LAUSR

Data-driven unified modeling for multiple energy-coupling device (MECD) with considering the changing of operation modes (OMs) triggered by seasonal variation or the alteration of production rhythm is a challenging task. To overcome this, in this article, an unified modeling framework by using four-order tensor-based generalized interval type-2 fuzzy neural network is proposed, which considers the multiple operation property of MECD generated in the production process. The unified modelling of MECDs lie in the aspects of input and output of multiple energy media, OMs, working mode, which are related to analyze the structure difference of various type, specification, the changing of OMs triggered by demand response, and the alteration of season affected on working mode. Moreover, a model-guided rule pruning and recalling strategy was presented for avoiding the catastrophic forgetting of fuzzy rules in the process of modeling. A series of real-world data collected from a steel industrial park are selected for verifying the effectiveness of the proposed method; the established unified model has a widespread applicability and its prediction result outperforms state-of-the-art approaches in aspect of flexibility and accuracy.