The paper presents two original and innovative contributions: 1) the model of machine learning (ML) based approach for predictive maintenance in manufacturing system based on machine status indications only, and 2) semi-Double-loop machine learning based intelligent Cyber-Physical System (I-CPS) architecture as a higher-level environment for ML based predictive maintenance execution. Considering only the machine status information provides rapid and very low investment-based implementation of an advanced predictive maintenance paradigm, especially important for SMEs. The model is validated in real-life situations, exploring different learning algorithms and strategies for learning maintenance predictive models. The findings show very high level of prediction accuracy.

该论文提出了两个原创性和创新性的贡献：1) 基于机器学习 (ML) 的模型，用于仅基于机器状态指示的制造系统预测性维护方法，以及 2)基于半双环机器学习的智能信息物理系统(I-CPS) 架构作为基于 ML 的预测性维护执行的更高级别环境。仅考虑机器状态信息可提供快速且基于投资的高级预测性维护范例​​的实施，这对中小企业尤其重要。该模型在现实生活中得到验证，探索不同的学习算法和学习维护预测模型的策略。研究结果表明预测准确度非常高。