Search

This paper provides a comprehensive review of tree-based models and their application in condition assessment and prediction of water, wastewater, and sewer pipe failures. Tree-based models have gained significant attention in recent years due to their effectiveness in capturing complex relationships between parameters of systems and their ability in handling large data sets. This study explores a range of tree-based models, including decision trees and ensemble trees utilizing bagging, boosting, and stacking strategies. The paper thoroughly examines the strengths and limitations of these models, specifically in the context of assessing the pipes' condition and predicting their failures. In most cases, tree-based algorithms outperformed other prevalent models. Random forest was found to be the most frequently used approach in this field. Moreover, the models successfully predicted the failures when augmented with a richer failure data set. Finally, it was identified that existing evaluation metrics might not be necessarily suitable for assessing the prediction models in the water and sewer networks. [ABSTRACT FROM AUTHOR]

Shake-table tests of a full-scale 3-story reinforced concrete frame building were performed at the E-Defense facility (Miki City, Japan) as part of the Tokyo Metropolitan Resilience Project. Building data were collected before and after each shake-table experiment, using light detection and ranging (lidar) scanning technology. Three-dimensional point clouds were generated and processed to investigate the damage features that can be extracted, which are important for postdisaster structural assessment purposes. To this end, this paper discusses aspects pertaining to lidar data processing and registration, calculation of residual displacements and interstory drifts, and concrete cracking and spalling identification and measurement. Most notably, this paper deals with very large data sets, processed to extract structural response information at both global and local scales. Limited multiscale work has been done in the past, and the different sets of challenges that arise have typically been addressed separately. Evidently, this gap needs to be filled before lidar can be effectively used in the field for structural assessment applications. A series of examples in which lidar data are processed and analyzed to assess specific structural components is illustrated, showing that point cloud data can indeed be processed to obtain measurements of global residual drifts or of the extent of visible damage (e.g., crack width and spalling). Although preliminary in nature, the results presented in this paper indicate that collecting postdisaster data using terrestrial lidar scanners has the potential to contribute improving how buildings are assessed in the aftermath of seismic events. [ABSTRACT FROM AUTHOR]

Big data and data-driven analysis could be utilized for traffic management to improve road safety and the performance of transportation systems. This paper introduces a web-based proactive traffic safety management (PATM) and real-time big data visualization tool, which is based on an award-winning system that won the US Department of Transportation (USDOT) Solving for Safety Visualization Challenge and was selected as one of the USDOT Safety Data Initiative (SDI) Beta Tools. State-of-the-art research, especially for real-time crash prediction and PATM, are deployed in this study. A significant amount of real-time data is accessed by the system in order to conduct data-driven analysis, such as traffic data, weather data, and video data from closed-circuit television (CCTV) live streams. Based on the data, multiple modules have been developed, including real-time crash/secondary crash prediction, CCTV-based expedited detection, PATM recommendation, data sharing, and report generation. Both real-time data and the system outputs are visualized at the front end using interactive maps and various types of figures to represent the data distribution and efficiently reveal hidden patterns. Evaluation of the real-time crash prediction outputs is conducted based on one-month real-world crash data and the prediction results from the system. The comparison results indicate excellent prediction performance. When considering spatial-temporal tolerance, the sensitivity and false alarm rate of the prediction results [i.e., high crash potential event (HCPE)] are 0.802 and 0.252, respectively. Current and potential implementation are also discussed in this paper. [ABSTRACT FROM AUTHOR]

In the Chengdu-Chongqing economic circle in China, the recommended bearing capacity of a medium-weathering mudstone foundation is less than the capacity of the rock structure to withstand loads. The aim of this paper was to perform plate loading tests (PLTs) with binocular vision measurement (BVM) to obtain the accurate deformation of the foundation under vertical loading. The results revealed that vertical load–bearing foundations have a failure mode that can be reduced to that of thrust failures for thin bedded structures of mudstone foundations. By analyzing the differences in results between the PLTs and the uniaxial compressive strength tests under the same conditions, a new formalism for calculating foundation-bearing capacity was proposed; there is only a 5% difference compared to in situ tests. The results of this research provide theoretical support and a practical benchmark for the full development of the load-bearing capability of soft red rocks. As the Chengdu-Chongqing economic circle progresses toward the international urban cluster, inevitably, soft rock is used as a foundation retaining layer in the construction of superhigh and super-rise buildings in the region. However, owing to antiquated testing methods and poor intelligibility of the testing equipment, the bearing capacity of such a mass of rock has yet to be fully developed and cannot meet the dual-carbon economic goals and sustainable development of architectural design. Based on this, the study introduces a noncontact method to measure the bearing capacity of soft rock within the traditional method of in situ testing and combines the concept of big data analysis to advance an empirical method for calculating the bearing capacity accurately. Based on the results of this study, the concept of natural foundations can be adopted and the engineering dilemma that the treatment of building foundations cannot be economical and practical can be solved. [ABSTRACT FROM AUTHOR]

