Abnormal Action Recognition with Lightweight Pose Estimation Network in Electric Power Training Scene

Electric power training is essential for ensuring the safety and reliability of the system.In this study,we introduce a novel Abnormal Action Recognition(AAR)system that utilizes a Lightweight Pose Estimation Network(LPEN)to efficiently and effectively detect abnormal fall-down and trespass incidents in electric power training scenarios.The LPEN network,comprising three stages—MobileNet,Initial Stage,and Refinement Stage—is employed to swiftly extract image features,detect human key points,and refine them for accurate analysis.Subsequently,a Pose-aware Action Analysis Module(PAAM)captures the positional coordinates of human skeletal points in each frame.Finally,an Abnormal Action Inference Module(AAIM)evaluates whether abnormal fall-down or unauthorized trespass behavior is occurring.For fall-down recognition,three criteria—falling speed,main angles of skeletal points,and the person’s bounding box—are considered.To identify unauthorized trespass,emphasis is placed on the position of the ankles.Extensive experiments validate the effectiveness and efficiency of the proposed system in ensuring the safety and reliability of electric power training.

Application of Optical Motion Capture Technology in Power Safety Entitative Simulation Training System

The safety production is critical to stable development of Chinese electric power industry. With the development of electric power enterprises, the requirements of its employees are also becoming higher and higher. In this paper, an optical motion capture system based on the virtual reality technology is proposed to meet the requirements of the power enterprise for the qualified business ability. Electric power equipment, power equipment model entitative operating environment and the human model are established by electric power simulation unit, ZigBee technology and OpenGL graphics library. The problem of missing feature points is solved by applying the human model driven algorithm and the Kalman filtering algorithm. The experimental results show that it is more accurate to use Kalman filtering algorithm to extract the feature point in tracking process of actual motion capture and real-time animation display. The average absolute error of 3D coordinates is 1.61 mm and the average relative error is 2.23%. The system can improve trainees’ sense of experience and immersion.

Dynamic Characteristics of Tracked Vehicle Power Train Considering Road Random Torsional Excitation

The road random torsional excitation is one type of torque rooted from the road roughness and vehicle drive system. This paper aims to study how the road random torsional excitation affects the dynamic characteristics of vehicle power train. The method of simulating the random torsional excitation of tracked vehicle is explored at first. Secondly,the road random torsional excitations under different road roughness,vehicle speeds and pre-tensions are obtained. Thirdly,the dynamic analysis model of tracked vehicle power train is constructed with the consideration of the road random torsional excitation. Eventually,the influences of this excitation on output torque,bearing support force,vibration acceleration and dynamic shear stress of transmission shafts are intensively studied.The research conclusions are helpful to correct and refine the present virtual prototype of tracked vehicle power train.