A Simulation-Based Decision Support System forManufacturingEnterpriseAbstract: The simulation-based decision support system (SBDSS) is designed to achieve a high levelof performance, flexibility and adaptability, in response to meet the special needs of production andlogistics management during the economic system reform era in China. It consists two subsystems:the object library modeler (OLM) and the simulation engine and its manager (SEM). Using SBDSSthe decision makers can work out their optimal production choice under certain circumstances throughscenario simulations. And they can test a set of virtual organizations reflecting systems reform beforea real reorganization has been taken, as well as perform a virtual manufacturing process for a newproductdesign.Keywords:Modeling, Simulation, Object-oriented programming,Decision support systems,Integration.1.INTRODUCTIONInrecent years,the Chinese industrial decision makers(DMs),facing with thechallengesto handlethe complex system convertibility problem in the dramatic social economic system reform process.require a major shift towards maturity of necessary identification of feasible reform strategies andtheir assessment. Tools of the decision support that help DMs to learn about possible decision optionsand their predicated results are of urgent need. Our research efforts devoted in constructing thesimulation-based decision support system (SBDSS) are aimed to provide such a tool, instead ofgetting help from an experts team through meetings discussion face to face in the past. DMs obtainanswers about the real world problems by working interactively with the SBDSS, in which anappropriatelygenerated model ormodel system performs (simulates)as an experimental substituteforthe modeled system itself in predetermined circumstances for observations. Although such a modelcan never be perfect and can not encompass all aspects of the very complicated decision situation, it isoften a great help to the DM in the process of learning cause and consequences about novel aspects ofthe decision problem and thus gaining expertise in handling problems of a given class.This is acost-effectiveapproach
A Simulation-Based Decision Support System for Manufacturing Enterprise Abstract: The simulation-based decision support system (SBDSS) is designed to achieve a high level of performance, flexibility and adaptability, in response to meet the special needs of production and logistics management during the economic system reform era in China. It consists two subsystems: the object library modeler (OLM) and the simulation engine and its manager (SEM). Using SBDSS the decision makers can work out their optimal production choice under certain circumstances through scenario simulations. And they can test a set of virtual organizations reflecting systems reform before a real reorganization has been taken, as well as perform a virtual manufacturing process for a new product design. Keywords: Modeling, Simulation, Object-oriented programming, Decision support systems, Integration. 1. INTRODUCTION In recent years, the Chinese industrial decision makers (DMs), facing with the challenges to handle the complex system convertibility problem in the dramatic social economic system reform process, require a major shift towards maturity of necessary identification of feasible reform strategies and their assessment. Tools of the decision support that help DMs to learn about possible decision options and their predicated results are of urgent need. Our research efforts devoted in constructing the simulation-based decision support system (SBDSS) are aimed to provide such a tool, instead of getting help from an experts team through meetings discussion face to face in the past. DMs obtain answers about the real world problems by working interactively with the SBDSS, in which an appropriately generated model or model system performs (simulates) as an experimental substitute for the modeled system itself in predetermined circumstances for observations. Although such a model can never be perfect and can not encompass all aspects of the very complicated decision situation, it is often a great help to the DM in the process of learning cause and consequences about novel aspects of the decision problem and thus gaining expertise in handling problems of a given class. This is a cost-effective approach
In configuration of SBDSS, we are given conditions as follows:1) The end-user of the system is a large state enterprise, engaged in small-batch and one-offproduction,2)There existed a hybrid computer management system, consisting many separately locatedsubsystems, each offering a certain functional service, either for a department or for the wholeenterprise. And these subsystems have been realized incrementally over many years in the past. Manysubsystems are technologically obsolete, such as the old financial management subsystem. On thecounterpart, some newly addedparts areof themost up-to-date class,such as CAD,CAMsubsystems;3) A cost-effective approach must be considered to keep the entire enterprise running on originalmanagement system, simultaneously to