Computer-aided evaluation of movement and appearance of vacuum arcs

High-speed imaging is a common method to investigate plasma phenomena in vacuum. This paper presents a new method to measure and characterize arc modes by means of object detection and image analysis. The subject of investigation is the vacuum interrupter (VI). The VI can perform breaking operations for rated and short circuit currents. At switching operation, the energy of the switching arc is distributed over the contact surface using magnetic fields to guarantee a constant performance over the period of service. The visual appearance of the vacuum arc consisting of metal vapor is classified in different modes depending on contact geometry, distance and current. Within these modes diffuse plasma or different kinds of constriction can appear. The comparison of the investigations can vary between different publications due to the high complex processes and the different interpretations of the resulting visual appearances. This paper suggests a reproducible approach to automate the evaluation of images. Here a custom-made vacuum chamber with butt electrodes with 35 Hz and 7 kA peak current is used. First results of arc mode characterization for long gap lengths and long arcing times are presented as a reference.The valuation of different modes, the tracking of the vacuum arc and the algorithms implementation are key contents.