared against other popular ANN models. The comparison shows that the TDSSNN outperforms other networks and remains very comparable with the SSNN. Future research is needed to analyze TDSSNN's ability in corridor prediction settings.     

Analytical compressive stress–strain model for concrete confined with high‐strength multiple‐tied‐spiral transverse reinforcement

The behavior of reinforced concrete members subjected to seismic loads is mainly based on the ultimate strength of concrete and its ductility. Based on this, an additional configuration of transverse reinforcement using high‐strength multiple‐tied‐spiral was proposed to improve the strength and ductility of concrete. In this paper, an experimental study of a number of axial loading tests on reinforced concrete columns confined with high‐strength multiple‐tied‐spiral transverse reinforcement is described. The effects of spacing of circular spiral and rectangular hoop, the confined area of circular spiral and concrete strength on axial behavior of confined concrete were investigated. The formulas of confined compressive strength and corresponding axial strain, factor to control the slope of descending branch, and stress in high‐strength circular spiral at confined strength are proposed based on the test results. A stress–strain model is also proposed that is found to give reasonably good prediction of the experimental behavior of reinforced concrete (RC) columns confined by high‐strength multiple‐tied‐spiral transverse reinforcement.     

Seismic Data‐Driven Identification of Linear Models for Building Structures Using Performance and Stabilizing Objectives

A two‐stage dual‐objective structural identification method is presented in this article. The complexity of the identification of story‐level physical models for large‐scale building structures is first addressed through a comparative study. A stiffness variation‐based stabilizing objective is proposed to be necessarily incorporated into iterative optimization with the classical performance objectives to improve the model feasibility, and an area‐type evaluation index is subsequently proposed for the stopping criteria. Accordingly, a two‐stage differential evolution‐based dual‐objective optimization framework is presented for the computation of Pareto fronts for nondominated candidate solutions. Then, the proposed method is investigated using two illustrative examples, including a nine‐story benchmark structure, and a real‐world seven‐story reinforced concrete structure. A series of condensed models are identified from the nondominated solutions on the Pareto front. The prediction performance of the single‐objective optimal model and the dual‐objective acceptable models is compared using the overall discrepancies of acceleration, interstory drift, and modal properties, within both estimation and validation cases. Incorporation of the noise effect into the method is finally studied and discussed.     

A spatio‐temporal measure of spatial dependence: An example using real estate data*

The computation of Moran's I index and his statistics test relies mainly on an exogenous specification of a spatial weights matrix. However, the exogenous weights matrix is usually developed in a strictly spatial context, even when data are collected over time. This paper develops a spatio‐temporal weights matrix and uses the new definition to evaluate spatial dependence using Moran's I index applied to real estate data for Québec City from 1986 to 1996. The results are compared with the original Moran's I index using a strictly spatial weights matrix specification based on Euclidian distance or contiguity. The findings suggest that ignoring the temporal dimension could lead to misinterpretation of the ‘real’ measure of spatial dependence over time. However, the time dimension cannot explain the total spatial autocorrelation since the Moran's I index is still significant even when adjusting for time consideration. The differences between the estimated indices and statistics depend on the structure of the spatial and the temporal weights matrices that are used to construct the complete spatio‐temporal weights matrix.     

Experimental validation of a deep neural network—Sparse representation classification ensemble method

In the current study, a new pattern recognition‐based damage detection technique is developed using the frequency response function of the structure. Principal component analysis is employed as an authoritative feature extraction method for dimensional reduction of the measured frequency response function data and constructing distinct feature patterns. Subsequently, as a novel approach, an ensemble of 2 powerful classifiers containing deep neural networks and couple sparse coding classification is utilized for damage prediction of the structure because there is no individual optimal classifier for all the problems. Verification of the proposed method is evaluated by an aluminum beam experimental setup besides a numerical 3D finite element model of a truss bridge. Damage detection results elucidate that the ensemble method decisions are much more accurate compared with the individual classifier decision. The proposed ensemble method verifies to be a novel, robust, and powerful damage detection process.     

Structure‐aware 3D reconstruction for cable‐stayed bridges: A learning‐based method

A powerful deep learning‐based three‐dimensional (3D) reconstruction method for reconstructing structure‐aware semantic 3D models of cable‐stayed bridges is proposed herein. Typically, conventional bridge semantic 3D model reconstruction methods are not robust when low‐quality point clouds are used. Furthermore, they are suited particularly for their respective fields and less generalized for cable‐stayed bridges. Hence, a structure‐aware learning‐based cable‐stayed bridge 3D reconstruction framework is proposed. The encoder part of the network uses both multiview images and a photogrammetric point cloud as input, whereas the decoder part uses a recursive binary tree network to model a high‐level structural relation graph and low‐level 3D geometric shapes. Two actual cable‐stayed bridges are employed as examples to evaluate the proposed method. Test results demonstrate that the proposed method successfully reconstructs the bridge model with structural components and their relations. Quantitative results indicate that the predicted models achieved an average F₁ score of 99.01%, a Chamfer distance of 0.0259, and a mesh‐to‐cloud distance of 1.78 m. The achieved result is similar to that obtained using the manual reconstruction approach in terms of component‐wise accuracy, and it is considerably better than that obtained using the manual approach in terms of spatial accuracy. In addition, the proposed recursive binary tree network is robust to noise and partial scans. The potential applications of the obtained 3D bridge models are discussed.     

