In order to promote the development of medical rehabilitation industry, the study deeply analyzes flexible wearable devices and utilizes joint moment estimation based on skeletal muscle model in order to calculate the joint moments of elbow and wrist joints, so as to carry out the design of flexible pneumatic wrist joint system. And a fuzzy-PI dual-mode control strategy is used in the position control of the flexible pneumatic wrist joint to construct an intelligent flexible rehabilitation device for the wrist joint. The wrist joint rehabilitation equipment is systematically tested to analyze its practical application effect. The response speed of the fuzzy-PI dual-mode control method is faster than that of the traditional PID control strategy, and it can effectively reduce the vibration noise. The accuracy of the hybrid recognition method in this paper is 97%, which is better than the single recognition model. The average time taken by the wrist rehabilitation device on the seven tasks of lifting, grasping, undertaking, pulling, pushing, probing down and probing up is between 2.06 and 2.67 seconds. The output moments of the wrist and elbow positions were 17.1 and 11.6 N.m respectively for the human body-worn wrist joint rehabilitation device with 50N driving force output, and the joint output moments decreased significantly, and the joint comfort of the human body was improved greatly.