A model predictive current control (MPCC) with
adaptive-adjusting method of timescales for permanent magnet
synchronous motors (PMSMs) is proposed in this paper to
improve the dynamic response and prediction accuracy in
transient-state, while lessening the computational burden and
improving the control performance in steady-state. The timescale
characteristics of different parts of MPCC, such as signal
sampling, prediction calculation, control output, model error
correction, are analyzed, and the algorithm architecture of MPCC
with multi-timescale is proposed. The difference between
reference and actual speed, and the change rate of actual speed
are utilized to discriminate the transient state of speed change and
load change, respectively. Adaptive-adjusting method of control
period and prediction stepsize are illustrated in detail after
operation condition discrimination. Experimental results of a
PMSM are presented to validate the effectiveness of proposed
MPCC. In addition, comparative evaluation of single-step MPCC
with fixed timescale and proposed MPCC is conducted, which
demonstrates the superiority of proposed control strategy.
2. INTRODUCTION
• The world has been shocked by the COVID-19 disease.
• Acute respiratory failure (ARF) is the leading cause of death.
• 40 percent of critically ill COVID-19 patients developed acute respiratory
distress syndrome (ARDS) which requires invasive incubation and
ventilation .
• Mechanical ventilators are provided in hospitals.
3. OBJECTIVE
• To balance the tradeoff between the low-gain pressure and the high-gain
pressure control in the mechanical ventilation.
4. MECHANICAL VENTILATION
• It is used to assist the patients who have difficulty breathing on their own.
• It increases the pressure during an inspiration.
• The pressure is decreased during an expiration.
6. MECHANICAL VENTILATION
• A good ventilator control design ensures that the target pressure is
accurately tracked by the ventilator.
• Patient-ventilator synchrony is important for the comfort of the patient.
• Asynchrony can lead to prolonged stay in the hospital and even associates
with higher mortality.
7. VARIABLE GAIN CONTROL
• High-gain pressure controller - To achieve a fast pressure buildup and
release during ventilation.
• Low-gain pressure controller - For keeping the amount of unwanted
oscillations in the patient flow signal small, because this can result in false
triggers.
• There is a tradeoff between applying a high-gain controller and a low-gain
controller.
• A variable-gain control strategy is proposed.
9. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• Fig. 1- Photograph of a respiratory system with a hose that connects to a patient.
• Fig. 2 - Schematic of a respiratory system showing the different pressures (red), flows
(blue), and resistances and compliance (black).
• Blower system - Pressurizes the ambient air in order to ventilate the patient.
• Hose - To connect the respiratory module to the patient.
• The flow Qout that leaves the system runs through the hose toward the patient.
• The patient exhales partly back through the blower and partly through a leak in the hose.
• The leak, with leak resistance Rleak, is used to refresh the air in the hose in order to
ensure that the patient does not inhale his/her own exhaled low-oxygen, CO2-rich air.
10. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• Using the conservation of flow, the output flow Qout, patient flow Qpat, and
leakage flow Qleak are related as follows:
Qpat = Qout − Qleak (1)
• Due to the hose resistance Rhose, the output pressure pout is not equal to airway
pressure paw at the patients mouth.
• The pressure inside the lungs [with lung compliance (elastance) Clung and
resistance Rlung] is defined as plung and cannot be measured in general.
11. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• Assuming linear resistances Rlung, Rleak, and Rhose, the pressure drop
across these resistances can be related to the flow through these
resistances
12. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• and where the lung pressure satisfies the following differential equation:
• Combining (2) and (3), the lung dynamics can be written as
13. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• Substituting and rewriting (2) in (1) result in the following relation for the airway
pressure:
• Substituting this expression for the airway pressure paw into the lung dynamics (4)
results in the following differential equation for the lung dynamics:
14. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• Given (2), (5), and (6), the patient and hose system can be written as a
linear state-space system with input pout, outputs [paw, Qpat] T , and state
plung.
16. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
Closed-loop control scheme with a linear controller C(s)
17. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• The unity feedforward in combination with the identified blower characteristic
ensures that the output pressure Pout can reasonably accurately track the target
pressure Pset feedforward alone.
• The feedback controller has to compensate for the pressure drop
along the hose.
• The pressure drop along the hose is difficult to predict due to several factors, so
feedback path needs to be employed.
18. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• The actual system can be modeled as a second-order low-pass filter.
• In the state-space format, (11) can be written as
20. MATHEMATICAL MODEL OF A
RESPIRATORY SYSTEM
• By coupling the patient hose system dynamics (7) and the blower dynamics (12), the
general state-space form of the plant P(s) (to be controlled by the feedback controller) can
be formulated as,
22. PRESSURE CONTROLLER
• The goal of the pressure controller can
be achieved by formulating some
assumptions.
