Decentralized Supervisory Control Strategy For Automated Manufacturing Systems Using Petri Nets

The Discrete Event System(DES)is an event-driven dynamic system that describes many phenomena and processes in nature and engineering.In a discrete event system,the state needs to be constrained to ensure that the system state meets expectations.For states that do not meet expectations,we call it inadmissible markings.The research object of this paper is Automated Manufacturing System(AMS).Due to its discrete event-driven characteristics,it is usually categorized as a discrete event system.In the liveness-enforcing supervisory control of an automated manufacturing system,the admissible markings is defined as a deadlocked state and an inevitable development to a deadlocked state.Because in industrial practice,automated manufacturing systems tend to be quite large-scale,there may be competition between different processing processes using limited resources and deadlocks.Research on the deadlock problem has always been a hot topic in automated manufacturing systems.The supervisory control studied in this paper refers to the monitoring of the system.By designing a reasonable and discrete supervisory control system,the system is far from the inadmissible markings.In the description of the system and modeling tools,Petri nets are chosen in this paper because Petri nets are random,concurrency,combined with their graphical expressions and mathematical analysis,can better describe such systems.In the Petri nets system,Generalized Mutual Exclusion Constraints(GMEC)is a kind of control specification that appears in the form of linear inequality.Under this specification,the state that satisfies the inequality is called legal state,and vice versa.In order to keep the system away from the inadmissible markings,this article selects the supervisory controller based on the control of the place,and directly controls the system according to the validity of the marking.The design of the monitor needs to consider many aspects,such as retaining as many of the system's licensing behaviors as possible,while minimizing the number of incoming monitors.In the traditional monitor implementation,a central monitor is required to receive the status and release the control strategy.However,this method will generate a large amount of communication traffic in a complex supervisory control system.At the same time,controlling the physical distribution of the place will also bring additional costs to the design of the actual system.In this regard,this paper introduces decentralized design ideas in the supervision controller design.Aiming at the activity control of Petri nets,combined with the knowledge of algebra and geometry,a more general constraint conversion algorithm is designed.The algorithm can convert the current constraint to any desired form of constraint.This paper also uses the constraint conversion algorithm to design a supervisory controller strategy that satisfies decentralized characteristics.The main results of this paper are as follows:(1)From the perspective of algebra and geometry,a constraint transformation algorithm for generalized mutual exclusion constraints is proposed.The algorithm avoids traversal and judgment of discrete states.According to the volume of the convex feasible region formed by the continuous inequalities group as the quantification standard of the system including the number of legal states,a combination of non-linear programming design can convert the specified constraints to arbitrary and low dimensions.The desired inequality group constraint algorithm.(2)Using the constraint transformation algorithm,do a detailed derivation and analysis of the constraint strategy optimization of the decentralized monitor.For a generalized mutual exclusion constraint with no more than three places,a complete example is given to interpret the process of constraint conversion of the decentralized supervisory controller.The combination of geometric analysis makes the result more intuitive.In addition,the control experiment is also designed to try to generalize the general directionality laws between the different generalized mutual exclusion constraints and the optimal expectation constraints.(3)Exploring the strategy transformation implementation of high-dimensional,complexconstrained decentralized supervisory controllers.The algorithm and definition of highdimensional constraint conversion are put forward.Including: A complete conversion idea of high-dimensional GMEC be given;Designs a feasible domain vertex algorithm for the high-dimensional constraint group;Gives the definition of the feasible domain volume of the high-dimensional constraint group;Presents the height of the n-dimensional single body based on the cutting combination.Dimensional Feasible Domain Volume Algorithm;Improvements to the cutting algorithm;simulations and verifications are given for all algorithms.The results also show that this paper is theoretically feasible for a complete decentralized supervisory controller implementation in the case of a complex multi-database.