目前限制灵巧假肢推广普及的最主要的原因之一是生机接口的传输性能不足。表面肌电信号（sEMG）是目前研究最广泛的肌群控制接口信号源，在实验室条件下采用模式识别法可获得高达95%的解码准确率。然而由于患者残肢肌电信号源不足与肌电信号的时变特性及接触不稳定性，模式控制肌电接口目前尚未进入实际应用。项目针对肢残患者肌电信号源的限制，拟从多源高通测量仪器研制、解码模型构建和多源信号融合三方面入手，突破多源高通量肌群运动信息在体无创测量技术，解决生物信号传感器设计、电光声多源肌群信号测量问题，以获得高通量生物信息。在此基础上，提出基于单传感模块sEMG阵列的运动单元动作电位序列（MUAPt）反解方法，并根据人体生物信号的不同特征，构建电/光/声多源生物信号的联合解码模型。目标是提出单模块sEMG阵列MUAPt反解模型及多模态信号的解耦方法，为高传输率增强式生机接口研制及其灵巧假肢应用积累科学基础。

One of the main reasons that impede the popularization of current prosthetic hand is the inadequate transmission performance of bio-mechanical interface (BMI). Surface electromyography (sEMG) is the most widely used signal for BMI research. It is now accepted that myoelectric pattern recognition (PR) can yield 95% of classification accuracy in lab conditions. However, sEMG PR control has not been realized practical usage because of the sEMG signal source missing or degradation of amputation stump, time-shift characteristics and unreliable sensor-skin contact interface of sEMG. To overcome the inherent drawbacks of sEMG, the project is proposed to measure and fuse multi-sensor muscle activities from the modalities of electrophysiology, metabolic physiology and acoustics, representing as sEMG, near-infrared spectroscopy (NIRS) and mechanomyography (MMG), respectively. The research work will be focused on the following issues with the development of multi-sensor system, theoretical analysis and multi-sensor fusion: 1) the sensing element design and optimization of electrophysiology, photics and acoustics for multi-sensor system to acquire high-throughput biological information; 2) the decomposition of motor unit action potential trains (MUAPt) from single-node multi-pin sEMG signals as MUAPt contains more reliable and stable neural control information; 3) the fusion approach of multi-sensor information to built decoding model from biological signals to motion intentions. The goal of this project is to propose the methods of multi-sensor fusion with sEMG array, NIRS and MMG, especially the decomposition of single-node multi-pin sEMG signals into its MUAPt and the decoupling of multi-sensor muscle information. The outcomes of this project will provide scientific basis for the development of enhanced bio-mechanical interface with high transmission rate and dexterous prosthetic hands.