Delay Prediction Model Research Articles

Towards a greener Extended-Arrival Manager in air traffic control: A heuristic approach for dynamic speed control using machine-learned delay prediction model

Extended Arrivals Manager (E-AMAN) is a concept that reduces congestion and holding time in the Terminal Maneuver Airspace (TMA) by managing the arrival aircraft during the en-route phase. However, current E-AMAN deployment is only limited to a horizon of 150–200 NM from the airport, restricting the window of opportunity for any early intervention, and the prediction of delay in TMA remains a challenge given the inherent uncertainties in the air traffic environment. In this context, this research work presents an approach for predicting, transferring and absorbing the flight delays and holdings from the highly constrained TMA to the en-route phase using both data-driven and optimisation techniques. First, a method is developed to estimate holding time and TMA delay from historical data. Next, a Machine Learning based prediction framework is developed to predict holdings and delays in the TMA, from an extended horizon of 300–500 NM from the airport. Finally, a heuristics-based optimisation model is developed for dynamic speed management to transfer TMA delays to the en-route phase. To demonstrate the model's efficacy, a case study for Singapore airspace is developed using associated one-day air-traffic data. Four sets of experiments are designed to evaluate the performance of the speed management framework under different flight cooperation levels. For the experiment with the highest number of cooperative flights, the implementation of dynamic speed shows a transfer of 179 min of TMA delay to the en-route phase, equivalent to 65% of the initial TMA delay. This results in an estimated fuel saving of 1524 kg along with a reduction in carbon dioxide emissions of 48000 kg. The findings demonstrate that E-AMAN, for extended horizon with predictive delay modelling and dynamic speed management, has the potential to manage TMA congestion and reduce fuel consumption and emissions, therefore mitigating the environmental impact.

A Two-Stage Train Delay Prediction Method Based on Data Smoothing and Multimodel Fusion Using Asymmetry Features in Urban Rail Systems

During the operation of urban rail transit trains, train delays have a great negative impact on subway safety and operation management. It is caused by various external environmental disturbances or internal equipment failures. In order to better grasp the situation of train delays, adjust train operation schedules in time, and improve the quality of rail transit command and transportation services, this paper proposes a two-stage train delay prediction method based on data smoothing and multimodel fusion. In the first stage, the Singular Spectrum Analysis (SSA) method is used to smooth the train operation data, and the smoothed components and residual components are extracted. In the second stage, we use different machine learning methods to train the smoothed data and use the K -nearest neighbor (KNN) method to fuse different trainers. Finally, the predicted values of the smoothed and residual components are combined to improve the overall performance of the train delay prediction model, especially under asymmetry features. The research results show that the prediction accuracy of the two-stage train delay prediction method based on KNN multimodel fusion is significantly better than that of the independent machine learning model.

A Computationally Efficient Model for FDSOI MOSFETs and Its Application for Delay Variability Analysis

This paper proposes a compact, physics-based current model for fully depleted silicon-on-insulator (FDSOI) MOSFETs and applies it to delay variability analysis. An analytical method is applied to avoid the numerical iterations required in the evaluation of surface potential, which directly improves the computational efficiency. The accuracy of the explicit surface potential approximation is 190.3 nV, which allows for fast convergence. Surface potential and current calculations achieve 1.8× and 1.4× acceleration compared with BSIM-IMG, respectively. To establish the relationship between delay and underlying process parameters, we introduce the effective current and propose a process variation-aware delay prediction model. Higher-order derivatives are calculated to compensate the nonlinearity of delay variations with respect to process parameters. Experiments show a significant improvement in the prediction accuracy with higher-order derivatives, which are proved to be able to handle nonlinearity under process variations. The front gate work function contributes the most to the nonlinearity of the delay variation and the accuracy of the third-order prediction is 4.07%. Under the variation in the channel length and width, front and back gate oxide thickness and body thickness, delay variations have similar characteristics and the second-order prediction is found to be sufficient to model the nonlinearity with a maximum relative error of 1.22%. The delay prediction model only requires a single-point HSPICE DC or transient simulation and is universal for different voltages and different cells. Compared with the Monte Carlo (MC) simulation, the accuracy of the first-order prediction in the above-threshold region (0.8 V) is 0.94%. In the sub-threshold region (0.3 V), a prediction accuracy of 2.01% can be obtained while achieving a 21× reduction in computational time.

