Multi-step passenger demand forecasting is a crucial task in on-demand vehicle sharing services. However, predicting passenger demand over multiple time horizons is generally challenging due to the nonlinear and dynamic spatial-temporal dependencies. In this work, we propose to model multistep citywide passenger demand prediction based on a graph and use a hierarchical graph convolutional structure to capture both spatial and temporal correlations simultaneously. Our model consists of three parts: 1) a long-term encoder to encode historical passenger demands; 2) a shortterm encoder to derive the next-step prediction for generating multi-step prediction; 3) an attentionbased output module to model the dynamic temporal and channel-wise information. Experiments on three real-world datasets show that our model consistently outperforms many baseline methods and state-of-the-art models.
In recent years, adaptive thermal comfort models have been integrated into several building design and operations regulatory documents. Although the theoretical background of the adaptive thermal comfort models is quite mature, still some ambiguities exist for their application. The objective of this study is to identify the main sources of uncertainty around application of adaptive models and to analyze quantitatively the difference between the adaptive comfort models proposed by the regulatory documents when applied across a spectrum of different climate zones. This paper analyzes the adaptive models in ASHRAE Standard 55, the European EN 15251 (and its revision prEN 16798), the Dutch ISSO 74 and the Chinese GB/T 50785. For each regulatory document, the major variations or sources of uncertainty are investigated: for ASHRAE 55, the length of the calculation period of the prevailing mean of outdoor temperature, and for EN 15251, prEN 16798, and GB/T 50785, the exponential decay weighting factors used in the calculation of the running mean outdoor temperature. This study shows that, although these regulatory documents have promoted the uptake of adaptive comfort models by practitioners and designers, uncertainties surrounding their application obstruct full exploitation. In response, this paper offers a fine-tuning of some of the adaptive comfort models. However, the issue of adaptive models' applicability in hybrid ventilation or mixed-mode buildings is still to be resolved, as is a rational basis for identifying the operational mode of such buildings when the adaptive models can be applied, because of their intermittent compliance during transition seasons and also extreme weather events.
With the widespread adoption of Internet of Things (IoT), billions of everyday objects are being connected to the Internet. Effective management of these devices to support reliable, secure and high quality applications becomes challenging due to the scale. As one of the key cornerstones of IoT device management, automatic cross-device classification aims to identify the semantic type of a device by analyzing its network traffic. It has the potential to underpin a broad range of novel features such as enhanced security (by imposing the appropriate rules for constraining the communications of certain types of devices) or context-awareness (by the utilization and interoperability of IoT devices and their high-level semantics) of IoT applications. We propose an automatic IoT device classification method to identify new and unseen devices. The method uses the rich information carried by the traffic flows of IoT networks to characterize the attributes of various devices. We first specify a set of discriminating features from raw network traffic flows, and then propose a LSTM-CNN cascade model to automatically identify the semantic type of a device. Our experimental results using a real-world IoT dataset demonstrate that our proposed method is capable of delivering satisfactory performance. We also present interesting insights and discuss the potential extensions and applications.
Accurate prediction of passenger demands for taxis is vital for reducing the waiting time of passengers and drivers in large cities as we move towards smart transportation systems. However, existing works are limited in fully utilizing multi-modal features. First, these models either include excessive data from weakly correlated regions or neglect the correlations with similar but spatially distant regions. Second, they incorporate the influence of external factors (e.g., weather, holidays) in a simplistic manner by directly mapping external features to demands through fully-connected layers and thus result in substantial bias as the influence of external factors is not unified. To tackle these problems, we propose an end-to-end multi-task deep learning model for passenger demand prediction. First, we select similar regions for each target region based on their Point-of-Interest (PoI) information or historical demand and utilize Convolutional Neural Networks (CNN) to extract their spatial correlations. Second, we map external factors to future demand levels as part of the multi-task learning framework to further boost prediction accuracy. We conduct experiments on a large-scale real-world dataset collected from a city in China with a population of 1.5 million. The results demonstrate that our model significantly outperforms the stateof-the-art and a set of baseline methods.
Irregularities in railway tracks are a key factor influencing the safety of trains. In this paper, rail track is considered to consist of consecutive track maintenance units whose individual defect states can be quantified in terms of a track quality index. A Markov stochastic process approach is used to evaluate the deterioration of a maintenance unit. A hazard model is formulated using the heterogeneity of the maintenance units, and a matrix of the Markov transition probabilities is constructed. The parameters of the developed models are estimated via a maximum log-likelihood function. The prediction model is validated with track irregularity data measured using track geometry cars.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.