Comparative analysis of M/G/K queueing model with artificial neural networks for service system dynamics
Open Access
- 11 May 2025
- journal article
- research article
- Published by Springer Nature in Discover Applied Sciences
- Vol. 7 (5) , 1-16
Abstract
Queues emerge when the demand for service exceeds the available capacity to provide it. The M/G/K queuing model is the typical and theoretical framework to model and analyze the system when arrival rates are not in constant mode and service rate follow exponential distribution. This study investigates traffic congestion at fuel stations, focusing on queue length, customer waiting time distribution, and operational efficiency. A system with four servers were modelled to evaluate the system's behavior. The governing differential equations using M/G/K model were developed and solve by Runge–Kutta (RK4) method to derive initial state probabilities. Additionally, an Artificial Neural Network (ANN) was implemented to predict these state probabilities using simulated data. A comparison of results revealed a close approximation between the RK4 solutions and ANN predictions, demonstrating the ANN's robustness and accuracy in replicating traditional queuing dynamics These findings emphasize the potential of artificial neural networks (ANNs) as a practical tool for planning and optimizing fuel station operations, allowing for better congestion management and customer experience. Future applications may involve the extension of this methodology to other service-oriented systems.Keywords
This publication has 13 references indexed in Scilit:
- A bivariate two-state Markov modulated Poisson process for failure modelingReliability Engineering & System Safety, 2021
- Modelling and understanding count processes through a Markov-modulated non-homogeneous Poisson process frameworkEuropean Journal of Operational Research, 2021
- A congested capacitated multi-level fuzzy facility location problem: An efficient drone delivery systemComputers & Operations Research, 2019
- Application of Neural Network Models for Mathematical Programming Problems: A State of Art ReviewArchives of Computational Methods in Engineering, 2019
- Performance of small-world feedforward neural networks for the diagnosis of diabetesApplied Mathematics and Computation, 2017
- Automated design parameter selection for neural networks solving coupled partial differential equations with discontinuitiesJournal of the Franklin Institute, 2012
- Decomposition of general queueing networks with MMPP inputs and customer lossesPerformance Evaluation, 2003
- A location model for a facility operating as an M/G/k queueNetworks, 1989
- Learning representations by back-propagating errorsNature, 1986
- The Generalized M/G/k Blocking System with Heterogeneous ServersJournal of the Operational Research Society, 1982