Timescales of delay propagation in airport networks

Flight delays persist and spread in airport networks due to high interconnectivity in the air transportation infrastructure. How quickly delay propagates between two airports is determined by factors such as the number of flights between airports, the duration of the flight, presence of disruptions, and schedule buffers. Accurate estimation of the time for delay propagation can improve system predictability and reliability. However, noisy airport delay data, along with a lack of visibility into airline scheduling and disruption management strategies, result in a challenging estimation problem for such propagation timescales. We present an algorithm to estimate statistically significant time lags between airport delays from noisy, aggregate operational data. The algorithm uses sliding correlation windows to extract the airport pairs with stable delay lags. We apply our method to identify different timescales of interactions for US airport delays in 2017. Our analysis yields two main results: (1) The most stable lags between airport delays involve the Northeast airports; (2) The stable lags between two airports are negatively correlated to the scheduled flight times between the same two airports. These results regarding delay propagation speeds have potential implications for delay prediction models and airline schedule design.

Flight delay forecasting and analysis of direct and indirect factors

The accurate prediction of flight delays is of great significance to airports, airlines and passengers. This paper presents a causal flight delay prediction model developed for a single airport. A long short-term memory network of delay prediction with an attention mechanism (LSTM-AM) is established to predict flight delays and analyse their primary causes. In this model, the direct and indirect factors related to delays are comprehensively considered. LSTM-AM can focus on input data combined with the attention vector to capture the critical time points, which can make the prediction more accurate. The model's performance is verified by actual operational data of Beijing International Airport, one of the busiest airports in the world. Experimental results show that LSTM-AM has better prediction accuracy than baseline algorithms such as some machine learning methods and deep learning methods. The mean absolute error of LSTM-AM is about 8.15 min on the test dataset. The study found that using the predicted results of this paper to release delayed information in advance can effectively alleviate the nervousness of passengers. The critical time point captured by LSTM-AM combined with runway and apron flow control can reduce or eliminate delays of one flight.

Real-Time Passenger Train Delay Prediction Using Machine Learning: A Case Study With Amtrak Passenger Train Routes

Passenger train delay significantly influences riders’ decision to choose rail transport as their mode choice. This article proposes real-time passenger train delay prediction (PTDP) models using the following machine learning techniques: random forest (RF), gradient boosting machine (GBM), and multi-layer perceptron (MLP). In this article, the impact on the PTPD models using Real-time based Data-frame Structure (RT-DFS) and Real-time with Historical based Data-frame Structure (RWH-DFS) is investigated. The results show that PTDP models using MLP with RWH-DFS outperformed all other models. The influence of the external variables such as historical delay profiles at the destination (HDPD), ridership, population, day of the week, geography, and weather information on the real-time PTPD models are also further analyzed and discussed.

A Novel Markov Model for Near-Term Railway Delay Prediction

Predicting the near-future delay with accuracy for trains is momentous for railway operations and passengers' traveling experience. This work aims to design prediction models for train delays based on Netherlands Railway data. We first develop a chi-square test to show that the delay evolution over stations follows a first-order Markov chain. We then propose a delay prediction model based on non-homogeneous Markov chains. To deal with the sparsity of the transition matrices of the Markov chains, we propose a novel matrix recovery approach that relies on Gaussian kernel density estimation. Our numerical tests show that this recovery approach outperforms other heuristic approaches in prediction accuracy. The Markov chain model we propose also shows to be better than other widely-used time series models with respect to both interpretability and prediction accuracy. Moreover, our proposed model does not require a complicated training process, which is capable of handling large-scale forecasting problems.

Machine Learning-based Optimal Framework for Internet of Things Networks

Deep neural networks (DNN) are widely employed in a wide range of intelligent applications, including image and video recognition. However, due to the enormous amount of computations required by DNN. Therefore, performing DNN inference tasks locally is problematic for resource-constrained Internet of Things (IoT) devices. Existing cloud approaches are sensitive to problems like erratic communication delays and unreliable remote server performance. The utilization of IoT device collaboration to create distributed and scalable DNN task inference is a very promising strategy. The existing research, on the other hand, exclusively looks at the static split method in the scenario of homogeneous IoT devices. As a result, there is a pressing need to investigate how to divide DNN tasks adaptively among IoT devices with varying capabilities and resource constraints, and execute the task inference cooperatively. Two major obstacles confront the aforementioned research problems: 1) In a heterogeneous dynamic multi-device environment, it is difficult to estimate the multi-layer inference delay of DNN tasks; 2) It is difficult to intelligently adapt the collaborative inference approach in real time. As a result, a multi-layer delay prediction model with fine-grained interpretability is proposed initially. Furthermore, for DNN inference tasks, evolutionary reinforcement learning (ERL) is employed to adaptively discover the approximate best split strategy. Experiments show that, in a heterogeneous dynamic environment, the proposed framework can provide considerable DNN inference acceleration. When the number of devices is 2, 3, and 4, the delay acceleration of the proposed algorithm is 1.81 times, 1.98 times and 5.28 times that of the EE algorithm, respectively.

