Investigating the validity of applying artificial neural networks to localise harmonic distortion sources [power quality management]

Power system equipment is designed to operate under sinusoidal conditions. Damage and even failure of equipment can result from the presence of a harmonically distorted supply waveform. The level of harmonic distortion on modern power systems is a concern as more non-linear devices are used in sophisticated energy conversion. Harmonic mitigation techniques are most effective when the location of the harmonic distortion source is known. Tariff systems cannot allocate distortion costs amongst the utility's consumers when the relative distortion contributions cannot be assigned to distinct sources. A reliable localisation technique is therefore required. In this paper, the application of artificial neural networks as a possible aid in the localisation of distortion sources, is investigated and demonstrated