Developing an IoT and Machine Learning-Based Monitoring System for Discrete Production Processes - Research Paper

Aug 3, 2026

Upload by Hazel

Purpose: Develop a tool for monitoring discrete manufacturing processes by combining IoT sensor data with machine learning models. Design/Methodology/Approach: Use machine learning to prepare and analyze measurements collected along the production line at different locations, supporting continuous analysis and earlier detection of multiple critical situations. Findings: Present an analysis framework for continuously monitored technological process data and a method to classify components on the production line to improve decision-making under uncertainty. Practical Implications: The data preparation and analysis approach supports improved production management and quality observation. Originality/Value: The novelty lies in the data preparation and processing pipeline, neural network model preparation, and production-line element classification.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Prepare Measurement

Production line modeling in JaamSim

Production line modeling in SimPy

Implementation of the Solution - STREAMLIT

European Research Studies Journal Volume XXVII, Special Issue 2, 2024 pp. 38-48 Developing an IoT and Machine Learning-Based Monitoring System for Discrete Production Processes Submitted 18/02/24, 1st revision 16/03/24, 2nd revision 20/04/24, accepted 16/05/24 Krzysztof Król 1, Michał Oleszek2, Grzegorz Bartnik3, Olena Ivashko4, Marek Rutkowski5, Adam Hernas6 Abstract: Purpose: This paper aims to develop a tool to support discrete manufacturing process monitoring using IoT sensors and machine learning systems. Design/Methodology/Approach: Machine learning was used to prepare and analyze data from the production line. In discrete manufacturing, measurements from sensors throughout the line at various locations are read for objects moving on the line. The measurements and related research allow for ongoing data analysis and earlier reactions to multiple critical situations. Findings: The study's result was the measurement data analysis in a discrete manufacturing process. Data was obtained from continuous monitoring of technological processes. It also shows how to classify components on the production line, allowing for better decisionmaking under uncertainty. Practical Implications: The presented method of preparation and analysis of measurement data will allow for better production management and observation of the quality of this production. Originality/Value: A novelty is using an approach to data preparation and processing, neural network systems preparation, and element classification on the production line. Keywords: Neural network, positive predictive, negative predictive, RMSE. JEL codes: C45, C61, E20, L23, O14. Paper type: Research article. 1Corresponding Author: Netrix S.A./WSEI University , Lublin, Poland, e-mail: [krzysztof. ](krzysztof. krol@netrix.com.pl)[krol@netrix.com.pl](krzysztof. krol@netrix.com.pl); 2Netrix S.A./WSEI University, Lublin, Poland, e-mail: michał.[oleszek@wsei.lublin.pl](oleszek@wsei.lublin.pl); 3WSEI University, Lublin, Poland, [e-mail: ](e-mail: Grzegorz.Bartnik@wsei.lublin.pl)[Grzegorz.Bartnik@wsei.lublin.pl](e-mail: Grzegorz.Bartnik@wsei.lublin.pl); 4WSEI University, Lublin, Poland, [e-mail: ](e-mail: Olena.Ivashko@wsei.lublin.pl)[Olena.Ivashko@wsei.lublin.pl](e-mail: Olena.Ivashko@wsei.lublin.pl); 5WSEI University, Lublin, Poland, [e-mail: ](e-mail: Marek.Rutkowski@wsei.lublin.pl)[Marek.Rutkowski@wsei.lublin.pl](e-mail: Marek.Rutkowski@wsei.lublin.pl); 6Wyższa Szkoła Biznesu-National Louis University, e-mail: [ahernas@wsb-nlu.edu.pl](ahernas@wsb-nlu.edu.pl); 1. Introduction In today's industrial era, effective monitoring of production processes is critical to ensuring high product quality, optimizing productivity, and minimizing production costs. Process monitoring plays a vital role in discrete manufacturing, where products are manufactured as separate units due to the complexity and dynamics of operations (Król, 2021) . One of the commonly used production systems is production based on production and assembly lines, which allows products to move continuously through individual stages of production (Król, 2023) . This article will analyze and develop a strategy for monitoring the discrete manufacturing process, particularly on belt lines. Our goal is to provide a comprehensive methodology that allows you to effectively track various aspects of the production process, identify potential problems, and ensure a quick response during irregularities. 2. Prepare Measurement The first model is the production line model created in the JaamSim environment. JaamSim (JaamSim, 2016) is a free discrete event simulation software that includes, but is not limited to, a drag-and-drop user interface, interactive 3D graphics, input and output processing, and modeling tools and editors. The designed model reads data from a file and a number and then squares it. The result and other important information, such as the time the line has been running, are stored

FAQs

What is the purpose of the proposed monitoring system?

The paper aims to support discrete manufacturing process monitoring by using IoT sensors together with machine learning systems to analyze production-line data and improve responsiveness to critical situations.

How is machine learning used in the workflow?

Machine learning is used to prepare and analyze data collected from sensor measurements across different locations on the production line, enabling continuous analysis and earlier reactions to irregular or critical events.

What practical outcomes are expected from the method?

The approach is intended to improve production management and enable better observation of production quality by classifying components and supporting decisions under uncertainty.

Aug 3, 2026

Upload by Hazel