A flexible manufacturing process mechanism model for realizing process execution and its construction method

Abstract

The present invention discloses a flexible manufacturing process mechanism model for implementing a process and a construction method thereof. The model is composed of three parts: a flexible manufacturing process knowledge graph, a parameter calculator, and an action executor. The flexible manufacturing process knowledge graph can be structurally divided into an abstract concept layer, a concept instance layer, and a capability layer, and from different perspectives of knowledge, a composition view, a machine resource view, a process view, and a mode view can be abstracted. The action executor is responsible for executing the robot-executable actions obtained through reasoning. The parameter calculator is responsible for calculating the parameter values required for various robot actions according to the process mechanism and the manufacturing resources in the specific process manufacturing scenario deduced. The above three parts cooperate with each other and complement each other, enabling the model to control the robot to realize the intelligent path planning of the manufacturing process according to the manufacturing requirements.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer services, and relates to a flexible manufacturing process mechanism model and a construction method thereof, in particular to a knowledge modeling method for flexible manufacturing to realize intelligent planning of the process path executed by a robot. Background Art

[0002] The current world presents a trend of diversification and personalization. As a new model of future "intelligent manufacturing", personalized customization has been applied to production manufacturing. However, the traditional production mode that focuses on standardization and mass production can no longer meet the pursuit of uniqueness and personalization by consumers, and the manufacturing capabilities of production equipment are difficult to estimate, which requires detailed modeling of process procedures. Taking the product assembly process as an example, the assembly link accounts for 40-60% of the total production time. Personalized customization means that the assembly process needs to be frequently modified during assembly. Manual assembly has low efficiency, and it is impossible to adapt to the environment of large-scale personalized customization even with more manpower. In addition, introducing automation technology in the production process to assist manual processing, assembly and other processes of products improves production efficiency, but it is applicable to the production process of products with almost unchanged processes. Therefore, establishing a flexible manufacturing process mechanism model for process execution, modeling industrial knowledge from multiple perspectives, and realizing intelligent recommendation of process execution paths is of milestone significance in mass customization. Summary of the Invention

[0003] In order to solve the above problems existing in the prior art, the present invention provides a flexible manufacturing process mechanism model for process execution and a construction method thereof. The present invention can realize intelligent planning of the process path executed by a robot according to the combination of a process knowledge graph and robot control codes, so as to provide services for flexible manufacturing. In the process of constructing the process knowledge graph, based on the multi-view industrial mechanism modeling method (Abstraction-Instance-Capability, AIC), industrial knowledge is modeled from perspectives such as the composition, process, resources, and mode of production equipment, and the robot control codes are encapsulated to realize the combination of the process knowledge graph and the robot control codes, providing agile services for flexible manufacturing.

[0004] The object of the present invention is achieved by the following technical solutions:

[0005] A flexible manufacturing process mechanism model for process execution is composed of three parts: a flexible manufacturing process knowledge graph, a parameter calculator, and an action executor, where:

[0006] The flexible manufacturing process knowledge graph is composed of triples, where the head entity and the tail entity correspond to Node, and the relationship corresponds to Relation. Among them, the categories of nodes represented by Node include Abstraction, Instance, Capability, Goal, Operation, Action, and Parameter; the relationships represented by Relation are the relationships between nodes, and the relationship categories it contains are include, has_instance, has_capability, achieve_goal, operation, installed_on, has_action, has_parameter, and after;

[0007] From the perspective of the composition structure of the knowledge graph, the flexible manufacturing process knowledge graph includes an abstract concept layer, a concept instance layer, and a capability layer. From the perspectives of the composition of the knowledge graph, machine resources, processes, and patterns, it includes a composition view, a machine resource view, a process view, and a pattern view;

[0008] The abstract concept layer is represented as follows:

