Abstract: Based on the summaries and analyses of various types of strand defects and their causes, the halfway crack, central crack and central segregation of special steel bloom were adopted as the samples for the establishment of a quality prediction model using BP neural network. Based on the metallurgical theory related to the defects introduction and the analyses to a great deal of statistical quality data of the bloom castings, 20 defectcausing parameters with the continuous casting process were selected as the input of the model, a 20153 network topology structure for the artificial neural network prediction model was developed with back propagation algorithm. The training sample sets and test sample sets for the ANN model were prepared from the real production of the continuous casting production. The BP network was trained by training sample sets and tested by test sample sets in sequence. Based on the well trained BP network, the online quality prediction system was developed, and the realtime quality of bloom was predicted on line.
,曾智,张家泉,何庆文,,申景霞,,张利平. 基于BP神经网络的大方坯质量在线预报模型常运合[J]. 钢铁, 2011, 46(5): 33-37.
CHANG Yunhe1,ZENG Zhi1,ZHANG Jiaquan1,HE Qingwen1,2,SHEN Jingxia1,2,ZHANG Liping2. Online Quality Prediction System of Bloom Castings Based on BP Neural Network. Iron and Steel, 2011, 46(5): 33-37.