Open AccessA new intelligent time series prediction technique for coherency identification performance enhancement
Author(s) -
Mohd Farid Mohd Ariff
Publication year - 2021
Publication title -
maǧallaẗ al-kuwayt li-l-ʿulūm
Language(s) - English
Resource type - Journals
eISSN - 2307-4116
pISSN - 2307-4108
DOI - 10.48129/kjs.v48i4.10383
Subject(s) - nonlinear autoregressive exogenous model , autoregressive model , computer science , identification (biology) , generator (circuit theory) , control theory (sociology) , process (computing) , electric power system , nonlinear system , series (stratigraphy) , artificial neural network , time series , power (physics) , algorithm , artificial intelligence , machine learning , mathematics , control (management) , statistics , paleontology , botany , physics , quantum mechanics , biology , operating system
The manuscript proposes a time series prediction technique to enhance the response time of the coherency identification technique. The proposed methodology utilizes the nonlinear autoregressive exogenous neural network (NARX) algorithm to predict the generator speed deviations following a disturbance in the system. Consequently, the coherency identification technique based on independent component analysis (ICA) is utilized on the predicted system responses. The effectiveness of the proposed approach is demonstrated on of the IEEE 16-generator 68-bus test system model. The result shows that the proposed technique can predict 0:2s following a disturbance in the system accurately. Therefore, the NARX allows a 0:2s head start for the ICA to determine the real-time coherent generator group in the system. Furthermore, the result shows that the proposed approach can identify the coherent group of generators based on the predicted generator speed deviation in all cases considered in this study, accurately. Conclusively, the result implies that the proposed technique can speed up the overall coherency identification process in a power system operation.