Merging of various indicators and levels of disturbances for optimized power quality characterization

Matching various types of disturbances helps to accurately assess the stress imposed on load and electrical network itself. Electrical equipment susceptibility to disturbances varies significantly from type to type. Additionally, not only one physical phenomenon (e.g. harmonics) can cause equipment malfunctioning or damage, but more likely a superposition of some different disturbances. On the contrary, certain disturbance levels, even higher than prescribed in standards - if not superimposed with others disturbances - can be harmless to some types of equipment. Logical and functional coupling between disturbances sources and sinks is needed. The authors consider a neuro-fuzzy architecture appropriate for power quality characterization, establishing a promising link between distortions levels and equipment susceptibility. Neuro-fuzzy system performance has been compared with radial basis function neural networks.