Many objects automatic control unsteady. This is manifested in the change of their parameters. Therefore,
Problem statementIn industry, the object of regulation (OR) is most often is a technological apparatus. To the strong influence of parametric perturbations. Unfortunately, PI and PID controllers compensate the changes of OR parameters insufficiently and that is why the adjustment of these controllers is required. As for Multi-Loop Automatic Control Systems (ACS), the quantity of which increases, adjustment of controllers is a difficult engineering challenge. If we add that the parameters of most OR change fast to this, ensuring the controllers' parameters compliance with OR parameters becomes a nearly impossible task (moreover if dealing with multidimensional OR with a great number of interconnections) with all consequences resulting from that fact to the extent of control quality decrease.
Problem solutionOne of the solutions for such a situation is an increase of ACS durability. That term requires specification. Within the scope of this article it can be construed as follows: ACS durability is its ability to provide execution of the assigned control tasks (compliance with regulations) during a long time with the OR being affected with parametrical and coordinate disturbances without the necessity of conduction of adjustment works at controllers.It is possible to decrease the acuteness of the specified issue by using hybrid controllers. That term requires specification. The hybrid controller shall be construed as that one the structure of which applies fuzzy logics algorithms and (or) artificial neural networks together with common dynamic integrating units and, if necessary, differentiators.There are lots of variants of construction of hybrid controllers. What index should be used for the most objective comparison of ACS with various controllers including common control algorithms? It appears that ACS durability can be such an index.The quantitative assessment of ACS durability can be performed by comparing the Normal Operating Limits (NOL) of various controllers. What is NOL of a controller?We shall proceed from the fact that upon development of a controller's structure one needs to use the advantages provided by artificial neural networks, namely the expressed cooperativity, at the same time avoiding the difficultness of software implementation within the industrial controller. It is such -balance of interests‖ that is native to the controller structure described below.The figure 1 shows the structure diagram of stabilizing ACS simulating model the hybrid controller of which uses a very simple artificial neural network. The model is realized in the environment of MATLAB with use of software packages of Simulink.
Fig. 1 -Model of automatic control system in SimulinkThe parameters of control interaction indexes of manually taught neural network used for determination of NOL of the controller shall have the numerical values in vector mode: k1 = -0.2; k2 = 2.8; k3 = 0.5; k4 = -2. It can be express...