Model Based-Testing of Spatial and Time Domain ArtificialIntelligence Smart Antenna for Ultra-High Frequency ElectricDischarge Detection in Digital Power Substations

Abstract

This paper presents a fifth-generation (5G) wireless smart antenna for performing bothpower substation communication (in space domain beam-steering) and electrostatic discharge (in timedomain Ultra-high Frequency “UHF” impulse) detection. The same smart antenna used to communicatewith other wireless antennas in the switchyard, as well as with the control room, is utilized to cyclicallygather data from power apparatus, busbars, and switches where electrostatic discharge (ESD) mayoccur. The ESD poses a major threat to electrical safety and lifetime of the apparatus as well asthe stability of the power system. The same smart antenna on which beam rotation in space-domainis designed by implementing an artificial neural network (ANN) is also trained in time-domain toidentify any of the received signals matching the ultra-high frequency band electrostatic dischargepulses that may be superimposed on the power frequency electric current. The proposed system ofelectrostatic discharge detection is tested for electrostatic pulses empirically simulated and representedin a trigonometric form for the training of the Perceptron Neural model. The working of the system isdemonstrated for electrostatic discharge pulses with rising times of the order of one nanosecond. Theartificial intelligence system driving the 5G smart antenna performs the dual roles of beam steeringfor 5G wireless communication (operating in the space domain) and for picking up any ESD generatedUHF pulses from any one of the apparatus or nearby lightning leaders (operating in the time domain).