Automatic Controllers for Marine Engineering Systems - Case Study Example

AUTOMATIC CONTROLLERS FOR MARINE ENGINEERING SYSTEMS The design and use of control logic is a common and a strategic scenario in most engineering systems and the ones that come under the purview of marine engineering are no exception. In fact, the quest for designing and implementing control logic that provides for the evolution of a dynamic system capable of enhancing a system’s stability in all situations is a common goal for most marine engineering design teams. The whole and sole goal for designing control logic and control systems therefore, is to bring in a sense of automation into the system that will enable it to analyze the various parameters of functioning through continuous feedback and take the appropriate action that will facilitate the smooth running of the system in the desired direction. It is in this regard that the use and implementation of various types of controllers in marine engineering systems shall be discussed and compared. Under the current paper, PID, Robust and Optimal controllers shall be discussed. PID stands for proportional integral derivative controller. Though the name seems to rely a bit on the mathematical logic used for control purposes, the system relies on the use of a loop feedback mechanism that provides continuous input into the state of the various parts of the system. Consider the example of an autopilot that is a common aspect in most marine systems. An autopilot relies heavily on such feedback based systems, wherein the correct and set course of the vehicle is maintained by the controller. Essentially speaking, the PID controller receives feedback of the current direction of the marine vehicle (in the form of the feedback parameter) and compares it with the actual value that needs to be followed. Depending on the outcome of such comparison, the PID controller is configured with mechanisms that enable to perform the necessary corrective action that will allow the vehicle to proceed in the correct direction (Bryson, 1999). The name PID derives from the fact that the algorithm that works as part of the logic behind the controller consists of three main ingredients namely the Proportional, Integral and the Derivative. Each of these functions has different criteria to look at. The proportional component of a PID controller works towards estimating the amount of reaction that needs to be expended with respect to the recently measured error. The integral component is used to estimate the overall summation of the set of recent errors. Thus, the integral component provides a historical perspective to the controller. The Derivative controller, on the other hand, is used to estimate the rate of change of the measured error with respect to time. The calculation of the derivative component is particularly important as it helps determine the speed at which the controller is supposed to take the corrective action in order to avoid any undesired actions. The proportional, integral and the derivative components are summed up to determine the amount of adjustment that needs to be affected by the controller system. The required correction is brought about by the issue of a command by the controller to a control element such as the valve of an oxygen inlet in an underwater marine vehicle (Oppenheim, Willsky, & Nawab, 1997). With the help of the measurement and calculation of the P, I & D parameters, a PID controller can be used to design specialized control systems that can be used for certain and specific processes alone. This means that in order to design a PID controller, the design team must develop the controller from scratch as most of the work goes into determining the parameters that need to be monitored and the manner in which these parameters must be translated into the determination of the proportional, integral and derivative components. On a similar note, the degree of efficiency of a PID controller is measured in two different ways. In the first approach, the PID controller’s efficiency is estimated by the degree of its responsiveness to a set of possible errors. On the other hand, the extent to which a controller can swerve from a given set point and the degree to which it oscillates from the exact desired standpoint is also an important criterion when judging the efficiency of a PID controller. In the presence of these vulnerabilities, it is obvious that a PID controller cannot guarantee any optimal efficiency or control of a system (Bryson, 1999). A common advantage with PID controllers lies in the way in which they can be configured to cater to specific control systems’ responsiveness requirements. Depending on the need of a control system, variations of a PID controller in the form of a PI, P, PD or I control controller can be used with ease. To provide for the vulnerabilities exhibited by PID based control systems, optimal Control theory based controllers determine the various strategies that allow for a system to be configured in such a way that the vulnerabilities discussed above can be minimized and the control system can instead focus on the minimization or maximization of a desired attribute. However, an optimal control system suffers from a major problem whenever the information provided by a feedback loop in uncertain or incorrect or whenever the control system is disturbed by external phenomenon. In such cases, the control system is restricted from taking the correct actions. This situation can be overcome by the PID controller to a certain extent given the fact that the presence of an integral component helps the system to calculate the required action depending on the previous historical results. An optimal controller, however, is configured to either maximize or minimize the specified criterion in the wake of such uncertainties (Sage, 1968). In situations that lead to random behaviour of the system, an optimal control system works by way of being able to recognize such random behaviour initially, and then work towards optimizing the response to the configured levels. A typical example in this regard would the use of an optimal control system in the use of navigation systems, wherein a typical scenario could be the presence of an obstacle in a ship’s path. In such situations, the work of an optimal control system would be to first determine the time before possible collision and then issue commands to a control namely the propulsion system so that corrective actions (reversing the direction of sailing or stopping the engines etc) can be taken on time. Another limitation of an optimal control system is that its functioning is limited by the fact that most of its actions are calculated by anticipating the behaviour of parameters ahead of time. In situations where such future estimations are not possible, optimal systems have not been able to perform to required efficiency, which is overcome by PID based control systems. Optimal control systems work by way of using a cost function that is made up of numerous state and control variables. The cost function is used to determine the value of an outcome, which is compared against the optimality criterion. The goal of the control system in every situation is to work towards minimizing the cost function and is calculated by the use of a set of differential equations. This approach differs from the one adopted by PID wherein the objective is to determine the disparity between the expected and observed values at an instance of time and take the necessary corrective action at the same instance. As such, there is no minimization involved in the PID approach (Sage, 1968). An Optimal controller also depends on two basic specifications namely, the necessary