[0009] T ACL ={(a i , r1, a j ) | a i , a j ∈Abstraction, r1 = include} ∪

[0010] {(a i , r6, a j ) | a i , a j ∈Abstraction, r6 = installed_on} ∪

[0011] {(c i , r1, c j ) | c i , c j ∈Capability, r1 = include} ∪

[0012] {(a, r3, c) | a ∈ Abstraction, c ∈ Capability, r3 = has_capability}

[0013] Among them, the include relationship represents an inclusion and composition relationship. Hierarchies can be divided among abstract concepts through include, and the abstract ability can also use the include relationship to represent more refined abilities. The has_capability relationship indicates that a concept of flexible manufacturing has a certain ability, and the installed_on relationship represents the installation relationship between a fixture and a robotic arm;

[0014] The concept instance layer is represented as follows:

[0015] T CIL ={(a, r2, i) | a ∈ Abstraction, i ∈ Instance, r2 = has_instance} ∪

[0016] {(i i , r1, i j ) | i i , i j ∈ Instance, r1 = include}

[0017] Among them, the has_instance relationship represents the relationship between an abstract concept and an instance of that concept, and the include relationship can be used to describe the composition relationship between instances;

[0018] The ability layer is represented as follows:

[0019] T CL ={(i, r3, c) | i ∈ Instance, c ∈ Capability, r3 = has_capability} ∪

[0020] {(i, r4, g) | i ∈ Instance, g ∈ Goal, r4 = achieve_goal} ∪

[0021] {(g, r5, o) | g ∈ Goal, o ∈ Operation, r5 = operation} ∪

[0022] {(o, r7, n) | o ∈ Operation, n ∈ Action, r7 = has_action} ∪

[0023] {(n, r8, p) | n ∈ Action, p ∈ Parameter, r8 = has_parameter} ∪

[0024] {(o i , r9, o j ) | o i , o j ∈ Operation, r9 = after} ∪

[0025] {(n i , r9, n j ) | n i , n j ∈ Action, r9 = after}

[0026] Among them, entities of type Operation represent production steps in the production process, entities of type Action represent the actions of the robot, entities of type Parameter represent control parameters related to the robot in the production process, the has_action relationship represents the relationship between operations and robot actions, the has_parameter relationship represents the connection between actions and parameters, and the after relationship represents the execution order of operations and actions;

[0027] The machine resource view is used to describe the robot that executes the process path, obtain the fixture and manipulator at the end of the robot, store the information of the robot, and reflect the matching relationship between the fixture and the manipulator;

[0028] The process view is used to describe the process manual, obtain the process objectives and process execution steps from it, that is, the process steps, and at the same time describe the relationships between different types of actions of the robot and the process steps, as well as the relationships between actions and parameters;

[0029] The mode view contains a mode overview among abstract concepts, instances, capabilities, and process objectives. The abstract concept and the capability are associated by the relationship r3, the abstract concept and the process objective are associated by the relationship r4, the capability and the process objective are implicitly associated through the process mechanism (Process Mechanism, PM), and the relationships between the abstract concept, the capability nodes, and between them map the relationships between instances of the abstract concept and between instances and instances of the capability;

[0030] The composition view is responsible for obtaining the data of the product's composition structure according to the product's composition structure;

[0031] The parameter calculator is responsible for calculating the parameter values required for various robot actions according to the manufacturing resources and process mechanisms inferred from the flexible manufacturing process knowledge graph;

[0032] The action executor is responsible for making a specific implementation of the actions described by the nodes of type Action in the flexible assembly process knowledge graph according to the knowledge inferred from the flexible manufacturing process knowledge graph and the parameter values calculated using the parameter calculator, obtaining the actions that the robot can execute, and controlling the robot to execute the actions.

