[Paper Review] Topography scanning as a part of process monitoring in power cable insulation process

Janne Harjuhahto, Jaakko Harjuhahto+3 more|arXiv (Cornell University)|Feb 6, 2026

High voltage insulation and dielectric phenomena0 citations

TL;DR

The paper presents a novel topography scanning system for XLPE cable core monitoring, building a 3D surface map and using deep learning for real-time surface defect detection.

ABSTRACT

We present a novel topography scanning system developed to XLPE cable core monitoring. Modern measurement technology is utilized together with embedded high-performance computing to build a complete and detailed 3D surface map of the insulated core. Cross sectional and lengthwise geometry errors are studied, and melt homogeneity is identified as one major factor for these errors. A surface defect detection system has been developed utilizing deep learning methods. Our results show that convolutional neural networks are well suited for real time analysis of surface measurement data enabling reliable detection of surface defects.

Motivation & Objective

• Motivate improved process monitoring in power cable insulation to detect geometry and melt-related defects.
• Develop a system that generates detailed 3D surface maps of insulated cable cores.
• Identify cross-sectional and lengthwise geometry errors and associate them with melt homogeneity.
• Incorporate a deep learning-based surface defect detector for real-time analysis.

Proposed method

• Combine modern measurement technology with embedded high-performance computing to create 3D surface maps of XLPE insulated cores.
• Analyze cross-sectional and lengthwise geometry errors to assess the influence of melt homogeneity.
• Develop a surface defect detection system using deep learning, evaluating convolutional neural networks for real-time data analysis.
• Apply CNN-based approaches to detect surface defects from the topography data.
• Provide results on the feasibility of real-time surface defect detection in process monitoring.

Experimental results

Research questions

• RQ1 Can topography scanning produce detailed 3D surface maps of XLPE cable cores for process monitoring?
• RQ2 What geometry errors (cross-sectional and lengthwise) arise in the insulation process and how are they related to melt homogeneity?
• RQ3 Can convolutional neural networks effectively and in real time detect surface defects from topography data?
• RQ4 How does the integrated system enable real-time monitoring and defect detection in the insulation process?

Key findings

• A novel topography scanning system can produce a detailed 3D surface map of the insulated cable core.
• Cross-sectional and lengthwise geometry errors are studied and linked to melt homogeneity as a major factor.
• A surface defect detection system based on deep learning enables reliable real-time detection of surface defects.
• Convolutional neural networks are well suited for real-time analysis of surface measurement data.