VI. RESEARCH WORK DONE IN ACADEMIC INSTITUTIONS RELATED TO THESIS WORK

D. TRAFFIC AND TRANSPORTATION
13. Development of Trip Generation Model using Artificial Neural Network
Date of Start

January 2006
School of Planning and Architecture, New Delhi (R)

 

Scope and Objectives

(i) To understand the role of neural network approach in trip generation stage of UTPS four stage modeling through literature review.

(ii) To compare ANN based trip generation model with the traditionally used multiple regression based Trip Generation Model for work and education trip using household data of Bangalore Metropolitan Area.

(iii) To develop suitable ANN model for work and Education trip using household data of Bangalore Metropolitan and examine its utility as operational model.

Scope
(i) The study was based on the secondary data collected for Bangalore Metropolitan Area and limited literature available.

(ii) The study was limited to the development of trip Generation Models for work and education trips only.

(iii) For comparison with conventional models, the study was restricted to comparing the findings of ANN models with that of MLR models.

 
Methodology
At the beginning of study the concept, advantages and utilities of Artificial Neural Network were made. Modeling for Artificial Neural Network was done with a case study of Hyderabad Urban Area. As a part of literature study, comparison of alternative trip generation models for hurricane evacuation were carried out. The secondary data of household travel survey was collected for Bangalore Metropolitan area. Trip generation models were developed for work and Education trips. Finally ANN models were developed and compared with the various MLR models.
 
Findings and Conclusions
(i) ANN model has the capability to choose the appropriate functions, that may necessarily not be linear in nature. As such they can incorporate both linear as well as non-linear functions with equal ease.

(ii) ANN models using Multiple Perceptron with back propagation function, have the inbuilt capability of iterations, thus ensuring that the models comes out with the best solution for any problem.

(iii) ANN based trip generation models offers more potential to capture variables to ensure high degree of accuracy.