This document discusses using a radial basis function neural network (RBFNN) to analyze harmonics in a power system due to a single-phase welding machine. The RBFNN is trained on data collected from the voltage and current waveforms of a welding machine connected to a power system. The RBFNN is able to accurately determine the harmonic components present in the power system due to the non-linear load of the welding machine. The RBFNN training and testing results demonstrate it is effective at measuring and identifying harmonics generated by welding machines connected to power distribution systems.