a self-adaptive control system in which adaptation to randomly varying conditions is provided by an automatic change in the alignment parameters or by an automatic search for the optimum alignment. In any nonself-adaptative automatic control system there are parameters that govern the stability and quality of the control processes and that can be changed by aligning the system. The control processes can deteriorate or even become unstable if these parameters remain unchanged when the operating conditions—the disturbing actions and the characteristics of the controlled system—change substantially. The manual alignment of a system is often difficult and sometimes impossible to carry out. The use of a self-aligning system in such cases is technically and economically advisable and may even be the only way of providing reliable control.
Self-aligning systems are divided into searching and non-searching types. In searching self-aligning systems, the required quality of control is achieved by an automatic search for the optimal (in some sense) alignment (seeSEARCHING SYSTEM). The quality of the alignment is characterized by a generalized index whose complicated relationship with the initial alignment parameters is generally not completely stable and usually is insufficiently known. The index is measured directly or is calculated from measured values of the initial parameters. Search, or test, changes are made in the alignment parameters of the self-aligning system. Analysis of the changes in the adjustment quality index that are caused by the search actions shows whether the alignment obtained is optimal, that is, whether it corresponds to an extremum (a maximum or minimum) of the quality index. When deviations from the extremum occur, changes are made in the alignment until the optimal adjustment is approached. Searching self-aligning systems are capable of operating under widely varying external conditions.
Nonsearching self-aligning systems have a certain advantage over searching systems because the search for an optimal state consumes a substantial amount of time; in other words, the self-aligning time of searching systems has a lower limit. Non-searching self-aligning systems make use of a monitored index of control quality, for example, the value of the derivative of a controlled parameter with respect to time. This index is maintained within specified limits by automatic alignment of the parameters. Depending on the type of index, such self-aligning systems are classified into various types—for example, systems that monitor transient processes, systems that monitor frequency responses, and systems having a reference model. These types are closed-loop nonsearching self-aligning systems; they have a closed self-alignment loop in which the alignment parameters are automatically changed when the quality index goes beyond the permissible limits. Some closed-loop nonsearching self-aligning systems are similar to ordinary nonlinear automatic control systems having a reduced sensitivity to the characteristics of the controlled system; relay systems and variable structure systems are examples of such ordinary nonlinear control systems. Besides closed-loop systems, there are also used open-loop self-aligning systems, which are known as parameter-compensation systems. Such systems monitor the actions producing a change in the properties of the controlled system, and the control system’s alignment parameters are altered in accordance with a program calculated beforehand; the self-alignment loop in this case is open. Such a self-alignment can be almost instantaneous, but its realization requires that the environment be controlled and that the laws governing the action of the environment on the controlled system be accurately known.
Self-alignment can be realized both by special apparatus in the form of self-alignment units or self-aligning optimizing controllers and by adaptive algorithms from central control computers. The incorporation of self-alignment, or adaptation, properties in control algorithms considerably increases the possibility of controlling a variety of processes. The use of a self-aligning system permits controlled systems to approach their optimal operating conditions, facilitates the standardization of control systems, reduces the time required for testing and debugging, lowers the technological requirements for the manufacture of some control units, and frees maintenance personnel from time-consuming alignment operations. The practical application of self-alignment systems and adaptive algorithms is one of the characteristic features of technical progress in the field of control.
REFERENCESFel’dbaum, A. A. Vychislitel’nye ustroistva ν avtomaticheskikh sistemakh. Moscow, 1959.
Krasovckii, A. A. Dinamika nepreryvnykh samonastraivaiushchikhsia sistem. Moscow, 1963.
Tsypkin, la. Z. Adaptatsiia i obuchenie ν avtomaticheskikh sistemakh. Moscow, 1968.
Kozlov, Iu. M., and R. M. Iusupov. Bespoiskovye samonastraivaiushchiesia sistemy. Moscow, 1969.
A. A. KRASOVSKII