Leveraging Real-Time Data Analysis and Multiple Kernel Learning for Manufacturing of Innovative Steels

TL;DR

The paper tackles quality variability in thermally sprayed coatings for innovative steels by introducing a real-time analytics framework that couples a data aggregator with a Simple and Efficient Multiple Kernel Learning (SEMKL) based predictor. The data aggregator gathers and preprocesses heterogeneous sensor data from the coating process, producing features for predictive modeling, while the SEMKL predictor delivers real-time quality estimates for multiple coating properties using a multi-kernel representation with nonnegative weights under an $L_p$-norm constraint. In a case study at voestalpine, the approach trained on $N=49$ samples and tested on $N_{test}=10$, achieving substantial improvements over a linear baseline across eight properties, and enabling proactive operator alerts. The study demonstrates practical deployment viability, including rapid online latency (on the order of a few milliseconds for data processing and a few seconds for live prediction display), and discusses scalability considerations for broader industrial adoption and future extensions to transfer and continual learning. Overall, the work offers a concrete pathway to enhance coating reliability and reduce downtime in steel manufacturing through tight integration of IIoT-based data collection and kernel-based predictive quality management.

Abstract

The implementation of thermally sprayed components in steel manufacturing presents challenges for production and plant maintenance. While enhancing performance through specialized surface properties, these components may encounter difficulties in meeting modified requirements due to standardization in the refurbishment process. This article proposes updating the established coating process for thermally spray coated components for steel manufacturing (TCCSM) by integrating real-time data analytics and predictive quality management. Two essential components--the data aggregator and the quality predictor--are designed through continuous process monitoring and the application of data-driven methodologies to meet the dynamic demands of the evolving steel landscape. The quality predictor is powered by the simple and effective multiple kernel learning strategy with the goal of realizing predictive quality. The data aggregator, designed with sensors, flow meters, and intelligent data processing for the thermal spray coating process, is proposed to facilitate real-time analytics. The performance of this combination was verified using small-scale tests that enabled not only the accurate prediction of coating quality based on the collected data but also proactive notification to the operator as soon as significant deviations are identified.

Figures (6)

• Figure 3.1: Framework of real-time analytics and quality prediction for thermally spray coated components in steel manufacturing (TCCSM).
• Figure 4.1: Evolution of the kernel weight values in SEMKL model for predicting particle in-flight velocity.
• Figure 4.2: Prediction performance comparison on test data. The bar charts depict the predicted (green) and measured (orange) values of TCCSM properties across ten test samples.
• Figure 4.3: Visual representation of the thermal spray process for TCCSM coating. The left panel shows the application of a thermal spray coating on a large-scale plate by a robotic system within the spray booth at voestalpine Stahl GmbH TSM2. The right panel depicts operator interaction with the system controller next to the spray booth, where the validated framework is operational.
• Figure 4.4: Interface of the web application within the validated framework for real-time monitoring of the thermal spray process, displaying live data on critical process variables, including propane, compressed air, and oxygen flow rates, pyrometer temperature readings, and lathe rotational speed, with an average display latency of 4.92 ms.

Theorems & Definitions (3)

• Remark 3.1
• Remark 3.2
• Remark 4.1