The uneven geography of US air traffic delays: Quantifying the impact of connecting passengers on delay propagation

Sustained airport congestion periods translate into delays, especially in hub-and-spoke networks in which delay propagation is more evident.We examine the impact of connecting passenger arrival delays on network delay propagation by using passenger level data combined with flight delay data that allow us to analyse the correlation between delayed incoming flights and departure delays at the 21 U.S. airports with most delays, in July 2018.Results show that correlation between daily arrival delays and daily carrier induced departure delays are statistically significant only for flights carrying high proportions of connecting passengers. Correlation values are also higher for short-to-moderate arrival delays. In addition, a Neural Network model was trained for six major airports to build a delay prediction model and map the potential delay propagation. The results of the propagation scenarios suggest that the presence of a unique dominant carrier at an airport translates into a stronger correlation between arrival and carrier delays than that at airports where different carriers compete for connecting passengers. Furthermore, airline hubs located near the areas of the network with more traffic density, independently of the hub's volume of traffic, are more likely to propagate the delay than hubs located in the periphery. The results of this study can be relevant for airline, airport, and traffic control policies aimed at mitigating airport and network congestion.

머신러닝(Machine Learning) 기법을 활용한 제주국제공항의 운항 지연과의 상관관계 분석 및 지연 여부 예측모형 개발

Due to the recent rapid increase in passenger and cargo air transport demand, the capacity of Jeju International Airport has been approaching its limit. Even though in COVID-19 crisis which has started from Nov 2019, Jeju International Airport still suffers from strong demand in terms of air passenger and cargo transportation. However, it is an undeniable fact that the delay has also increased in Jeju International Airport. In this study, we analyze the correlation between weather and delayed departure operation based on both datum collected from the historical airline operation information and aviation weather statistics of Jeju International Airport. Adopting machine learning techniques, we then analyze weather condition Jeju International Airport and construct a delay prediction model. The model presented in this study is expected to play a useful role to predict aircraft departure delay and contribute to enhance aircraft operation efficiency and punctuality in the Jeju International Airport.

Investigation on the ignition delay prediction model of multi-component surrogates based on back propagation (BP) neural network

The ignition delay prediction model of three-component surrogates was established based on the back propagation (BP) neural network. The ambient temperature, ambient pressure, molar fraction of n-heptane, iso-octane and toluene were utilized as the input parameters. The ignition delay was utilized as the output parameter. The training and validation set only contained the 0-D simulation ignition delay of single- and two-component surrogates. But the trained BP neural network also showed a strong predictive ability towards the ignition delay of three component toluene primary reference fuel (TPRF) surrogates. Results show that the BP neural network with two hidden layers performs better than that with single hidden layer. With the optimization of genetic algorithm (GA) and particle swarm optimization (PSO) algorithm, the correlation coefficient is higher than 0.9996. The mean relative error (MRE) and the mean square error (MSE) are also maintained at a low level. The computational costs of 0-D simulation and BP neural network methods are compared. In 0-D simulation, the computational time of one case is 28 s. When the BP neural network is utilized, the computational time of 176 cases is just 3.2 s, which shows a significant improvement in the computation time. Through the mean impact value (MIV) analysis, the significance of each input variable to the output results is investigated. The input parameters of ambient temperature and molar fraction of iso-octane obtain the highest and lowest absolute value of MIV, respectively, indicating the major and minor effects on the ignition delay. Based on the predicted ignition delay of three-component TPRF surrogates, the research octane number (RON) and motor octane number (MON) can also be accurately predicted with the maximum deviance no more than 3 units. For real fuels of fuels for advanced combustion engines (FACE) gasolines, such as FACE A and FACE C gasolines, the surrogate fuel which has the same ignition delay at the specific pressure and temperature can be determined through the construction of the ignition delay look-up table of TPRF surrogates by the BP neural network. Following this method, the ignition delay of the real fuels can be matched accurately and the molar fraction of each component of the corresponding TPRF surrogates can also be acquired.

