The Prediction of Steel Bar Corrosion Based on BP Neural Networks or Multivariable Gray Models

Table 4

Comparison of prediction values and errors between models.

	Actual values	BPNN	GM (1, N)	OGM (1, N)
Accelerated corrosion test data	0.148	0.151	−1.59	0.143
	0.163	0.147	0.386	0.162
	0.173	0.1556	0.157	0.17
The average prediction error		7.29%	>100%	1.9%
Project data	7.68	8.381	−32.395	5.998
	8.32	7.943	103.206	2.314
	5.85	8.78	0.996	5.21
The average prediction error		8.2%	>100%	48%