Neural network models to detect airplane near-collision situations

Hielscher, Rafael Palacios; Doshi, Anuja; Gupta, Amar; Orlando, Vince; Midwood, Brent R.

doi:10.1080/03081061003732300

Cited by 8 publications

(11 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Typically, a feedback neural network is only used for datasets that do not change and thus is not adaptable to the changing dataset. The feed-forward model was also used because it employs a set of unsupervised classifiers to summarize the probability distribution (Palacios, Doshi, Gupta, Orlando, & Midwood, 2010). As noted earlier, the feed-forward neural network is preferred in most predictive model applications because the WEKA tool can be employed to make predictions about the nonlinear data using classification and auto-associative learning (Palacios et al, 2010).…”

Section: Methodsmentioning

confidence: 99%

“…The feed-forward model was also used because it employs a set of unsupervised classifiers to summarize the probability distribution (Palacios, Doshi, Gupta, Orlando, & Midwood, 2010). As noted earlier, the feed-forward neural network is preferred in most predictive model applications because the WEKA tool can be employed to make predictions about the nonlinear data using classification and auto-associative learning (Palacios et al, 2010). WEKA tools are used because not all records have values in every column of the dataset.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Intelligent Modeling of Risk Factors Contributing to Runway Incursion

Replogle

Bischoff

Willy

et al. 2014

International Journal of Aviation Systems, Operations and Training

View full text Add to dashboard Cite

The purpose of this paper is to validate the use of an intelligent neural network model to identify the risk factors contributing to runway incursions. The study utilized multi-dataset fusion and a neural network model to identify risk factors. Historical runway safety data, weather data, and data on the physical characteristics of airports were obtained from multiple publicly available government websites. The results of the analysis showed that a neural network model was able to determine the factors most strongly associated with runway incursions, without the need for subjective weighting by safety experts used in most previous runway incursion studies. The Federal Aviation Administration could use a cyber-physical system, which combines human and computer processes, to analyze the runway incursion factors identified in the present study to determine which aspects of runway safety could be improved to reduce future incursions and save lives.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Intelligent Modeling of Risk Factors Contributing to Runway Incursion

Replogle

Bischoff

Willy

et al. 2014

International Journal of Aviation Systems, Operations and Training

View full text Add to dashboard Cite

show abstract

“…Several prediction models can be used to estimate the value of any magnitude in the future. The algorithm presented in this paper is based on a neural network model, because previous studies demonstrated that neural networks can provide better accuracy than other models [5]. Neural networks (NN) are able to model non-linear processes accurately and are tolerant of certain degree of inaccuracy in the inputs.…”

Section: Description Of the Prediction System Designedmentioning

confidence: 99%

“…The neural network is used to estimate X movement, Y movement or actual movement, using 10 seconds of past history to predict values up to 30 seconds in the future. One can feed the model with absolute X or Y coordinates to estimate the absolute coordinates in the future, but it was found more accurate to use delta values (distance made during the last seconds) to estimate the distance that the airplane will make in the future; then the futures coordinate is computed by adding estimated distance to the current position [5]. Fig.…”

mentioning

confidence: 99%

Computing Aircraft Position Prediction~!2008-08-05~!2008-10-18~!2008-11-28~!

Hielscher¹,

Doshi²,

Gupta³

2008

TOTJ

Self Cite

View full text Add to dashboard Cite

Air traffic is increasing world wide at a steady annual rate, and airport congestion is already a major issue for air traffic managers. This paper presents a model based on neural networks to predict the position of aircraft on the airport, during landing or takeoff. The same model can also be used to predict the behavior of other vehicles moving on the airport. The predictions help to detect near-collision situations earlier, giving air traffic controllers additional time to take remedial actions. The system uses the list of coordinates produced by the airport radar system, and obtains a prediction of the future position of each object. It is only necessary to store a short history of positions for each object in order to perform the estimation. This estimation has an average error comparable to the size of the airplane when the algorithm is adjusted for 20 second look ahead. The proposed model has been evaluated using data from Chicago O'Hare International Airport, which is the airport with the highest number of movements (from 2001 to 2004).

show abstract

“…In a LAHSO, an airport increases the rate of operations by using two intersecting runways simultaneously. Due to the increased risk of aircraft crossing active runways, additional safety restrictions are activated, including requiring aircraft to stop short of an intersecting runway or taxiway, or a point on the runway when landing (Palacios et al, 2010;Singh & Meier, 2004). These designated points, called the ''hold short point'' by the Federal Aviation Administration (FAA), are identified on airport charts and implemented on long runways at some large airports (Palacios et al, 2010).…”

Section: Introductionmentioning

confidence: 99%

Modeling Land and Hold Short Operations: Balancing Safety and Arrival Rate

Ward¹,

Owen-Perry²

2020

Journal of Aviation Technology and Engineering

View full text Add to dashboard Cite

Many airports conduct simultaneous operations on intersecting runways to increase the rate of takeoffs and landings. This requires landing aircraft to hold short of the intersecting runway, which incurs a safety risk of runway incursions in the process. A Monte Carlo simulation was conducted to analyze the traffic load at maximum operational capacity at Charlotte-Douglas International Airport in order to analyze the fleet types and the rate of those landing aircraft unable to stop short of the intersecting runway. The researchers used the actual and four alternative compositions of the subject airline's aircraft arrivals, interspersed among other airport traffic, to assess how such changes affect the rate of runway incursions, the rate of operations at the airport, and the mean number of passengers the subject airline can land per hour. The simulation revealed that runway length up to the hold short point was the biggest determinant of aircraft being unable to hold short. The total airport rate of operations decreased when heavy wake turbulence category aircraft were introduced. Despite heavy wake turbulence category aircraft carrying more passengers individually, the decreased operations rate also led to fewer passengers per hour that the subject airline could carry.

show abstract

Neural network models to detect airplane near-collision situations

Cited by 8 publications

References 7 publications

Intelligent Modeling of Risk Factors Contributing to Runway Incursion

Intelligent Modeling of Risk Factors Contributing to Runway Incursion

Computing Aircraft Position Prediction~!2008-08-05~!2008-10-18~!2008-11-28~!

Modeling Land and Hold Short Operations: Balancing Safety and Arrival Rate

Contact Info

Product

Resources

About