We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Overload Protection of Marine Power Generators Using Supervised Machine Learning-Based Kalman Predictor Algorithm.
- Authors
Narayanswamy, Vedachalam; Jyothi Vandavasi, Bala Naga; Vittal, Doss Prakash; Raju, Ramesh; Arumugam, Vadivelan
- Abstract
Reliable protection is the key requirement for modern multi-megawatt–capacity marine power systems with multiple synchronous generators of various capacities operated in parallel. Outage of the mains or unavailability of the other generators results in momentary overload to the healthy generator(s). Conventional load surge protection relays based on a zero-crossing detection algorithm require a minimum of 10 ms to detect a momentary overload, which could be detrimental for lower inertia/capacity generators. This article presents a novel supervised machine-learning–based Kalman predictor algorithm that could detect momentary overloads in <1 ms by estimating error covariance. The algorithm is also found to be effective for IEC 61000 Class C power networks with total voltage harmonic distortions up to 10%.
- Subjects
MARINE resources conservation; SYNCHRONOUS generators; SUPERVISED learning; ALGORITHMS
- Publication
Marine Technology Society Journal, 2021, Vol 55, Issue 5, p109
- ISSN
0025-3324
- Publication type
Article
- DOI
10.4031/mtsj.55.5.16