What is the best way to control the speed during the cruise phase of an airplane?
I'm wondering which is the best way, because in theory I can control the speed manipulating the thrust given by the engine. However I don't know if a real engine can be manipulate the throttle with an quantification for example of 0.01 (suppose that throttle varies 0 to 1).
So, is best to make a "fine" control of the speed using manipulating the aerodynamic surfaces (elevator)? I mean in the multivariable system I have one variable to control and two inputs. Which is the best strategy, uses only the throttle variable, elevator or both?
And second, if I wanna control the altitude and speed simultaneous , which is the best strategy? Coupling or decoupling the input, I mean the feedback control law must be coupled or decoupled?
well it depends on what you want to do. if you want to measure relative speed of airplane, you would better to use pitot tube. if you want to measure absolute speed you can use GPS, grouns station radar for tracking your plane.
What which I can mention is that in reality the required precision of tracking the commands affects the precision of control inputs. For example if you want to track the M=0.85 with accuracy of DM=0.1 the required throttle increment is different from the accuracy of DM=0.01. So you can change your first question to "Which velocity increment is meaningful in real application?"
The second point is that as aircraft performance offers, you should follow a specific velocity to minimize cost (like fuel consumption). Using aerodynamic control can reduce the flight efficiency.
For your control law you have (in a simplified view) two parameters for controlling, that is the thrust and the pitch (which itself is controlled by the aerodynamic surfaces such as rudders, flaps etc). The manipulation of each of them leads to a a change in your both controlled variables (the velocity and the altitude). For example, when you increase thrust, that increases speed, but also the aircraft lift and hence the altitude.
Therefore, in my opinion, you should use a coupled control.
Thank you for your answers.
From control point o f view, I wanna control the altitude and speed. Before I make my question I will give an introduction of the problem.
Firstly, suppose the longitudinal and lateral dynamics of the aircraft are weakly coupled and the states variables of the longitudinal control are (Vt, alpha, q , theta , h). However, I do only measure the (Vt, q, theta, h).
This was I've done so far.
In my first try I did as mention in the following. Analyzing the short period and phugoid mode I noticed the both are weakly coupled. So, first I augmented the stability of the short-period mode by adding and Kd (proportional gain of angular velocity). Second, I closed the loop around the pitch and finally around the altitude.
Actually the I closed q and theta loops simultaneously (doing a loop shaping (more precisely specifying the wcg (crossover frequency) and phase margin) take in account out known dynamics (actuator, computational delays, sensor, etc).
The speed mode gives me much more trouble, since my objectives are robust against uncertainty (engine delay, thrust and drag) and at same time I wanna low control activity I do not need to track the speed precisely
The simplest control structure I've tried was a control gain proportional to velocity error. As I excepted to decrease the tracking error below the requirement, I have to increase the proportional gain, which decreases the stability margins and at same the time increase the bandwidth (not good, high control activity).
The second structure I've test was to put a feedforward gain (function of the speed reference), where the gain was chosen to the control signal be correspondent to the thrust which cancels the drag at this specific given flight condition.
As my model has a moderate uncertainty related to thrust and drag, I put together with the feedforward gain an feedback loop (only gain proportional to velocity error). As expected this decrease the control activity and recovery the stability margin) because the feedback gain was lower than the first design. However, if my estimative of drag and thrust are not so good, the feedforward gain won't let the system in specific flight condition (velocity) and the low proportional gain will result an moderate steady error.
Then, I tried an third option. Feedforward + PI. This seems creates, despite of a little reduction of the stability margins. However, due the integral term I should implement a anti-windup.
First question. What is the appropriate anti-windup logic to be implemented in this type of the problem?
Last question. Which could be the most affordable technique (output feedback, LQG/LTR, Hinfinity, rlocus (one loop per time), loop shaping, etc) to design a speed and altitude hold in terms of tuning, real implementation and robust against uncertainty.
Thank you at advance.