Attention boosted autoencoder for building energy anomaly detection

作     者：Durga Prasad Pydi S.Advaith 

作者机构：Department of Mechanical EngineeringIndian Institute of Technology MadrasChennai600036India 

出 版 物：《Energy and AI》 (能源与人工智能（英文）)

年 卷 期：2023年第14卷第4期

页      面：479-489页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：This research was supported by the Prime Minister’s Research Fel-lows(PMRF)[project ID:SB22230925MEPMRF008998]under the Min-istry of Education Government of India 

主　　题：Artificial intelligence UN Sustainable Development Goals Interpretable model Multivariate time series HVAC 

摘      要：Significant energy savings can be realised from buildings if deviations from the usual operating conditions are detected early,and appropriate measures are *** anomaly detection techniques automate identifying such instances by leveraging the high dimensional data collected from the installed smart *** allow for dimensionality reduction and also model the underlying data ***,these models treat features as independent *** contrast,the current work investigates an attention mechanism with an autoencoder to include the correlations among the *** value addition from the attention mechanism is demonstrated by comparing the model’s reconstruction ability with an ANN-based autoencoder on synthetic *** study identifies that adding an attention layer enables the encoder–decoder architecture to be robust to outliers in training data,thereby reducing the preprocessing ***,the model is tested on a real-world dataset,and the attention maps generated from the model are used to interpret the correlations among the features and across the time dimension,thereby establishing a human-interpretable way to understand the reconstruction from the model.