| There are many type of protection device that are currently installed in all distribution network. Those protection devices react when a fault occurs in the network. However, some faults like an electrical arc or high impedance fault do not always activate those devices. Therefore, those faults can cause serious damage such as electrocution.;This thesis proposes a system able to detect electrical arc faults that occur in a distribution network. It is composed mainly with three current sensors and a Digital Signal Processing (DSP) of type TMS320F2812. The concept is based on digital filters running in the DSP. Those filters are created with the Signal Processing Toolbox in Matlab/Simulink. The Embedded Target for TI C2000 DSP toolbox in Matlab/Simulink was also used in order to create a complete model and to directly implement this model in the DSP. By doing a comparison with a threshold, the DSP can detect the presence of undesired frequencies in the network. When the DSP detects a predetermined number of those frequencies in a certain amount of time it will trigger a relay to open a three-phase switch in the distribution network.;Key words. Electrical fault, electrical arc, high impedance, distribution network, detection, protection, Matlab/Simulink, DSP.;In the last two decades, many researches have been done regarding electrical arc faults. Unfortunately, most of those works did not give a high efficiency. The main reason was in the difficulty to extract the information during the electrical faults. The generated current in the network during those faults was too low in magnitude and had random frequencies. |
