Linear quadratic control with feedback-linearized models

The paper considers two questions. Firstly, to adjudicate whether the feedback linearization is actually better for control than the plain Jacobian linearization. This issue is settled by using linear quadratic control strategy for above-mentioned linear models. The second question results from the efforts of joining feedback linearization and linear quadratic control: how should the weights of the cost function be calculated for the model obtained from feedback linearization, when the variables are not direct physical values? This paper shows a method of adjusting those weights, which is based on the weights of the quadratic cost functions for Jacobian-linearized model. The two questions are therefore interrelated because the method of adjusting weights gives us similar control strategies for the two different linear models. This is used for the comparison of the two linearization methods. The obtained results show that combining linear quadratic control with feedback linearization yields the performance index usually better, or at least not worse, than when combining it with Jacobian linearization. The former combination can be used successfully for many nonlinear systems.