A Multi-Step Anomaly Detection Strategy Based on Robust Distances for the Steel Industry

Kisan Sarda,Leonardo Manfredi,Vittorio Nole,Antonio Acernese,Luigi Glielmo,Carmen Del Vecchio,Luca Greco

doi:10.1109/access.2021.3070659

Kisan Sarda, Leonardo Manfredi + Show 5 more

Open Access

https://doi.org/10.1109/access.2021.3070659

Copy DOI

Abstract

Steel making industries exhibit extreme working conditions characterized by high temperature, pressure, and production speed as well as intense throughput. Due to high economic and energy investments of the overall production process, an intense and expensive preventive maintenance program is adopted to avoid breakdowns. Steel making process would greatly benefit from a predictive maintenance module able to detect incoming faults from data process analysis. However, due to intense preventive maintenance, available data recording process operations enclose only a few samples of fault events, avoiding the efficient application of classical data driven anomaly detection models. In an attempt to overcome the above mentioned limits, we report the outcome of an industrial research project on data-driven anomaly detection in a steel making production process. The study assesses a fault detection strategy for rotating machines in the hot rolling mill line: we developed an automatic two-step strategy, which combines two statistical methods over the available data set: more precisely, the combination of Re-weighted Minimum Covariance Determinant estimator and Hidden Markov Models helped identify working conditions in a drive reducer of a hot steel rolling mill line and automatically isolate signs of decreasing performance or upcoming failures. The proposed strategy has been validated on real data collected in a steel making plant placed in the South of Italy.

Highlights

The steel making process forges a collection of raw materials, including steel scrap, carbon, and limestone, into steel bars with different diameter sizes
The two-stage scheme we developed relies on robust measures; the Hidden Markov Model (HMM) is identified over the distances of the data points from the robust fit stemming obtained through reweighted minimum covariance determinant (RMCD) estimation, allowing a dynamic strategy that takes into account the time-varying nature of the data and infers several latent states in the production process
During the observation period lasting three months, we observed only one significant downtime caused by a fault on the 18th cage; the overall number of billets produced during the observation time is equal to 5634

Summary

Introduction

The steel making process forges a collection of raw materials, including steel scrap, carbon, and limestone, into steel bars with different diameter sizes. In order to obtain the highest production rates, today’s steel industries handle heavier loads and faster velocity than ever before. To reduce breakdown impacts on production, steel making companies adopt preventive maintenance strategies substituting equipment long before the end of their useful life [2], [3]. Apart from scheduled maintenance, production line reliability may be increased adopting fully automatic monitoring and fault diagnosis systems. A major task in steel making industries is the detection of faults, whereas neither an analytical description of faults and process models nor the collection of typical breakdown patterns exists [4]

Objectives

Results

Conclusion