An Application of Neural Networks in Predictive Construction Equipment Maintenance

Abstract

Construction project equipment are subject to several types of breakdowns throughout the project duration. As a result, contractors and equipment operators are keen to establish and adopt effective equipment maintenance strategies. Adopting a maintenance strategy that minimizes the downtime of construction equipment and allows for the progression of works in a timely manner is essential to satisfy the increasingly stringent constraints set by project owners. The availability of several types of equipment data is crucial to understand the breakdown patterns of construction equipment. However, in many cases, projects operating with tight profit margins, and particularly projects in developing countries, access to such data is not always readily available. The aim of this research study is to establish a predictive maintenance framework based on machine learning (ML) that leverages historical breakdown data with the absence of information relating to the condition of the equipment and any output extracted from monitoring devices and sensors. The proposed model for accomplishing this task is the multilayer perceptron (MLP) neural network, which is applied to a real-life multi-million-dollar infrastructure project in the Middle East region. The collected data includes an equipment maintenance log database.

The results obtained are promising, with significant improvements shown in accuracy in terms of mean absolute error (MAE) compared to the baseline models: Linear Regression and Non-linear Regression. An improvement of 185% compared to the Linear Regression model, and an improvement of 26% compared to the Non-linear Regression model in the case of equipment of type excavator was witnessed. Moreover, an improvement of 173% compared to the Linear Regression model, and an improvement of 23% compared to the Non-linear Regression model in the case of equipment of type articulated haulers was witnessed. This framework could be of significant value to the industry practitioner, as it could play a role in enhancing the overall productivity of construction equipment by minimizing their breakdown rate and criticality, in turn reducing the associated equipment operating costs and expediting the rate at which works are performed.