Arcing faults are a major problem in the electrical networks of aircraft. Installing a protection system, activated when a fault is detected, is one of the most effective solutions for avoiding the fires and accidents that they cause. Thanks to recent advances in artificial intelligence (AI), we can now envisage the development of reliable detectors that meet aeronautical specifications. However, in order to train an AI unit, it is necessary to create a large database of arc signals. This article proposes a method for generating thousands of arc signals, using the state-space representation formalism, according to Andrea’s arc model. We will show how it is possible to simulate arc-fault signals by using the studies of two cases of arc faults. We will compare the results of the simulations with experimental measurements in order to demonstrate that the arc signals obtained are identical to those of the experiment with root-mean-square deviation errors of 0.048 V and 0.005 A.