Cutting the Apron Strings
Emerging Intelligent Process Control Systems
General Systems Theory (GST) is concerned with developing a systematic framework of inter-relationships among the components of the natural and human-made world. In 1956, economist Kenneth E. Boulding envisioned GST as a hierarchy of increasingly complex levels of behavior studied by the various fields of scholarly inquiry. Boulding’s first three levels of complexity termed “frameworks,” “clockworks,” and “thermostat,” are the purview of physics, chemistry; and information science, respectively. Frameworks examines static elementary particles and physical laws; clockworks, the simple dynamic systems of molecules and mechanics; and thermostat encompasses the transmission and interpretation of information. The next levels of “cell,” “plant,” and “animal” increase complexity to include ability to sense and react to environment, genetic replication, and self-awareness. The remaining levels of “human” and “social organization” incorporate self-consciousness and collective association for a common purpose.
A typical data acquisition (DAQ) system operates at the clockworks level. Although information and knowledge representing measured values flow through the system, the information is presented as an output to a high-level human operator who takes action based on the acquired information. This concept of passing the task of control to a human is known as open loop control. For DAQ to mature to the thermostat level, the DAQ system alone must act on the information measured.
A quick trip to howstuffworks.com illustrates the operational principles of an analog household thermostat. A quantity of liquid mercury encapsulated in a glass ampoule can complete or interrupt an electronic circuit depending on the ampoule’s pitch. This mercury switch is affixed to the end of a thermometer coil comprised of an alloy having a large coefficient of thermal expansion. Moving a lever attached to the coil energizes the heating system by tilting the mercury switch such that the circuit is completed and “on.” As the temperature of the room increases, the thermometer coil elongates and eventually tilts the mercury switch back to the “off” position. This temperature feedback mechanism infuses the thermostat with the ability to act autonomously on the temperature it measures. In this fashion, the simple DAQ system evolves to a closed-loop acquisition and control system (ACS).
The analog house thermostat functions admirably. However, it is only capable of turning the heating system full on or full off. It is akin to living in a world with traffic lights operating only green and red and automobiles having only a clutch and a parking break. The acceleration and break pedals of actual automobiles permit the driver to gradually respond to a yellow light without the jerk associated with abrupt acceleration changes. With proper design, the thermostat can react smoothly based on information acquired from its environment at the same hierarchal level of a living cell. If only slightly away from the set-point temperature, the advanced thermostat can signal a heating rate proportional to the amount of heat required. The definition of “slightly away from the set-point” must be checked periodically by examining the actual temperature over time through integration so the controller does not drift away from the set point. To avoid rapid temperature cycling above and below the set point, the rate of heating/cooling must be examined through differentiation of the temperature as a function of time. This combination of proportional, integral, and derivative functions is known as PID control.
The operation of typical PID controls must be monitored by a trained technician and adjusted when necessary thus returning the process to an open-loop system. To obtain autonomous closed-loop process control, so- called “intelligent controllers” are created by feeding additional sensor data traditionally utilized by the human technician into the controller directly. Artificially intelligent control programs employing knowledge bases, expert systems, neural networks, fuzzy logic, and generic algorithms utilize the increased number of sensor streams to adjust PID parameters automatically. Equipped with the appropriate number of sensor streams, they hold the promise of exhibiting animal-level self-awareness with the ability to self-optimize and adapt to the environment.
To spur the development of these futuristic control systems, the National Institute of Standards and Technology (NIST) maintains its Intelligent Systems Division (ISD) providing the measurement and standards infrastructure needed for application of intelligent systems. As our factories and machines climb Boulding’s hierarchy of intelligence, let us not forget that demands for collective bargaining may be just over the horizon.
This material originally appeared as a Contributed Editorial in Scientific Computing and Instrumentation 19:11 October 2002, pg. 18.
William L. Weaver is an Associate Professor in the Department of Integrated Science, Business, and Technology at La Salle University in Philadelphia, PA USA. He holds a B.S. Degree with Double Majors in Chemistry and Physics and earned his Ph.D. in Analytical Chemistry with expertise in Ultrafast LASER Spectroscopy. He teaches, writes, and speaks on the application of Systems Thinking to the development of New Products and Innovation.
