A method to estimate speed droop ratios for primary frequency control responses in power generation units. (July 2020)

• Record Type: Journal Article
• Authors: Yang, Zijiang; Wang, Jiandong; Gao, Song; Zhou, Donghua
• Is Part Of: International journal of electrical power & energy systems. Volume 119(2020)
• Publication Date: 2020-07
• Subjects: Primary frequency control -- Speed droop ratio -- Power generation unit -- Hidden Markov model -- Piece-wise linear representation; Electrical engineering -- Periodicals; Electric power systems -- Periodicals; Électrotechnique -- Périodiques; Réseaux électriques (Énergie) -- Périodiques; Electric power systems; Electrical engineering; Periodicals; 621.3
• DOI: 10.1016/j.ijepes.2020.105868
• Languages: English
• ISSNs: 0142-0615

Abstract

Highlights: Primary frequency control (PFC) responses of power generation units are considered. A method is proposed to estimate speed droop ratios for PFC responses. PFC response segments are selected from a trained hidden Markov model. Three stages are formulated from PLRs of selected PFC response segments. Numerical and industrial examples are provided to support the proposed method. Abstract: Primary frequency control (PFC) in power generation units is very important for power grids to suppress frequency fluctuations and preserve load-generation balance. The speed droop ratio is the most crucial performance index for a PFC response. This paper proposes a method to estimate speed droop ratios from historical data for PFC responses of power generation units. The proposed method is composed by three steps. The first step selects data segments containing PFC responses through a hidden Markov model describing underlying frequency states. The second step obtains piece-wise linear representations (PLRs) of the generated power and power grid frequency in the selected PFC data segments. The third step formulates different PFC response stages based on the PLRs, from which speed droop ratios are calculated. Numerical and industrial examples are provided to support the proposed method and compare with two existing ones.