Integrated aircraft fleet planning using fuzzy logic and artificial neural network approach

Abstract

Aviation industry is one of the most sophisticated and commercially challenging industries, providing both passenger and freight services. In a highly competitive environment, it is essential to dynamically control their operating costs by managing their flights, selecting aircraft, and scheduling crews effectively. This is a continuous process of fleet planning and route planning. In aircraft fleet planning problem, large number of parameters or input variables needs to be considered, which are often conflicting in nature. Thus, this complex decision making process requires the use of analytical as well as optimization techniques for a solution with high degree of optimality. Among many such parameters, some are design and capacity (in terms of seating, flying, mileage) of the aircraft, demand, cost, carbon emission, issues related to customer satisfaction, number of crews, flying distance, etc. The probable outputs are also multi-fold, such as selection of aircraft type, capacity (size) and number of aircrafts, fleet assignment and schedule, fleet routing, crew schedule, etc. This dissertation formulates and analyzes six different models, each of which examines a composition of certain pertinent airline fleet planning problems. First one is the profit maximization model. Second one is fuzzy customer satisfaction model, third one is the carbon emission model, fourth one is the integration of all three models using Artificial Neural Network. Fifth and sixth models are used to verify trip cost using Genetic algorithm and Particle Swarm Optimization respectively. In this dissertation, cognitive computing fuzzy logic has been used to rank a set of selected aircraft according to the level of satisfaction on the basis of the reviews made by the passengers in a linguistic pattern. This research applied artificial neural network to solve optimization sub-problems, such as profit maximization, satisfaction maximization and carbon emission minimization. Analyses involving many optimization requirements, taking into account so many input variables/parameters, and with targets of so many outputs make the complete problem NP-hard. Additionally, while analytical methods might suffer from slow convergence and the curse of dimensionality, heuristics-based swarm intelligence can be an efficient alternative. This necessitates the use of meta-heuristics. Particle Swarm Optimization (PSO), part of the swarm intelligence family, is known to effectively solve large-scale nonlinear optimization problems. This dissertation used Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) techniques for optimizing trip cost for short haul and long haul operation. A comparative analysis has also been made for trip costs, obtained from different methodologies. The proposed models, methods of implementation of those models and proposed analytical techniques can be of tremendous help for not only aviation industry, but also for other similar situations in logistics management in general.