Disclosed is a multi-dimensional surface electromyographic signal based artificial hand control method based on a principal component analysis method, the control method comprising the following steps: first, an armband provided with a 24-channel array electromyography sensor being worn on a forearm of a subject, and five finger joint attitude sensors being respectively worn on a distal phalanx of a thumb and middle phalanxes of other fingers of the subject; the subject being subjected to independent bending and stretching training of five fingers, and acquiring electromyography sensing array data and finger joint attitude sensor data at the same time; decoupling the electromyography sensing data by using a principal component analysis method to constitute a finger movement training set; after the training is completed, removing the sensors worn on the fingers; carrying out data fitting on the finger movement training set by using a neural network method to build a finger continuous-movement prediction model; and predicting a bending angle of the current finger by using the finger continuous-movement model. The control method can overcome the non-continuity of modal classification of discrete actions, thus ultimately achieving smoother control over an artificial hand.
展开▼