Using the gradient boosting decision tree (GBDT) algorithm for a train delay prediction model considering the delay propagation feature

Accurate prediction of train delay is an important basis for the intelligent adjustment of train operation plans. This paper proposes a train delay prediction model that considers the delay propagation feature. The model consists of two parts. The first part is the extraction of delay propagation feature. The best delay classification scheme is determined through the clustering method of delay types for historical data based on the density-based spatial clustering of applications with noise algorithm (DBSCAN), and combining the best delay classification scheme and the k-nearest neighbor (KNN) algorithm to design the classification method of delay type for online data. The delay propagation factor is used to quantify the delay propagation relationship, and on this basis, the horizontal and vertical delay propagation feature are constructed. The second part is the delay prediction, which takes the train operation status feature and delay propagation feature as input feature, and use the gradient boosting decision tree (GBDT) algorithm to complete the prediction. The model was tested and simulated using the actual train operation data, and compared with random forest (RF), support vector regression (SVR) and multilayer perceptron (MLP). The results show that considering the delay propagation feature in the train delay prediction model can further improve the accuracy of train delay prediction. The delay prediction model proposed in this paper can provide a theoretical basis for the intelligentization of railway dispatching, enabling dispatchers to control delays more reasonably, and improve the quality of railway transportation services.

Graph to sequence learning with attention mechanism for network-wide multi-step-ahead flight delay prediction

The primary objective of this study is to predict network-wide multi-step-ahead flight delay. A novel graph-to-sequence learning architecture with attention mechanism (AG2S-Net) is developed to predict the multi-step-ahead hourly departure and arrival delay of the entire network. The proposed AG2S-Net consists of a graph convolution neural network, a bi-LSTM neural network, and a sequence-to-sequence framework with embedded attention mechanism. Five-year flight data of 75 airports and 242 links are collected from the National Airspace System to illustrate the procedure. First, K-means clustering algorithm is applied to classify the hourly flight delay states of the entire network into four typical delay patterns, which are further used as explanatory variables in prediction model. Then, the network-wide multi-step-ahead flight delay prediction model is built with the proposed AG2S-Net. The result indicates that the model achieves better performance on large-scale airports than small-scale airports. The delay pattern variables and weather variables can greatly improve the prediction performance. In addition, compared with several benchmark methods, the AG2S-Net performs the best for all three airport types in terms of the lowest RMSE and MAE values. Finally, the graph network analysis further reveals that the proposed AG2S-Net can capture the hidden correlations between airports without links and collect information from more airports in the same community to enhance the prediction performance. The results of this study could provide insightful suggestions for aviation authorities and airport regulators to develop effective air traffic control strategies for alleviating flight delays and improving operation efficiency of the entire network.

A novel hybrid method for flight departure delay prediction using Random Forest Regression and Maximal Information Coefficient

Flight departure delay prediction is one of the most critical components of intelligent aviation systems. The accurate prediction of flight departure delays can provide passengers with reliable travel schedules and enhance the service performance of airports and airlines. This article proposes a hybrid method of Random Forest Regression and Maximal Information Coefficient (RFR-MIC) for flight departure delay prediction. Random Forest Regression and Maximal Information Coefficient are inherently fused in terms of Information Consistency. Furthermore, this article focuses on utilizing flight information on multiple air routes for flight departure delay prediction. To validate the proposed flight departure delay prediction model, a numerical study is conducted using flight data collected from Beijing Capital International Airport (PEK). The proposed RFR-MIC model exhibits good performance compared with linear regression (LR), k-nearest neighbors (k-NN), artificial neural network (ANN), and standard Random Forest Regression (RFR). The results also show that flight information on multiple air routes can certainly improve the accuracy of flight departure delay prediction.

A multi‐step airport delay prediction model based on spatial‐temporal correlation and auxiliary features

A multi‐step airport delay prediction model based on spatial‐temporal correlation and auxiliary features

Predicting delays in prefabricated projects: SD-BP neural network to define effects of risk disruption

PurposePrefabricated construction is often hindered by scheduling delays. This paper aims to propose a schedule delay prediction model system, which can provide the key information for controlling the delay effects of risk-related factors on scheduling in prefabricated construction.Design/methodology/approachThis paper combines SD (System Dynamics) and BP (Back Propagation) neural network to predict risk related delays. The SD-based prediction model focuses on dynamically presenting the interrelated impacts of risk events and activities along with workflow. While BP neural network model is proposed to evaluate the delay effect for a single risk event disrupting a single job, which is the necessary input parameter of SD-based model.FindingsThe established model system is validated through a structural test, an extreme condition test, a sensitivity test, and an error test, and shows an excellent performance on aspect of reliability and accuracy. Furthermore, 5 scenarios of case application during 3 different projects located in separate cities prove the prediction model system can be applied in a wide range.Originality/valueThis paper contributes to academic research on combination of SD and BP neural network at the operational level prediction, and a practical prediction tool supporting managers to take decision-making in a timely manner against delays.

