Demand Profiling Oriented Nuisance Tripping Forecasting for Residential Power Network

Heating, ventilation, and air conditioning (HVAC) loads contribute significantly to the energy consumption of a residential power network. Environmental factors, such as ambient temperature, humidity, wind, and solar radiation, determine the power demand of the HVAC loads. Due to the constant variation of these factors throughout the day, the daily power demand curve of HVAC loads exhibits significant fluctuations, especially during peak hours. These fluctuations can impact the power quality of the entire distribution network. Additionally, many small-scale residential HVAC units are operated by on/off controllers, resulting in load curves characterized by sharp transitions rather than smooth variations. As a result, traditional expected-value-based power demand estimation methods may be less effective for residential power networks, particularly those with a limited number of loads, such as microgrids and small residential complexes. Furthermore, certain power distribution systems are deployed in outdoor environments, for example, outdated residential buildings and expeditionary power networks. In these scenarios, extreme operating temperatures can lead to circuit breakers tripping at unexpected load levels, posing risks to the resilience of the power distribution system. To address these challenges, this paper proposes a comprehensive approach. First, autoregression-based time series methods are used to forecast ambient environmental trends. Building upon this forecast, general regression neural network tools are employed to predict the expected value of the HVAC power demands. Subsequently, the on/off probabilities of the HVAC units are estimated based on their expected power demands at different hours of the day. This systemic model allows for the detection of potential unexpected trips in residential power networks.