Application of Machine Learning to Express Measurement Uncertainty

Abstract

The continuing increase in data processing power in modern devices and the availability of a vast amount of data via the internet and the internet of things (sensors, monitoring systems, financial records, health records, social media, etc.) enabled the accelerated development of machine learning techniques. However, the collected data can be inconsistent, incomplete, and noisy, leading to a decreased confidence in data analysis. The paper proposes a novel “judgmental” approach to evaluating the measurement uncertainty of the machine learning model that implements the dropout additive regression trees algorithm. The considered method uses the procedure for expressing the type B measurement uncertainty and the maximal value of the empirical absolute loss function of the model. It is related to the testing and monitoring of power equipment and determining partial discharge location by the non-iterative, all-acoustic method. The example uses the dataset representing the correlation of the mean distance of partial discharge and acoustic sensors and the temperature coefficient of the sensitivity of the non-iterative algorithm. The dropout additive regression trees algorithm achieved the best performance based on the highest coefficient of determination value. Most of the model’s predictions (>97%) fell into the proposed standard measurement uncertainty interval for both “seen” and “unseen” data.

Towards Online Ageing Detection in Transformer Oil: A Review

“…The dependability of data analysis is increased by managing and quantifying model uncertainty. Additionally, it strengthens confidence in the model’s judgments [ 84 ]. A detailed review of uncertainty quantification for machine learning models and deep learning models highlights the use of Bayesian neural networks and Bayesian physics informed networks for deep learning uncertainty quantification, as well as gaussian process regression (GPR) and physics-informed neural networks for traditional machine learning [ 85 ].…”

Switchgear Digitalization—Research Path, Status, and Future Work

“…Consequently, the measurement uncertainty caused by sensors significantly affects the accuracy of machine/deep learning models, especially in safety-critical applications. To combat this recognized problem, several studies have been conducted with positive results, such as [ 39 ], which is limited to supervised machine learning regression techniques, and [ 40 ], which proposed a method for calibrating uncertainty prediction for regression tasks. Ultimately, the data obtained from the sensors, considering the measurement uncertainty, combined with IEDs using the IEC 61850 communication protocol, provided the basis for defining digital switchgear.…”