The automated durian sorting system on the conveyor belt uses acoustic resonance analysis and artificial intelligence in the AI ​​device at the border.

Abstract

The invention relates to an automated system for sorting durian fruit on a conveyor belt using acoustic resonance analysis and edge AI devices. The system comprises a transport and positioning unit, a physical stimulation unit, edge AI devices, and an actuator sorting unit. The edge AI device includes an acoustic signal acquisition unit, a signal filter, a microprocessor, memory, and an artificial intelligence module, configured to acquire the acoustic resonance signal generated when a hammer mechanism strikes the durian fruit, extract 13 MFCC coefficients of the sound, determine the resonance characteristics of each fruit, process the signal, and perform inference using a convolutional neural network (CNN) model to determine the fruit's ripeness. Based on the inference results and corresponding reliability, the microprocessor controls an adaptive measurement cycle, including adjusting measurement parameters and re-measuring as necessary before controlling the actuator sorting unit.The solution allows for improved accuracy, reduced impact of environmental noise, and meets the requirements for real-time automated sorting on industrial production lines.

Images