Exploring Machine Learning Frameworks to Formulate Thermal Profile for Reflow Oven Through Multizone Temperature Prediction on Production Data

Cheh Phey Tan and Jasy Suet Yan Liew ()
Additional contact information
Cheh Phey Tan: Universiti Sains Malaysia
Jasy Suet Yan Liew: Universiti Sains Malaysia

SN Operations Research Forum, 2025, vol. 6, issue 2, 1-23

Abstract: Abstract In reflow soldering, the precise control of oven zone temperatures is crucial for ensuring product quality and production efficiency. The current practice of implementing the manual trial-and-error method in oven temperature recipe (reflow thermal recipe) development raises the concern of product and process limitations, thus garnering significant interest in the development of prediction models for oven zone temperatures. Additionally, predicting conveyor belt speed, which plays a vital role in controlling heat transfer within the oven, is equally important to acquire a key feature for zone temperature prediction. Using actual production data, we first train a conveyor belt speed prediction model using random forest regressor. Subsequently, the predicted conveyor speed is utilized as input for predicting zone temperatures in the oven. Our primary goal is to examine the performance of using two different learning frameworks for predicting zone temperatures: (1) single task learning using random forest regressor (STL) that incorporates separate individual models to predict temperature for each zone independently and (2) multi-output learning (MOL) using multi-layer perceptron (MLP) that employs a unified MLP model capable of predicting multiple zone temperatures simultaneously. Based on model performance evaluation across all 17 zone temperatures using root mean squared error (RMSE), our results indicate that STL (mean RMSE = 0.4329) outperforms MOL (mean RMSE = 2.9170), suggesting that independent zone-specific models are more effective for the multizone temperature prediction task. Our novel findings on comparing STL versus MOL using a data-driven approach on real-world production data contribute to a greater understanding of the efficacy of different modeling approaches in predicting oven zone temperatures, thereby facilitating informed decision-making in industrial processes and automation.

Keywords: Reflow soldering; Thermal profile; Multizone temperature prediction; Machine learning; Artificial neural network; Regression (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s43069-025-00477-2 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:snopef:v:6:y:2025:i:2:d:10.1007_s43069-025-00477-2

Ordering information: This journal article can be ordered from
https://www.springer.com/journal/43069

DOI: 10.1007/s43069-025-00477-2

Access Statistics for this article

SN Operations Research Forum is currently edited by Marco Lübbecke

More articles in SN Operations Research Forum from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().