| Popis: |
Variation in short-circuit levels (SCLs) caused by the occasional operation of on-load tap-changers (OLTCs) presents a significant challenge in maintaining protection coordination in distribution networks (DNs), especially where the integration of distributed energy resources (DERs) increases complexities. To tackle this complex issue, this paper first develops an AI-assisted protection scheme utilizing smart relays (SRs) in conjunction with conventional overcurrent relays (CRs). In the proposed scheme, the SR logic is trained by deployment of a multilayer perceptron (MLP) model which enables the detection and classification of faults across diverse locations, accounting for a wide range of fault resistances, DERs outages, capacitor bank switching conditions, and on/off grid states, while also considering various OLTC tap levels. To maintain security and functionality in case of SR operation failures and to reduce model errors, the scheme designates SRs as the primary protection mechanism while CRs serving as a backup. Next, a techno-economic model employing a fuzzy decision-making approach is introduced to optimize the placement of SRs, minimizing relay replacement costs while ensuring technical coordination among SRs and CRs. The conventional and proposed protection frameworks are implemented and extensively evaluated on the IEEE 14-bus and PG&E 69-bus test systems. The obtained results demonstrate a 99.5% fault detection accuracy by SRs and an impressive 74.2% reduction in overall relay operation time. Moreover, complete miscoordination rectification is achieved by 25% replacement rate of SRs which confirm the technical superiority and economic optimality of the proposed framework. |
