Model identification study on micro robot mobile in liquid based on support vector machine

How to identify the frequency-velocity model of micro robot mobile in liquid which was driven by intelligent material was studied in this paper. The sample data set that reflected the relationship between the vibrating frequency of driving flanks and moving velocity of micro robot was obtained by experiments. After comparing the of regression support vector machine (RSVM) with the least square support vector machine (LS-SVM), this paper selected the latter method to identify the model. Then the LS-SVM was trained and testified by the sample data set and contrast experiments had been done. The primary testing result shows that the identified model is accurate if the training information is enough. The general performance testing result shows that the identified model reflects the nonlinear mapping relationship between the flanks driven frequency and moving velocity of robot even though the setting frequency is out of range of sample data set. The dynamic performance result shows that the identified model is not only quite accurate during the positive step period but also precise during the negative step period . The study is the important basic to control the micro robot mobile in liquid.