Scholarlink

<p>With the advent of the fourth industrial revolution, the traditional manufacturing industries’ practices were replaced with smart manufacturing technologies. This paper addresses the predictive maintenance (PM) of Flexible Unit Systems (FUS) to predict the Remaining Useful Life (RUL) of machines. This is an important topic and alternative solution since industries are in a dilemma to shift from their existing technologies to recently emergent practices. Within this context, we present a novel integrated workload adjustment strategy proposed with a Meta-learning-based intelligent cyber-physical system approach for predicting the health status of the machines in turn minimizing the throughput time. Primarily, the model framework is developed based on a machine learning (ML) approach for the considered FUS. A semi-double loop ML-based I-CPS architecture has been used to develop an ML-based PM approach to predict the RUL of the machines. Thereafter, with simulations the workload on machines are adjusted based on predicted RUL to process the number of jobs to reduce the throughput time. With real-life configurations, the developed approach is evaluated with ML parameters and found that the predicted algorithms performed with above 95% accuracy. The feasibility of our results is validated by a comparison study with the other two benchmark strategies i.e. equal and random workload for throughput time minimization. The results stated that the proposed method provided minimum throughput time for processing the number of jobs compared to the other two benchmark strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine-Learning Based Predictive Maintenance for Flexible Systems: A Cyber-Physical Approach to Minimize Throughput time

  • 期刊论文

  • 发表日期：

    2026年1月28日

查看全文
  • Thirupathi Samala,
  • Kanthala Uma Reddy

摘要

With the advent of the fourth industrial revolution, the traditional manufacturing industries’ practices were replaced with smart manufacturing technologies. This paper addresses the predictive maintenance (PM) of Flexible Unit Systems (FUS) to predict the Remaining Useful Life (RUL) of machines. This is an important topic and alternative solution since industries are in a dilemma to shift from their existing technologies to recently emergent practices. Within this context, we present a novel integrated workload adjustment strategy proposed with a Meta-learning-based intelligent cyber-physical system approach for predicting the health status of the machines in turn minimizing the throughput time. Primarily, the model framework is developed based on a machine learning (ML) approach for the considered FUS. A semi-double loop ML-based I-CPS architecture has been used to develop an ML-based PM approach to predict the RUL of the machines. Thereafter, with simulations the workload on machines are adjusted based on predicted RUL to process the number of jobs to reduce the throughput time. With real-life configurations, the developed approach is evaluated with ML parameters and found that the predicted algorithms performed with above 95% accuracy. The feasibility of our results is validated by a comparison study with the other two benchmark strategies i.e. equal and random workload for throughput time minimization. The results stated that the proposed method provided minimum throughput time for processing the number of jobs compared to the other two benchmark strategies.