Feature active learning-feedback digital twinning method for mechanical arm performance monitoring and error compensation

The invention provides a feature active learning-feedback digital twinning method for mechanical arm performance monitoring and error compensation, and belongs to the technical field of digital twinning. According to the twinning method, a physical entity mechanical arm, sensor real-time data and numerical simulation data are combined, the flexibility error of the mechanical arm in work can be calculated in real time in the digital world, the flexibility error is fed back to a physical entity in real time, and the physical entity is compensated; the twinning method is achieved based on a hardware system and a 3D model of the mechanical arm. Wherein the mechanical arm is a carrier of various sensors, controllers and communication modules, a certain number of sensors are installed on the mechanical arm, the controller and the communication system are carried on the mechanical arm, and the mechanical arm is connected with the upper computer. Real-time flexibility error compensation can be rapidly and efficiently