Open AccessA data-driven fusion model for energy consumption prediction of hot extrusion forming
Author(s) -
Dan Yang,
Hongcheng Li,
Rong Luo,
Yitong Liu,
Huajun Cao
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1939/1/012073
Subject(s) - extreme learning machine , kriging , outlier , computer science , ground penetrating radar , energy consumption , extrusion , gaussian process , process (computing) , artificial intelligence , algorithm , entropy (arrow of time) , data mining , gaussian , machine learning , engineering , radar , artificial neural network , materials science , telecommunications , physics , quantum mechanics , electrical engineering , metallurgy , operating system
Hot extrusion forming is widely used to process lightweight parts in automobile, aircraft manufacturing, and subway industry. However, hot extrusion forming is an energy-intensive process with high environmental emissions. Because hot extrusion forming is a complex thermal-mechanical coupling process, it is difficult to develop its energy consumption prediction model based on the process mechanism. With the increasing application of data intelligent acquisition, this paper proposed a fusion data model integrating bagging enhanced ELM (Extreme Learning Machine) and GPR (Gaussian Process Regression) using their entropy weight. The energy-related data collected by energy management system are purified using the local outlier factor (LOF) algorithm and Rrelief F method. The bootstrap sampling method is used to obtain the training set and test set from the processed data. Then, the ELM and GPR are improved by the bagging algorithm for constructing the B-ELM learning model and the B-GPR learning model. Furthermore, the entropy weight method is used to further integrate the B-ELM and the B-GPR. Finally, an experiment validates the accuracy and reliability of the proposed model.