Metaheuristic data fitting methods to estimate Weibull parameters for wind speed data: a case study of Hasan Polatkan Airport

Abstract

Flight delays may be decreased in a predictable way if the Weibull wind speed parameters of a runway, which are an important aspect of safety during the take-off and landing phases of aircraft, can be determined. One aim of this work is to determine the wind profile of Hasan Polatkan Airport (HPA) as a case study. Numerical methods for Weibull parameter determination perform better when the average wind speed estimation is the main objective. In this paper, a novel objective function that minimises the root-mean-square error by employing the cumulative distribution function is proposed based on the genetic algorithm and particle swarm optimisation. The results are compared with well-known numerical methods, such as maximum-likelihood estimation, the empirical method, the graphical method and the equivalent energy method, as well as the available objective function. Various statistical tests in the literature are applied, such as R2, Root-Mean-Square Error (RMSE) and $\chi$ 2. In addition, the Mean Absolute Error (MAE) and total elapsed time calculated using the algorithms are compared. According to the results of the statistical tests, the proposed methods outperform others, achieving scores as high as 0.9789 and 0.9996 for the R2 test, as low as 0.0058 and 0.0057 for the RMSE test, 0.0036 and 0.0045 for the MAE test and 3.53 × 10−5 and 3.50 × 10−5 for the $\chi$ 2 test. In addition, the determination of the wind speed characteristics at HPA show that low wind speed characteristics and regimes throughout the year offer safer take-off and landing schedules for target aircraft. The principle aim of this paper is to help establish the correct orientation of new runways at HPA and maximise the capacity of the airport by minimising flight delays, which represent a significant impediment to air traffic flow.

Simulation Study on Modified Weibull Distribution for Modelling of Investment Return

“…Guedes et al (2020) compared the performance of four metaheuristic optimisation algorithms; Migrating Birds Optimisation (MBO), Cuckoo Search (CS), Harmony Search (HS) and Imperialist Competitive Algorithm (ICA). Furthermore, a genetic algorithm (GA) with particle swarm optimisation (PSO) was recently used by Kaba and Suzer (2021) in searching for the root-mean-square error using the cumulative distribution function, and many more are increasingly becoming effective tools for optimisation and parameter estimation problem.…”