The rise time of a pressure set point
should be approximately 200 ms.
The pressure at the end of an
inspiration, plateau pressure, should
be within a pressure band of ±2 mbar.
The overshoot in the flow during an
expiration should have a typical value
of 2 L/min.
24. PERFORMANCE TRADEOFF USING
LINEAR CONTROL
• Paw/Pcontrol<1
• There will be a leakage flow through
the leakage hole in the hose, which
results in a pressure drop.
• To cope with such leakage-induced
disturbances, the following linear
integral feedback controller has been
designed:
25. PERFORMANCE TRADEOFF USING
LINEAR CONTROL
• Ki=0: The need for a feedback controller
in order to achieve the target pressure
accurately.
• Ki=0.4: The low-gain controller does
achieve the target pressure albeit slowly,
but it induces a flow response without
overshoot (favorable to avoid false
inspiration triggers).
• Ki=10: The high-gain controller achieves
the target pressure quickly, but it induces
an unwanted oscillation in the patient
flow, which may result in false patient
flow triggering.
26. VARIABLE-GAIN CONTROL SCHEME
• The variable-gain controller consists of a linear controller C(S) and a
variable-gain element and
28. VARIABLE-GAIN CONTROL SCHEME
• Switching law between low- and high-gain control ?
When the patient flow is small, the pressure drop along the hose is almost
constant, and the low-gain controller is favorable, since this minimizes the
oscillations in the flow response.
For large flows, however, during pressure buildup and pressure release, a
higher controller gain is desired in order to compensate for the pressure
drop along the hose quickly.
29. VARIABLE-GAIN CONTROL SCHEME
Variable gain controller is proposed to switch the controller gain based on the patient
flow.
ϕ(e, Qpat) = φ(Qpat)e
φ(Qpat) should be designed to be large for large flows (high-gain) and small for small
flows (low gain).
• δ : switching length
• α : additional gain
31. RESULTS
• Clearly, the (linear) high-gain controller has a fast pressure buildup (and pressure
release) performance.
• The overshoot exceeds the flow trigger threshold of 2 L/min and would induce
the false patient triggers.
• The low-gain controller, has a stable flow response, but it is too slow in the
pressure buildup (and pressure release) in order to meet the desired rise-time
performance specification.
• The variable-gain controller uses the nonlinearity to differentiate between
low/high gains based on the patient flow information provided by the flow
sensors.
32. RESULTS
• If the patient flow is larger than δ = 6 L/min, the additional gain α applied
in order to compensate for the pressure drop along the hose and quickly
reach the desired pressure target, similarly as the high-gain controller.
• C˜ = C(s)(1 + α)
• However, once the patient lung becomes full, and the patient flow
becomes close to zero (and, hence, the pressure drop along the hose is
approximately constant), the low-gain controller is used in order to reach a
stable flow response without overshoot.
34. INFLUENCE OF SWITCHING LENGTH
• A smaller δ value results in a small rise time, because the additional gain is active
for a longer period of time.
• However, it results in a larger overshoot.
• The variable-gain controller can balance this tradeoff in a more desirable manner,
by choosing, for example, δ ≈ 10 L/min, which results in a response with a good
rise time and also a small overshoot in the flow.
• The region in which overshoot and rise time are both small is quite large (e.g.,
any δ between 5 and 50 L/min would perform well), which means that the design
is robust with respect to the switching length δ
35. CONCLUSION
• Variable-gain control strategy for a mechanical ventilator is used in order to
balance the tradeoff between accurate and fast pressure buildup and a stable flow
response.
• The results confirm that indeed a variable-gain controller, which switches gains
on the basis of the magnitude of the patient-flow.
• As a benefit for the patient, trigger levels can be set to a small level, to allow for
best possible machine patient synchronization for the comfort of the patient,
while still achieving sufficiently fast pressure buildup that guarantees good
breathing support.
36. REFERENCES
• Hunnekens, B., Kamps, S. and Van De Wouw, N., 2018. Variable-gain
control for respiratory systems. IEEE Transactions on Control Systems
Technology, 28(1), pp.163-171.
• E. Akoumianaki et al., “Mechanical ventilation-induced reverse triggered
breaths: A frequently unrecognized form of neuromechanical coupling,”
Chest, vol. 143, no. 4, pp. 927–938, 2013.
• M. Borrello, “Modeling and control of systems for critical care
ventilation,” in Proc. Amer. Control Conf., Jun. 2005, p