develop an integrated new system, which fully utilizes theadvanced information technology, leading to the efficiency of decision makingA simulation-based approach is chosen, because the modeling and simulation are fundamentalapproaches to meet the high flexibility needs as given in above, and to develop applied systems ofobject-oriented paradigm, we have our platform and tool kit AIOOM.2.CONCEPTUALDESIGN2.1 Distributed Heterogeneous Computing and Networking PlatformsAs it has been described in the Section 1, the desired application system might involve multiple users(agents) participating in a computing environment consisting of heterogeneous and autonomousinformation resources, e.g. database management systems, might also be heterogeneous andautonomous.Theseapplications might be supported by distributed and heterogeneous computing andnetworking platforms (both hardware and software components) and have multiple administrative andaccess control authorities.2.2 AFrame-Work for Controlling Cooperative AgentsA software paradigm that can support such applications flexibly and reliably is a distributedcooperative task. In this paradigm, an agent supports a user, represents the user to the system, andhandles complex interactions with other cooperating agents and system resources.A critical issue in such a paradigm is controlling interactions among the cooperating agents to meetthe application objective,despite unpredictable user interventions and system failures
In configuration of SBDSS, we are given conditions as follows: 1) The end-user of the system is a large state enterprise, engaged in small-batch and one-off production; 2) There existed a hybrid computer management system, consisting many separately located subsystems, each offering a certain functional service, either for a department or for the whole enterprise. And these subsystems have been realized incrementally over many years in the past. Many subsystems are technologically obsolete, such as the old financial management subsystem. On the counterpart, some newly added parts are of the most up-to-date class, such as CAD, CAM subsystems; 3) A cost-effective approach must be considered to keep the entire enterprise running on original management system, simultaneously to develop an integrated new system, which fully utilizes the advanced information technology, leading to the efficiency of decision making. A simulation-based approach is chosen, because the modeling and simulation are fundamental approaches to meet the high flexibility needs as given in above, and to develop applied systems of object-oriented paradigm, we have our platform and tool kit AIOOM. 2. CONCEPTUAL DESIGN 2.1 Distributed Heterogeneous Computing and Networking Platforms As it has been described in the Section 1, the desired application system might involve multiple users (agents) participating in a computing environment consisting of heterogeneous and autonomous information resources, e.g. database management systems, might also be heterogeneous and autonomous. These applications might be supported by distributed and heterogeneous computing and networking platforms (both hardware and software components) and have multiple administrative and access control authorities. 2.2 A Frame-Work for Controlling Cooperative Agents A software paradigm that can support such applications flexibly and reliably is a distributed cooperative task. In this paradigm, an agent supports a user, represents the user to the system, and handles complex interactions with other cooperating agents and system resources. A critical issue in such a paradigm is controlling interactions among the cooperating agents to meet the application objective, despite unpredictable user interventions and system failures
User1User2Usern++1Agent]Agent2AgentnITB1ITB2ITBnSharedObjectSystemExistingRealManagementFig.1 Multi-Agent cooperation process simulationIn the Fig. I, each distributed user represented by a software agent, equips with an informationinteractivetransactionblock(ITB).TheITBs are agent specific,and usefixed-point criterion to definea stable state of the system with respect to an agent. Thus, the system in the Fig. I supports adaptivecooperation among agents and reliable operations on the shared objects stored in heterogeneous andautonomouscomponentssystems.Group processed and decision-making can be supported as a process and process structure or as atask and task structure.Theprocess and process structure support includes the provision of languagecommunication channels and means for process structuring and conducting. Tasks are traditionallysupported by quantitative models and decision modeling tools, and recently also by qualitativereasoning models. The difference between these two domains can be identified by two terms: groupprocess support systems (GPSS)and group decision support systems(GDSS).The researchoforganizational activities, presented in the paper, concerns problems within the domain of groupprocess and process-structure support, such as Group Ware, aimed to figure out an efficient way tofinalizeagroupdecisionmakingprocess,And a software agent can represent each member of thegroup.2.3 Computational Infrastructure and A10OMIt has been described that SBDSS is designed tocope withvarious problems,which is not identifiablein detail, but its existing heuristic solutions.We need to combine the deep knowledge of scientificprinciples in simulation with the superficial what-to-do knowledge in expert systems effectively toform a real-time learning simulation schema. We further expect that SBDSS can formulate afoundation for an integrated advisory system in CIMS environment with an autonomous goal-driventechnological character
