A multi-resolution model description method and system for complex product simulation

Abstract

The present invention proposes a multi-resolution model description method and system for complex product simulation, which belongs to the field of computer modeling and simulation technology. The method comprises: constructing basic models corresponding to each entity in the simulation object at each resolution, and the description of the basic model is constructed based on the DEVS atomic model and the Agent metamodel; according to the basic model, constructing a multi-resolution atomic model of the entity, and the atomic model contains the description of the basic model of different resolutions for any entity in the simulation object; according to the atomic model, constructing a multi-entity network model, and the network model is a multi-entity network formed by coupling the atomic models of each entity; according to the network model, constructing a coupling model for the simulation object in the current simulation, and the coupling model contains the network models corresponding to the simulation object under different network structures. The present invention can realize multi-level model description and model construction for complex products or systems, and meet the modeling needs of simulation scenes at different abstract levels.

Description

Technical Field

[0001] The invention belongs to the technical field of computer modeling and simulation, and particularly proposes a multi-resolution model description method and system for complex product simulation. Background Art

[0002] Complex products refer to high-cost, large-scale, high-tech, engineering-intensive products, subsystems, systems or facilities, covering the fields of automobiles, ships, aerospace vehicles, robots, etc., with complex composition, complex functions and complex behaviors. In the product design process, it is necessary to complete a variety of simulation tasks such as solution verification, performance inspection, refinement iteration, and optimization design. Multi-resolution modeling provides different resolution model switching and consistency control mechanisms, which can well cope with its multi-field, multi-level, multi-stage, and multi-type modeling needs, and is of great significance to the design and development of complex products.

[0003] Multi-resolution modeling does not involve the construction of specific models, but provides a modeling framework for simulation tasks at different abstract levels. Multi-resolution modeling includes modeling of multiple levels such as systems, subsystems, components, and sub-components, as well as modeling of different models of the same modeling object, and focuses on resolution model switching and control, and model consistency maintenance. Existing research on multi-resolution model description based on Discrete Event System Specification (DEVS) has universal significance for simulation analysis of complex systems.

[0004] Discrete Event System Specification (DEVS) is a general formal modeling description specification for modeling and simulation of complex dynamic systems. As a modular model description method, it realizes interoperability and system response by defining port communication mechanisms and abstract functions based on the description of atomic models and coupled models. Its advantage lies in its support for system composition structure, communication mechanism, and time concept, but it lacks corresponding description of intelligent behavior within the module and collaboration between modules. The entity behavior is relatively simple, the function is limited, and there is a lack of learning and cognitive reasoning related mechanisms.

[0005] Agent-based modeling (ABM) supports the modeling of subject behavior, where Agent, as the basic component unit, is the model generated by abstract modeling of active individuals in complex systems. By modeling individual behavior, control, learning, and communication and collaboration between individuals, it is possible to study macroscopic emergent phenomena from microscopic behavior and support the study of group collaboration. Agent-based modeling and simulation methods are mainly used to model models with subject behavior characteristics, but lack the ability to model simulation objects without subject behavior, and this method has not yet been applied to multi-resolution modeling. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a multi-resolution model description method and system for complex product simulation. The present invention combines DEVS with an agent-based modeling method to achieve multi-level model description and model construction for complex products or complex systems, and can especially model subjects with characteristics such as perception behavior, communication collaboration, and learning evolution in complex systems, and can meet the modeling needs of simulation scenarios at different abstraction levels.

[0007] The embodiment of the present invention provides a multi-resolution model description method for complex product simulation, including:

[0008] Constructing basic models corresponding to each entity in the simulation object at each resolution, wherein the description of the basic model is constructed based on the DEVS atomic model and the Agent meta-model;

[0009] According to the basic model, construct a multi-resolution atomic model of the entity, wherein the atomic model includes descriptions of basic models of different resolutions for any entity in the simulation object;

[0010] According to the atomic model, a multi-entity network model is constructed, wherein the network model is a multi-entity network formed by coupling the atomic models of each entity;

[0011] According to the network model, a coupling model for the simulation object in the current simulation is constructed, and the coupling model includes network models corresponding to the simulation object in different network structures.

[0012] In a specific embodiment of the present invention, the method further comprises:

[0013] The coupled model is converted into an atomic model for system simulation of the simulation object as a multi-resolution entity.

