A robust approach for the decomposition of high-energy-consuming industrial loads with deep learning

The knowledge of the users' electricity consumption pattern is an important coordinating mechanism between utility company and consumers in terms key decision makings. load decomposition therefore crucial to reveal underlying relationship its characteristics. However, conventionally performed on residential commercial loads, adequate consideration has not been given high-energy-consuming industrial loads leading inefficient results. This paper thus focuses park (IPL). commonly used parameters a conventional method are however inapplicable loads. Therefore, more robust approach developed comprising three-algorithm model achieve this goal IPL. First, improved variational mode (IVMD) algorithm introduced denoise training data IPL improve stability. Secondly, convolutional neural network (CNN) simple recurrent units (SRU) joint algorithms non-intrusive non-invasive process using double-layer deep learning based Specifically, CNN extract characteristics while long short-term memory (LSTM) network, SRU, adopted develop further train data. Through process, extracted. results obtained from numerical examples show that outperforms state-of-the-art process.