ntc info page

Natural Tracking

NATURAL TRACKING CONTROL: THEORY, ALGORITHMS AND CONTROLLERS

U.S. Patent 5,379,210, January 3, 1995

This invention pertains to control systems, particularly to controllers which can operate with relatively minimal knowledge of the structure or function of the system being controlled.

The general objective of a control system is to minimize the effects of external disturbances and internal variations while satisfying some performance criterion. The difference between the desired output and the actual output of the system is called the "tracking error." With perfect tracking, the tracking error is always equal to zero, i.e. the output of the system exactly follows or "tracks" the desired output, at all times, regardless of the desired output and regardless of external disturbances.

We have developed apparatus, methods, and conditions for the effective engineering implementation of natural tracking control for both linear and nonlinear systems. Natural tracking control has been achieved for analog (continuous-time), digital (discrete-time), or hybrid (continuous-time measurements combined with discrete-time processing) implementations. These controllers may be synthesized without using information about the system's state and internal dynamics, or about the effects of disturbances, hence the name "natural" tracking control.

The methods of this invention force a system to exhibit high-quality tracking of a desired output while the system is subject to disturbances whose values and effects on the system may be completely unknown. The high-quality tracking property is expressed in terms of the output error its derivatives, and/or its integral -- independently of the characteristics of the system.

The characteristics of natural tracking control include:

	forces the output tracking error to zero in finite time
	incorporates prespecified tracking qualities for the convergence of the output tracking error; i.e. exponential, second order, stablewise with finite settling time, etc.

	may be applied to linear and nonlinear systems
	may be applied to multiple-input, multiple-output systems of any size
	uses a minimum amount of information about the internal dynamics of the system
	may be implemented with analog and digital controllers or a combination of both
	does not require the knowledge of the external disturbances
	does not require the a priori knowledge of the desired outputs (setpoints and trajectories)
	has been applied to aircraft, chemical, robotic, HVAC, etc. systems
	may be expanded to include neural networks to learn system characteristic.