Trip Generation Model: Advanced Transportation Modeling Model

1129 Words5 Pages
CEL 785 – Advanced Transportation Modelling
Research Paper Review
By- Neetu Yadav
2012CE10368
NEURAL NETWORKS FOR TRIP GENERATION MODEL
Daehyon KIM
(Lecturer, Transportation & Logistics System Engineering, Yosu National University, Korea)

1. Abstract:

The research paper aims to estimate the number of trips generated in an area which is the first step of the process for transportation planning. A trip generation model depicts the correlation between the socioeconomic factors and the number of trips generated from that particular area. Till present, General Linear Regression Model and Category Analysis has been the most common methods for modeling trip generation. In this research paper, the author has tried to examine how the Neural Network
…show more content…
This step consists of the estimation of total number of trips produced or attracted by a zone as a function of socioeconomic, locational, and land-use characteristics of that zone. Till now, the most used method for trip generation is Linear Regression Model which carries problems like imposition of linearity. This is an accepted truth that socioeconomic parameters like population, employment, income and car ownership are not linearly proportional to number of trips generated. One more method used for trip generation modelling is Category Analysis which carries the assumption that trip generation rates are relatively stable over time for certain household stratifications and we have to estimate number of households in each stratum and hence needs large numbers of…show more content…
For the number of hidden neurons, two different architectures have been used according to the number of input units. The author used an enhanced Backpropagation model (Kim,
1998), Backpropagation with Momentum and Prime offset since it has been showing better performances in terms of computing cost and predictive accuracy as compare to standard Backpropagation.
The research was carried out on the three layer network with one hidden layer. The number of neurons on the hidden layer was given differently based on the number of input units that have been used as dependent variables on the regression models.
For example: l (input unit)-5(hidden neurons)-1(output unit) and 2(input units)-10(hidden neurons) - 1(output unit).
The research has accounted only vehicle trips and the trip data were taken from the freeway tollgates in three major cities of Korea - Seoul, Chonan and Pusan. 5 year data (1991 -1995) were used for both regression and neural network model, and 1 year data (1996) has been used for testing the prediction accuracy of each model. For the input units of the neural networks, the same variables as regression models have been used in order to compare two different models directly. For the sake of efficiency, all the input and output data were scaled into values that ranged between 0.1 and 0.9, and the networks were initialized to random values between +0.5 and -0.5 before learning.

More about Trip Generation Model: Advanced Transportation Modeling Model

Open Document