MODELING OF INTELLECTUAL CONTROL SYSTEMS WITH APPLICATION OF MODIFIED FUZZY COLORED PETRI NETS

Abstract

A modification of the Petri nets is proposed  a Fuzzy Colored Petri Net (FCPN), leading to the integration of Fuzzy Petri Nets (FPN) and Colored Petri Nets (CPN). Separately, the shortcomings of FPN and CPN and the advantages the developed FCPN for modeling intelligent control systems are identified and justified. In the developed modified FCPN the membership functions of the terms of a linguistic variable are applied to markers of the CPN as color and fuzzy existence conditions are assigned to arcs depending on the values of the linguistic variable. As a result, FCPN with enhanced capabilities was obtained, which eliminates the shortcomings of CPN and FPN. The FCPN and a simulation approach are defined, which includes a reasoning algorithm, as well as a detailed procedure for modeling and analysis of nonlinear discrete objects. The structure is organized in the CPN Tools system with the synchronization of the CPN ML (Markup Language) language with the MatLab package. The choice of membership function and fuzzification of term values is performed in the Fuzzy Toolbox application of the MatLab system. The proposed approach is illustrated by simple example, including the control of water pumps, to maintain the required water level in the pumping well. A model is developed for the automation of adaptive fuzzy control of water pumps based on modified FCPN. Based on the criteria for the operation of water pumps according to the water levels in the pumping well, many positions and transitions of the FCPN are formed. Describing the necessary behavior of the system by the relations between the positions and transitions of the FCPN using the logic “If ... Then ...” an adaptation algorithm is developed. Based on the algorithm, the matrices of input and output incidents are determined. The graph model of the FCPN is developed. Visualization of the model is implemented in the CPN Tools system. The computer simulations and net analysis experiments demonstrate the convenience of the developed approach when modeling the intelligent control of dynamic systems.