[0014] In a specific embodiment of the present invention, the basic model is described as follows:

[0015] Basic=<X,Y,S,s0,δ int ,δ con ,δ ext ,λ,ta,C,AG,CG>

[0016] The input set X is the set of input events and input messages, X = X e ∪X m , X e is the input event set, X m is the input message set; similarly, Y=Y e ∪Y m , Y e is the output event set, Y m is the output message set; S = S s ∪S d is the basic model state feature set, identified by static attributes S s and the state variable S d composition; s0 is the initial state; δ int :S d →S d is the internal state transfer function; δ con :S d ×X e b →S d is the concurrent conflict handling function, and the superscript b represents the concurrent event δ ext :Q×X e →S d is the external state transfer function, where Q = {(s, t) | s∈S d ,0≤t≤ta(s)} is the total state set, t is the time the entity stays in state s; λ:S→Y b Output function for the system; is the time advancement function, which generates the time series T; C is the Agent rule set; CG: (T×X×S→C)∪(T×S→C) is the control law generation operator; AG: (T×C→S)∪(T×C→Y)∪(T×C→II) is the action generation operator used to represent the Agent's external behavior generation method, and II represents internal information feedback.

[0017] In a specific embodiment of the present invention, the structure of the atomic model is:

[0018]

[0019] Among them, X Res , Y Res is the resolution port, used to send and receive resolution change information between different entities; R is the Agent identification set of multiple resolutions of the entity; ψ={2 R-φ} is the resolution mode set; is the initial resolution mode, The resolution is The basic model when M r is the basic model set of the entity when the resolution is r; is the model consistency mapping function; Resolution mode The association between different modules at the same resolution;

[0020] PSC = PS ∪ PC is the entity common set, where PS is the common state set, which represents the entity state shared by basic models of different resolutions; PC is the common control law, which is the knowledge set shared by agents of different resolutions for the same entity; Generate operators for common control laws;PAG generates operators for common actions;

[0021] The atomic model with resolution r is described as:

[0022] m r =<X,Y,S,δ int ,δ con ,δ ext ,λ,ta,C,AG,CG,{ρ r→i},λ ρ >

[0023] Where r is the model resolution, ρ r→i :S r ×X Res →M i is the resolution switching function to achieve model switching between different resolutions; ρ :S→Y Res It is a resolution change notification function, used to generate and send resolution change information.

[0024] In a specific embodiment of the present invention, the network model is described as follows:

[0025] n=<X,Y,D,{M d},{I d},Z i,d ,ρ N ,λ N ,Σ,G Σ ,Σ G >

[0026] Where D is the set of entities in the network model, self represents the coupling model itself; M d is the atomic model or coupling model of entity d; I dIt is the collection of all components affected by entity d, used to reflect structural information; Z i,d ,i∈D∪self,d∈I d is the output-input conversion operator from component i to d, including event conversion operators and message passing operators, which are used to reflect the output-input coupling relationship between components; ρ N :S n →X NRes is the network resolution structure transfer function, where λ N :S n →Y NRes Output function for resolution change;

[0027] Σ represents the nature of the network model; G Σ :{M d}→Σ is the network model property generation operator; Σ G :self→self is a network model operation operator used for learning, control and reorganization of the network model level.

[0028] In a specific embodiment of the present invention, the coupling model structure is:

[0029] Coupled=<X NRes ,Y NRes ,n init ,N>

[0030] Among them, X NRes , Y NRes is the network model resolution port, X NRes Receives resolution information from other network models and transmits it to all entities in its own network model; when the resolution of any entity in the network model changes, the resolution information is transmitted to all entities in the network model. Res The port notifies other entities in the network model of resolution changes, and at the same time passes Y NRes Ports sent to other network models' X NRes Port, transmits resolution change information; n init ∈N is the initial network model configuration, and N is the minimum set of network structures composed of entities.

[0031] The embodiment of the present invention further proposes a multi-resolution model description system for complex product simulation based on the above method, comprising:

[0032] A basic model construction module is used to construct basic models corresponding to each entity in the simulation object at each resolution. The description of the basic model is constructed based on the DEVS atomic model and the Agent metamodel.

[0033] An atomic model construction module, used to construct a multi-resolution atomic model of the entity according to the basic model, wherein the atomic model includes a description of a basic model of different resolutions for any entity in the simulation object;

[0034] A network model construction module, used to construct a multi-entity network model according to the atomic model, wherein the network model is a multi-entity network formed by coupling the atomic models of each entity;

[0035] The coupling model construction module is used to construct a coupling model for the simulation object in the current simulation according to the network model, and the coupling model includes network models corresponding to the simulation object under different network structures.

[0036] Features and beneficial effects of the present invention:

[0037] The present invention is a model description method with coupling closure, multi-resolution variable structure modeling capability and agent modeling capability. The method is based on the multi-resolution discrete event description specification (MR-DEVS) and combines the description capabilities of the intelligent agent's perception behavior, communication collaboration and learning evolution in the agent-based modeling method (ABM), providing the modeling capability of multi-resolution complex products or complex systems.

