A new model reference adaptive control approach
During normal operation, a controlled dynamic process occasionally experiences a change in its physical characteristics due to environmental disturbances. Temperature, pressure or variations in the structure of the process are some common examples of such disturbances. This situation may produce undesirable response and, in severe cases, the stability of the process may be compromised. In order to achieve a desired response in the presence of disturbances, researchers have developed a new class of control systems called adaptive control systems. Many adaptive control systems involve two central process control stages: 1) estimating the time-varying process parameters and 2) modifying the controller's parameters so as to achieve a desired system response. Such schemes include the Model Reference Adaptive Control (MRAC) systems where a model of the controlled process is used to specify the desired performance of the actual process. This work will introduce two new MRAC design approaches to control linear, time-variant systems. In the first approach, the actual controlled process and its model are described by two difference equations. The individual terms of the actual process are equated to the corresponding terms of the model. The unknown process parameters are recursively estimated using the least squares method. From these equations, one can solve for the appropriate controller's parameters that will force the actual process to be consistant with its model. In the second scheme, the controller is designed using the output-to-input relationship. The second scheme updates the controller's parameter values whenever the process parameters change.