An improved feedforward neural network with barnacles mating optimizer for traffic class prediction

Abstract Traffic management plays a crucial role in modern urban planning, as accurate traffic prediction is essential for reducing congestion, improving road safety, and enhancing transportation efficiency. In this study, a hybrid model combining a Feedforward Neural Network (FFNN) with the Barnacles Mating Optimizer (BMO) is proposed for traffic class prediction. BMO is utilized to optimize the FFNN parameters, specifically the weights and biases, in order to enhance its predictive performance. The model is evaluated using a traffic dataset consisting of four traffic classes: low, normal, high, and heavy. To assess the effectiveness of the proposed BMO-FFNN model, comparisons were made against other hybrid algorithms, namely FFNN optimized by Particle Swarm Optimization (PSO-FFNN), Whale Optimization Algorithm (WOA-FFNN), and Harmony Search Algorithm (HSA-FFNN). The results demonstrated that BMO-FFNN not only achieved the highest overall accuracy, precision, recall, and F1-score across all four classes, but also exhibited the fastest convergence, attaining the lowest final objective value. This superior convergence performance indicates that BMO more efficiently navigates the search space and fine-tunes the network parameters, yielding more reliable and stable traffic class predictions than the alternative metaheuristic optimizers.