How well can HMM model load signals

Finding models that can efficiently represent load signals is one key issue in non-intrusive load monitoring (NILM) because they are the foundation of most load disaggregation algorithms. In the past, the factorial hidden Markov model (fHMM) has been proposed as one probabilistic model for the aggregate real power measurement. It is assumed that each load can be represented as one hidden Markov model (HMM) and the HMMs of all loads have been learned successfully before disaggregation. Although fHMM showed some promising results for eventless disaggregation, a detailed investigation on how well HMM is suited to model load signals is still missing till today. In this paper, we study the feasibility of HMM modeling for different categories of loads by using the UK-DALE dataset and propose a method for model adaptation across different houses.

[1]  Alex Rogers,et al.  Non-Intrusive Load Monitoring Using Prior Models of General Appliance Types , 2012, AAAI.

[2]  L. Rabiner,et al.  An introduction to hidden Markov models , 1986, IEEE ASSP Magazine.

[3]  Dietmar Bruckner,et al.  A hidden Markov model based procedure for identifying household electric loads , 2011, IECON 2011 - 37th Annual Conference of the IEEE Industrial Electronics Society.

[4]  Daniel P. W. Ellis,et al.  Multi-channel source separation by factorial HMMs , 2003, 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03)..

[5]  M. Baranski,et al.  Genetic algorithm for pattern detection in NIALM systems , 2004, 2004 IEEE International Conference on Systems, Man and Cybernetics (IEEE Cat. No.04CH37583).

[6]  Eric C. Larson,et al.  Disaggregated End-Use Energy Sensing for the Smart Grid , 2011, IEEE Pervasive Computing.

[7]  Bin Yang,et al.  A novel DNN-HMM-based approach for extracting single loads from aggregate power signals , 2016, 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[8]  Jack Kelly,et al.  The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes , 2014, Scientific Data.

[9]  T. Suzuki,et al.  Nonintrusive appliance load monitoring based on integer programming , 2008, 2008 SICE Annual Conference.

[10]  Lucio Soibelman,et al.  Robust adaptive event detection in non-intrusive load monitoring for energy aware smart facilities , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).