Accurate power load forecasting could provide a scientific basis for the rapid response and stable operation of a modern power system. To take advantage of the meteorological big data to improve short-term forecasting accuracy, while considering the nonlinear and spatiotemporal correlation characteristics of the power load data, this paper proposes a short-term power load forecasting method based on meteorological data dimensionality reduction and a hybrid deep neural network. First, the elastic network is used to reduce the dimensions of high-dimensional meteorological big data, eliminate irrelevant meteorological factors, and improve the quality of input data. Then, taking the dimension-reduced meteorological data and historical load data as input, a load forecasting model based on a novel deep neural network is established. This model uses a convolution neural network (CNN) and a bi-directional long short-term memory (BiLSTM) neural network to extract the spatial and temporal correlation features of power load related data, and combines the attention mechanism to enhance the learning weight of the load series in important periods, and adopts residual connection (RC) to optimize the network training speed and alleviate the overfitting problem. Finally, taking the open data set of the New York Independent System Operating Agency (NYISO) as an example, single-step and multi-step advance prediction experiments are carried out to verify the advantages of the proposed method. [ABSTRACT FROM AUTHOR]

As increasing attention is given to the impact of the built environment on citizens' well-being and health, the role and performance of urban parks have to be carefully considered. The research presented in this paper focuses on the efficiency of urban parks concerning their visitor flow and explores an aspect that has not been studied so far: the impact of the built environment surrounding urban parks in attracting visitors to parks. To assess the visitor flow, the research took advantage of the large amount of accessible and accurate data connecting users and spaces: points of interest data (POI) data and Weibo check-in data. The research correlated them to investigate how the land uses and their mix in a buffer area of 400 m around each park affect the quantity of visitor flow into the parks of the city of Suzhou, China. The research first quantified the POI mixed degree of land uses and the visitor quantity based on the Hill Number Index and the social media data; then, it analyzed the sensitivity of different types of parks to different types of POI. After clarifying the quadratic function between visitor intensity and the POI mixed degree, the research found: (1) a clear correlation between the visitor intensity and the land-use mix around the parks; and (2) that different kinds of parks are affected by different groups of uses. Specifically, in the case of community parks, the visitor flow is affected by the land use for daily activities, while the visitor flow of comprehensive parks is affected by uses such as offices and services. These findings provide clear indications for improving urban parks and land-use efficiency. [ABSTRACT FROM AUTHOR]

During subway tunnel construction, the consumed time for each ring (unit of construction progress) is highly dependent on objective factors such as geological conditions and shield performances, as well as subjective reasons including workers' proficiency and contractors' management skills. It is nontrivial to score each ring's efficiency from an objective angle, so that workers and contractors can receive fair evaluations. In this paper, a scoring method driven by construction big data is proposed. First, to resolve the high-dimensionality problem, a Gaussian mixture model (GMM) was employed to cluster rings of similar conditions in a probabilistic style so that detailed information can be retained. Second, the standard time to finish one ring was analyzed, so that each ring can be labeled as fail or pass, and our task can be considered as a classification problem. Third, for each cluster, a broad learning system (BLS) was developed as a classifier due to its advantages of fast computation and incremental learning. Finally, the BLS was trained with real tunneling data of 23,822 rings, and then scorecards were developed, where results of validation and statistical tests suggested that our method outperforms conventional ones. Feedback from the subway company and two compared contractors suggested that the proposed method is fair and practical, and it could reveal management problems that were easily overlooked. For subway shield tunnels, this study proposes a construction progress scoring method driven by big data for subway companies to evaluate the performance of each construction unit, e.g., each ring of the shield tunnel is treated as a sample analysis. First, in view of the difficulty of many geological types and risk variables, the Gaussian mixture method is used to classify the samples and obtain probability that each ring belongs to a certain geological type. Statistical analysis is used to calculate the shield standard time of each geological type, so that the time interval of pass can be determined. Second, a broad learning model with low calculation and updating cost is constructed to fit the mapping relationship between geological factors and construction progress. With this model, each ring can be scored. Finally, through the model training and verification of 23,822 ring data in seven subway intervals, results suggest that the method is practical. Our method can objectively reflect the work efficiency and management ability of different contractors and workers. In addition, suggestions can be elicited to improve the management level of construction units. [ABSTRACT FROM AUTHOR]

