What is an Adaptive Controller?

An adaptive controller is one that continually evaluates its own performance in regulating industrial processes to achieve specific goals, and adjusts itself to better achieve these goals. An example would be a commercial food cooking process where maintaining a precise cooking temperature is essential to consistently producing a high-quality product. Variability in the product being processed, or in the processing plant itself, frustrate this consistency. Such variability can either be predicted or unpredicted. Adaptive control systems have been developed to deal with either or both types.

There are two groupings of setpoint control, where the setpoint, or the desired operating value, and process variable, the actual value, are compared, yielding an output decision. The first is traditional on-off control such as that used in home thermostats, where the heat is either on or off. The second type is throttling control, the same type used to control the speed of an automobile by proportionately applying more or less power through the accelerator. Proportional-Integral-Derivative (PID) or three-mode control is the most common algorithm used by process controllers to execute throttling control action. Briefly, a PID controller provides immediate, long term, and anticipatory intervention to manipulate a control actuator, such as an engine throttle, to make the process variable equal the desired setpoint, and keep it there.

While the first adaptive controller strategies were developed for aviation autopilot systems and early space vehicles, the most prolific uses of adaptive control have been in the industrial process control realm and in transportation. With the widespread use of microcomputers, adaptive control has also found its way into everyday consumer-oriented systems. Automotive cruise control is one example of this.

Cruise controls augment PID throttling with adaptive control tuning adjustments. If a driver sets the cruise control for 60 mph (96.5 kph), the system continually senses the actual speed, compares it with the 60 mph (96.5 kph) setpoint, and modulates the accelerator to maintain the speed. This system is designed to work consistently on level ground.

If the same automobile were towing a trailer, going up a steep incline, or into a stiff headwind, the PID model would require more aggressive tuning to allow it to still provide the same relative responsiveness to changing conditions while keeping the actual cruise speed constant. The adaptive section of the controller would sense the change in responsiveness and adjust the PID tuning aggressively to keep speed consistent. This is unpredicted, since the adjustment is purely based on system responsiveness.

A predicted variability adaptive controller is preprogrammed to change its PID tuning with a known variable. For example, this type of adaptive controller could be used in a heating control on an industrial process that normally and predictably goes exothermic at a point during its run. Adaptive controllers help industrial, transportation, and consumer systems to cope with variability in all its forms.