The artificial limb bionic technology project under intelligent medical treatment aims to realize high-precision and natural bionic manipulator control by using surface electromyography (sEMG) signal recognition and artificial intelligence algorithm. The system collects the residual limb muscle electrical signals through non-invasive electrodes, and after feature extraction and classification recognition, drives the modular bionic hand to complete the grasping action with multi-degree of freedom and high accuracy. The project uses STM32 as the core control chip, combined with inductive and angle force sensors, so that the action recognition rate is increased to more than 90%, and has the characteristics of real-time feedback and stable response. The manipulator adopts three-stage drive and modular design, which can be flexibly disassembled and expanded, and is suitable for rehabilitation training, daily life and industrial automation scenarios. This achievement not only improves the flexibility and functionality of prostheses, but also provides innovative human-computer interaction solutions for smart medical, smart home, industrial manufacturing and service robots.