| Horseback riding has long been the king of the sports world. However, limited by thevenue, keeping costs and other factors, it is difficult to promote. Therefore, the study onanalog riding device is very significant. Horse riding has effect on exercise and fitness,and also can be applied to rehabilitation. Clinical trials have confirmed its effectiveness. Inthis paper, we analog ride with a new orthogonal6-PSS machine horse parallel platformand collect the rider’s physiological indicators information in motion. After the fusion ofphysiological indicators, the rider’s physiological reflects are analyzed, researched andjudged, as the feedback quantity to control the trajectory, posture and speed of the machinehorse. By the simulation and experimental, it has good effect on safe riding and healthimprovement. The specific research mainly includes the following aspects:(1) Analyzed the relationship between the physiological and exercise effects, andused the changes of heart rate, blood pressure, oxygen saturation to judge exerciseeffects.Used different algorithms to realize time registration, so improved data accuracyand stability.(2) Analyzed the biominetic robot horse’s different needs of information fusion ininitial stages and practical application of the follow-up stage. Used a fusion algorithmbased on the combination of Neural Network and SVM to the riding position seekingcontrol, to ensure the accuracy of the informaiton fusion in the practical of the subsequentstages, on which the sample gradually perfect.(3) Introduced the improved Newton’s descent method into the conventional BFGSaltorithm. In this algorithm, we do not have to pre-calculate the descent direction of theobjective function, and can search the value of the next descent point. So this algorithmmay reduce the amount of calculation and have a faster convergence speed and betterstability. The improved BFGS algorithm of gradient descent method to learn BP networkweights, which replaced the traditional one, so improved the speed and accuracy ofinformation fusion.(4) For the initial parameters (including weights and thresholds) vulnerable to have impact on convergence speed and accuracy of BP network, used an improved ant colonyalgorithm to select the optimal initial parameters. Propose a new method to search optimalparameters of BP networks based on improved ant colony algorithm. The proposedalgorithm is based on each ant searches only around the best solution of the previousiteration with parameter, which can reduce search space fast. Parameter is theproposed for improving ACO’s solution performance to reach global optimum fairlyquickly. Simulation results indicate that optimize parameters of BP networks with thismethod can not only overcome the limitations both the slow convergence and the localextreme values by basic BP algorithm, but also improve the learning ability andgeneralization ability.(5) In this study, quantum principles is introduced in particle swarm optimizationalgorithm, which can improves traverse property of particle of particle, thus can overcomethe limitation of local extreme values and get the optimal parameters of SVM. Simulationresults indicate that quantum particle swarm optimization-based SVM classifier has higherclassification accuracy than common optimization algorithm.(6) Basing on the feasibility analysis of design controller by the motion platform’sdynamic model, we used a fuzzy adaptive PID controller, which can realize the dynamic,real-time control of the moving platform. Using high performance trajectory device tocollect horse trajectory, so that6-DOF parallel machine horse sports platform can morefactually reproduce a variety of horse sports, while taking advantage of the horse ridersphysiological information fusion results to control the movement of the machine horsespeed and posture, in order to achieve the best sports effection. Simulation experimentsand field tests show that the machine horse posture optimization control system has a goodperformance. |
PDF Full Text Request