Fig. l Multi-Agent cooperation process simulation In the Fig. l, each distributed user represented by a software agent, equips with an information interactive transaction block (ITB). The ITBs are agent specific, and use fixed-point criterion to define a stable state of the system with respect to an agent. Thus, the system in the Fig. l supports adaptive cooperation among agents and reliable operations on the shared objects stored in heterogeneous and autonomous components systems. Group processed and decision-making can be supported as a process and process structure or as a task and task structure. The process and process structure support includes the provision of language, communication channels and means for process structuring and conducting. Tasks are traditionally supported by quantitative models and decision modeling tools, and recently also by qualitative reasoning models. The difference between these two domains can be identified by two terms: group process support systems (GPSS) and group decision support systems (GDSS). The research of organizational activities, presented in the paper, concerns problems within the domain of group process and process-structure support, such as Group Ware, aimed to figure out an efficient way to finalize a group decision making process. And a software agent can represent each member of the group. 2.3 Computational Infrastructure and AIOOM It has been described that SBDSS is designed to cope with various problems, which is not identifiable in detail, but its existing heuristic solutions. We need to combine the deep knowledge of scientific principles in simulation with the superficial what-to-do knowledge in expert systems effectively to form a real-time learning simulation schema. We further expect that SBDSS can formulate a foundation for an integrated advisory system in CIMS environment with an autonomous goal-driven technological character
We are confident of success, because our previous development activities has constructed theplatform, artificial intelligent embedded object-oriented methodology (AIOOM), which is powerful tosupport the current development of SBDSS with its tools and enablers. AIOOM written in C++,naturally accommodates the heterogeneity and autonomy of large-scale, distributed systems: Theformer is because messages sent to distributed components depend only on the component's interfaces,not on their internals, And the latter is because components can change independently andtransparently, provided they maintain their interfaces.AIOOM's equipped with tools to produceknowledge-based systems, Inductive Inference Systerns, Autonomous Activity Systems, SociallyOrganizedSystemsaswellastheirblendingAIOOMhasbeenintroducedelsewhere3.SBDSSCONFIGURATION3.1 The Application System SBDSSThe simulation-based decision support system has been developed with our AI Embeddedobject-oriented platform.AIOOM SBDSs is designed to achieve a high level of performance,flexibility and adaptabilityIt consists two functional separate subsystems: the object-oriented modeler and the simulationengine and its manager. OLM enhances real world system abstraction by providing a natural mappingbetween system components being modeled and objects of the library. Object behavior can be definedusing state transition diagrams. Thus a set of objects associated with the various linkages among them,completely change responses to in-coming messages. While the SEM is the simulation kernelresponsible for managing the progress of simulation on one host machine.The host language of theSEM is C++, the object-oriented nature of C++ assists in realizing the correctness and softwarequality.3.2 The Object Library ModelerThe OLM of SBDSS provides a hierarchical basis within which a library of general-purposesimulation objects can be defined.These objects can be interconnected and specialized to construct anaccurate model of any target real world system. The object-hierarchy is founded on the observationthat most management processes can be viewed as composed of nodes that are interconnected in someway by communication links.At this level a node may be viewed as a state machine. State transitions are triggered by thearriving messages and this activity may produce one or more messages. The function of each node in a