Wavelet‐Based Denoising for Traffic Volume Time Series Forecasting with Self‐Organizing Neural Networks

Abstract: In their goal to effectively manage the use of existing infrastructures, intelligent transportation systems require precise forecasting of near‐term traffic volumes to feed real‐time analytical models and traffic surveillance tools that alert of network links reaching their capacity. This article proposes a new methodological approach for short‐term predictions of time series of volume data at isolated cross sections. The originality in the computational modeling stems from the fit of threshold values used in the stationary wavelet‐based denoising process applied on the time series, and from the determination of patterns that characterize the evolution of its samples over a fixed prediction horizon. A self‐organizing fuzzy neural network is optimized in its configuration parameters for learning and recognition of these patterns. Four real‐world data sets from three interstate roads are considered for evaluating the performance of the proposed model. A quantitative comparison made with the results obtained by four other relevant prediction models shows a favorable outcome.     

A Two‐Dimension Dynamic Bayesian Network for Large‐Scale Degradation Modeling with an Application to a Bridges Network

Modeling the stochastic evolution of a large‐scale fleet or network generally proves to be challenging. This difficulty may be compounded through complex relationships between various assets in the network. Although a great number of probabilistic graph‐based models (e.g., Bayesian networks) have been developed recently to describe the behavior of single assets, one can find significantly fewer approaches addressing a fully integrated network. It is proposed an extension to the standard dynamic Bayesian network (DBN) by introducing an additional dimension for multiple elements. These elements are then linked through a set of covariates that translate the probabilistic dependencies. A Markov chain is utilized to model the elements and develop a distribution‐free mathematical framework to parameterize the transition probabilities without previous data. This is achieved by borrowing from Cooke's method for structured expert judgment and also applied to the quantification of the covariate relationships. Some metrics are also presented for evaluating the sensitivity of information inserted into the covariate DBN where the focus is given on two specific types of configurations. The model is applied to a real‐world example of steel bridge network in the Netherlands. Numerical examples highlight the inference mechanism and show the sensitivity of information inserted in various ways. It is shown that information is most valuable very early and decreases substantially over time. Resulting observations entail the reduction of inference combinations and by extension a computational gain to select the most sensitive pieces of information.     

Roadway Asset Inspection Sampling Using High‐Dimensional Clustering and Locality‐Sensitivity Hashing

A high‐dimensional clustering‐based sampling method for roadway asset condition inspection is proposed in this study. The method complements existing literature by selecting sample roadway segments that contain multiple types of assets (e.g., signage, shoulder work, pavement marking, etc.) for the accurate estimation of their respective level of maintenance (LOMs). This is consistent with the standard maintenance procedure as inspection activities are often conducted on roadway segment basis. The proposed method consists of three components: current condition estimation, similarity matrix construction, and stratification. Current condition estimation predicts assets’ “current condition” by considering historical inspection records. Similarity matrix construction represents the core piece of the sampling framework, which employs locality‐sensitive hashing algorithm to define the similarity between segments. The stratification process is implemented with spectral clustering, which assigns segments into clusters based on the similarity matrix. The proposed method outperforms simple random sampling, which is widely used in practice, especially under the circumstances where LOM varies greatly across assets. The main highlight of the proposed method is the ability to select sample segments with multiple types of assets that are representative of their respective LOMs of the full inventory, which directly translates into an efficient maintenance activity management. The method is implemented using asset inspection records in the state of Utah from September, 2014 to March 2016. It represents a potentially useful tool for agencies to effectively conduct asset inspection and can be easily adopted for choosing samples containing multiple features.     

Mode shape linearization and correction in coupled dynamic analysis of wind‐excited tall buildings

The three‐dimensional mode shapes found in modern tall buildings complicate the use of the high‐frequency base balance (HFBB) technique in wind tunnel testing for predicting their wind‐induced loads and effects. The linearized‐mode‐shape (LMS) method was recently proposed to address some of the complications in the calculation of the generalized wind forces, which serve as the input to modal analysis for predicting wind‐induced dynamic responses of tall buildings. An improved LMS method, called the advanced linearized‐mode‐shape (ALMS) method, is developed in this paper by introducing torsional mode shape corrections to account for the partial correlation of torques over building height. The ALMS method has been incorporated into the accurate complete quadratic combination method in the coupled dynamic analysis to form a comprehensive procedure for the determination of equivalent static wind loads (ESWLs) for structural design of complex tall buildings. The improved accuracy in the prediction of generalized forces by the ALMS method has been validated by a 60‐storey benchmark building with multiple‐point simultaneous pressure measurements. A practical 40‐storey residential building with significant swaying and torsional effects is presented to demonstrate the effectiveness of the proposed wind load and response analysis procedure based on the HFBB data. Copyright © 2010 John Wiley & Sons, Ltd.     

