The progress of technology on energy and IoT fields has led to an increasingly complicated\nelectric environment in low-voltage local microgrid, along with the extensions of electric vehicle,\nmicro-generation, and local storage. It is required to establish a home energy management system\n(HEMS) to efficiently integrate and manage household energy micro-generation, consumption and\nstorage, in order to realize decentralized local energy systems at the community level. Domestic\npower demand prediction is of great importance for establishing HEMS on realizing load balancing\nas well as other smart energy solutions with the support of IoT techniques. Artificial neural networks\nwith various network types (e.g., DNN, LSTM/GRU based RNN) and other configurations are\nwidely utilized on energy predictions. However, the selection of network configuration for each\nresearch is generally a case by case study achieved through empirical or enumerative approaches.\nMoreover, the commonly utilized network initialization methods assign parameter values based\non random numbers, which cause diversity on model performance, including learning efficiency,\nforecast accuracy, etc. In this paper, an evolutionary ensemble neural network pool (EENNP)\nmethod is proposed to achieve a population of well-performing networks with proper combinations\nof configuration and initialization automatically. In the experimental study, power demand\npredictions of multiple households are explored in three application scenarios: optimizing potential\nnetwork configuration set, forecasting single household power demand, and refilling missing data.\nThe impacts of evolutionary parameters on model performance are investigated. The experimental\nresults illustrate that the proposed method achieves better solutions on the considered scenarios.\nThe optimized potential network configuration set using EENNP achieves a similar result to manual\noptimization. The results of household demand prediction and missing data refilling perform better\nthan the naïve and simple predictors.
Loading....