Integrating Artificial Intelligence and Sensor Fusion for Real-Time Quality Control of Dried Spice Products in Nigerian Export Processing Zones

Nigerian dried spices including ginger, pepper, cinnamon, and cloves have significant export potential, yet variability in quality arising from inconsistent drying conditions and post-harvest handling limits market access and price competitiveness in international markets. Real-time quality monitoring using intelligent sensor systems offers a transformative approach to in-line quality control in export processing. This study develops and validates an artificial intelligence-enabled sensor fusion platform for real-time quality prediction and control in dried spice processing, with application at a spice export facility in Kaduna, North West Nigeria. The theoretical contribution is a multi-sensor data fusion architecture that integrates signals from near-infrared spectroscopy, electronic nose, machine vision, and process monitoring sensors into a unified quality prediction model using deep learning. The deep learning architecture employs a multi-branch convolutional neural network for spectral and image data streams, combined with a recurrent neural network for time-series process parameter data, fused at a dense layer for final quality prediction. Quality attributes predicted include moisture content, essential oil content, aflatoxin level, colour, and overall quality grade. Training data were collected from 1,800 spice production batches over 24 months. The optimised model achieved prediction accuracy for moisture content within 0.3 percent wet basis and for aflatoxin above the regulatory limit with 97 percent sensitivity and 95 percent specificity. Integration with the production line control system enabled real-time rejection of below-specification batches, reducing export rejection rates by 64 percent in a 6-month validation period. The study contributes an original deep learning sensor fusion architecture and the first validated real-time quality control system for Nigerian export spices. Keywords: artificial intelligence, sensor fusion, dried spices, real-time quality control, export processing