Application Of Information Fusion For Dynamical Physiology In The Mimetic Robotic Horse

Over the years, with the development of pluralistic information, computer, micro-electronic and multi-sensor technology, information fusion, as an effective information process method, gained rapidly development both in theory and application research. However, the application research of information fusion for physiological information is still in a very early stage, which mostly focuses on the qualitative medical diagnosis while the research for dynamical physiological is insufficient. As the information technology develops, information fusion for dynamical physiology is bound to gain more and more attention, especially in sports health, physical education, military training and astronaut simulation training and so on.Firstly, an introduction for the control of biomimetic robotic horse, which is a special example for the physiological information fusion research, is presented in this paper. Physiological information variation of human body during the sport process and assessment standard of sport state is analyzed, then base on the character of information fusion theory and physiological information, a system model for real-time-dynamical physiological information fusion is established to illustrate the functions of the four-layers model and their interrelations. In addition, this system model is applied to the virtual speed prediction for the biomimetic robotic horse.Based on the research above, an improved physiological information fusion algorithm, based on the ant colony neural network, is proposed. With the disadvantage of easily local minimization, a simple BP algorithm is necessary to optimized by ant colony algorithm for its distributed parallel computation, positive feedback and avaricious globally search. Simulation results show that local minimization of simple BP fusion algorithm is able to be effectively avoided by choosing proper parameter. Meanwhile, the information process ability of this algorithm is expanded and the training speed and precision is greatly improved.