Intelligent Sorting Method for Smartphones Based on Agents

Automated disassembly of discarded smartphones is a critical step in electronic waste processing, yet existing automated systems often lack adaptability when handling various types of discarded smartphones. To address the issue of poor adaptability in sorting processes, this study proposes an agent-based sorting method for discarded smartphones. This method leverages large language model(LLM) for natural language understanding and generation, enhancing user-friendliness. It integrates information from knowledge graphs and historical sorting records to build a comprehensive knowledge base. By combining intelligent agents that track historical data and utilize external knowledge bases, the proposed method improves adaptability for sorting multiple phone types. This allows the system to dynamically formulate sorting strategies based on the specific characteristics of discarded smartphones, significantly enhancing the adaptability of disassembly lines. Furthermore, the mobile phone classification module draws inspiration from facial recognition model architectures. A deep learning-based classification method for discarded smartphones is proposed, achieving highly adaptable classification functionality for various phone types. Experimental results demonstrate that the proposed classification algorithm maintains high accuracy in recognizing phone models not included in the training set.