Predicting longitudinal crack of slab based on dynamic time warping and k-Nearest Neighbor
DUAN Hai-yang1,2, WANG Xu-dong1,2, YAO Man1,2
1. School of Materials Science and Engineering, Dalian University of Technology, Dalian 116024, Liaoning, China; 2. Key Laboratory of Solidification Control and Digital Preparation Technology (Liaoning Province), Dalian 116024, Liaoning, China
摘要 纵裂纹是典型的铸坯表面缺陷,严重影响过程顺行和连铸坯质量。铸坯表面纵裂纹的识别和预测对于铸坯质量的提升具有重要意义。针对纵裂纹形成与扩展过程中,结晶器热电偶温度在时间序列上的动态演化和一维传播特征,捕获和提取了纵裂纹和正常工况下热电偶时序温度的典型变化趋势;运用动态时间弯曲(Dynamic Time Warping,DTW)方法度量不同工况下时序温度的相似性和差异性,结合k近邻(k-Nearest Neighbor,kNN)分类算法,建立了针对连铸坯纵裂的在线识别和分类模型。结果表明,建立的模型对纵裂纹和正常工况样本具有较高的识别准确率。提出的基于温度时序特征动态时间弯曲及kNN方法的纵裂预测模型,为铸坯纵裂的实时检测和准确预报提供了可靠途径。
Abstract:Longitudinal cracks are typical surface defects of slabs, which seriously affect the quality of casting slabs and process progression. The identification and prediction of longitudinal cracks is of great significance to improve the quality of the cast slabs. According to the dynamic evolution and one-dimensional propagation characteristics of the mould thermocouple temperature in time series during the formation and propagation of longitudinal cracks, the typical variation trends of the thermocouple temperature in time series under longitudinal cracks and normal conditions were captured and extracted. Dynamic Time Warping (DTW) method is used to measure the similarity and difference of time series temperature under different working conditions, and k-Nearest Neighbor (kNN) classification algorithm is used to establish online identification and classification model for longitudinal crack of continuous casting billet.The results show that the model has a high accuracy rate for the identification of longitudinal crack and normal condition samples, which provides a reliable way for real-time detection and accurate prediction of slab longitudinal cracks.
段海洋, 王旭东, 姚曼. 基于动态时间弯曲和k近邻分类预测铸坯纵裂纹[J]. 连铸, 2021, 40(4): 66-71.
DUAN Hai-yang, WANG Xu-dong, YAO Man. Predicting longitudinal crack of slab based on dynamic time warping and k-Nearest Neighbor. CONTINUOUS CASTING, 2021, 40(4): 66-71.
Schwerdtfeger K, Spitzer K H. Application of reduction of rrea-temperature diagrams to the prediction of surface crack formation in continuous casting of steel[J].ISIJ International, 2009,49(4):512.
Nakato H, Ozawa M, Kinoshita K, et al. Factors affecting the formation of shell and longitudinal cracks in mold during high-speed continuous-casting of slabs[J].Transactions of the Iron and Steel Institute of Japan, 1984,24(11):957.
[9]
YU H L, LIU X H. Longitudinal crack on slab surface at straightening stage during continuous casting using finite element method[J].Journal of Central South University of Technology, 2010,17(2):235.
[10]
DU F M, WANG X D, LIU Y, et al. Prediction of longitudinal cracks based on a full-scale finite-element model coupled inverse algorithm for a continuously cast slab[J].Steel Research International, 2017,88(10):1700013.
[11]
LIU Y, WANG X D, SUN Y, et al. Research on a new detection method of slab surface crack in mould during continuous casting[J].Metallurgical Research and Technology, 2018,115(1):108.
[12]
LI N, KONG H P, MA Y F, et al. Human performance modeling for manufacturing based on an improved KNN algorithm[J].International Journal of Advanced Manufacturing Technology, 2016,84(1-4):473.
[13]
GOU J P, MA H X, OU W H, et al. A generalized mean distance-based k-nearest neighbor classifier[J].Expert Systems with Applications, 2019,115:356.
[14]
Abu A, Haneen A, Hassanat Ahmad B A, et al. Effects of distance measure choice on k-nearest neighbor classifier performance: a review[J].Big Data, 2019,7(4):221.
[15]
Tuan T M, Le X M T, Nguyen H T, et al. A novel non-parametric method for time series classification based on k-nearest neighbors and dynamic time warping barycenter averaging[J].Engineering Applications of Artificial Intelligence, 2019,78:173.
[16]
Kate R J. Using dynamic time warping distances as features for improved time series classification[J].Data Mining and Knowledge Discovery, 2016,30(2):283.