does someone know the difference PID & Robust & Optimal controllers for marine engineering systems?

Question

Marine Engineering

Answer 1

Robust controls are designed to account for an unknown or an approximated value. Basically a difference between a theoretical vs actual value is compensated for by the controller. An example would be a boiler airflow control which computes airflow using an theoretical model based upon a single physical measurement of differential pressure. The controller model is ignoring other factors which would effect the computation such as temperature and atmospheric pressure. Instead the robust control algorithm is designed to provide an adequate output signal based solely upon the single input variable in such a way that the influences of the other physical factors are insignificant to the controller action.


Optimal Control base or adjust controller action upon one or more limiting factors. For example boiler firing rate is influenced by drum pressure. Lower drum pressure cause a call for more fuel increasing the firing rate. An optimized control may place limits on firing rate so that if the boiler pressure continues to drop below a minimum operating pressure the firing rate is reduced thereby protecting the boiler from being overfired during a situation where water circulation may be adversly effected by low pressure operation. Another example of optimization would be a three element feed water regulator where the controller monitors drum level, feed flow into and steam flow out of the boiler drum. Changes in steam demand which cause water levels to shrink and swell are compensated for. A sudden call for steam initially lowers drum pressure which causes the drum level to temporarily rise. Optimizing the feed flow with a three element regulator the increase in steam demand over rides the drum level measurment and increases feed rate to the boiler. Without this optimization a rise in drum level would cause a reduction in feed flow at a time when more steam is leaving the drum.


PID is Proportional, Intergral Derravitive Control. PID takes into account not only the amount of deviation from setpoint but includes rates of the deviation and duration of deviation. These three controller actions combine so that the process variable is controlled and maintaied at setpoint without deviation and a minimum of over/under shoot.

Consider Proportional action by itself in a temperature controller of an oil cooler with a set point of 120F. If the temperature rises the amount of cooling water flow is increased. And conversly if the temp falls the cooling water flow is reduced.

A temperature rise in the oil caused by an increase in speed causes the cooling water value to open a little more. The amount of this increased flow depends on how much hotter the oil gets abouve the setpoint. The hotter the oil gets the flow of cooling water is proportionally increased to increase cooling. Since the engine is now running faster the heat load has increased causing the oil temp to rise. It is this rise in temperature above the setpoint which causes the proportional control to call for more cooling water. The cooing valve is kept open more as long as the oil is hotter than the set point. But the valve only opens by an amount proportional to the amount of deviation between the oil temp and its set point. Temp raises a little the valve open a little, temp raises a lot the valve opens a lot. If the valve opens too much in response to an increase in temp the oil will be over cooled and the temp will fall below the setpoint at which time the valve will close causing a rise in oil temp. This condition of repeated opening and closing due to over control is known as hunting and is undesirable.

In a purely proportional control with response rate tuned to avoid hunting, an offset will ocurr between the process variable (oil temp now @ 122F) and the set point 120F because of the increase in load(speed increase generating more heat) In order to cool the oil in response to the speed increase the oil had to rise in temp to caue the cooling value to open. Additional cooling flow would be needed to further lower the oil temp back to its setpoint.
This is where in Integral control comes into play. The integral control monitors how long and how much of an offset is present. The Integral control provides the additional corrective action to return the 122F oil back to its 120F setpoint by adding additional signal to open the cooling valve a little more ultimately removing the offset. Derrivative control adds aditional controller action to help anticipate where the otput signal to the cooling valve needs to be in response to how fast the temperature increases. So for a sudden large speed change which causes a fast rise in oil temp The cooling valve is kicked open in anticipation of the ultimate temperature rise the controller is now taking a proactive measure to limit the temperature rise anticipating the higher temps to follow rather than just wainting on the proportional control to react. Derrivative action limits the overshoot, minimizing the amplitude of excursion of the process variable

Answer 2

Anywhere there is a closed control loop, be it marine or otherwise, there must be a controller. A controller is just a means, an algorithm, to adjust the response of the closed loop.

Although you do not mention any specific loop, my guess is that you are talking about boiler related control loops. If that is the case, consult the quoted reference book.

Finding the "best" response is a somewhat subjective task In theory, the response can be either "fast" (somewhat oscillatory) or "slow" (damped). In reality, the nature of the loop and its many elements do not allow a "nice and clean" response, and one must be happy with a "reasonable" response, whatever that comes out to be.

PID tuning implies a controller which has three parameters that can be set to correct the error, Proportional (or gain), time Integral, or time Derivative, of the error.

Robust control methods deal with small differences between the true system and the nominal model used for design. It is not a specific controller, but the means to an end. It is a somewhat loosely defined method.

Optimal control implies we have a global model of a system in the form of a matrix, to which a specific optimizing technique can be applied, like the time integral of the error squared. Despite all the theory available, obtaining the matrix of a system, also called "modeling and identification", is by no means straightforward.

I.o.w., PID & Robust & Optimal controllers are completely different approaches to deal with the control issue.

In the end, most control loops are just single-loop PID type. Basically, because there is a very poor understanding of the process´ dynamic behavior and an accurate model is not available. PID allows to achieve a decent response by the trial-and-error method, which satisfies most situations.
.
.

Answer 3

The PID (Proportional, Integral, Derivative) controller drovides a control signal designed to minimize the difference between the state of the controlled system and the setpoint value for the system by utilizing not only the linear difference (the proportional part) bit also the long term integral and the derivative of hte error (which used to be called 'rate feedback' in older control systems). Basically, the controller can be adjusted to compensate for perturbations to a system described by a simple, 2'nd order linear differential equation.
The newer 'Robust' and 'Optimal' control systems use a lot of microprocessor muscle to continuously analyse the response of the controlled system to the correction outputs of the controller in order to create a 'model' of the system being controlled.
One word of warning: Be careful if you use several of these 'Robust' controllers. If they're coupled through the system so that they can affect each other, you can develop some -really- strange oscillations if they start to 'argue' with each other over what needs to be done.

Doug

Answer 4

okay, here is the skinny.The Navy and Marines are the expeditionary force (people tend to think of the Marines as a branch even though they are a department of the Navy) . There job is to go out and take the enemy areas. The Navy will commence with aerial and sea bombardments while the Marines will over take compounds, building and construction. Your Marines are the boots on group type of people. The Army and Air force or more of a policing faction these days. Army and Air force personnel more or less watch over the areas that have been taken over and root out the rebellious nature that may linger in the area. But with that being said either branch can do what the other does. IE the navy sending troops to Iraq and Afghanistan to help support the army