Open Access

ARTICLE

A Multi-Sensor and PCSV Asymptotic Classification Method for Additive Manufacturing High Precision and Efficient Fault Diagnosis

Lingfeng Wang¹, Dongbiao Li², Fei Xing^1,3,*, Qiang Wang⁴, Jianjun Shi⁵

1 School of Mechanical Engineering, Shenyang University of Technology, Shenyang, 110870, China
2 CCCC Tunnel and Bridge (NANJING) Technology Co., Ltd., Nanjing, 211800, China
3 Nanjing Zhongke Raycham Laser Technology Co., Ltd., Nanjing, 210038, China
4 College of Automation & College of Artificial Intelligence, Nanjing University of Posts and Telecommunications, Nanjing, 210003, China
5 Industrial Center, Nanjing Institute of Technology, Nanjing, 211167, China

* Corresponding Author: Fei Xing. Email:

(This article belongs to the Special Issue: Sensing Data Based Structural Health Monitoring in Engineering)

Structural Durability & Health Monitoring 2025, 19(5), 1183-1201. https://doi.org/10.32604/sdhm.2025.063701

Received 21 January 2025; Accepted 14 April 2025; Issue published 05 September 2025

Abstract

With the intelligent upgrading of manufacturing equipment, achieving high-precision and efficient fault diagnosis is essential to enhance equipment stability and increase productivity. Online monitoring and fault diagnosis technology play a critical role in improving the stability of metal additive manufacturing equipment. However, the limited proportion of fault data during operation challenges the accuracy and efficiency of multi-classification models due to excessive redundant data. A multi-sensor and principal component analysis (PCA) and support vector machine (SVM) asymptotic classification (PCSV) for additive manufacturing fault diagnosis method is proposed, and it divides the fault diagnosis into two steps. In the first step, real-time data are evaluated using the T² and Q statistical parameters of the PCA model to identify potential faults while filtering non-fault data, thereby reducing redundancy and enhancing real-time efficiency. In the second step, the identified fault data are input into the SVM model for precise multi-class classification of fault categories. The PCSV method advances the field by significantly improving diagnostic accuracy and efficiency, achieving an accuracy of 99%, a diagnosis time of 0.65 s, and a training time of 503 s. The experimental results demonstrate the sophistication of the PCSV method for high-precision and high-efficiency fault diagnosis of small fault samples.

---

Keywords

---