چكيده به لاتين
Myoelectric control has been the reference for the upper Prosthesis control for decades. Other uses of myoelectric control include remote operation and robot assist. Various artificial hand control techniques have been based mainly on pattern recognition to perform a series of finite motion sequences. These methods were far from the normal hand operation. For example, the pattern recognition method based on the myoelectric control system employs a sequential strategy. That at one time, only one artificial organ function could be performed. Therefore, some recent studies have used regression methods to solve this problem. Since it has been used online in clinical applications of myoelectric control, this study has also used online regression to achieve a simultaneous and proportional control. One of the unsolved challenges is the simultaneous control of myoelectric proportionality of the electrode displacement problem. The purpose of this study was to investigate and reduce the effect of electrode displacement in online regression-based experiments. In order to investigate the effect of electrode displacement, offline and online experiments were performed at three different electrode positions. Four different methods have been proposed and evaluated to reduce the effect of electrode displacement in this study, as follows: I- Composition (combining all offline data to create a regressor) II- selection (selection of the nearest electrode position) III- Modular composition (approximate combination of two regressors) IV- Finding the nearest electrode position using EMG data. In order to estimate the position of the electrodes in this study, an impedance meter with intra-tissue impedance measurement has been developed that has been able to detect the position of the electrodes. In order to estimate the position of the electrodes in this study, an impedance gauge with intra-tissue impedance measurement has been developed that has been able to detect the position of the electrodes. Electrode displacement has significantly reduced performance. The Lasso regressor with a performance of 0/42 and 0/74 in the displaced and original position had more stable electrode displacement than other regressor methods. The proposed modular blend method with 96% completion rate has the best performance in various online test methods. Also, using EMG data, the electrode position is estimated at 90% accuracy.
