Model reference adaptive control MRAC

If the process is not known well, we need to evaluate the objective function on-line (while the process is operating) using the values of the controlled output. Then the adaptation mechanism will change the controller parameters in such a way as to optimize (maximize or minimize) the value of the objective function (criterion). In the following two examples we examine the logic of two special self-adaptive control systems model reference adaptive control (MRAC) and self-tuning regulators (STRs). 

Example 22.3 Model-Reference Adaptive Control (MRAC)... 

Neural networks have also been used in model reference adaptive control (MRAC) structures (Figure 12.4) [Naranedra and Parthasarathy, 1990 Narendra, 1992]. This approach builds upon established techniques for adaptive linear control and incorporates neural networks to address the problem of controlling nonlinear systems. The MRAC approach is directed at adapting the controlled system such... 

In this study, adaptive control algorithms have been utilized for designing active controllers for smart structure test articles. Adaptive control schemes require only a limited a priori knowledge about the system in order to be controlled. The availability of limited control force and inherent deadband and saturation effects of shape memory actuators are incorporated in the selection of the reference model. The vibration suppression properties of smart structures were successfully demonstrated by implementing the conventional model reference adaptive controllers on the smart structure test articles. The controller parameters converged to steady state values within 8 s for both direct and indirect MRACs. 

Various neural network-based adaptive control techniques were discussed in this study. A major problem in implementing neural network-based MRACs is the translation of the output error between the plant and the reference model so as to train the neural controller. A technique called iterative inversion, which inverts the neural identification model of the plant for calculating neural controller gains, has been used. Due to the real-time computer hardware limitations, the performance of neural network-based adaptive control systems is verified using simulation studies only. These results show that neural-network based MRACs can be designed and implemented on smart structures. 

MRACs are composed of a reference model that specifies the desired performance an adjustable controller whose performance should be as close as possible to that of the reference model and an adaptation mechanism. This adaptation mechanism processes the error between the reference model and the real process to modify the parameter of the adjustable controller accordingly. 

The first controller implemented on the structure was the direct MRAC shown in Fig. 4.5. This gives a basis for comparison between direct and indirect control. Figure 4.6a shows a plot of the open-loop response envelope, the desired response envelope, and the closed-loop response achieved. As can be seen, the closed-loop system adapts to the reference-model response until the deadband is reached (after approximately 11s), at which point adaptation is turned off. The deadband is inherent to the Nitinol wire actuators. 

---