Projects

This project presents an industrial multi-agent system designed for intelligent manufacturing. The platform integrates real-time monitoring, predictive maintenance, and multimodal analysis by leveraging a smart intent recognition engine and a prototype-based router to efficiently distribute complex tasks. Built upon a foundation of large language models (e.g., Qwen) and retrieval-augmented generation (RAG), the system enables intelligent interaction through chat interfaces and large industrial displays. It features self-supervised retrieval and generation chains for automated QA and decision support, supported by technologies like FAISS for vector search and LoRA for model adaptability, creating a comprehensive AI-powered solution for modern industrial applications.

This project aims to establish a functional evaluation system for facial paralysis and cranial nerve disorders. Addressing the limitations of traditional assessment methods, such as high subjectivity, lack of objective standards, and difficulties in processing multimodal data. It proposes core algorithms including intelligent image denoising and enhancement, high-dynamic feature extraction and classification, and multimodal data alignment and adaptive fusion. This forms a standardized, multi-scenario, multi-disease-adaptive functional assessment model with interpretability. At the application level, this system serves both clinical diagnosis and primary screening at the grassroots level. It also enables public self-assessment and rehabilitation tracking via mobile devices. Ultimately, by integrating with digital standard development, it provides unified norms for the objective evaluation of facial paralysis and related diseases, driving the digital and intelligent advancement of medical diagnosis, health management, and scientific research collaboration.

This project aims to generate code for PLCs using Multi-Agent system.