果蝇算法在基于LSSVM智能水下机器人操纵运动模型辨识中的应用  

Application of Fruit Fly Optimization Algorithm in Model Identification of Maneuverability Motion of Underwater Vehicle Based on LSSVM

作　　者：孙玉山[1] 徐昊 曹东东 庞永杰[1] 

机构地区：[1]哈尔滨工程大学水下机器人技术重点实验室,哈尔滨150001

出　　处：《船舶工程》2017年第2期94-98,共5页Ship Engineering

基　　金：中国博士后科学基金资助项目(2012T50331);国家"863"计划资助项目(2011AA09A106)

摘　　要：基于水下机器人的Z型仿真试验,应用支持向量机对水下机器人的操纵运动模型进行辨识,从核函数结构中得到水动力系数,并建立水下机器人的预报模型,引入果蝇算法对惩罚因子C进行寻优,以减少基于经验而选择的参数对辨识精度产生的影响。通过预报与仿真比较,验证了该方法的有效性。Support vector machine is applied to identify the maneuverability motion model of underwater vehicle based on the zig-zag simulation test in the paper. The hydrodynamic parameters are obtained from the structure of kernel function. The forecasting model of underwater vehicle is established. In order to reduce the effect of choosing parameters based on experience, fruit fly optimization algorithm is used to optimize the penalty factor C. Comparing the prediction results and simulation data verify the effectiveness of the method.

关 键 词：智能水下机器人 辨识 支持向量机 果蝇算法 

分 类 号：U661.33[交通运输工程—船舶及航道工程]