Research On Identification Technology Of Electric Workers' Working Conditions Based On Multi-sensor

With the increasing complexity of the installation environment of power equipment,there are hidden dangers in life during the maintenance and inspection of power equipment.In order to improve the safety of workers,more and more research and development techniques for electric power warning devices.However,the alarm distance is affected by the voltage level,and the electric field distribution is different under different environments,which causes the early warning device to be prone to false positives and false negatives.Research on voltage level identification technology in literature [1].However,the installation environment of the power equipment is different,resulting in different moving conditions near the charged body,affecting the accuracy of the voltage level identification.Therefore,it is necessary to distinguish the installation environment of the power equipment.Because of the electric field distortion in the electric field distribution,this paper establishes the idea of distinguishing the environment of power equipment based on the recognition of moving conditions.Due to there are many type of motion conditions when close to the charged body.At present,there are mainly image acquisition and inertial sensor extraction for motion attitude information extraction technology.Image acquisition has the disadvantages of limited use range and inconvenient carrying.Inertial sensor extracts information,which has the disadvantages of single data and vulnerability to interference.Based on the above problems,this paper proposes a multi-sensor based power operator's motion condition identification technology research.The main work content is as follows:(1)Aiming at the drift of the relative height of the air pressure sensor due to factors such as temperature,illumination and airflow,this paper proposes a multi-source information cooperative fusion algorithm.Using the acceleration value and the relative height value to fuse together,extract the effective relative height value,effectively eliminate the drift problem;(2)In order to accurately identify the motion conditions,this paper first extracts the effective feature parameters.Using mathematical analysis methods to extract the mean and variance of acceleration values.Fit the effective relative height values to obtain the fitting parameters.Secondly,based on the mean,variance and fitting parameters,two classification models based on cross-validation improved BP neural network and SVM support vector machine are established.Finally,the classification results of the head,waist and foot of the operator are combined to determine the final identification type of condition;(3)In order to solve the problems of inconvenient sensor and difficult data reception during data acquisition,this paper designs the data acquisition terminal with AltiumDesigner software and SolidWorks software to realize portable sensor and convenient data acquisition;(4)In order to realize the effective information visualization,this paper uses MATLAB software to design the GUI interface.In the online verification process of the model,the GUI information is used to quickly see the working condition identification result and the number of tests.