[0038] Specifically, the method includes multi-resolution entity-level atomic model description and multi-resolution system-level coupled model description: at the entity level, the basic models of different resolutions in the atomic model can share the entity model information, and the basic model of a single resolution is jointly characterized by the intelligent agent characteristics and the DEVS atomic model characteristics, which retains the ability to respond to events while supporting the ability to characterize the subject's intelligent characteristics; at the system level, the coupled model is composed of a network structure composed of various entity atomic models, which supports model structure changes and reorganization when the external resolution changes.

[0039] This model description method is coupled and closed, that is, the coupled model can be converted into an atomic model description, thereby supporting the construction of a larger-scale simulation system model.

[0040] This model description method has the ability of variable structure modeling: the structural changes at the sub-component level can be expressed by the internal transfer function and external transfer function of the atomic model. There are two levels of structural changes at the component level and system level:

[0041] (i) Structural changes caused by resolution changes are characterized by the network structure transfer function;

[0042] (ii) Structural changes caused by changes in the composition of the atomic model in the coupled model are characterized by network model operation operators.

[0043] This model description method has multi-resolution modeling capabilities, and realizes multi-resolution model construction and maintenance through resolution switching control related functions and consistency maintenance functions; it realizes resolution status monitoring and information reception and transmission through the resolution change notification ports of the atomic model and the coupling model, and provides them to the internal operation operators to realize resolution switching. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is an overall flow chart of a multi-resolution model description method for complex product simulation proposed in an embodiment of the present invention.

[0045] Figure 2 It is a schematic diagram of a basic model in a specific embodiment of the present invention.

[0046] Figure 3 The figure is a schematic diagram of an atomic model in a specific embodiment of the present invention.

[0047] Figure 4 It is a schematic diagram of a network model in a specific embodiment of the present invention.

[0048] Figure 5 It is a schematic diagram of a coupling model in a specific embodiment of the present invention.

[0049] Figure 6 The figure is a simulation modeling flow chart in a specific embodiment of the present invention.

[0050] Figure 7 It is a schematic diagram of a multi-resolution atomic model of a car at the entity level in a specific embodiment of the present invention.

[0051] Figure 8 It is a schematic diagram of the deaggregation and aggregation process of a multi-resolution model in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0052] The present invention proposes a multi-resolution model description method and system for complex product simulation, and the present invention is described in detail below with reference to specific drawings:

[0053] This embodiment proposes a multi-resolution model description method for complex product simulation. The overall process is as follows: Figure 1 As shown, the following steps are included:

[0054] 1) According to the simulation object, construct basic models corresponding to each entity in the simulation object at multiple resolutions;

[0055] In this embodiment, the description of each basic model is constructed based on the DEVS atomic model and the Agent metamodel, and then each basic model will be incorporated into the multi-resolution modeling framework to construct the atomic model; wherein each basic model is a single-resolution model, and the model description combines the characteristics of the DEVS atomic model and the Agent metamodel, which can effectively characterize the individual's perceptual behavior, communication collaboration, and learning evolution.

[0056] Definition 1: Agent is a computing entity that has autonomy, can sense and respond to changes in the environment, and can interact through messages. Agent is the description object of the Agent metamodel.

[0057] Definition 2: A message is a description of the communication between agents. Its formal definition is:

[0058] Message=<MsgID,From,To,MsgType,Content,SendTime,ValidTime,TransTime>

[0059] Among them, MsgID is the message number, From is the message source, To is the message recipient, MsgType is the message type, Content is the message content, SendTime is the message sending time, ValidTime is the message validity time, and TransTime is the message delivery time.

[0060] Definition 3: The basic model does not focus on the model resolution, and uses the Agent metamodel to describe the model extension of the DEVS atomic model. The structure is as follows:

[0061] Basic=<X,Y,S,s0,δ int ,δ con ,δ ext ,λ,ta,C,AG,CG>

[0062] The input set X is the set of input events and input messages, X = X e ∪X m , X e is the input event set, X m is the input message set; output set Y = Y e ∪Y m , Y e is the output event set, Y m is the output message set; S = S s ∪S d is the basic model state feature set, identified by static attributes S s and the state variable S d composition; s0 is the initial state; δ int :Sd →S d is the internal state transfer function; δ con :S d ×X e b →S d is the concurrent conflict handling function, and the superscript b represents the concurrent event δ ext :Q×X e →S d is the external state transfer function, where Q = {(s, t) | s∈S d ,0≤t≤ta(s)} is the total state set, t is the time the entity stays in state s; λ:S→Y b Output function for the system; is the time advancement function, which generates the time series T.

[0063] In this embodiment, the Agent meta-model is combined with the DEVS atomic model as the basic model description.<C,AG,CG> and the state feature set S together constitute the Agent model set, where C is the Agent rule set, that is, the knowledge base owned by the Agent; CG: (T×X×S→C)∪(T×S→C) is the control law generation operator; AG: (T×C→S)∪(T×C→Y)∪(T×C→II) is the action generation operator used to represent the Agent's external behavior generation method, and II represents internal information feedback.

