SOFTWARE DEVELOPMENT FOR MULTIVARIABLE SELF TUNING CONTROLLER

Abstract

The problem of controlling a system with constant, but unknown parameters is considered. The control algorithm is presented which combines the minimum variance strategy with -a. Dahlin's control algorithm... The analysis is . first considered -for. single- input ':single-output case. . The minimum variance --strategy and Dahlin's - control- algorithm are first discussed both the algorithms are combined to obtained` the algorithm for single input-single output case. The algorithm is then extended to multivariable system. This type of cont ol..algorithm. ' overcomes the shortcoming. of original version of minimum variance control. It tracks the set point . changes with the desired speed of response:, penalizes the exessive control action. Since in this a)gorithm process zeroes on negative real axis are not cancelled- and so can be applied to non-minimum phase, systems. A very important feature is ° provided in that ` =unlike the other multivariable versions of the minimum variance self-tuners which have been developed for problems where Bo is non-singular, the ;structure , of reported. control algorithm allows for different time ' delays as at permits the singularity of "B0. Software for this, work has been 'developed in `C' language on TATA ELXI in. unix environment