Enhancing Supply Chain Efficiency Resilience Using Predictive Analytics and Computational Intelligence Techniques

Abstract

This study addresses critical challenges in supply chain management, particularly focusing on enhancing forecast accuracy and optimizing inventory management. Traditional methods often fall short in accuracy, leading to inventory imbalances and inefficiencies. To overcome these limitations, the study employs a combination of Transformer models for demand forecasting and Particle Swarm Optimization (PSO) for inventory parameter optimization. The methodology involves a comprehensive approach: data collection includes historical sales data and inventory levels, which are preprocessed through cleaning, normalization, and feature extraction. Transformer models are used for predicting demand, leveraging their ability to capture complex patterns in time-series data. PSO is applied to optimize inventory parameters, addressing multi-objective optimization problems in the supply chain. Results from the study indicate significant improvements. The Transformer model achieved a reduction in Mean Absolute Error (MAE) from 15.8 to 8.2 and Root Mean Squared Error (RMSE) from 22.3 to 11.5, demonstrating enhanced forecasting accuracy. The application of PSO led to a 12% reduction in overall operational costs and a 25% improvement in order fulfillment times. Additionally, inventory holding costs decreased by 18%, and transportation costs were reduced by 10%. Integrating Transformer models with PSO presents a robust solution for modern supply chains, offering substantial improvements in efficiency and cost-effectiveness. The study recommends adopting these advanced methodologies for better forecasting and inventory management, and suggests further research into additional machine learning techniques and real-time data integration to enhance supply chain performance.

Keywords

• Predictive analytics
• transformer models
• particle swarm optimization (PSO)
• supply chain optimization
• inventory management