Identifying power system events using a long short-term memory neural network

• Grant number: 19/08200-5
• Support Opportunities: Scholarships abroad - Research Internship - Master's degree
• Effective date (Start): August 09, 2019
• Effective date (End): February 08, 2020
• Field of knowledge: Engineering - Electrical Engineering - Power Systems
• Principal Investigator: Daniel Dotta
• Grantee: Orlem Lima dos Santos
• Supervisor: Meng Wang
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Research place: Rensselaers Polytechnic Institute, United States
• Associated to the scholarship: 17/25425-5 - Analysis of artificial neural networks methodologies for event classification using Synchrophasors, BP.MS

Abstract

Nowadays, the real-time identification of events is crucial to the operation of Electrical Power Systems (EPS). In this context, the WAMS (Wide Area Measurement System) networks are capable to realize simultaneous measurements of voltage and current, known as synchrophasors. These features allow the real-time monitoring of the dynamic response of the large power systems. The WAMS networks capture and store a large amount of data that must be analyzed in order to extract relevant information about the EPS performance. Thus, there is a need to explore algorithms of data mining, such as Neural Network (NN), since they may allow fast and efficient extraction of significant information about the EPS. The research realized so far has focused on the application of static, non-recurrent, NN that do not aggregate the ability to identify the non-stationary (or dynamic) characteristics present in an EPS. Considering this, the main concern of this BEPE research project is to develop a data-driven event identification method, focusing on a dynamic NN called Long Short-Term memory (LSTM), that could accurately identify different types of events. In addition, we propose to train a LSTM on the dominant eigenvalues and singular values as features instead of training on time series directly. The method will be evaluated on real events of Brazilian Interconnected Power System (BIPS). (AU)

Publicações científicas

(4)

(Referências obtidas automaticamente do Web of Science e do SciELO, por meio da informação sobre o financiamento pela FAPESP e o número do processo correspondente, incluída na publicação pelos autores)

ALCAHUAMAN, HEVER; LOPEZ, JUAN CAMILO; DOTTA, DANIEL; RIDER, MARCOS J.; GHIOCEL, SCOTT. Optimized Reactive Power Capability of Wind Power Plants With Tap-Changing Transformers. IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, v. 12, n. 4, p. 1935-1946, OCT 2021. (16/08645-9, 19/08200-5, 15/21972-6, 19/01906-0, 17/21752-1, 19/10033-0, 18/20104-9, 17/25425-5)

PINHEIRO, BRUNO; LUGNANI, LUCAS; DOTTA, DANIEL; IEEE. A Procedure for the Estimation of Frequency Response using a Data-Driven Method. 2021 IEEE POWER & ENERGY SOCIETY GENERAL MEETING (PESGM), v. N/A, p. 5-pg., 2021-01-01. (18/20104-9, 19/08200-5, 16/08645-9, 19/10033-0, 17/25425-5)

LOPES, GABRIEL V. DE S.; MORAES, GUIDO R.; ISSICABA, DIEGO; DOTTA, DANIEL. WAMS-based two-level robust detection methodology of power system events. SUSTAINABLE ENERGY GRIDS & NETWORKS, v. 31, p. 13-pg., 2022-09-01. (19/08200-5, 19/10033-0, 18/20104-9, 17/25425-5, 16/08645-9)

SANTOS, ORLEM L. D.; DOTTA, DANIEL; WANG, MENG; CHOW, JOE H.; DECKER, ILDEMAR C.. Performance analysis of a DNN classifier for power system events using an interpretability method. INTERNATIONAL JOURNAL OF ELECTRICAL POWER & ENERGY SYSTEMS, v. 136, p. 15-pg., 2022-03-01. (19/08200-5, 18/20104-9, 17/25425-5, 16/08645-9, 19/10033-0)