[0064] In a specific embodiment of the present invention, the internal structure of the basic model is as follows Figure 2 shown.

[0065] The basic model describes the characteristics of the object, which is characterized by the state, which is structured information. The basic model receives external input information and responds to it. Based on the input event, the external transfer function is used to update its own state information; the internal transfer function is used to realize the internal state transfer of the model; the output function outputs the event according to the updated state information. The sensor is used to perceive the external environment, receive and process input messages; the controller is a control rule processing system, which manipulates the control rule library, generates, selects and evaluates the rules, thereby realizing the learning and evolution capabilities as the model subject; the effector produces actions and generates output information according to the control rules and action generation operators given by the controller, and updates it to the environment.

[0066] 2) Based on the basic model in step 1), construct a multi-resolution atomic model of each entity;

[0067] The atomic model is used to describe a multi-resolution entity, and contains descriptions of basic models of different resolutions of the entity, so that switching between different resolutions can be achieved.

[0068] In a specific embodiment of the present invention, the internal structure of the atomic model is as follows Figure 3 shown.

[0069] An atomic model contains basic models of different resolutions for the same entity. The difference between basic models of different resolutions lies in the different levels of abstraction in describing the entity object. Low-resolution basic models have a higher level of abstraction and more macroscopic information, while high-resolution basic models have richer details and more accurate and detailed descriptions of physical mechanisms, internal structures, etc. Since these basic models describe the same object, different basic models often have common information. Therefore, the entity-level model description information exists as a common set in the atomic model, which is used to represent the entity state, control laws, and behaviors, and is common to models of all resolutions.

[0070] For entities, since the basic models of each resolution describe the same object, they often contain some attributes, states, and structures of the object. Therefore, the entity state, control law, and behavior are retained as the common set of each resolution model of the entity, which is common to each resolution basic model, saved in the atomic model, and maintained by the entity property update function. The entity common set and the basic model at the current resolution together constitute the description of the entity at the current resolution. The consistency maintenance function ensures the information consistency between the basic models of different resolutions. Each multi-resolution atomic model of the entity receives external resolution change information and implements switching of different resolutions for the corresponding entity. When the entity being described has only one basic model of one resolution, the entity common state set is an empty set, and the atomic model degenerates into a basic model.

[0071] In a specific embodiment of the present invention,

[0072] Definition 4: The MR-Agent-DEVS atomic model is a dynamic structure DEVS with reflective and agent characteristics, and is a meta-model for multi-resolution entity description. The structure is as follows:

[0073]

[0074] Where X Res , Y Res is the resolution port, used to send and receive resolution change information between different entities; R is the Agent identification set of multiple resolutions of the entity; ψ={2 R -φ} is the resolution mode set; is the initial resolution mode, is the basic model at this resolution; M r is the basic model set of the entity when the resolution is r; is the model consistency mapping function; Resolution mode The association between different modules at the same resolution.

[0075] PSC = PS ∪ PC is the entity common set, where PS is the common state set, which represents the entity-level state between models with different resolutions, and PC is the common control law, which is the knowledge set shared by agents with different resolutions of the same entity; The public control law generation operator represents the control law generation at the entity level; PAG is the public action generation operator. The entity-level properties and their operations reflect the agent characteristics of the entity at a higher level of abstraction.

[0076] The resolution mode in the multi-resolution modeling framework is Compared with the Agent-DEVS basic model, the description of the model resolution is added.

[0077] Description of the atomic model with resolution r in the multi-resolution framework:

[0078] m r =<X,Y,S,δ int ,δ con ,δ ext ,λ,ta,C,AG,CG,{ρ r→i},λ ρ >

[0079] Where r is the model resolution, ρ r→i :S r ×X Res →M i is the resolution switching function, which completes the model switching between different resolutions and maintains the continuity of the model; ρ :S→Y Res It is a resolution change notification function, used to generate and send resolution change information.

[0080] 3) Based on the atomic model in step 2), a multi-entity network model is constructed.

[0081] In a specific embodiment of the present invention, its internal structure is as follows Figure 4As shown. The network model contains atomic models of different entities, and the atomic models can interact with each other. The transmission and conversion of events and messages are realized through the pre-set input-output conversion algorithm; the network model describes the properties that a single entity does not have after the entity coupling produces the emergence phenomenon, and is generated and updated by the network property generation function. There are two kinds of structural changes inside the network model, namely resolution change and internal entity coupling relationship change. The latter is realized by the network model operation function to reorganize, increase or decrease the internal structure of the network model. When the current resolution changes, the corresponding resolution switching of each entity atomic model is realized through the pre-set network structure resolution change function, and the resolution change information is passed to the resolution change output function, which transmits the network resolution change information to the outside of the network model.

