Random Forest Regressor-Based Approach for Detecting Fault Location and Duration in Power Systems.

Zakaria El Mrabet,Niroop Sugunaraj,Prakash Ranganathan,Shrirang Abhyankar

doi:10.3390/s22020458

Zakaria El Mrabet, Niroop Sugunaraj + Show 2 more

Open Access

https://doi.org/10.3390/s22020458

Copy DOI

Abstract

Power system failures or outages due to short-circuits or “faults” can result in long service interruptions leading to significant socio-economic consequences. It is critical for electrical utilities to quickly ascertain fault characteristics, including location, type, and duration, to reduce the service time of an outage. Existing fault detection mechanisms (relays and digital fault recorders) are slow to communicate the fault characteristics upstream to the substations and control centers for action to be taken quickly. Fortunately, due to availability of high-resolution phasor measurement units (PMUs), more event-driven solutions can be captured in real time. In this paper, we propose a data-driven approach for determining fault characteristics using samples of fault trajectories. A random forest regressor (RFR)-based model is used to detect real-time fault location and its duration simultaneously. This model is based on combining multiple uncorrelated trees with state-of-the-art boosting and aggregating techniques in order to obtain robust generalizations and greater accuracy without overfitting or underfitting. Four cases were studied to evaluate the performance of RFR: 1. Detecting fault location (case 1), 2. Predicting fault duration (case 2), 3. Handling missing data (case 3), and 4. Identifying fault location and length in a real-time streaming environment (case 4). A comparative analysis was conducted between the RFR algorithm and state-of-the-art models, including deep neural network, Hoeffding tree, neural network, support vector machine, decision tree, naive Bayesian, and K-nearest neighborhood. Experiments revealed that RFR consistently outperformed the other models in detection accuracy, prediction error, and processing time.

Highlights

Fault identification is critical for seamless power grid operation
These results suggest that deep neural network (DNN), Hoeffding tree (HT), and random forest regressor (RFR) are the optimal models for predicting fault duration as the difference between the actual and predicted duration for the entire testing dataset was less than 0.6 s
The accuracy of the RFR model increases to 91% detection for the medium duration, followed by HT with 22%, DNN with 16%

Summary

Introduction

Fault identification is critical for seamless power grid operation. Utilities are working around the clock to reduce outage rates from interruptions such as contact with natural vegetation, animals, or weather events [1,2,3]. The cost to various consumers for a one-hour outage during a summer afternoon was estimated to be approximately USD 3 for a typical customer, USD 1200 for small and medium organizations, and USD 82,000 for large organizations [4]. These outage costs increased substantially depending on the time of year and outage duration, especially when they occur during winter. Predicting faults in the system along with their duration is the first step towards reducing the number of unplanned outages and providing a prediction-based plan to the utility for deploying the appropriate maintenance crews and the sequence of operations [5,6]

Methods

Results

Discussion

Conclusion