Non-intrusive load monitoring based on graph signal processing

NALM (Non-intrusive Appliance Load Monitoring) is effective method to disaggregate a particular appliance from the aggregate active power measurements acquired from a common measurement point. Based on appliance level energy consumption data optimum energy utilisation can be obtained by formulating load scheduling strategies. Power consumption of an appliance load is smooth and follows a certain pattern, this motivates the use of graph based signal processing (GSP) for NALM. In the presented work we propose a different NALM method by incorporating on GSP, by representing active power dataset on a graph. Proposed regularization method is applied for maximizing the smoothness of the graph signal, which allow us to perform energy disaggregation in simpler manner. Proposed method is evaluated on publically available iAWE dataset.