Fault Diagnosis Method of Polymerization Kettle Equipment Based on Rough Sets and BP Neural Network

Table 11

Performance contrast of fault diagnosis results under 10-fold cross validation.

Attribute set	Average training error	Average training step	Average accuracy (%)
{𝑆1, 𝑆4, 𝑆5,𝑆6, 𝑆7}	7.12×10^−5	21	98.8
{𝑆2, 𝑆4, 𝑆5, 𝑆6, 𝑆7}	2.37×10^−4	37	97.2
{𝑆1 ~𝑆8}	5.15×10^−5	134	99.1