and the sufficient condition. Thus, the output of the cost function is used to determine whether it satisfies the necessity and sufficiency parameters, which differs from the PID approach, which only specifies a necessary condition that needs to be followed. A robust controller refers as defined by Chandrasekhar (1998) aims to do the following: "Robust control refers to the control of unknown plants with unknown dynamics subject to unknown disturbances" From the above definition, it is clear that robust control systems are used in situations that require control even in the presence of uncertainty. As such, the efficiency of a robust control system is determined by the degree to which it is able to handle a given uncertainty. As such, it differs from PID and optimal control systems, whose performance is dependent on correct feedback and the minimization of uncertainties. When talking about a robust control system, the various uncertainties that the control system has to deal with contribute to the additive or the multiplicative component of the control logic (Zames, 2001). It is also deemed necessary to mention that in the other two types of controllers discussed previously, the output of the system is dependent on the feedback as such these systems can be termed to be responsive in nature. On the other hand, a robust control system works towards creating a bound for the uncertainty thereby leading to a system that works towards minimizing the uncertainties generated by every iteration of the control logic. One such example of a robust controller is an adaptive control system where the output of the system is attempted to be brought closer and closer to the desired value with every sequence of operation. Additionally, a robust controller is also dependent on effective feedback, using which it continues to learn about changes in the system parameters and act accordingly. Thus it can be seen that a robust controller is restrictive of uncertainties in nature. Optimal controllers otherwise work towards minimizing the cost function (Ackermann, 1993). PID and optimal controllers cannot measure any non performance or malfunctioning of the sensors and feedback mechanisms. Robust control systems, on the other hand, are complex to build and are capable of measuring the uncertainties associated with every sensory equipment as well. Given the additional constraints and the precision associated with robust controllers, designing and implementing them require a multi-faceted approach. In fact, the robustness of a system is determined by how well a control system reacts to erroneous or failed inputs as well as stressful conditions. One of the best scenarios where PID controllers find use are automatic cruise control systems. These systems are configured to work in such a way that they are capable of guiding the course of a marine vessel along a pre-determined path. The task of a cruise control system is to navigate the course of the vessel and in doing so; the control system has to periodically compare the actual position and direction of the vessel against the coordinates and the direction that needs to be followed. After estimation of the amount of correction needed, the PID controller issues the required signals to the control elements responsible for guiding the vessel on the correct path. To any one examining the working of this controller, it seems as if the vessel has been on the correct path all along, but in reality, the sense of correct path has been brought about by the PID controller through a series of corrective adjustments of the cruise system. Such a cruise system does not employ the other two systems as the question of optimization does not arise (Ackermann, 1993). A simple situation that requires the use of robust controllers is in the automated control of diesel engines. The amount of power that the diesel engines need to generate is dependent on the amount of speed that needs to be maintained in conjunction with the amount of resistance from the water that interacts with a marine vessel’s rudders. The question before the robust controller would be to reduce the consumption of fuel and yet utilize the least possible travel time. Given the uncertain nature of the latter constraint, the robust controller tries to achieve efficiency by ensuring that both these parameters are contained to the greatest degree possible. It also means that the robust controller would have to be configured with an estimation function that would take both these factors into consideration. Given these constraints that need to be maintained, a PID controller would not find place as it is not capable to handle problems of minimization and is dependent on a set of predetermined parameters that need to be followed. The system cannot also be called optimal as such an example requires the use of precision based robust control rather than optimal controllers as the latter are not very good in cases of uncertainties and unpredictable behavior (Oppenheim, Willsky, & Nawab, 1997). Unmanned underwater vehicles take the help of optimal controllers for waypoint tracking and obstacle avoidance. The task of the optimum controller is to monitor the path to a destination and given the several constraints underwater, such a vehicle must be capable of negotiating any obstacles in the least possible time and effort. This requirement of minimization leads to the use of optimality as the prime criterion for designing the controller for such underwater vehicular systems. In such vehicles, the controller determines a way around the obstacles and upon carting out a clear path, issues commands to the actuators for affecting the requisite movement. In cases of faults, the fault diagnosis part detects any such occurrences and isolates such faulty components from being used for further actuation. Thus, such vehicles come with fault tolerant optimal controllers (Sage, 1968). CONCLUSION The preceding paragraphs have discussed in great detail some of the different types of controllers that are in use today under marine engineering, with each one being used for a specific problem situation. The choice of using a specific form of controller however depends on the preferred complexity, observability, stability and the controllability of the system. For systems that rely on some sort of historical lookup, the PID controllers are the best bet as they provide well for systems that rely on a set of historical results in certain situations. Whenever controllers are supposed to be designed with a view to achieving a great degree of precision (which in turn requires that such controllers contain additional elements that filter out all possible disturbances), robust controllers are the most suitable options as they are not only capable of withstanding uncertainties in most cases, but are also additionally capable of being adaptive in nature. As such, given the complexity of the equipments involved, robust controllers are expensive and time consuming to construct and deploy. Optimal controllers can be regarded as being a partial superset of the former two types as they provide a certain level of accuracy in addition to relying on some form of feedback for their results. REFERENCES 1. Chandrasekhar (1998), Robust Control of Linear Dynamical Systems, Academic Press. 2. Bryson (1999), Optimal Control -- 1950 to 1985, IEEE Control Systems 3. Oppenheim, Willsky, & Nawab (1997), Signals and Systems, Second Edition, Prentice Hall. 4. Sage (1968), Optimal Systems Control, Prentice Hall, Inc. 5. Zames, G (2001), Input-Output Feedback Stability and Robustness, 1959-85, IEEE Control Systems. 6. Ackermann (1993), Robust Control, Systems with Uncertain Physical Parameters, Springer-Verlag. Read More