Non-intrusive Load Monitoring Using Factorial Hidden Markov Model Based on Gaussian Mixture Model

Load disaggregation is a difficult problem in non-intrusive load monitoring. This paper proposes a load disaggregation method by the combination of supervised learning and unsupervised learning. Firstly, the proposed model uses the sum of squared errors to determine the optimal clustering number. Secondly, Gaussian mixture model is applied to achieve robustness to noise and outliers. Thirdly, the hidden Markov model and factorial hidden Markov model are established for each load and total loads, respectively. By disaggregating the window of state change, the computational complexity is effectively reduced. The simulation experiment is carried out with the AMPds data set. The effect of changing the clustering numbers of the third type load on the disaggregation accuracy is studied. Moreover, the proposed model has been compared with other models. The simulation results show that the proposed model can not only reduce the computational complexity but also effectively improve decomposition accuracy.