Machine learning for comparing energy appliance usage across different demographics

Primary supervisor

David Dowe

Co-supervisors

Using relevant available data-sets, we compare appliance usage across households of different demographics.  We then use machine learning techniques to infer how different households use different appliances at different times, resulting in diverse energy consumption behaviours. 


 

Student cohort

Double Semester

Aim/outline

We intend to endeavour to improve accuracy by taking machine learning techniques and integrating additional features of known patterns of use (such as time-of-day associated with activities) and/or characteristic load patterns (such as, e.g., transient load decay associated with refrigerator cooling cycles).  Such machine learning techniques might include Hidden Markov Models (HMMs), Long Short-Term Memory (LSTM) deep learning networks and/or applications of the Bayesian information-theoretic Minimum Message Length (MML) principle. 

URLs/references

    Fang, Z., D. L. Dowe, S. Peiris and D. Rosadi (2021), "Minimum Message Length in Hybrid ARMA and LSTM Model Forecasting", Entropy 2021, 23 (12), 1601; https://doi.org/10.3390/e23121601 (29 Nov 2021).

    Wallace, C.S. (2005), Statistical and Inductive Inference by Minimum Message Length, Springer (Series: Information Science and Statistics), 2005, XVI, 432 pp [see also preface - and p vi].
    Wallace, C.S. and D.L. Dowe (1999a). “Minimum Message Length and Kolmogorov Complexity, Computer Journal” (special issue on Kolmogorov complexity), Vol. 42, No. 4, pp270-283
    Wallace, C.S. and D.L. Dowe (2000). “MML clustering of multi-state, Poisson, von Mises circular and Gaussian distributions”, Statistics and Computing, Vol. 10, No. 1, Jan. 2000, pp73-83

    Wei, Z. and Wang H., Characterizing Residential Load Patterns by Household Demographic and Socioeconomic Factors, ACM e-Energy 2021 (The Twelfth ACM International Conference on Future Energy Systems, 2021. https://doi.org/10.1145/3447555.3464867
    Wong, Yung Fei, Lachlan LH Andrew, and Y. Ahmet Sekercioglu. "Hidden semi-Markov models for electricity load disaggregation." ACM SIGMETRICS Performance Evaluation Review 46.3 (2019): 18-23. https://dl.acm.org/doi/abs/10.1145/3308897.3308906

    Non-intrusive load monitoring toolkit: https://github.com/nilmtk/nilmtk
    Pecan Street Dataset - https://www.pecanstreet.org/dataport/

 

Required knowledge

Required knowledge


Candidates should have a strong computer science background with good programming skills, particularly in Java, Python and/or MATLAB. A good understanding of probability theory is a big plus, as is knowledge of at least one of mathematics, statistics and/or machine learning principles.