Abstract Acquiring knowledge from the growing amount of Building Automation and Control Systems (BACS) data is becoming a more and more challenging and complex engineering task. However, it is also a prerequisite for smart and sustainable energy management as well as improving energy efficiency and comfort of building users. This report analyses the prospects of applying selected supervised learning methods for time series classification in BACS. Our training and testing data covered multivariate time series from 5,142 data points located in E.ON Energy Research Center building, describing observations from 22 classes, such as temperatures of gaseous fluid, CO2 concentrations, heat flows, and operating messages. We trained thirteen types of classifiers: complex tree, medium tree, simple tree, linear Support Vector Machines, quadratic Support Vector Machines, boosted trees, bagged trees, subspace discriminant, subspace KNN, RUSBoosted Trees, Fine KNN, Coarse KNN and random forests. The highest demonstrated average classification accuracy concerned bagged trees (56.76%), with the maximum accuracy level of 76.54%. However, the maximum accuracy achieved by random forests was even higher, reaching 78.95%. Finally, we identified factors that may have a substantial influence on performance of particular methods.
[1]
Jason Lines,et al.
Classification of Household Devices by Electricity Usage Profiles
,
2011,
IDEAL.
[2]
Yannis Manolopoulos,et al.
Feature-based classification of time-series data
,
2001
.
[3]
Arno Schlueter,et al.
A review of unsupervised statistical learning and visual analytics techniques applied to performance analysis of non-residential buildings
,
2018
.
[4]
Ricardo Matsumura de Araújo,et al.
Towards skeleton biometric identification using the microsoft kinect sensor
,
2013,
SAC '13.
[5]
Eamonn J. Keogh,et al.
Real-Time Classification of Streaming Sensor Data
,
2008,
2008 20th IEEE International Conference on Tools with Artificial Intelligence.
[6]
Qiang Wang,et al.
Application of time series techniques to data mining and analysis of spatial patterns in 3D images
,
2004,
2004 IEEE International Conference on Acoustics, Speech, and Signal Processing.
[7]
Huey-miin Hsueh,et al.
Random forests-based differential analysis of gene sets for gene expression data.
,
2013,
Gene.