A BP Neural Network Based Method for Upper Limb Motion Intensity Evaluation

Motion intensity is a comprehensive property to reflect to the motion speed, motion frequency and motion explosive power. Motion intensity evaluation plays a very important role in both the stroke patients' rehabilitation program development and competitive athletes' daily training. The traditional motion intensity evaluation takes the heart rate or rate of perceived exertion as evaluation parameters, which can't determine the motion intensity of peoples because everyone has subtile differences in these aspects. This paper proposes a new upper limb motion intensity evaluation model based on BP neural network, whose inputs are the change rate of angle and motion amplitudes which are computed according to the measured values from the three-axis acceleration sensor, and whose output is the motion intensity grade. This new model is verified via the MATLAB neural network toolbox, and the simulation experiment shows that the model has higher efficiency in evaluating the upper motion intensity grade than the traditional method and the accuracy rate reaches 93.75%.