Contents lists available at ScienceDirect Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes An adaptive filters based PMU algorithm for both steady-state and dynamic conditions in distribution networks Yinfeng Wang a , Chao Lu a, ⁎ , Innocent Kamwa b , Chen Fang c , Ping Ling c a Department of Electrical Engineering, Tsinghua University, Beijing, China b Hydro-Québec Research Institute (IREQ), Varennes, Quebec, Canada c Shanghai Electric Power Research Institute, State Grid Corporation of China, Shanghai, China ARTICLE INFO Keywords: Phasor estimation ESPRIT Finite impulse response filter Taylor Fourier transform Transient recognition Distribution network ABSTRACT Although synchronized phasor measurement unit (PMU) technology is widely applied in transmission networks, conventional PMU algorithms suffer from new challenges (e.g. large frequency deviation, multiple interferences and dynamic behaviors, etc.) if directly applied to distribution networks. On the basis of predecessors' research, this paper presents a comprehensive PMU algorithm, using adaptive filters to simultaneously satisfy the high- precision and dynamic-response requirements of phasor estimation in distribution networks. In this algorithm, out-of-band (OOB) interferences that are difficult to be filtered out, are specifically considered in both steady- state and dynamic signal modelling processes. Based on the models, a finite impulse response (FIR) filter and a Taylor–Fourier transform (TFT) filter that both contains notches at the OOB interferences are elaborately de- signed and implemented in two parallel channels. With the support of a super-resolution frequency analysis method for signal components, the filters’ central frequency and notches’ position can be online adaptively adjusted. Finally, the outputs are switched between the two-channel estimations according to a designed tran- sient recognition strategy. Test results in IEEE standard scenarios and some other realistic but severe scenarios demonstrate the effectiveness and reliability of this proposed algorithm. 1. Introduction With the continuously growing penetration of distributed genera- tions (DGs), electric vehicles (EVs) and other interactive loads, certain new features gradually emerge in distribution networks. Such features include intermittency and randomness of power sources, changeable operating modes of power grids, diversity and interactivity of loads, and potential islanding morphology, all of which complicate the op- eration and control of distribution networks. With the development of wide area measurements (WAMS) technology, phasor measurement units (PMUs) are widely used in bulk power systems, such as state es- timation, dynamic monitoring, security assessment, stability control and other fields [1]. Theoretical research and application results have demonstrated its effectiveness and advancement. Compared to the traditional measurement methods, such as supervisory control and data acquisition (SCADA), PMUs have higher measurement accuracy, better dynamic performance, and unique global synchronization capability, which could help to address the new challenges in distribution net- works. Typical applications based on the synchrophasor can be pri- marily divided into three categories, i.e., monitoring, diagnostic and control, as discussed in [2–4], showing that the PMUs have good de- velopment prospects at the distribution level. It is well-known that the phasor estimation algorithm determines the precision and quality of PMU data under various measurement conditions, which is the cornerstone of all types of PMU-based tech- nologies and applications. However, the previous algorithms designed for transmission network are not suitable for distribution network ap- plications. Actually, measurement scenarios in distribution networks are more complicated. Firstly, distribution network lines are so short that the voltage phasor angle differences (tenths of a degree or less) are typically two orders of magnitude smaller than those in transmission networks, demanding the improvement of the angular resolution [2,5]. Even worse, there are more non-linear loads (e.g., rectifier, arc furnace, and frequency converter), distributed renewable generations (wind power, photovoltaic) and energy storages in distribution networks, which introduce a large number of interharmonics and strong noise [6–8]. Besides, distribution networks are closer to the customers than the transmission networks, resulting in more sophisticated dynamic behaviours [9,10], e.g. waveform oscillation, voltage flicker and fre- quency change. In summary, the signals in distribution networks have https://doi.org/10.1016/j.ijepes.2019.105714 Received 15 October 2018; Received in revised form 21 October 2019; Accepted 18 November 2019 ⁎ Corresponding author. E-mail address: luchao@tsinghua.edu.cn (C. Lu). Electrical Power and Energy Systems 117 (2020) 105714 Available online 04 December 2019 0142-0615/ © 2019 Elsevier Ltd. All rights reserved. T