About Us Contact Us Help


Archives

Contribute

 

Technology - Automatic Feedback Control

Dr. Anuradha Annaswamy
04/07/2003

The thermostat in our homes is a simple device that regulates the temperature at a set value. It does this by measuring the temperature and either turning on the furnace if the measured temperature is higher than the set value or turning off the furnace if the measured temperature is lower than the set value. This is the essence of automatic feedback control which underlies a range of engineering applications all the way from water-clocks in 300 B.C. to robots and nanomachines in the twenty-first century. Two thousand years ago, the Chinese and Greeks used in their water-clocks, a simple sensing device that measured the water-level and a valve to regulate the flow of water thereby making the pointer's motion and the time it indicates more precise. Since then, the world has witnessed numerous remarkable engineering innovations that utilized automatic control in one form or another. In 1788, James Watt used a fly-valve to regulate the steam engine, where if the engine went too fast, the flyballs swung outward causing to slow the engine down, and conversely, if the engine went too slow, the inward swing of the flyballs caused the engine to speed up by opening an input valve. Yet another example occurred in 1914 when Lawrence Sperry demonstrated how the aircraft is capable of regulating its attitude by standing up and placed his hands over his head while his assistant walked on the lower wing, even as the aircraft was in flight. The plane continued to maintain level flight by moving its ailerons automatically. The positive feedback amplifier designed by Armstrong in the turn of the twentieth century was a core component in many of the instruments used in World War II. And the list goes on.

Over the last century, control has played an essential role in several branches of engineering including aerospace, transportation, manufacturing, robotics, power, communication, biology, and medicine. Feedback is an enabling technology in a variety of application areas and reinvented and patented several times in different contexts. The success of Wright brothers' demonstration in 1903 is largely due to the fact that their flight was controllable using movable control surfaces such as vertical fins and canards. This early example has been followed by a plethora of demonstrations and improvements in flight control technology, culminating in the very sophisticated and highly reliable automatic flight control systems we have on current commercial and military aircraft. Driven by the oppressing demands of urban traffic, a large program is underway in California to introduce an automatic highway system where a platoon of specially equipped cars, trucks and buses could travel together under computer control free of congestion, by exchanging a range of information about vehicle speed, location, and destination. Starting with the regulation of the processing of cement, paper, and paper, the process manufacturing operations have evolved to very high levels of performance of quality and low cost by using advanced information and actuation strategies. Examples of such process abound in the semi-conductor and pharmaceutical industries. The field of robotics is rife with examples of control where intelligent machines that are capable of highly dextrous and flexible tasks have been designed through a combination of measurement, computation, and actuation. The explosive growth of the internet has necessitated the introduction of a feedback control component in order to maintain quality of service, reliability, and reduced delay in the presence of uncertainty and variation in the network.

At the core of these multitudes of examples is the principle of feedback which consists of the triumvirate - measure, compare, and correct. You measure the quantity that is to be controlled, compare it with the desired value, and use the difference to correct the inputs into the system. This procedure is, however, not easy as it sounds. The reason is that it is not always clear how the system inputs should be changed. In a simple system as a thermostat, when the actual temperature is higher than the set temperature, it is clear that a positive error between the actual and the set temperature dictates an opposite action at the input side. That is, reduce the amount of heat that the furnace is putting out. In a complex system such as an airplane, such a simple algorithm need not necessarily be the right one. For example, if the feedback controller (which is the autopilot in this case)'s job is to keep the airplane flying level, if the airplane starts to pitch down, it does not immediately follow that this negative difference translates into a positive deflection of the elevators (which are flaps that contribute a pitching moment). The inertia of the airplane and subtle coupling effects between its lateral and longitudinal motion can cause an elevator motion to result in fairly complex motions of the aircraft. Much more careful analysis of the working of an airplane, the effects of motion of the elevator, and the devices that provide the measurements is called for. It is this kind of a careful analysis and synthesis that has gone into all of the aforementioned applications of feedback control in various branches of engineering.

(Dr. Annaswamy is a Principal Research Scientist in the Department of Mechanical Engineering at MIT. Her research interests are in the areas of Adaptive Control Theory and Applications; Control of Thermo-fluid Systems; Active Combustion Control; Dynamic Instability; Control using Neural Networks. In 2002, she was elected an IEEE Fellow. )

Bookmark and Share |

You may also access this article through our web-site http://www.lokvani.com/




Home | About Us | Contact Us | Copyrights Help