Learning control is one of the earliest research areas of intelligent control. The concepts of learning and learning control were introduced about a quarter of century ago [17,18,64,68]. During the past ten years, learning control with applications to dynamic systems, such as robot manipulation and flight guidance, has become an increasingly important research topic. A set of learning control schemes and better control processes have also been investigated and proposed [4,5,9,24,25,36,65].In this chapter, some basic concepts including definition, reasons and history of learning control followed by machine learning and its application in control will be first introduced. Then the schemes of learning control, which involve pattern recognition-based learning control, iterative learning control, repetitive learning control, NN-based learning control and rule-based learning control, etc., will be discussed. After that, some issues for learning control, which deal with design principles, control rules, stability analysis and convergence analysis of learning control, will be analyzed. Finally, an example of learning control systems, the self-learning fuzzy neurocontrol for arc welding process, will be presented. Intelligent Control: Principles, Techniques and Applications Downloaded from www.worldscientific.com by UNIVERSITY OF BIRMINGHAM on 06/25/16. For personal use only.Learning Control System 303 material that can be influenced by the natural selection. If the change comes out in a type of behavior and provided this behavior is harmless, then this change would be continued generation after generation. This change of behavioral type from one generation to its next generation is called racial learning or system growth learning, and the behavior change or behavior learning that happens in a specific individual is called individual growth learning. C. Shannon defined learning with more limitation in 1953. Definition 8.2 (Shannon). Suppose: (1) an organism or a machine is in a kind of environment; (2) There exists a "successful" measure or "adaptive" measure to the environment; (3) This measure is a relative local measure with respect to time, i.e., people can test this successful measure in a time shorter than the life time of the organism. If this local successful measure is improved with the time for the environment considered, then it is said that, for the selected successful measure, the organism or machine is in learning for adapting this kind of environment.Osgood presented learning from the point of view of psychology in 1953. Definition 8.3. In repetitive circumstances with same features, the organism individual continuously changes its own behaviors and enhances its selection in competitive reaction depending upon its own adaptability. This kind of selective variation formed by the individual experience is the so-called learning.Tsypkin gave learning and self-learning a more general definition as follows [68]. Definition 8.4. Learning can be understood as a process that repeats various input signals, corrects s...