[0082] In a specific embodiment of the present invention, the description method of the network model is defined as:

[0083] n=<X,Y,D,{M d},{I d},Z i,d ,ρ N ,λ N ,Σ,G Σ ,Σ G >

[0084] Self represents the coupling model itself, X, Y are structured input and output events and message sets; D is the set of entities in the network model, M d is the MR-Agent-DEVS model of entity d, which can be an atomic model or a coupled model; d It is the set of all components affected by entity d, reflecting the system structure information; Z i,d ,i∈D∪self,d∈I d is the output-input conversion operator from component i to d, including event conversion operators and message passing operators, reflecting the output-input coupling relationship between components; ρ N :S n →X NRes is the network resolution structure transfer function, where λ N :S n →Y NRes Output function for resolution change.

[0085] Σ represents the properties of the network model, which is a property that a single atom model does not have; G Σ :{M d}→Σ is the network model property generation operator; Σ G:self→self is a network model operation operator, which is used for learning, controlling and reorganizing the network model level. The properties and operators of the network model reflect the characteristics of each resolution state in the coupled model.

[0086] 4) constructing a coupling model for the current simulation based on the network model constructed in step 3);

[0087] The coupling model is used to describe a multi-resolution system, including network models representing different network structures of the system, and can realize switching of network models with different resolutions.

[0088] In this embodiment, the internal structure of the coupling model is as follows: Figure 5 As shown. The coupling model contains the network model in the initial state and a set of candidate network models, which are generated by step 3). The resolution change information can be received by the resolution port to realize the switching of different network structures of the system.

[0089] Furthermore, the coupled model can be transformed into an atomic model, that is, the multi-resolution system is regarded as a multi-resolution entity, and the description corresponding to the coupled model is transformed into an atomic model description, which serves as the basis for building a larger-scale system, reflecting the coupled closure of the model.

[0090] In a specific embodiment of the present invention, Definition 5: The MR-Agent-DEVS coupling model is a dynamic network with reflective and agent characteristics, and is a meta-model for multi-resolution system description. The structure is as follows:

[0091] Coupled=<X NRes ,Y NRes ,n init ,N>

[0092] Among them, X NRes , Y NRes is the network model resolution port, X NRes Receives resolution information from other network models and transmits it to all entities in its own network model; when the resolution of any entity in the network model changes, the resolution information is transmitted to all entities in the network model. Res The port notifies other entities in the network model of resolution changes, and at the same time passes Y NRes Ports sent to other network models' X NRes Port, transmits resolution change information; n init ∈N is the initial network model configuration. N is the minimum set of network structures composed of entity members in the system.

[0093] 5) The coupled model can be transformed into an atomic model as the basis for building a larger-scale simulation system.

[0094] The model description method of the present invention has coupling closure, that is, the coupling model can be converted into an atomic model, thereby meeting the needs of multi-level modeling. Since the MR-DEVS description specification is coupled and closed, the proof of the coupling closure of the MR-Agent-DEVS model only needs to prove the coupling closure of the relevant descriptions of the Agent characteristics in the atomic model and the coupling model on this basis. Network model properties Σ and entity common set PSC, network property generation operator G Σ With the common control law generation operator PCG and the network model operation operator Σ G Corresponding to the public action generation operator PAG, the conversion from coupling model description to atomic model description can be realized. The description related to Agent in the basic model description is irrelevant to the coupling closure proof. The coupling closure of the model description still holds.

[0095] The specific implementation method is:

[0096] Resolution of coupled model input and output ports X NRes , Y NRes As the input and output ports of the atomic model respectively; that is, let X Res =X NRes ,Y Res =Y NRes ; Let R = U d∈D R d ,R d is the resolution set of entity d; let ψ={2 R -φ}, Where D init To form the initial network n init The set of all entities, is the initial resolution (mode) set of entity d, i.e. the initial resolution mode of the atomic model is the set of initial resolutions (modes) of all entities in the initial configuration of the network; let is the initial model representation of entity d, that is, the initial model representation of the atomic model is the set of initial model representations of all entities in the initial configuration of the network; let is the consistency mapping function of entity d in parallel mode; let is the correlation function between modules of entity d in the disaggregated state operation mode. Let λ ρ =λ N . The network transfer function ρ N and the resolution switching function ρ r→i The functions are the same, and both need to follow the rules of value preservation and initialization of new components and their variables. ΣWith the common control law generation operator PCG and the network model operation operator Σ G The corresponding transformations are completed with the public action generation operator PAG, thereby realizing the representation of the coupled model description as an atomic model description.

[0097] Furthermore, the method of the present invention can be applied to the simulation modeling of complex products. The modeling process is mainly divided into three parts: model abstraction, model construction and model verification. The overall process is as follows: Figure 6 As shown, including:

[0098] 1) Model abstraction;

[0099] 1-1) System level division

[0100] When faced with specific simulation scenarios, we can clarify the simulation requirements for each resolution level from dimensions such as data resolution, process resolution, and structural resolution, and divide the system into levels.

