Scholarlink

<p>Hydraulic cylinders widely employed in manufacturing equipment are prone to internal wear and leakage anomalies under dynamic loads, leading to low efficiency and quality. Digital twin-based condition monitoring can be considered a more effective solution than a traditional physical approach or a data-driven approach alone. However, constructing high-precision digital twin models that can accurately reflect the dynamic behavior of hydraulic cylinders to support anomaly recognition remains a critical challenge. This paper presents a digital twin modeling method that comprises parametric dynamic, geometric, and behavioral models of hydraulic cylinders. A trust-region reflective (TRR)-based algorithm is developed to identify key parameters in these models, ensuring consistency between the physical and digital twin outputs. Leveraging high-fidelity digital twin models, twin-generated data is integrated with physically acquired data using an improved interactive multi-model Kalman filtering algorithm, thereby enhancing the validity of anomaly recognition. To validate the effectiveness, an experimental manufacturing platform for a 200 kN hydraulic press is built in the case study. It implements the digital twin model using a multi-body dynamics platform with the identified parameters. The observable outputs, i.e., pressure and displacement, are collected under three different working paths. Results show that the digital twin model is highly consistent with the physical production, with average absolute errors in piston’s displacement and rod-chamber pressure of 2.78&#xa0;mm and 0.092&#xa0;MPa, respectively, corresponding to percentages of 0.74% and 2.14%, respectively. The accuracy of leakage anomaly recognition reaches 98.8%, a 2.3% improvement over the prior physical-data-only approach, with a baseline of 96.5%. It indicates that high-fidelity digital twin models are effective for hydraulic cylinder monitoring, and the fusion of virtual and real data can enhance identification performance. This work assists in building high-fidelity digital twin models of hydraulic systems and in enabling accurate condition monitoring for anomaly recognition in manufacturing equipment.</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

Digital twin modeling of hydraulic cylinders in manufacturing equipment: methodology, validation, and application on condition monitoring of hydraulic press

  • 期刊论文

  • 发表日期：

    2026年5月8日

查看全文
  • Lei Li,
  • Yuan Huang,
  • Rui Jin,
  • Tao Li,
  • Yuhang Xu,
  • Yunchao Liu,
  • Haihong Huang

摘要

Hydraulic cylinders widely employed in manufacturing equipment are prone to internal wear and leakage anomalies under dynamic loads, leading to low efficiency and quality. Digital twin-based condition monitoring can be considered a more effective solution than a traditional physical approach or a data-driven approach alone. However, constructing high-precision digital twin models that can accurately reflect the dynamic behavior of hydraulic cylinders to support anomaly recognition remains a critical challenge. This paper presents a digital twin modeling method that comprises parametric dynamic, geometric, and behavioral models of hydraulic cylinders. A trust-region reflective (TRR)-based algorithm is developed to identify key parameters in these models, ensuring consistency between the physical and digital twin outputs. Leveraging high-fidelity digital twin models, twin-generated data is integrated with physically acquired data using an improved interactive multi-model Kalman filtering algorithm, thereby enhancing the validity of anomaly recognition. To validate the effectiveness, an experimental manufacturing platform for a 200 kN hydraulic press is built in the case study. It implements the digital twin model using a multi-body dynamics platform with the identified parameters. The observable outputs, i.e., pressure and displacement, are collected under three different working paths. Results show that the digital twin model is highly consistent with the physical production, with average absolute errors in piston’s displacement and rod-chamber pressure of 2.78 mm and 0.092 MPa, respectively, corresponding to percentages of 0.74% and 2.14%, respectively. The accuracy of leakage anomaly recognition reaches 98.8%, a 2.3% improvement over the prior physical-data-only approach, with a baseline of 96.5%. It indicates that high-fidelity digital twin models are effective for hydraulic cylinder monitoring, and the fusion of virtual and real data can enhance identification performance. This work assists in building high-fidelity digital twin models of hydraulic systems and in enabling accurate condition monitoring for anomaly recognition in manufacturing equipment.