基于K-means算法的连铸漏钢预报方法研究
|
摘要
漏钢现象是连铸过程主要操作故障之一,影响连铸生产效率及其设备寿命,漏钢预报一直是连铸研究领域的热门研究课题。本文针对当前连铸漏钢预报系统误报率高和预报准确率低的问题,提出了一种基于K-means算法连铸漏钢预报方法。分析了漏钢预报系统的架构及其连接方式;采用K-means算法对热电偶采取的结晶器坯壳温度数据进行降噪和聚类处理;对某钢厂板坯220 mm×1600 mm连铸机浇铸过程采集到的数据进行测试,测试结果表明,利用K-means算法对正常浇铸数据、温度上升数据和漏钢时刻数据可以正确区分,从而证明了提出连铸漏钢预报方法的有效性。
Breakout is one of the main operational failures in the continuous casting process,it reduces the production efficiency and equipment life. Breakout prediction has always been a hot research issue in the field of continuous casting. This paper proposed a breakout prediction method based on the K-means algorithm,it could solve the problem on low prediction accuracy of current breakout prediction system. Firstly,the structure and the connection method of the breakout prediction system are analyzed. Secondly,making noise reduction and clustering on the temperature data of continuous casting mold using the K-means algorithm. Finally,the proposed method was tested using the data of a steel plant,the results show that the K-means algorithm can correctly distinguish the normal casting data,temperature data and breakout data,which proves the effectiveness of the proposed breakout prediction method.
Breakout is one of the main operational failures in the continuous casting process,it reduces the production efficiency and equipment life. Breakout prediction has always been a hot research issue in the field of continuous casting. This paper proposed a breakout prediction method based on the K-means algorithm,it could solve the problem on low prediction accuracy of current breakout prediction system. Firstly,the structure and the connection method of the breakout prediction system are analyzed. Secondly,making noise reduction and clustering on the temperature data of continuous casting mold using the K-means algorithm. Finally,the proposed method was tested using the data of a steel plant,the results show that the K-means algorithm can correctly distinguish the normal casting data,temperature data and breakout data,which proves the effectiveness of the proposed breakout prediction method.
引文
[1] Mills K C,Billany TJH. Causes of Sticker Breakout during Continuous Casting[J]. Iron Making and Steelmaking,1991,18(04):253-265.
[2]秦旭,朱超甫,高光河等.板坯漏钢机理与基于热电偶检测的漏钢预报技术研究[J].钢铁研究学报,2011,23(02):7-11.
[3]厉英,王正,敖志广等. BP神经网络漏钢预报系统优化[J].控制与决策,2010,25(03):453-456.
[4]刘永贞,王旭东,贾启忠等.结晶器内粘结漏钢机器传播行为的研究[J].炼钢,2009,25(03):45
[5]张本国,李强,王葛等.基于改进BP神经网络的漏钢预报[J].中国机械工程,2012,23(02):204-207.
[6]王葛,尹少锋,李强,等.基于支持向量机算法的连铸:204-207漏钢预报模型研究[J].燕山大学学报,2015,39(01):22-29.
[7] Peter Harrington. Machine learning in action[M].Manning Publications,2012.
[8]张学梅,汪洪峰.连铸结晶器专家系统的研发与生产实践[J].连铸2018,01(4):1-7.
[9]张瑞忠,刘耀辉,刘建树.连铸漏钢预报数据分析系统的开发与应用[J].连铸,2017,(3):10-15.
[10]何冰,米进周,王旭英等.连铸生产线智能化方向的初步探究[J].重型机械2017:1-5.
[11]翟莹莹.连铸漏钢预报系统中的热成像分析[J].控制工程,2015,22(6):1251-1254.
[12]张勇,张惠等.板坯漏钢预报系统的研究及应用[J].钢铁研究,2013,41(5):23-27.
[13]史辰阳,孟阳.提高结晶器漏钢预报警准确率的措施[J].天津冶金,2018,S1:99-101.
