IPFOA-MKSVM and BA-MLP models for predicting closed busbar temperatures in high voltage nuclear power plants in different vacuum environments

The nuclear power closed busbar is a key power transmission component in the power system, and its high temperature may cause equipment failure. In this paper, for the temperature prediction of nuclear power closed busbar under vacuum environment, a multi-core support vector machine model optimized by the improved falcon predation algorithm and a multi-layer perceptron model enhanced by the back propagation algorithm are proposed. The vacuum pump is used to reduce the air pressure in the closed space to achieve a vacuum state. The collected data are preprocessed to improve the accuracy and stability of the model. In addition, the PSO and SVM models are used to compare and verify the superiority of the proposed model. The data set is divided into a training set and a test set. The results show that under different vacuum degrees, the prediction accuracy and stability of the IPFOA-MKSVM model are better than those of other models, but its error is slightly higher than the physical calculation result. Finally, the performance of the model in wind speed prediction is verified, and compared with several models to verify the accuracy of IPFOA-MKSVM under different vacuum and wind speed conditions. • Select a suitable vacuum pump, adjust its speed and power, and create vacuum environments with different concentrations in the nuclear power closed busbar system. Collect historical temperature data under medium and rough vacuum conditions. • Preprocess the collected data and propose IPFOA-MKSVM and BA-MLP models. Train and debug the models using processed data to enhance prediction accuracy and stability. Compare these models with others, using working state data from a regional power plant. Divide the dataset into training and test sets. • Establish a physical model to analyze heat transfer mechanisms: thermal convection dominates in rough vacuum conditions, while thermal radiation is primary in medium vacuum. • Experimental results demonstrate that IPFOA-MKSVM and BA-MLP achieve superior prediction accuracy and stability in both vacuum conditions, outperforming other models.