AI-based Service Lead Generator for Hoisting Wire-rope
Abass, Adeyinka (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202505089920
https://urn.fi/URN:NBN:fi:amk-202505089920
Tiivistelmä
Hoisting wire ropes are critical components in crane systems, essential for ensuring reliability and safety during operations. Periodic inspections are mandated by international standards to prevent wire rope failures, which can lead to severe accidents and productivity losses. The commissioning company, a global leader in crane manufacturing, aims to enhance its predictive maintenance capabilities by integrating machine learning models to predict wire rope conditions based on actual usage data. This thesis aimed to develop machine learning models capable of predicting the condition of steel wire ropes using data collected from crane systems. The goal was to move away from fixed periodic inspections towards a more dynamic, data-driven approach that optimizes maintenance schedules based on real-time usage data.
Data was collected from various sources, including wire rope service history, crane usage, and crane configuration. The data was cleaned, merged, and subjected to rigorous Exploratory Data Analysis (EDA) to identify significant features and potential feature thresholds for wire rope failure. Four machine learning models—Logistic Regression, Support Vector Machine (SVM), Decision Tree (Random Forest), and Deep Neural Network—were developed and fine-tuned. The models were trained using the cleaned dataset and evaluated based on their ability to minimize false positives while maintaining high recall rates. All machine learning models outperformed the baseline rule-based approach derived from EDA.
The commissioning company found the rule-based model most promising for initial pilot implementation due to its simplicity and ease of internal adoption. The developed machine learning models offer a promising solution to enhance predictive maintenance capabilities for wire ropes, potentially reducing downtime and improving safety. Future research should explore additional features and advanced machine learning techniques to further improve predictive power. The methodologies can be adapted for predictive maintenance in other industries, providing broader insights into model generalizability.
Data was collected from various sources, including wire rope service history, crane usage, and crane configuration. The data was cleaned, merged, and subjected to rigorous Exploratory Data Analysis (EDA) to identify significant features and potential feature thresholds for wire rope failure. Four machine learning models—Logistic Regression, Support Vector Machine (SVM), Decision Tree (Random Forest), and Deep Neural Network—were developed and fine-tuned. The models were trained using the cleaned dataset and evaluated based on their ability to minimize false positives while maintaining high recall rates. All machine learning models outperformed the baseline rule-based approach derived from EDA.
The commissioning company found the rule-based model most promising for initial pilot implementation due to its simplicity and ease of internal adoption. The developed machine learning models offer a promising solution to enhance predictive maintenance capabilities for wire ropes, potentially reducing downtime and improving safety. Future research should explore additional features and advanced machine learning techniques to further improve predictive power. The methodologies can be adapted for predictive maintenance in other industries, providing broader insights into model generalizability.