[14]宋志斌.基于RBF算法的漏钢预报系统开发[J],世界有色金属,2018,5:287-289.
[15]方一鸣,郑贺军等.基于模拟退火-粒子群算法优化支出向量机参数的连铸漏钢预报[J],中国机械工程2017,28(12):1462-1467.
[16]郑键.连铸漏钢预报系统应用及改进[J].现代冶金,2016,44(4):54-56.
[17]张本国,张新疆,夏建生等.基于PSO-BP神经网络的连铸漏钢预报[J].铸造技术,2016,37(07):1481-1484.
[18]张本国,展邦华,刘军等.薄板连铸GA-LMBP漏钢预报模型研究[J].连铸技术,2017,38(08):1936-1939.
[19]方一鸣,胡春洋,刘乐等.基于主动学习GASVM分类器的连铸漏钢预报[J].中国机械工程,2016,27(12):1609-1614.
[20]董长征,朱志强等.薄板坯连铸漏钢预报系统的优化与应用[J].武钢技术,2016,54(6):34-30.
[21] Peter Harrington. Machine learning in action[M].Manning Publications,2012.
[2]秦旭,朱超甫,高光河等.板坯漏钢机理与基于热电偶检测的漏钢预报技术研究[J].钢铁研究学报,2011,23(02):7-11.
[3]厉英,王正,敖志广等. BP神经网络漏钢预报系统优化[J].控制与决策,2010,25(03):453-456.
[4]刘永贞,王旭东,贾启忠等.结晶器内粘结漏钢机器传播行为的研究[J].炼钢,2009,25(03):45
[5]张本国,李强,王葛等.基于改进BP神经网络的漏钢预报[J].中国机械工程,2012,23(02):204-207.
[6]王葛,尹少锋,李强,等.基于支持向量机算法的连铸:204-207漏钢预报模型研究[J].燕山大学学报,2015,39(01):22-29.
[7] Peter Harrington. Machine learning in action[M].Manning Publications,2012.
[8]张学梅,汪洪峰.连铸结晶器专家系统的研发与生产实践[J].连铸2018,01(4):1-7.
[9]张瑞忠,刘耀辉,刘建树.连铸漏钢预报数据分析系统的开发与应用[J].连铸,2017,(3):10-15.
[10]何冰,米进周,王旭英等.连铸生产线智能化方向的初步探究[J].重型机械2017:1-5.
[11]翟莹莹.连铸漏钢预报系统中的热成像分析[J].控制工程,2015,22(6):1251-1254.
[12]张勇,张惠等.板坯漏钢预报系统的研究及应用[J].钢铁研究,2013,41(5):23-27.
[13]史辰阳,孟阳.提高结晶器漏钢预报警准确率的措施[J].天津冶金,2018,S1:99-101.
[14]宋志斌.基于RBF算法的漏钢预报系统开发[J],世界有色金属,2018,5:287-289.
[15]方一鸣,郑贺军等.基于模拟退火-粒子群算法优化支出向量机参数的连铸漏钢预报[J],中国机械工程2017,28(12):1462-1467.
[16]郑键.连铸漏钢预报系统应用及改进[J].现代冶金,2016,44(4):54-56.
[17]张本国,张新疆,夏建生等.基于PSO-BP神经网络的连铸漏钢预报[J].铸造技术,2016,37(07):1481-1484.
[18]张本国,展邦华,刘军等.薄板连铸GA-LMBP漏钢预报模型研究[J].连铸技术,2017,38(08):1936-1939.
[19]方一鸣,胡春洋,刘乐等.基于主动学习GASVM分类器的连铸漏钢预报[J].中国机械工程,2016,27(12):1609-1614.
[20]董长征,朱志强等.薄板坯连铸漏钢预报系统的优化与应用[J].武钢技术,2016,54(6):34-30.
[21] Peter Harrington. Machine learning in action[M].Manning Publications,2012.