A Two‐Dimensional Plum‐Blossom Sensor Array‐Based Multiple Signal Classification Method for Impact Localization in Composite Structures

The growing use of composite materials on aircraft structures has attracted much attention for the impact monitoring as a kind of structural health monitoring method. Uniform linear sensor array (ULSA)‐based multiple signal classification (MUSIC) technology is a promising method because of its directional scanning ability and easy arrangement of the sensor array. However, the monitoring range of ULSA‐based MUSIC method is 0°–180°, and its beamforming properties degrade at angles close to 0° and 180°. Besides, the ULSA‐based MUSIC methods proposed require the knowledge of the direction dependent velocity profile obtained by additional experiments. This article presents a novel two‐dimensional (2‐D) plum‐blossom sensor array (PBSA)‐based MUSIC method. First, the velocity propagating at the specific direction is estimated by impact signal itself using PBSA directly. Second, 2‐D PBSA‐based MUSIC method well realizes omnidirectional 0°–360° impact localization of composite structures. Experimental results show its successful performance on epoxy laminate plate and complex composite structure.     

Slope stability prediction using ensemble learning techniques: A case study in Yunyang County,Chongqing, China

Slope stability prediction plays a significant role in landslide disaster prevention and mitigation. This study develops an ensemble learning-based method to predict the slope stability by introducing the random forest(RF) and extreme gradient boosting(XGBoost). As an illustration, the proposed approach is applied to the stability prediction of 786 landslide cases in Yunyang County, Chongqing, China. For comparison, the predictive performance of RF, XGBoost, support vector machine(SVM), and logi...     

A multi‐time‐delay compensation controller using a Takagi–Sugeno fuzzy neural network method for high‐rise buildings with an active mass damper/driver system

An active mass damper/driver (AMD) control system with a single mass has such problems as the excessive weight of the auxiliary mass and the insufficient capacity of its driving equipment. It is necessary to work through multiple subsystems to achieve effective control of high‐rise buildings. However, the time‐delay effect in each subsystem impedes its application in engineering practices. In the paper, an augmented system based on a zero‐order hold is proposed for discrete‐time systems with multiple time delays, and then the system is designed according to the compensation strategy using a classical linear quadratic regulator algorithm. After that, the sample data obtained from the zero‐order hold compensation controller is trained through a Takagi–Sugeno fuzzy neural network method. Finally, a new simplified compensation controller is designed to further shorten the time consuming calculation on the premise of guaranteeing its control effects and parameters. To verify its effectiveness, an AMD system in a high‐rise building is regarded as an example, and the proposed methodology is also applied to an experiment of a four‐story frame. Both results demonstrate that the method can enhance the performance of an AMD system with multiple time delays.     

Fuzzy‐Valued Intensity Measures for Near‐Fault Pulse‐Like Ground Motions

Two fuzzy‐valued (FV) structure‐specific intensity measures (IMs), one based on squared spectral velocity and the other on inelastic spectral displacement, are presented to characterize near‐fault pulse‐like ground motions for performance‐based seismic design and assessment of concrete frame structures. The first IM is designed through fuzzying structural fundamental period to account for the period shift effect due to stiffness degradation, whereas the second IM is developed to take into account higher mode contribution in high‐rise buildings by employing a fuzzy combination of the first two or three modes for the lateral loading pattern in pushover analysis. A benchmark study of three example reinforced concrete frame structures shows that for moderate‐ to medium‐period structures, both of the proposed IMs improve prediction accuracy in comparison with the existing IMs. For short‐period structures, the FV inelastic spectral displacement is the best.     

A Grey System Theory‐Based Default Prediction Model for Construction Firms

As the prediction of construction firm failure is of great importance for owners, contractors, investors, banks, insurance firms, and creditors, previous studies have developed several models for predicting the probability of construction firm default based on financial ratio analysis. However, to be applied, these models require a considerable quantity of data, including normally distributed data, and the models cannot tolerate too many changing factors. Furthermore, most of the approaches produce sample selection biases. To avoid these disadvantages, this study is the first to integrate the grey system theory with all available firm‐year samples during the sample period to provide a new method for predicting the probability of construction firm default. This method not only offers an improved rate of prediction accuracy, but it also offers simpler and clearer procedures as a reference for examining firm default probability and ranks all financial ratios in terms of their level of importance. The research collects and analyzes the financial reports of 92 construction firms in the United States. The proposed model includes only eight ranked variables (financial ratios), and it achieves an 84.8% level of accuracy for predicting construction firm default probability. As a result, practitioners may directly use the model as a means of quickly and conveniently examining their firm default probability with the simple procedures.