Prediction of air traffic delays: An agent-based model introducing refined parameter estimation methods.

We propose an agent-based model for predicting individual flight delays in an entire air traffic network. In contrast to previous work, more detailed parameter estimation methods were incorporated into the agent-based model, acting on the state transitions of agents. Specifically, a conditional probability model was proposed for modifying the expected departure time, which was used to indicate whether a flight had experienced the necessary waiting due to Ground Delay Programs (GDPs) or carrier-related reasons. Additionally, two random forest regression models were presented for estimating the turnaround time and the elapsed time of flight agents in the agent-based delay prediction model. The parameter models were trained and fitted using the flight data for 2017 in the United States. The performance of the delay prediction model was tested for thirty days with three types of delay levels (low, medium, and high), which were randomly selected from 2018. The experimental results showed that the average absolute error in the test days was 6.8 min, and the classification accuracy with a 15 min threshold for a two-hour forecast horizon was 89.5%. The performance of our model outperformed that of existing research. Additionally, the positive effect of introducing parameter models and the negative impact of increasing the prediction horizon on the prediction performance were further studied.

A Deep Graph-Embedded LSTM Neural Network Approach for Airport Delay Prediction

Due to the strong propagation causality of delays between airports, this paper proposes a delay prediction model based on a deep graph neural network to study delay prediction from the perspective of an airport network. We regard airports as nodes of a graph network and use a directed graph network to construct airports’ relationship. For adjacent airports, weights of edges are measured by the spherical distance between them, while the number of flight pairs between them is utilized for airports connected by flights. On this basis, a diffusion convolution kernel is constructed to capture characteristics of delay propagation between airports, and it is further integrated into the sequence-to-sequence LSTM neural network to establish a deep learning framework for delay prediction. We name this model as deep graph-embedded LSTM (DGLSTM). To verify the model’s effectiveness and superiority, we utilize the historical delay data of 325 airports in the United States from 2015 to 2018 as the model training set and test set. The experimental results suggest that the proposed method is superior to the existing mainstream methods in terms of accuracy and robustness.

An Improved Artificial Neural Network Model for Flights Delay Prediction

With the limitation of air traffic and the rapid increase in the number of flights, flight delay is becoming more frequent. Flight delay leads to financial and time losses for passengers and increases operating costs for airlines. Therefore, the establishment of an accurate prediction model for flight delay becomes vital to build an efficient airline transportation system. The air transportation system has a huge amount of data and complex operation modes, which is suitable for analysis by using machine learning methods. This paper discusses the factors that may affect the flight delay, and presents a new flight delay prediction model. The five warning levels are defined based on flight delay database by using K-means clustering algorithm. After extracting the key factors related to flight operation by the grey relational analysis (GRA) algorithm, an improved machine learning algorithm called GRA — Genetic algorithm (GA) — back propagation neural network, GRA-GA-BP, is introduced, which is optimized by GA. The calculation results show that, compared with models before optimization and other two algorithms in previous papers, the proposed prediction model based on GRA-GA-BP algorithm shows a higher prediction accuracy and more stability. In terms of operation efficiency and memory consumption, it also has good performance. The analysis presented in this paper indicates that this model can provide effective early warnings for flight delay, and can help airlines to intervene in flights with abnormal trend in advance.

Accuracy Improvement Model for Predicting Propagation Delay of Loran-C Signal Over a Long Distance

We previously used back propagation neural network (BPNN) with the meteorological factors of the receiver point to establish a model for predicting propagation delay of Loran-C signal over a short distance. Nevertheless, for a long propagation path, it is not proper to use only the meteorological factors of the receiver point. In this letter, a propagation delay prediction model that considers multiweather and multipoint was established by using a more suitable generalized regression neural network (GRNN) over a long distance. We first compared three meteorological factors of six points on the propagation path, and found that they have obvious differences. Then, a propagation delay prediction model based on three meteorological factors of the six points is established with GRNN. Finally, the further comparison shows that on the one hand, GRNN is more suitable to establish the prediction model of propagation delay than BPNN; on the other hand, more points on the propagation path are considered, and the prediction accuracy of the model is higher.