RSS Based Localization for Partial Discharge Source Using Received Signal Strength Only

Partial discharge (PD) localization has been performed on a periodic or on a request basis to assess the health of high-voltage (HV) systems mainly due to lack of feasibility of techniques for continuous monitoring and localization. Advancements in the field of communication technology have made it possible to detect and locate PD activity in HV systems on a continuous basis. Existing PD localization techniques mainly include the time of arrival (TOA), time difference of arrival (TDOA) and angle of arrival (AOA) methods. These techniques require time-based synchronization of sensor nodes that are involved in the receiver system resulting in expensive and complex hardware and software solutions.

In this thesis, a received signal strength (RSS) based localization of PD is proposed. It is demonstrated that RSS based localization can be used under anonymous and harsh industrial environments for PD localization. RSS based localization does not require synchronization because unlike TOA, TDOA and AOA, it processes the amplitude of the received signal and not its phase.

A theoretical model of the algorithm has been developed based on the path loss model equation. Simulations have been performed to prove the principle in noiseless and noisy conditions before the experimental study was conducted. Artificial noise has been generated to test the performance of the algorithm in different noise conditions.

To explore the algorithm in real substation environments, an empirical study was performed in indoor and outdoor environments. Artificial PD signal is generated by using a high voltage partial discharge (HVPD) Pico Coulomb (pc) calibrator to perform the field trials at two different sites i.e., power network distribution centre (PNDC) at the University of Strathclyde and TATA Steel at Port Talbot, Wales. A specialised radiometer sensor is used to measure PD signals. Received signals from voltage levels are converted into power signals (dBm) as input to the location algorithm. Various sensors configurations in indoor and outdoor environments were used. The algorithm’s performance was evaluated based on four parameters which include, the estimated location, localization error, the path loss exponent (PLE) optimisation and the scalability. Simulation and experimental studies show that there is sufficient agreement and RSS based localization is a promising technique that can be used autonomously in future condition monitoring of HV systems on a continuous basis.