Synthetic Power Consumption Data Generation For Appliance Operation Modes

Realistic appliance power consumption data plays a pivotal role in the development of smart home energy management systems and the foundational algorithms for appliance data analysis. However, publicly available datasets are often limited in availability and time-consuming to collect. Consequently, the creation of simulation models for generating synthetic appliance data becomes needed. In this research, a novel approach is designed to simulate power consumption data that is tailored towards appliance operation modes. This model leverages existing public datasets and employs stochastic methods to enhance data variability and consistency. Usage profile characteristics are extracted and used to generate base usage profiles. To synthesize usage profiles while ensuring realism, a probabilistic model is employed and tuning parameters, encompassing components such as white noise, switch-on surges are added to refine the synthetic profiles. The DTW algorithm is then utilized to assess the proximity of the synthetic profiles to the existing ones. Remarkably, our results reveal that the average differences among these profiles can be as low as ten samples, even with a 1Hz sampling frequency.