[0101] 1-2) Modeling object abstraction at each level

[0102] According to the simulation task requirements at different levels, abstract models of entities at each level are established to determine the resolution and agent characteristics of each entity.

[0103] 1-3) Model relationship abstraction

[0104] According to the requirements of multi-resolution modeling, it is divided into:

[0105] (i) Determine the logic, structure, and interaction relationships between model objects at the same level; (ii) Determine the mapping relationship between models at different levels of resolution; (iii) Determine the construction method and consistency maintenance strategy for models of different resolutions for the same entity.

[0106] 2) Model construction;

[0107] 2-1) Message protocol construction;

[0108] According to the simulation requirements, define the message transmission mechanism and protocol between agents, and then define the message format.

[0109] 2-2) Model object construction;

[0110] According to the resolution characteristics and agent characteristics of the entity, the model description of each entity is constructed. For entities with agent characteristics, behavior modeling, control rule generation and knowledge update can be performed in the construction of each resolution model.

[0111] 2-3) Model mapping relationship construction;

[0112] Based on the abstract mapping relationship constructed in steps 1-3), a mapping relationship in dimensions such as structure, function, data, process, and attribute is constructed between models at different layers and between models of different resolutions of the same entity.

[0113] 2-4) Resolution controller construction;

[0114] Construct a system structure resolution controller and a model resolution controller.

[0115] 3) Model verification;

[0116] 3-1)Logical verification;

[0117] Verification at each level: verify whether the model satisfies the abstract model relationship and whether it can meet the simulation purposes at each level; aggregation verification and disaggregation verification: verify whether the high-resolution model aggregation and low-resolution model disaggregation satisfy the abstract model relationship; consistency verification: verify the consistency maintenance relationship of each multi-resolution entity.

[0118] 3-2) Simulation operation

[0119] Run the simulation to check whether the model can meet the simulation requirements at different levels. If it can, you can proceed to the simulation stage. Otherwise, you need to consider adjusting the abstraction level or model relationship and return to step 1).

[0120] The present invention is further described in detail below in conjunction with a specific embodiment:

[0121] As a typical complex product, different simulation scenarios in the design, development and application process of automobiles often involve the construction of models at multiple different levels. Taking the model description of the vehicle-road cooperative traffic system as an example, taking an unmanned vehicle with vehicle-to-vehicle interaction as an example, considering the multi-resolution models of the traffic system layer and the vehicle-road cooperative layer and typical application scenarios, this method is explained.

[0122] Figure 7 This is a structural diagram of the vehicle-road cooperative system shown in this embodiment. The system consists of a multi-resolution vehicle entity V and a multi-resolution roadside equipment entity RS. U represents a roadside unit, CC represents a traffic control center, L represents a traffic signal device, and C represents a vehicle-road cooperative scene.

[0123] Multi-resolution atomic models

[0124] (1) Multi-resolution vehicle entity model; including:

[0125] a. MR-Agent-DEVS atomic model of vehicle entity:

[0126]

[0127] in:

[0128] The port for receiving resolution changes for vehicle entities, C ,V,CV represent the input events that the entity may receive, representing the vehicle-road cooperative mode, vehicle mode and the concurrent mode of the two respectively.

[0129] It is the resolution change notification port. DesRN and AggD represent the road network level system deaggregation and traffic control device aggregation events respectively. The former deaggregates traffic control devices, and the latter aggregates into the road network level system.

[0130] R V = {r C ,r V} is the set of entity resolutions, where r C is the vehicle-road collaboration level, r V For bicycle level.

[0131] ψ V = {r C ,r V ,(r C ,r v )} is a collection of entity resolution modes.

[0132] It is the initial resolution mode of the vehicle entity, which is the vehicle-road collaboration mode here.

[0133] It is a collection of entity models in the initial resolution mode, here it is the vehicle-road cooperative model.

[0134] A collection of vehicle entities in all resolution modes. It is a collection of models in the vehicle-road collaboration mode, including lane-changing model, car-following model, intersection model, etc. is the model set of the vehicle entity in the single vehicle mode, here it is the model set of vehicles 1, 2, …, n.

[0135] b. Vehicle high-resolution Agent-DEVS basic model:

[0136]

[0137] in:

[0138] S V Represents bicycle information, which may include descriptive variables such as its position, speed, acceleration, and status.

[0139] It is a resolution conversion function. When vehicle-to-vehicle or vehicle-to-road collaborative relationships are established between vehicles, the entity resolution will change, resulting in deaggregation or aggregation.

[0140] Notification function for resolution change events.

[0141] λ V :S V →Y V is the output function.

