| This thesis deals with the problems of myoelectric (ME) and force channel performance, which are important for comparing and optimizing the design of the signal processor for powered prostheses. Based on muscle physiology and electric activity associated with muscle contraction, mathematical models for describing the SNR characteristics of the myoelectric channel (MEC) are formulated using a system theory and statistical approach. Emphasis is placed on the study of the effects of physiological parameters on the SNR of the myoelectric channel.;In order to enhance the performance of the conventional myoelectric channel, several models are discussed in which the schemes for acquiring and processing the signal differ. It is shown that, in the case of a conventional MEC system employing a nonlinearity followed by integrator, the SNR is enhanced by increasing the integration time T; that, in the case of multi-MEC model, the SNR is correlation coefficient dependent and the SNR is improved over a single MEC by a factor of N, the number of uncorrelated channels; and that, in the case of a bipolar configuration model, the SNR can be maximized by appropriately choosing the spacing between the two electrodes. All these results should provide useful information for optimizing the design of the control system in myoelectric prostheses.;The influence on SNR of additive interference, including system noise and power line interference, is studied for the myoelectric channel. The results show a complex interaction among the physiological parameters and interference. A particular point of interest is the anomalous SNR values that can occur under certain physiological and interference conditions.;Also, it is shown throughout the performance analyses in this thesis that the parallel structures of motor units in skeletal muscle provide an inherent mechanism for enhancing the performance of information transmission via the myoelectric channel.;The derived SNR expressions, verified by simulation and experiment, relate to electric and physiological parameters such as motor unit recruitment, mean firing rate, action potential shapes and amplitude variation coefficients. It is found that, for the conventional MEC, the SNR will increase and eventually saturate with increasing muscle contraction level. Tissue filtering and time delay dispersion are discussed and noted to have the effect of a low pass filter, which influences the rate of increasing SNR.;Modifications of the ME signal power density spectrum, induced by various physiological parameters, are examined in the thesis. In particular, propagation dispersion is found to be low-pass in nature with a 3dB frequency inversely proportional to the variance of the dispersion, and to result in modification of the high frequency end of the spectrum. In contrast, changes in the firing statistics, forming a high-pass filter, are demonstrated to affect the low frequency end of the spectrum. It is shown that the interaction among the fiber action potential, tissue filtering, dispersion and firing statistics determines the characteristic of the ME power spectrum as a band-pass. These results should provide an aid in the correct interpretation of the ME power density spectrum. |
