Artificial Neural Network Approximation of Transient State Probabilities in a Finite-Capacity Multi-Server Queue with Customer abandonment
Year: 2026
Type: conference-paper
Abstract: This paper investigates the transient behavior of an M/M/C/N finite-capacity queuing framework with customer abandonment. The system dynamics are modeled using a continuous-time Markov chain, and the corresponding Kolmogorov forward differential equations are derived. The time-dependent probability distribution of the system is evaluated using a matrix exponential framework, providing an exact benchmark solution. To develop an efficient predictive framework, an Artificial Neural Network (ANN) is trained using the benchmark probability data. Important performance measures, including mean system size, server utilization, blocking probability, and abandonment rate, are evaluated. Numerical results demonstrate that the ANN accurately approximates the transient probability distributions and achieves close agreement with the matrix exponential solution. The proposed approach combines stochastic queuing analysis with machine learning techniques for efficient transient performance evaluation.
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7204100703