[0142] The resolution mode is It is the coupling relationship between different vehicle modules.

[0143] C V Develop driving control decision-making rules for unmanned vehicles.

[0144] CG V Represents the driverless car driving rule generation operator, which can represent the training model generated by deep neural network, deep reinforcement learning and other methods.

[0145] AG V :(T×C V →S V )∪(T×C V →Y V )∪(T×C V →II V ) represents the driverless car driving action generation operator, which represents the autonomous driving behavior generation and changes its own state S V , generating external output Y V Interact with the external environment and other vehicles while generating internal feedback information II V .

[0146] c. Message Model

[0147] Communication occurs between vehicles and between vehicles and roads, and messages are defined by the corresponding communication protocol. The following is an emergency brake message sent by the front vehicle Car30 to the rear vehicle Car25 within a range of 100 meters. The message number is 1086, where the message type is defined as vehicle-to-vehicle, the sending time is the effective time 01:00, and there is a 3-second delay in message transmission. It is expressed as:

[0148] Message=<1086,Car30,Car25,Car2Car,"Emergency Brake",01:00,01:00,00:03>

[0149] (2) Multi-resolution atomic model of roadside equipment

[0150] RS is used to represent the multi-resolution atomic model of roadside equipment, and RS deaggregation generates roadside unit M U 、Traffic Control Center M CC 、Traffic signal equipment ML .

[0151] Multi-resolution vehicle-road cooperative system model

[0152] The vehicle entity and the roadside equipment entity are coupled to form a network model of the vehicle-road cooperative system. The coupling model is described by the MR-Agent-DEVS coupling model:

[0153] VR-S=<X NRes ,Y NRes ,n init ,N>

[0154] in:

[0155] X NRes It is the system resolution change input port, corresponding to the resolution mode change of vehicle entity and roadside equipment entity.

[0156] Y NRes This is the output port for system resolution change. In this scenario, the empty set is used.

[0157] N is the minimum set of variable structure networks.

[0158] n init is the initial network model, and n init ∈N.

[0159] The elements in N are described as follows:

[0160] n=<X N ,Y N ,D,{M d},{I d},{Z i,j},ρ N ,λ N ,Σ,G Σ ,Σ G >

[0161] in:

[0162] D = {V, RS} is the entity index set in the network.

[0163] {M d} is the set of MR-Agent-DEVS atomic models corresponding to each entity in the network, d∈D.

[0164] {I d} is the set of models affected by entity d in the network, and the model members will also affect the network itself.

[0165] {Z i,j}={Z V ,Z RS ,Z S} represents the coupling relationship between models, where vehicles, roadside equipment and the network itself are coupled with each other.

[0166] Indicates that the coupling network itself does not generate resolution change events externally.

[0167] ρ N Represents the resolution structure transfer function, which includes system-level disaggregation function and entity-level aggregation function.

[0168] Σ is the system-level property of the vehicle-road cooperative system, such as vehicle traffic efficiency, accident probability, etc.

[0169] G Σ Represents a network-level property generation operator.

[0170] Represents a network structure change operator. In this embodiment, changes in the network structure caused by the increase or decrease of entities are not considered.

[0171] Figure 8 The figure shows the process of deaggregation and aggregation of the vehicle-road cooperative system in this embodiment. The system VR-S is coupled by the vehicle V and the roadside equipment RS, and a higher resolution model is generated after deaggregation, in which the multi-resolution vehicle entity V is deaggregated to generate the vehicle-road cooperative vehicle M. rc and other vehicles M V , RS deaggregation generates roadside unit M U 、Traffic Control Center M CC 、Traffic signal equipment M L For M V , the number of vehicles n is 1, so the atomic model degenerates into the Agent-DEVS basic model.

[0172] This embodiment also proposes a multi-resolution model description system for complex product simulation based on the above method, including:

[0173] A basic model construction module is used to construct basic models corresponding to each entity in the simulation object at each resolution. The description of the basic model is constructed based on the DEVS atomic model and the Agent metamodel.

[0174] An atomic model construction module, used to construct a multi-resolution atomic model of the entity according to the basic model, wherein the atomic model includes a description of a basic model of different resolutions for any entity in the simulation object;

[0175] A network model construction module, used to construct a multi-entity network model according to the atomic model, wherein the network model is a multi-entity network formed by coupling the atomic models of each entity;

[0176] The coupling model construction module is used to construct a coupling model for the simulation object in the current simulation according to the network model, and the coupling model includes network models corresponding to the simulation object under different network structures.