We are confident of success, because our previous development activities has constructed the platform, artificial intelligent embedded object-oriented methodology (AIOOM), which is powerful to support the current development of SBDSS with its tools and enablers. AIOOM written in C++, naturally accommodates the heterogeneity and autonomy of large-scale, distributed systems: The former is because messages sent to distributed components depend only on the component's interfaces, not on their internals; And the latter is because components can change independently and transparently, provided they maintain their interfaces. AIOOM's equipped with tools to produce knowledge-based systems, Inductive Inference Systerns, Autonomous Activity Systems, Socially Organized Systems as well as their blending. AIOOM has been introduced elsewhere . 3. SBDSS CONFIGURATION 3.1 The Application System SBDSS The simulation-based decision support system has been developed with our AI Embedded object-oriented platform. AIOOM SBDSS is designed to achieve a high level of performance, flexibility and adaptability. It consists two functional separate subsystems: the object-oriented modeler and the simulation engine and its manager. OLM enhances real world system abstraction by providing a natural mapping between system components being modeled and objects of the library. 0bject behavior can be defined using state transition diagrams. Thus a set of objects associated with the various linkages among them, completely change responses to in-coming messages. While the SEM is the simulation kernel, responsible for managing the progress of simulation on one host machine. The host language of the SEM is C++, the object-oriented nature of C++ assists in realizing the correctness and software quality. 3.2 The Object Library Modeler The OLM of SBDSS provides a hierarchical basis within which a library of general-purpose simulation objects can be defined. These objects can be interconnected and specialized to construct an accurate model of any target real world system. The object-hierarchy is founded on the observation that most management processes can be viewed as composed of nodes that are interconnected in some way by communication links. At this level a node may be viewed as a state machine. State transitions are triggered by the arriving messages and this activity may produce one or more messages. The function of each node in a
simulation is defined by the structure of its state machine, ie., its nodes. Nodes are classified intothree classes according to their primary function in a system, namely Source, Forward and Sink. Thethree classes of nodes withtheir attributes aredescribed in the Table 1.WeuseOBFRAMEto showthethreeclassesTable 1 The three classes of OBFRAMES (Objects)NameForward (i)Sink (k)SlotsSource (i)VBlankVAInput ChsVAVBlankOutput ChsVArbitraryAOthersArbitraryVVDBlankDataStructureVVVMReceiveVMVSendVVVMGetBlankVMBlankConsumeVMVVGenerateVMVBlankRountVVVMOthersVVBlankD.F.DrivingVVBlankTraceD.F.VPossibleBlankD.F.StochasticVVBlankD.F.ExecutionPossiblepossibleBlankD.F.OthersNote: A denotes Attribute; D denotes Data; M denotes Message; and D.F. denotes Driving Factor.3.3SimulationEngineand Its ManagerSEM contains a scheduler object and a route object, which manages the simulation nodes and an eventlist. The scheduler knows the nodes relevant to the SEM and provides the engine with interfaces toinvoke the method in the node that matches the youngest outstanding event. Similarly, every futureevent is generated by the nodes through invoking a method in the scheduler. The router stores theinterconnection scheme, whichdefines how the nodes of a simulation are connected together.ThisallowstheSEMtohandletheproperdeliveryofmessagessentbyoneobjecttoanotherIn parallel simulations, there are a set of host machines, and on each machine there is a mappingfunction used to forward events and messages, to the SEM containing the objects they are addressedto. At the first stage of SBDSS development, only one host machine is used
simulation is defined by the structure of its state machine, i.e., its nodes. Nodes are classified into three classes according to their primary function in a system, namely Source, Forward and Sink. The three classes of nodes with their attributes are described in the Table 1. We use OBFRAME to show the three classes. Table 1 The three classes of OBFRAMES (Objects) Note: A denotes Attribute; D denotes Data; M denotes Message; and D.F. denotes Driving Factor. 3.3 Simulation Engine and Its Manager SEM contains a scheduler object and a route object, which manages the simulation nodes and an event list. The scheduler knows the nodes relevant to the SEM and provides the engine with interfaces to invoke the method in the node that matches the youngest outstanding event. Similarly, every future event is generated by the nodes through invoking a method in the scheduler. The router stores the interconnection scheme, which defines how the nodes of a simulation are connected together. This allows the SEM to handle the proper delivery of messages sent by one object to another. In parallel simulations, there are a set of host machines, and on each machine there is a mapping function used to forward events and messages, to the SEM containing the objects they are addressed to. At the first stage of SBDSS development, only one host machine is used