A Data-Driven Fault Prediction Method for Nuclear Power Systems Based on End-to-End Deep Learning Framework

Table 4

Accuracy comparison of fault prediction and diagnosis methods.

Class ID	EDN-NPSP (%)	PCA + KNN (%)	PCA + SVM (%)	CNN10 (%)	Wavelet CNN (%)	Encoder-decoder (%)
1	100.00	97.00	100.00	100.00	100.00	100.00
2	100.00	92.00	96.33	100.00	100.00	100.00
3	98.33	93.67	97.33	99.33	98.67	99.00
4	97.67	91.33	93.33	97.33	97.33	96.00
5	95.67	89.67	90.33	95.33	94.67	95.67
6	94.67	88.33	89.00	96.00	93.33	93.67
7	95.00	91.33	90.33	93.67	95.33	94.67
8	96.33	90.66	88.66	99.00	93.67	93.67
Total	96.82	91.27	92.55	96.59	96.09	96.27