2018 · 16 citations · 3 references
EngineeringRespiratory TherapyFluid DynamicsVentilationVentilator ManagementControl ValvesFlow-volume LoopsMechanical VentilationVentilation Control SystemsVentilation System DesignVolume-control VentilationAdaptive Flow ControlFeedback Control For VentilationGas FlowGas Flow Control SystemsModel Reference Adaptive ControlRobust Tracking ControlVentilatory DynamicsVolume Control
Stable and precise control of air and oxygen gas flow is an essential technology for ventilators; devices that enable breathing life support functions for hospital patients. Flow control provides for oxygen mix, pressure support, and volume control functions in these devices. Ventilator flow controllers typically employ electrically controlled, proportional solenoid valves in a feedback loop with flow sensors - primarily to meet flow accuracy. But to serve the broad functionality mentioned above, flow control needs to further provide robust tracking of often dynamic flow reference signals. Simple linear controls cannot achieve this requirement and so present methods often rely on calibrating individual valves before use and `extending' a PID control structure using variable gain and/or inverse feedforward models. But a calibration-free approach is more preferred, both by clinicians and in providing resiliency to calibration drift during use. This paper describes specific challenges surrounding flow control in ventilation and presents a new method that was developed and applied using synthesis methods of model reference adaptive control (MRAC). The controller demonstrates robust performance to valve nonlinearity and variance and automatically accommodates inlet gas pressure changes, eliminating any need for inlet gas regulators or valve inlet pressure measurements. The new controller offers uniform tracking of dynamic flow reference signals across the full range of patient load from large adult to small infant thus requiring no manual (a-priori) selection in the control structure or its gains to adjust for patient size.
3