A Machine Learning Approach to Event Analysis in Distribution Feeders Using Distribution Synchrophasors

Abstract

This paper proposes a machine learning (ML) approach to detect, identify, and analyze the events that occur on distribution networks using data streams from real-world distribution-level phasor measurement units (PMUs). First, we develop two statistical event detection methods. One is based on testing absolute values around median and the other one is based on testing residuals on a non-linear estimation. Both methods use moving windows as well as dynamic window size. This allows us to detect events of different types and durations. Next, we use field expert knowledge to assign labels to the detected events and subsequently develop a multi-class support vector machine classifier to classify power quality events. Finally, we apply the above developed techniques to detect, identify, and analyze the events in a micro-PMU data stream from a real-world test site in Riverside, CA. We particularly study the oscillation events that occur somewhere across the distribution feeder itself, where their impacts are observed remotely by the available micro-PMUs.

Machine learning for protection of distribution networks and power electronics-interfaced systems

“…On one hand, the unsupervised event detection methods do not require labeled dataset compared to supervised methods. On the other hand, the unsupervised algorithms are highly sensitive to the choice of model parameters, which results in either overlooking some of the power quality events, or increasing the rate of false positive, i.e., predicting normal operating samples as events [10].…”

“…The main challenge toward achieving the situational awareness is the lack of high-resolution measurements in DNs, which has recently been resolved by the advent of distribution-level phasor measurement units (PMUs), commonly referred to as D-PMUs [10], [11], [12], [13], [14], [15]. An important step in leveraging high-resolution measurements for situational awareness is to recognize valuable portions out of the extremely large datasets from raw time-series streams [10], [11]. In the context of data analysis in DNs, the valuable portions of data mostly refer to the events.…”

“…Different techniques have been introduced in the literature to detect events in distribution grids. The proposed techniques in prior studies can be categorized into two broad categories: statistical methods [10], [16], [17], [18], [19], [20], [21], and ML methods [22], [23], [24], [25], [26]. On the other hand, the proposed methods in the literature can be categorized as either model-based methods [27] or model-free techniques.…”

Unsupervised Event Detection, Clustering, and Use Case Exposition in Micro-PMU Measurements

“…One of the emerging applications of micro-PMUs is to study "events" in power distribution systems. Event-based studies of micro-PMU measurements have a wide range of use cases, such as in situational awareness [2], equipment health diagnostics, such as for inverters [3], capacitor banks [4], transformers [5], distribution-level oscillation detection and analysis [6], fault analysis and fault location [7].…”

Research on situation assessment of active distribution networks considering cyberattacks

“…At present, situation assessment as the core content of power grid situation awareness, the study of situation assessment can be mainly divided into three categories. 1) From the perspective of power grid dispatching control center, the situation awareness technology to the power grid operation control, and an intelligent dispatching system based on situation awareness are applied by Lai et al (2020), Shahsavari et al (2019) and Li et al (2015). 2) The operation situation assessment and projection methods of power grid based on massive data collected by wide-area measurement system are studied by Li et al (2020), Liu et al (2018), ), Jena et al (2017 and Ren et al (2019).…”