Claims

1. A multi-resolution model description method for complex product simulation, characterized in that: include: Constructing basic models corresponding to each entity in the simulation object at each resolution, wherein the description of the basic model is constructed based on the DEVS atomic model and the Agent meta-model; According to the basic model, construct a multi-resolution atomic model of the entity, wherein the atomic model includes descriptions of basic models of different resolutions for any entity in the simulation object; According to the atomic model, a multi-entity network model is constructed, wherein the network model is a multi-entity network formed by coupling the atomic models of each entity; According to the network model, construct a coupling model for the simulation object in the current simulation, wherein the coupling model includes network models corresponding to the simulation object under different network structures; The basic model is described as follows: Basic=<X,Y,S,s0,δ int ,d con ,d ext ,λ,ta,C,AG,CG> The input set X is the set of input events and input messages, X = X e ∪X m , X e is the input event set, X m is the input message set; similarly, Y=Y e ∪Y m , Y e is the output event set, Y m is the output message set; S = S s ∪S d is the basic model state feature set, identified by static attributes S s and the state variable S d composition; s0 is the initial state; δ int :S d →S d is the internal state transfer function; δ con :S d ×X e b →S d is the concurrent conflict handling function, and the superscript b represents the concurrent event δ ext :Q×X e →S d is the external state transfer function, where Q = {(s, t) | s∈S d ,0≤t≤ta(s)} is the total state set, t is the time the entity stays in state s; λ:S→Y b is the system output function; ta: is the time advancement function, which generates the time series T; C is the Agent rule set; CG: (T×X×S→C)∪(T×S→C) is the control law generation operator; AG: (T×C→S)∪(T×C→Y)∪(T×C→II) is the action generation operator used to represent the Agent's external behavior generation method, and II represents internal information feedback; The structure of the atomic model is: Among them, X Res , Y Res is the resolution port, used to send and receive resolution change information between different entities; R is the Agent identification set of multiple resolutions of the entity; ψ={2 R -φ} is the resolution mode set; is the initial resolution mode, The resolution is The basic model when M r is the basic model set of the entity when the resolution is r; is the model consistency mapping function; Resolution mode The association between different modules at the same resolution; PSC=PS∪PC is the entity common set, where PS is the common state set, which represents the entity state shared by basic models of different resolutions; PC is the common control law, which is the knowledge set shared by agents of different resolutions of the same entity; PCG: Generate operators for common control laws;PAG generates operators for common actions; The atomic model with resolution r is described as: m r =<X,Y,S,δ int ,d con ,d ext ,λ,ta,C,AG,CG,{ρ r→i },l ρ > Where r is the model resolution, ρ r→i :S r ×X Res →M i is the resolution switching function to achieve model switching between different resolutions; ρ :S→Y Res It is a resolution change notification function, used to generate and send resolution change information to the outside; The network model is described as follows: n=<X,Y,D,{M d },{I d },Z i,d ,r N ,l N ,S,G Σ ,S G > Where D is the set of entities in the network model, self represents the coupling model itself; M d is the atomic model or coupling model of entity d; I d It is the collection of all components affected by entity d, used to reflect structural information; Z i,d ,i∈D∪self,d∈I d is the output-input conversion operator from component i to d, including event conversion operators and message passing operators, which are used to reflect the output-input coupling relationship between components; ρ N :S n →X NRes is the network resolution structure transfer function, where λ N :S n →Y NRes Output function for resolution change; Σ represents the nature of the network model; G Σ :{M d }→Σ is the network model property generation operator; Σ G :self→self is a network model operation operator used for learning, control and reorganization of the network model level.

2. The method according to claim 1, characterized in that The method further comprises: The coupled model is converted into an atomic model for system simulation of the simulation object as a multi-resolution entity.

3. The method according to claim 1, characterized in that The coupling model structure is: Coupled=<X NRes ,Y NRes ,n init ,N> Among them, X NRes , Y NRes is the network model resolution port, X NRes Receives resolution information from other network models and transmits it to all entities in its own network model; when the resolution of any entity in the network model changes, the resolution information is transmitted to all entities in the network model. Res The port notifies other entities in the network model of resolution changes, and at the same time passes Y NRes Ports sent to other network models' X NRes Port, transmits resolution change information; n init ∈N is the initial network model configuration, and N is the minimum set of network structures composed of entities.

4. A multi-resolution model description system for complex product simulation based on the method according to any one of claims 1 to 3, characterized in that: include: A basic model construction module is used to construct basic models corresponding to each entity in the simulation object at each resolution. The description of the basic model is constructed based on the DEVS atomic model and the Agent metamodel. An atomic model construction module, used to construct a multi-resolution atomic model of the entity according to the basic model, wherein the atomic model includes a description of a basic model of different resolutions for any entity in the simulation object; A network model construction module, used to construct a multi-entity network model according to the atomic model, wherein the network model is a multi-entity network formed by coupling the atomic models of each entity; The coupling model construction module is used to construct a coupling model for the simulation object in the current simulation according to the network model, and the coupling model includes network models corresponding to the simulation object under different network structures.

Images