Prediction of the Loss of Feed Water Fault Signatures Using Machine Learning Techniques

Table 2

Performance indices of the ANFIS, LSTM, and RBFN models.

Scheme	Performance indicator	Core inlet mass flow	SG U-tube temperature	Severity estimation mod.
ANFIS—single output	MAPE (%)	0.4526	0.1351	79.7684
	RMSE	20.8193	1.4038	31.7503
	R^2	0.9869	0.9843	0.1667
LSTM—single output	MAPE (%)	0.9645	0.5806	30.3695
	RMSE	57.1781	5.2899	18.3586
	R^2	0.8104	0.8162	0.8412
RBFN—single output	MAPE (%)	0.3238	0.1610	101.0111
	RMSE	20.1910	1.5662	35.4367
	R^2	0.9800	0.9834	0.1904
LSTM—multiple outputs	MAPE (%)	2.0249	1.0244	33.2772
	RMSE	122.3393	23.3238	20.0842
	R^2	0.3645	0.1993	0.7833
RBFN—multiple outputs	MAPE (%)	0.3238	0.1610	101.0111
	RMSE	20.1910	1.5662	35.4367
	R^2	0.9800	0.9834	0.1904