Identification of Industrial Furnace Temperature for Sintering Process in Nuclear Fuel Fabrication Using NARX Neural Networks

Table 2

Loss function (LF) and AIC, BIC, and MDL criterion obtained by trial and error with difference NARX neural network structure and number of hidden layers for soaking temperature step.

Hidden node number	Model order 𝑛𝑎𝑛𝑏𝑛𝑘	LF	Criterion
			AIC	BIC	MDL
5	219	1.5𝑒−05	−5.7613	−5.7314	−5.6865
	226	7.2𝑒−04	−1.9165	−1.8766	−1.8167
	238	4.4𝑒−07	−9.3200	−9.2701	−9.1952
	318	6.7𝑒−05	−4.2955	−4.2556	−4.1957
	328	5.0𝑒−07	−9.1794	−9.1295	−9.0546
10	216	4.2𝑒−05	−4.7765	−4.7466	−4.7016
	229	6.9𝑒−07	−8.8683	−8.8283	−8.7684
	238	1.8𝑒−05	−5.6130	−5.5631	−5.4882
	317	8.6𝑒−02	2.8603	2.9002	2.9601
	323	2.1𝑒−06	−7.7598	−7.7099	−7.6350
15	219	6.7𝑒−06	−6.5927	−6.5627	−6.5178
	227	3.5𝑒−04	−2.6251	−2.5852	−2.5252
	238	2.3𝑒−04	−3.0369	−2.9870	−2.9121
	319	6.1𝑒−06	−6.6938	−6.6539	−6.5940
	333	2.5𝑒−06	−7.5759	−7.5260	−7.4511