Data Processing Methods of Flow Field Based on Artificial Lateral Line Pressure Sensors

作     者：Bing Sun Yi Xu Shuhang Xie Dong Xu Yupu Liang Bing Sun;Yi Xu;Shuhang Xie;Dong Xu;Yupu Liang

作者机构：School of Electronics and Information EngineeringBeihang UniversityBeijing100191BeijingChina School of Automation Science and Electrical EngineeringBeihang UniversityBeijing100191BeijingChina 

出 版 物：《Journal of Bionic Engineering》 (仿生工程学报（英文版）)

年 卷 期：2022年第19卷第6期

页      面：1797-1815页

核心收录：

学科分类：08[工学] 080202[工学-机械电子工程] 0802[工学-机械工程] 0811[工学-控制科学与工程] 

基　　金：National Natural Science Foundation of China(NSFC)under Grant 62073017. 

主　　题：Robotic fish Artificial lateral line Data processing Neural network 

摘      要：The estimation of the type and parameter of flow field is important for robotic fish.Recent estimation methods cannot meet the requirements of the robotic fish due to the lack of prior knowledge or the under-fitting of the model.A processing method including data preprocessing,feature extraction,feature selection,flow type classification and flow field parameters estimation,is proposed based on the data of the pressure sensors in an artificial lateral line.Probabilistic Neural Network(PNN)is used to classify the flow field type and the Generalized Regressive Neural Network(GRNN)is the best choice for estimating the flow field parameters.Also,a few filtering methods for data preprocessing,three methods for feature selection and nine parameters estimation methods are analysis for choosing better method.The proposed method is verified by the experiments with both simulation and real data.