Masters of DonNTU

RUS

UKR

 

 My biography     Dissertation abstarct   Library   Links   Search results   The individual task


Dissertation abstract

Theme of master's work:

"Development of decision support system for passenger airlines management in "DonbassAero" airline"

 

 

1.Introduction

The key problem of analytical department in modern airline is prior estimation and further qualitative forecast of own and competitor's traffic flow. Number of flights, planned in advance allows to redistribute company’s resources rationally, not infringing upon passengers interests. The similar policy leads to exception of non-profitable "idle" flights and to company’s rating increase owing to the full accordance with clients’ needs.

Traffic flow forecasting is an integral part of the decision-making process. It is a regular check of company’s resources that allows to use all its advantages and to reveal potential threats. The company always have to keep up with traffic flow dynamics and look for alternative opportunities of situation development on the airline market. This will help to redistribute available resources in the best way and to choose the most expedient directions of company’s activity.

Unlike the enterprises where the size of an end-product can be defined, proceeding from presence of raw material and their usage rate in manufacture of concrete products, in any airline this process is more complicated because of the presence of the subjective factor – the passenger’s desire to take advantage of airline services (instead of any other type of transport). Besides in the airline market there is a huge number of airlines, and the passenger has the right of choice. Therefore to determine the amount of the end-product - a traffic flow of airline - the forecast for future periods must be made.

Forecasts can be short-range, intermediate range and long-range. Now there is no agreement on understanding of time borders of forecasting. Following representation about ranges of forecasts is most widespread: short-range - about one year, intermediate range - 3-5 years, long-range - till 10 years.

Traffic flow forecasting (both long-range and short-range) is one of aspects of administrative activity for modern airline. At long-range forecasting tendencies of the world airline market are considered. They consist in increase or reduction of flight frequencies on various directions during flight scheduling. Short-range forecasting allows to react operatively to the change of situation in the airline market and to build strategy of airline, proceeding from assumed traffic flow (for example to make special actions for attraction of passengers).

Traffic flow forecasting can be made on various levels of data aggregation : both at a level of whole airline, and at a level of single regions, flights and directions.

The problem of traffic forecasting is a basis for solving many optimization problems of air-transport system by the criteria directly connected with a parameter of profitability (incomes, expenses, profit) as the optimality of the received plans, depends first of all on accuracy of forecasts.

The main task of the given work is the development of decision support system for passenger airlines management which will allow to do forecasts for traffic flow, proceeds and profitability on the basis of the analysis of statistical data about passenger transportations. Thus, this system will enable airline’s management to make decisions about quantity of flights, requirements in transport,ticket prices, etc.

2. The purpose and problems of the work

The purposes:
Development of a computer decision support subsystem providing airline administration making most effective decisions on the airline organization and management.

Problems:

Theoretical researches:

  • analyzing of modern methods of forecasting
  • researching of various models and architectures of neural networks
  • choice of optimum structure of neural network with reference to a problem of traffic flow forecasting
  • researching and analyzing of the most effective methods of neural networks training for solving selected problem

Experimental researches:

  • gathering of statistical data for training neural network
  • experimental selection of parameters of neural network
  • training neural network on statistical data of airline
  • development of the decision-support system for the airline management on the basis of traffic flow forecast using neural networks

3. The practical and scientific importance of the work, results:

  • The theoretical data concerning statistics of aviatransportations are investigated and systematized;
  • The most popular statistical methods of forecasting are selected and analyzed;
  • The modern methods of forecasting based on neural networks are investigated;
  • Existing software packages of statistical and neural forecasting are investigated;
  • The practical importance of this work consists in development of the computer program that will make traffic flow forecast using neural networks.


Êîðíåâ Ñåðãåé Àëåêñàíäðîâè÷ Masters of DonNTU