Investigation of methods for identifying dynamic characteristics of control objects

Abstract

The paper proposes an approach for constructing a mathematical model of a control object, where the transient response of the object is used as the initial data. The advantages of this approach include the use of objective data generated by the control object itself, relative simplicity of implementation, and the ability to obtain an adequate and accurate model due to the utilization of comprehensive dynamic characteristics of the system. In addition, the approach improves identification reliability under conditions of limited experimental data.

The study is aimed at optimizing the considered technological process.

The results show that limiting the order of the differential equation (or transfer function) to the second order significantly simplifies the development of the mathematical model. In some cases, high-order models can be reduced to lower-order models, including first- or second-order ones, without a significant loss of accuracy in describing their characteristics. This can be explained by the following factors: analysis and synthesis tasks are much easier for low-order models; computational accuracy is inversely proportional to the model order; first- and second-order models contain sufficient information for system analysis and synthesis; moreover, increasing the model order does not always improve its accuracy. It is also shown that reducing the model order decreases computational costs and enhances numerical stability.

The obtained identification error is within acceptable limits for this type of problem.

The paper addresses the following issues: selection of the number of points on the step response curve of the control object; choice of an appropriate identification algorithm; determination of point distribution along the curve; and analysis of the influence of the number and placement of points on approximation accuracy. The obtained results can be applied in the design and tuning of automatic control systems.