Implementation and evaluation of digital twin framework for Internet of Things based healthcare systems DOI Creative Commons
Ahmed K. Jameil, H. S. Al‐Raweshidy

IET Wireless Sensor Systems, Год журнала: 2024, Номер unknown

Опубликована: Дек. 3, 2024

Abstract The integration of digital twins (DTs) in healthcare is critical but remains limited real‐time patient monitoring due to challenges achieving low‐latency telemetry transmission and efficient resource management. This paper addresses these limitations by presenting a novel cloud‐based DT framework that optimises monitoring, providing timely solution for needs. incorporates Pyomo‐based dynamic optimisation model, which reduces latency 32% improves response time 52%, surpassing existing systems. Leveraging low‐cost, multimodal sensors, the system continuously monitors physiological parameters, including SpO2, heart rate, body temperature, enabling proactive health interventions. A definition language (Digital Twin Definition Language)‐based series analysis twin graph platform further enhance sensor connectivity scalability. Additionally, machine learning (ML) strengthens predictive accuracy, 98% accuracy 99.58% under cross‐validation (cv = 20) using XGBoost algorithm. Empirical results demonstrate substantial improvements processing time, stability, capacity, with predictions completed 17 ms. represents significant advancement offering responsive scalable constraints applications. Future research could explore incorporating additional sensors advanced ML models expand its impact

Язык: Английский

A Multi-Variable Coupled Control Strategy Based on a Deep Deterministic Policy Gradient Reinforcement Learning Algorithm for a Small Pressurized Water Reactor DOI Creative Commons
Jie Chen, Kai Xiao, Huang Ke

и другие.

Energies, Год журнала: 2025, Номер 18(6), С. 1517 - 1517

Опубликована: Март 19, 2025

The reactor system has multivariate, nonlinear, and strongly coupled dynamic characteristics, which puts high demands on the robustness, real-time demand, accuracy of control strategy. Conventional approaches depend mathematical model being controlled, making it challenging to handle system’s complexity uncertainties. This paper proposes a multi-variable strategy for nuclear steam supply based Deep Deterministic Policy Gradient reinforcement learning algorithm, designs trains intelligent controller simultaneously realize coordinated multiple parameters, such as power, average coolant temperature, pressure, etc., performs simulation validation under typical transient variable load working conditions. Simulation results show that effect is better than PID ±10% FP step condition, linear dumping power overshooting amount regulation time, maximum deviation pressure pressurizer relative liquid level, time are improved by at least 15.5% compared with traditional method. Therefore, this study offers theoretical framework utilizing in field control.

Язык: Английский

Процитировано

0
Implementation and evaluation of digital twin framework for Internet of Things based healthcare systems DOI Creative Commons
Ahmed K. Jameil, H. S. Al‐Raweshidy

IET Wireless Sensor Systems, Год журнала: 2024, Номер unknown

Опубликована: Дек. 3, 2024

Abstract The integration of digital twins (DTs) in healthcare is critical but remains limited real‐time patient monitoring due to challenges achieving low‐latency telemetry transmission and efficient resource management. This paper addresses these limitations by presenting a novel cloud‐based DT framework that optimises monitoring, providing timely solution for needs. incorporates Pyomo‐based dynamic optimisation model, which reduces latency 32% improves response time 52%, surpassing existing systems. Leveraging low‐cost, multimodal sensors, the system continuously monitors physiological parameters, including SpO2, heart rate, body temperature, enabling proactive health interventions. A definition language (Digital Twin Definition Language)‐based series analysis twin graph platform further enhance sensor connectivity scalability. Additionally, machine learning (ML) strengthens predictive accuracy, 98% accuracy 99.58% under cross‐validation (cv = 20) using XGBoost algorithm. Empirical results demonstrate substantial improvements processing time, stability, capacity, with predictions completed 17 ms. represents significant advancement offering responsive scalable constraints applications. Future research could explore incorporating additional sensors advanced ML models expand its impact

Язык: Английский

Процитировано

2