我国不同地区钢材大气腐蚀预测算法评估与筛选

沈坚, 吴柯娴, 何晓宇, 方兴龙

Evaluation and Screening of Atmospheric Corrosion Prediction Algorithms of Steels in Different Regions of China

SHEN Jian, WU Kexian, HE Xiaoyu, FANG Xinglong

表2 不同机器学习模型预测结果的误差指标

Table 2 Error metrics of prediction outcomes for different machine learning models

Atmospheric station	Error indicator	Sample set	SVM	RF	RBFNN	BPNN	LSTM	CNN
WN	R²	Training set	0.81	0.91	0.70	0.95	0.91	0.33
	R²	Test set	0.74	0.65	0.48	0.85	0.87	0.38
	RMSE	Training set	152	105	178	68	100	287
	RMSE	Test set	100	109	206	137	92	163
	MAE	Training set	45	52	103	43	61	131
	MAE	Test set	59	69	140	75	58	94
	MAPE	Training set	265%	252%	521%	264%	254%	1511%
	MAPE	Test set	263%	272%	2220%	127%	1454%	304%
Others	R²	Training set	0.90	0.94	0.80	0.85	0.85	0.49
	R²	Test set	0.86	0.84	0.76	0.77	0.82	0.65
	RMSE	Training set	27	21	36	33	31	59
	RMSE	Test set	21	23	33	29	31	38
	MAE	Training set	10	7	20	19	15	28
	MAE	Test set	15	13	24	20	18	23
	MAPE	Training set	130%	29%	202%	215%	60%	569%
	MAPE	Test set	141%	57%	267%	156%	133%	237%