Identification of Motion Conditions Based on Self-organizing Competitive Neural Network Algorithm

2018 International Conference on Advanced Electronic Materials, Computers and Materials Engineering

材料科学;无线电电子学;计算机科学

Guolei, Xu^1,2 ; Nianrong, Zhou^1,3 ; Lijun, Tang^1 ; Wenbin, Zhang^2

Yunnan Power Grid Electric Power Research Institute of LLC, Kunming

650236, China^1

Faculty of Mechanical and Electrical Engineering, Kunming University of Science and Technology, Kunming

650500, China^2

School of Electrical Engineer, Chongqing University, Chongqing

400044, China^3

关键词: Acceleration sensors;    Barometric pressure;    Condition identification;    Experimental platform;    Fitting parameters;    Multi-source informations;    Outdoor environment;    Self-organizing competitive neural network;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/439/3/032092/pdf
DOI  :  10.1088/1757-899X/439/3/032092

来源: IOP

【 摘 要 】

Against the false alarms and false negatives caused by the error of alarm threshold selection because different working conditions when the early warning device is close to the same charged body, this paper proposes self-organizing competitive neural networ modle for Identification working conditions of worker, which climbing towers, climbing slopes and horizontal walking. Firstly, acceleration sensors and barometric pressure sensors are used to collect the acceleration value and barometric data of the head during the exercise of the experimenter. Secondly, Multi-source information collaborative filtering processes data to obtain effective relative height values and obtain fitting parameters by first-order fitting. Finally, building self-organizing competitive neural network model based on parameters. This paper selects outdoor towers, slopes with a slope of about 30° and horizontal roads as experimental platforms and collect 400 sets of data for each platform. Then randomly select 900 sets of data for training, 300 sets of data for verification. The experimental results show that the accuracy of the training sample reaches 94.67%, and the accuracy of the test sample reaches 92.73%, which meets the requirements for working condition identification in the outdoor environment.