Abstract:
This research work presents a robust, scalable and novel cloud based internet of things
(IoT) system for localization and identification of faults in power distribution systems
(PDS). Efficient fault localization techniques in power distribution systems can result in
reduced power recovery times. Target of this research work is to design a new algorithm
based on the modern IoT communication infrastructure that can detect simultaneous and
individual faults in power distribution network. The algorithm employs a zone-based
approach to detect and localize faults. To reveal different faults, current sensing devices
(CSD) are installed at zone boundaries. The objective of CSD is to yield time
synchronized current measurements. The current measurements are communicated to
cloud server through an edge device (ED). Furthermore, context aware policy (CAP) was
introduced in ED. Using CAP only that data was sent to the cloud which was different
from previously transmitted measurements. As a result, significant improvement was
achieved in terms of communication channel bandwidth. Furthermore, a device failure
mechanism was added to detect single or multiple simultaneous device failures. In case
of device failures or unavailability of complete sensor data, the proposed system can
modify the underlying algorithm in cloud server automatically to provide optimal results
based on the available information. Fault data was stored in a relational database model
also hosted in the cloud server. In order to substantiate the performance of presented
algorithm, an IEEE 37 node test feeder was selected as PDS. Three test cases were
implemented to observe the performance under different conditions. First two test cases
were designed to validate the performance of our system under individual and multiple
simultaneous faults in PDS. In the third test case single as well as multiple device failures
were introduced to establish the robustness and scalability of the proposed solution. It was found that the new algorithm successfully detected the faults for all three test cases.
Consequently, due to the adoption of CAP, significant reductions were noticed in the
volume of information sent over the network. In the end, a comparative study was
presented with existing methods to further highlight the benefits of our technique.
