Application of Time Series Data Anomaly Detection Based on Deep Learning in Continuous Casting Process

The inclusion is a crucial factor affecting the quality of cord steel. formation inclusions closely related to abnormal production process in continuous casting process. Automatic anomaly detection algorithms are proposed replace manual visual screening according smart manufacturing paradigm, and then relationship between product mined through data-driven methods this paper. Convolutional neural networks autoencoder models employed detect various types anomalies time-dependent parameters. A new idea detecting intervals from time series implemented instead conventional monitoring based on univariate control limit specifications. including starting time, duration type detected. Furthermore, scheme progresses multi-variable monitoring, which considers nonlinear coupling Finally, results fused analyze whether exist cast slab. applied automatic verified be effective via plenty actual data, with recall rate 93.06%. It prominent significance for improvement