Artificial intelligence (AI) in supply chain management models are designed to optimize operations by analyzing data patterns, enabling smarter predicting, inventory control and logistics planning. However, AI models in the supply chain face the challenge of accurately integrating diverse and often unstructured data from multiple sources to ensure reliable and actionable insights. In this manuscript, artificial intelligence in supply chain optimization using comparative analysis of techniques and applications(AI-ACO-CATA-BSCNN) is proposed. Firstly, input data is collected from Smart Logistics Supply Chain Dataset. Then the input data is pre-processed using Implicit Unscented Particle Filter (IUPF) which is used to data cleaning and normalization. After that, the pre-processed data are fed into Red-billed blue magpie optimizer (RBMO) for feature selection. RBMO is employed to choose the applicable features. After that, the selected features are fed into Binarized Simplicial Convolutional Neural Network (BSCNN) method is used to predict logistics delays and classify as Delay and No Delay. The proposed method implemented in python, demonstrates substantial improvements in accuracy and Mean Absolute Error. The proposed AI-ACO-CATA-BSCNN achieves the best results with accuracy up to 95%, precision and 0.01 Mean Absolute Error with existing methods such as AI-based risk management for improving supply chain agility: A DL-driven dual-stage PLS-SEM-ANN analysis (AID-RMSC), Deep learning techniques to recognize order status in a complex supply chain (DLA-CSC) and Improved Commodity Supply Chain Efficacy Through AI and Computer Vision Techniques (ICSC-AICV).