Equity is a critical field of study in transportation. The built transportation network does not serve the needs of the population to achieve equal levels of economic vitality and prosperity. Because of these concerns, there has been a recent effort to address these equity issues in the transportation network. This effort coincides with a massive growth in the data that are available for transportation practitioners, known as big data. The growth of data has led to data-driven decision-making to allow for more effective transportation policies and decisions than were afforded with classical methods. However, as the amount of data available has grown, the understanding of the biases within that data and the equity implications of those biases has not. Equity biases are any bias in a data source that produces a negative equity outcome by underrepresenting historically disadvantaged populations. This reframes the concept of data biases by focusing on the equity outcomes of data biases as opposed to the precipitating causes of bias. Understanding, quantifying, and mitigating these equity biases are critical to ensuring the practice of equity is maintained in the transportation field. This paper addresses this concept by first showing the historical and current practices related to equity in transportation, second by reviewing the growth of big data in transportation, and finally by reviewing the current practices related to data biases in transportation. [ABSTRACT FROM AUTHOR]

Dense three-dimensional (3D) point clouds collected from rapidly evolving data acquisition techniques such as light detection and ranging (lidar) and structure from motion (SfM) multiview stereo (MVS) photogrammetry contain detailed geometric information of a scene suitable for a wide variety of applications. Among the many processes within a typical point cloud processing workflow, segmentation is often a crucial step to group points with similar attributes to support more advanced modeling and analysis. Segmenting large point cloud data sets (i.e., hundreds of millions to billions of points) can be extremely time consuming and tedious with current tools, which primarily rely on significant manual effort. While many automated methods have been proposed, the practicality, scalability, and versatility of these approaches remain a bottleneck stifling processing of large data sets. To overcome these challenges, this paper introduces a novel, generalized segmentation framework called Vo-Norvana, which incorporates a new voxelization technique, a normal variation analysis considering the positioning uncertainty of the point cloud, and a custom region growing process for clustering. The proposed framework was tested with several large-volume data sets collected in diverse scene types using several data acquisition platforms including terrestrial lidar, mobile lidar, airborne lidar, and drone-based SfM-MVS photogrammetry. In evaluating the accuracy of models generated from Vo-Norvana against manual segmentation, the average error of the position, orientation, and dimensions are 2.7 mm, 0.083°, and 0.9 mm, respectively. Over 0.2 million points per second and 36 thousand voxels per second can be achieved when segmenting an airborne lidar data set containing over 639 million points to about 1 million segments. [ABSTRACT FROM AUTHOR]

With the large amounts of available traffic data, it becomes necessary to develop tools that can perform several tasks related to the collected data. These tasks include storing the data in a standard format, filtering the data/flagging suspicious records, processing the data and calculating useful quantitative traffic indices, and finally, visualizing the outcomes. In this paper, a data-driven, yet novel, data-filtering approach was proposed to flag outliers in daily cycling counts at automatic traffic counters (ATCs). The approach was motivated by the spatiotemporal relationship of cycling counts collected at permanent count stations. The proposed approach is flexible because it assumes no prior knowledge about which locations may experience sensor malfunction (i.e., outliers). The approach was tested using a large data set of more than 111,000 daily bicycle volumes collected in 4 years (2016-2019) at more than 60 different permanent count stations in the City of Vancouver, Canada. The approach was validated using complete annual sets of data at four count stations in 2016. Scenarios of undercounting and overcounting were simulated using different percentages of the actual counts. The results showed that the proposed approach has a strong ability in detecting and removing most outliers, especially for cases of substantial undercounting. [ABSTRACT FROM AUTHOR]

Congestion detection is one of the key steps in reducing delays and associated costs in traffic management. With the increasing use of global positioning system (GPS)-based navigation, promising speed data are now available. This study used extensive historical probe data (year 2016) in Des Moines, Iowa. We used Bayesian change point detection to segment the speed signal and detect temporal congestion. The detected congestion events were then classified as recurrent congestion (RC) or nonrecurrent congestion (NRC). This paper thus presents a robust statistical, big-data-driven expert system and a big-data-mining methodology for identifying both recurrent and nonrecurrent congestion. [ABSTRACT FROM AUTHOR]