Development of simulation models for the study of systems with probabilistic interactions of elements

Abstract

One of the topical areas of theoretical and experimental research is the analysis, modeling and synthesis of technical systems with probabilistic interactions of elements. The work is devoted to the development of simulation models for the study of one class of such systems - network management systems. As data transmission channels, such systems use industrial or public digital computer networks. The use of computer networks in control systems leads to new problems: random time delay in the transmission process, the likely loss of the data packet, the possibility of asynchronous operation of the elements of the system. Not taking these factors into account can lead to a loss of sustainability of the management system. To study these problems, it is proposed to use simulation models of appropriate control systems. In the work, simulation models for a digital control system without delay in the transmission channel, a network management system with a network communication channel between the sensor and the controller have been developed. In this case, cases are considered where the transmission time on the network channel does not exceed the quantization clock time and when the transmission time over the network channel may exceed the quantization clock time. For the cases considered, appropriate simulation models have been developed. When simulating a network management system in which the transmission time of a data packet over a network channel could exceed the time of a quantization clock, it was assumed that the sensor and the controller have an infinitely large buffer. For simplicity of modeling, it was assumed that if a data packet from a digital sensor is not transmitted in one clock cycle to the controller, it is placed in the sensor buffer. With the possible transfer of data over the channel, all packets from the sensor buffer are transferred to the controller buffer. It was assumed that the controller processes incoming data from the buffer sequentially – one packet per quantization cycle. To simulate such situations it was suggested to use the corresponding developed random processes. Simulation of the functioning of digital and network control systems was carried out in the Simulink environment of the Matlab system.