Patent US 3,934,124

The word self-adaptive has been used to describe the capability of human beings and machines to adjust their characteristics in accordance with a changing environment. Recently a more advanced form of self-adaptive system has been theorized and implemented called "self-organizing" controller because it performs even more like a human being. Its control action has a "dual" character. It studies the response from the system being controlled and then brings about its desired state by adjusting its own characteristics. Its performance can be somewhat likened to a blind man's actions as he attempts to follow a desired path. The controller can perform 250 or more solutions per second to make the sensed output response agree with the command signal with the desired limits. Each controller consists of a performance assessment part and a probability state variable (PSV) part. On-line biasing of the probability states of the PSV part results in the continual restructuring of the statistical control law. Information about the system's behavior is obtained from the error signal and its derivatives, while information about actions of the PSV part is obtained by monitoring its output. At the heart of the PSV part is a gaussian noise generator, which enables the system to perform cautiously, while simultaneously eliminating undesirable noise. Such a system can be described as having the ability to learn very rapidly. The improvements in this concept and technique involve a reduction in the number of parts, the replacement of binary multipliers, with analog type multipliers and the introduction of an error squarer device to enable plants having lower frequency responses to be controlled. This control system has the inherent capability to handle situations for which rigidly programmed systems are not particularly suited. BACKGROUND OF THE INVENTION