IPFOA-MKSVM and BA-MLP models for predicting closed busbar temperatures in high voltage nuclear power plants in different vacuum environments
Year: 2024
Type: article
Abstract: In high-voltage nuclear power systems, the nuclear closed busbar plays a crucial role in power transmission and distribution. Excessive temperature can cause equipment failures like overloads and short circuits. This paper introduces two predictive models for busbar temperature: a multicore support vector machine (MKSVM) optimized by an improved falcon predation algorithm (IPFOA), and a multilayer perceptron (MLP) improved by a back-propagation algorithm (BA). To create a vacuum environment, a rotary vane mechanical pump and a molecular pump are used. The data, which include temperature and wind speed, are preprocessed and split into training and test sets. For comparison, the study also uses a Particle Swarm Optimization (PSO) algorithm with an SVM model. The results show that the IPFOA-MKSVM model achieves higher accuracy and stability under varying vacuum conditions compared to other models. The models were validated using wind speed prediction, and comparisons between BA-MLP, IPFOA-MKSVM, SVM, BA, MLP, PSO, and PSO-SVM demonstrated the superiority of IPFOA-MKSVM and BA-MLP in temperature prediction.
Source: Vacuum
Authors: Zuoxun Wang, Guojian Zhao, Jinxue Sui, Wangyao Wu, Chuanzhe Pang, Liteng Xu
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,990
OpenAlex ID: W4404355793