Open Access

ARTICLE

Combining Principal Component Analysis and Multilayer Perceptron to Establish a Construction Quality Prediction Model

Ching-Lung Fan^*

Department of Civil Engineering, Republic of China Military Academy, Kaohsiung, Taiwan

* Corresponding Author: Ching-Lung Fan. Email:

(This article belongs to the Special Issue: AI-driven Monitoring, Condition Assessment, and Data Analytics for Enhancing Infrastructure Resilience)

Structural Durability & Health Monitoring 2026, 20(5), 24 https://doi.org/10.32604/sdhm.2026.081046

Received 22 February 2026; Accepted 03 June 2026; Issue published 24 August 2026

Abstract

Quality attainment in public construction projects is paramount for effective project management. This study employs Principal Component Analysis (PCA) in the initial phase to discern noncollinear critical defects from inspections across 1015 projects. The identified components are categorized into three aspects: Inspection Records and Occupational Safety and Health (Aspect I), Concrete Quality (Aspect II), and Construction Team Quality Management (Aspect III). Subsequently, a Multilayer Perceptron (MLP) network, trained on 13 PCA-identified critical defects, transforms input data into a probability, indicating project quality. The MLP model exhibits exceptional performance with 91.3% accuracy, 90.2% precision, and 96.1% recall. This hybrid machine learning approach effectively extracts and correlates defect-related information with construction quality, providing a robust tool for accurate quality prediction in construction projects.

---

Keywords

---