How does a control system work?
The components of any control system, apart from the process or system itself, are its sensors, controller and actuators.
The sensors are responsible for acquiring measurements about the process to be able to handle it properly. The controller examines the information obtained and determines the precise actions to correct errors regarding the desired behavior.
The actuators receive the commands from the controller and transform them into inputs that alter the evolution of the system.
Thus, in a refrigerator, a temperature sensor continuously informs the controller. If the measurement is above the desired setpoint, the controller will start a compressor. This is an actuator that activates the refrigeration cycle. When the temperature is below the desired temperature, the controller will deactivate the compressor.
Control mechanisms
In the previous example it can be seen that the main control mechanism is feedback: the controller is fed back with the measurements of the system outputs which are, in turn, the result of the actions carried out previously.
But there is also the pre-feeding mechanism, which allows anticipating the expected behavior of the process. Thus, a car can be accelerated just before a very steep hill to try to maintain its speed.
In practice, both mechanisms are combined. Feedback enables current errors to be corrected, and feedforward tries to prevent errors from occurring in advance. In a rally car, for example, the driver keeps the vehicle on the section that is in view (feedback), while the co-driver provides information on the next section that is not yet seen (preview).
The pandemic as a control system
The components of the COVID-19 control system can be viewed in a closed loop in Figure 2.
- Instead of sensors, there are indicators to know the extent of the pandemic such as antibody tests, people with symptoms or hospitalized patients.
- The controller of this system would correspond to the Executive Power of each country, that is, its Government advised by committees of experts.
- Instead of actuators would be the interventions adopted such as social distancing or strengthening the health system with the acquisition of protective material and treatments, including the vaccine when available.
The Spanish case
It is possible to use a simplified model to simulate the Spanish case in the face of the first wave of infections derived from the spread of the SARS-CoV-2 coronavirus.
The Spanish Government, as the main intervention, ordered a very strict confinement of the population that was quite late, since the WHO raised the world alert to very high two weeks before.
Figure 2 shows the percentage of the population without contagion over time with a blue line, in which three stages are distinguished:
In the first stage, where no action was taken, there is an exponential contagion curve. This indicates that the process is unstable if left unchecked.
Right after the adoption of the confinement, a slowdown of infections is observed that do not stop completely. This is characteristic of dynamic systems where there is a certain inertia, as in an emergency braking of a car, which continues to move despite applying maximum performance.
Finally, in the third stage, a smooth contagion curve appears, confirming the effectiveness of confinement.
A fictitious scenario where confinement had been adopted a week earlier has also been contemplated in this same figure with a red line. This would have resulted in a considerable reduction in infections.
In Figure 3, an estimate, according to the model, of the severely infected who would need hospital care has been represented for both scenarios. A very high peak can be observed (blue line), which would be quite attenuated if it had acted a week before (red line).
The main problem with both curves is that they exceed the estimated response capacity of the health system for a long time (black line). This translates into a very high number of deaths.
Difficulties in managing COVID-19
As a control problem, COVID-19 is complex to address because there are numerous uncertainties in the control loop. Some of them are:
- The reliable detection of infected (in the indicators of evolution).
- The exact effect of each specific intervention (in the actions carried out).
- The rate and form of transmission of the virus (in the dynamics of the system).
Likewise, the time delay introduced by the incubation time, aggravated by asymptomatic cases, is another relevant problem that tends to unstable the process. Consider, for example, that robots on Mars cannot be directed directly from Earth due to communication delays between the two planets and they have to move autonomously.
Still, it can be helpful to apply control engineering principles in managing pandemics like COVID-19. It would be about using control mechanisms to maintain the economic activity of a country without collapsing its hospital capacity.
Spain has accredited experts worldwide in process control who could participate in decision-making to try to mitigate the effects of this and future infectious diseases.
Jorge Luis Martínez Rodríguez, University Professor. Systems Engineering and Automation Department, University of Malaga
This article was originally published on The Conversation. Read the original.
