Probabilistic Forecast Combination for Anomaly Detection in Building Heat Load Time Series

TL;DR

This paper introduces a probabilistic forecast combination method for anomaly detection in building heat load time series, improving robustness and accuracy across diverse building profiles with complex seasonalities and noise.

Contribution

It proposes a novel ensemble-based probabilistic approach for anomaly detection that enhances detection accuracy in challenging heat load time series.

Findings

Forecast combination improves anomaly detection accuracy.

Probabilistic approach handles diverse building profiles.

Method robust to noise and multiple seasonalities.

Abstract

We consider the problem of automated anomaly detection for building level heat load time series. An anomaly detection model must be applicable to a diverse group of buildings and provide robust results on heat load time series with low signal-to-noise ratios, several seasonalities, and significant exogenous effects. We propose to employ a probabilistic forecast combination approach based on an ensemble of deterministic forecasts in an anomaly detection scheme that classifies observed values based on their probability under a predictive distribution. We show empirically that forecast based anomaly detection provides improved accuracy when employing a forecast combination approach.