Scholarlink

<p>Industrial machinery (IM) holds significant importance in the field of research especially for prototyping and experimentation, large-scale data generation, testing and quality control, simulation and modeling, precision (P) and accuracy (A), automation and efficiency. Thus, it is important to keep track of the health of IM at regular intervals for the purpose of tracking the equipment lifespan, maintenance costs, unscheduled downtime, efficiency and performance, and safety hazards, etc. for taking preventive measures at the earliest. So, different approaches such as federated learning (FL), quantum computing (QC), machine learning (ML), deep learning (DL), edge computing, etc. can be utilized to track the health status of IM. In this work, Quantum ML (QML) based approach is focused to monitor the fitness of industrial grinding machine (IGM) as ok (K) or not ok (NK). This work is focused on Quantum Gradient Boosting Model (QGB), Quantum Feature Mapping with Support Vector Machine (QSM) and Hybrid Quantum Model (HQM) to track the fitness of IGM. These models are also compared with classical SVM, Random Forest (RF), GB and XGBoost (XGB) for effective comparison. These models are compared in terms of classification A (in %), P ( in %), Recall (R in %), F1-Score (F1S in %), Training Time (TT in Sec.), Memory Usage (M in MB) and Model Size (MS in KB) by considering the training and testing ratio (TTR in %) as 80:20, 70:30 and 60:40 for hold-out validation and also using the 5-Fold cross validation (CV) mechanism. Here, receiver operating characteristic (ROC) curve is plotted by computing the true positive rate (TPR) and false positive rate (FPR) for each model, and the box plot and heatmap graphs are plotted by focusing on these parameters to visualize the results. The overall results (both hold-out and 5-Fold CV) indicate that HQM achieves higher performance in terms of A, P, R, F1S and TT with average values of 99.45, 99.50, 99.45, 99.47, 0.05 respectively, GB in terms of M with average value of 59.09 and RF in terms of MS with average value of 7.67 as compared to other models. This work is carried out in Python environment.</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

Quantum Inspired Data Driven Fitness Monitoring Approach for Industrial Machinery

  • 期刊论文

  • 发表日期：

    2026年4月3日

查看全文
  • Asish Kumar Roy,
  • Kalyan Kumar Jena,
  • Debasis Mohapatra

摘要

Industrial machinery (IM) holds significant importance in the field of research especially for prototyping and experimentation, large-scale data generation, testing and quality control, simulation and modeling, precision (P) and accuracy (A), automation and efficiency. Thus, it is important to keep track of the health of IM at regular intervals for the purpose of tracking the equipment lifespan, maintenance costs, unscheduled downtime, efficiency and performance, and safety hazards, etc. for taking preventive measures at the earliest. So, different approaches such as federated learning (FL), quantum computing (QC), machine learning (ML), deep learning (DL), edge computing, etc. can be utilized to track the health status of IM. In this work, Quantum ML (QML) based approach is focused to monitor the fitness of industrial grinding machine (IGM) as ok (K) or not ok (NK). This work is focused on Quantum Gradient Boosting Model (QGB), Quantum Feature Mapping with Support Vector Machine (QSM) and Hybrid Quantum Model (HQM) to track the fitness of IGM. These models are also compared with classical SVM, Random Forest (RF), GB and XGBoost (XGB) for effective comparison. These models are compared in terms of classification A (in %), P ( in %), Recall (R in %), F1-Score (F1S in %), Training Time (TT in Sec.), Memory Usage (M in MB) and Model Size (MS in KB) by considering the training and testing ratio (TTR in %) as 80:20, 70:30 and 60:40 for hold-out validation and also using the 5-Fold cross validation (CV) mechanism. Here, receiver operating characteristic (ROC) curve is plotted by computing the true positive rate (TPR) and false positive rate (FPR) for each model, and the box plot and heatmap graphs are plotted by focusing on these parameters to visualize the results. The overall results (both hold-out and 5-Fold CV) indicate that HQM achieves higher performance in terms of A, P, R, F1S and TT with average values of 99.45, 99.50, 99.45, 99.47, 0.05 respectively, GB in terms of M with average value of 59.09 and RF in terms of MS with average value of 7.67 as compared to other models. This work is carried out in Python environment.