| Adaptive Closed-Loop Neural Network Control of Artificial Ventilator |
| Paper ID : 1078-ICEEM2025 (R1) |
| Authors |
|
Mina Samer Habashy *1, Tarek Ahmed Mahmoud2, Mohamed Esmail Karar3 1Menofia University - Faculty of electronic engineering Menouf 2Menofia university- Faculty of electronic engineering 3Menofia university - Faculty of electronic engineering |
| Abstract |
| Artificial mechanical ventilator is a critical life-support system used in intensive care units (ICUs) to assist patients with lung diseases and respiratory failure. This paper proposes an adaptive neural network-based PID controller for an artificial ventilator. The traditional PID controller is commonly used in many medical applications; however, it cannot regulate the varying patterns of a critical patient's breathing. The proposed control system showed robust performance in the pressure-controlled ventilation (PCV) scenario, maintaining precise pressure regulation in the simulated case of sedated patients. Compared to traditional PID control, our proposed ventilator controller demonstrated superior responsiveness and accuracy in maintaining target ventilation pressures, achieving a minimal overshot of approximately 2.86% and the shortest peak time and settling time of 0.12 and 0.19 seconds, respectively, in one breathing cycle. Adaptive adjustment of the PID control parameters using the neural network resulted in stable air pressure delivery, thereby minimizing the clinical risk of ventilation-related issues. This proposed controller represents a significant step towards fully automated and personalized respiratory support system. |
| Keywords |
| Mechanical ventilator, adaptive neural network, intelligent control |
| Status: Accepted |