Non-intrusive load identification method based on XGBoost and GRU by stacking ensemble learning
by Lingzhi Yi; Ning Liu; Jiangyong Liu; Lv Fan; Jiankang Liu; Huina Song
International Journal of Advanced Mechatronic Systems (IJAMECHS), Vol. 9, No. 2, 2021

Abstract: Load identification method is one of the major technical difficulties of non-intrusive composite monitoring. In this paper, aiming at the low accuracy of single model load identification, a non-invasive load identification method based on the combination of XGBoost and GRU model is proposed. Considering the difference of data observation and training principles, the stacking based load identification model embedded various machine learning algorithms was proposed to utilise their diversified strength. Firstly, to reduce the impact of unbalanced samples on load identification, SMOTE algorithm is used to balance the samples; secondly, a stacking integrated model is constructed based on the idea of ensemble learning and model combination, which is established to improve the poor performance of load identification for single model. Finally, the results indicate the proposed stacking ensemble learning model has better identification performance compared with the traditional single models on the public data set PLAID.

Online publication date: Mon, 26-Jul-2021

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Advanced Mechatronic Systems (IJAMECHS):
Login with your Inderscience username and password:

    Username:        Password:         

Forgotten your password?


Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com