Optimizing LSTM Hyperparameters with Whale Optimization Algorithm for Efficient Freight Distribution in Smart Cities

Abstract

Smart cities save logistics and operational expenses by optimizing freight distribution. This paper presents LSTM hyperparameter adjustment to optimise freight allocation using the Whale Optimization Algorithm (WOA). Traditional hyperparameter tuning struggles with freight logistics’ complexity and dynamism. WOA, a revolutionary bio-inspired optimization approach, finds optimal LSTM network hyperparameters. Our integrated solution fine-tunes LSTM hyperparameters using WOA to increase forecast accuracy and efficiency. The solution is tested on many smart city freight distribution scenarios. To prove the method works, prediction accuracy, computing efficiency, and convergence rate are measured. To determine how well the model detects data patterns and variations, the authors compare anticipated and real traffic flows using MAE, MSE, RMSE, etc. The proposed model’s root mean squared error is 0.23912122600654664 and achieved MAE value of 0.17255859883764077. The WOA-optimized LSTM model outperforms hyperparameter tuning in prediction accuracy and convergence speed. This optimises resource allocation and reduces environmental effect in freight distribution, enabling smart city concepts. These findings affect urban logistics and encourage more investigation.