Prediction of Automatic Scram during Abnormal Conditions of Nuclear Power Plants Based on Long Short-Term Memory (LSTM) and Dropout

Table 2

Training and testing RMSE of LSTM prediction model trained by 100,000 epochs with different dropout rates.

Dropout rate	Training RMSE	Testing RMSE
		No noise	Noise ±0.01	Noise ±0.03	Noise ±0.05	Noise ±0.07
0	0.1258	1.8098	2.0543	4.1357	6.4225	8.7408
0.05	0.7406	1.9228	2.2484	3.5289	5.2413	7.2842
0.10	0.8566	1.9896	2.2555	3.5443	5.1793	6.5903
0.20	1.3063	2.1803	2.3142	3.2418	4.5737	5.9841
0.30	1.7377	2.6401	2.8563	3.6682	4.9724	6.3992
0.40	2.7155	3.4187	3.6943	3.9189	5.1976	6.7855
0.50	3.6668	4.2122	4.4845	4.9057	5.7148	7.0513