[0033] A method for constructing a flexible manufacturing process mechanism model for realizing process execution includes the following steps:

[0034] Step S1: Construct the ontology of the flexible manufacturing process knowledge graph:

[0035] Model industrial knowledge from the perspectives of the composition, process, resources, and mode of process production equipment. According to the existing ontology construction methods, construct the ontology of the flexible manufacturing process knowledge graph. The specific steps are as follows:

[0036] Step S11: Determine that the domain of the ontology of the flexible manufacturing process knowledge graph is the flexible manufacturing domain, and use this ontology for process reasoning in flexible manufacturing;

[0037] Step S12: Reuse the existing ontology construction method for the industrial domain (AIC model) and expand it on this basis. For the AIC model, retain the original entities and relationships of the model, and add entity types: Action represents the entity of an action, and Parameter represents the entity of a parameter; add relationship types: installed_on, has_action, has_parameter, after, which represent the installation relationship of the robotic arm fixture, the association relationship between an operation and an action, the requirement relationship between an action and a parameter, and the sequential relationship between operations and between actions; in addition, add an implicit relationship of the process mechanism (PM) association between an ability and a process goal;

[0038] Step S13: List the concepts of various entities in the flexible manufacturing domain, and define the classes representing entities in the flexible manufacturing domain, the attributes of the classes, and the relationships between classes;

[0039] Step S2: Construct the flexible manufacturing process knowledge graph based on the ontology:

[0040] Construct the flexible manufacturing process knowledge graph based on the AIC model, give specific examples of the concepts and relationships mentioned in the ontology, and write the entity classes corresponding to the flexible manufacturing process knowledge graph;

[0041] Step S3: Use a parameter calculator to calculate control parameters:

[0042] Define and configure a parameter calculator to calculate the parameters required for the inferred actions. Substitute a set of calculated parameters into the action executor to obtain an executable action. Among them: before the inferred action is converted into an executable action, it represents a framework of an action. After the parameter calculator calculates the required parameter values for it, it can be called an executable action. The implementation method of the parameter calculator adopts the simple factory pattern, and a configuration file is introduced to cooperate with the parameter calculator to calculate parameters. The specific steps are as follows:

[0043] Step 31: Define the parent class (Parameter class) of parameter calculators for all types of parameters, the parameter calculator classes for robot-related parameters and inherit the Parameter class, the parameter calculator classes for resource-related parameters and inherit the Parameter class, define the parameter calculator classes for specific different parameters and inherit the corresponding parent classes according to categories respectively, and write the implementation of calculating parameter values by parameter calculators respectively;

[0044] Step 32: Define a factory for production parameter calculators. The defined factory retrieves the configuration file according to the input parameter type and produces a matching parameter calculator;

[0045] Step 33: The parameter calculator calculates parameters according to the production resources. The parameters are divided into two major types, namely, robot-related parameters (such as the speed of joint movement) and manufacturing resource-related parameters (such as the pose of the end effector movement);

[0046] Step S4: Use the motion actuator to execute the robot actions:

[0047] Define and configure the motion actuator, and execute the actions of the executable robot deduced. The executable actions represent the actions that can plan specific motion trajectories. The implementation of the motion actuator adopts a combination of the single factory pattern and the configuration file. The specific steps for the motion actuator to execute actions are as follows:

[0048] Step S41: Define the parent class of all motion actuators, and create motion actuators for corresponding different actions according to the categories of robot actions respectively;

[0049] Step S42: Define a factory for producing motion actuators. The factory retrieves the configuration file according to the input action category and produces a motion actuator that matches the action category;

[0050] Step S43: The motion actuator executes specific actions according to the parameter set calculated by the input parameter calculator;

[0051] Step S5: According to Steps S1 to S4, the flexible manufacturing process knowledge graph, parameter calculator, and motion actuator together form a flexible manufacturing process mechanism model. The user inputs a manufacturing task and uses the flexible manufacturing process mechanism model to infer the motion trajectory of process execution, including the following steps:

[0052] Step S51: Analyze the user input to obtain the task list required for the manufacturing task. Each task includes the operation object, process name, and the robot used;

[0053] Step S52: Process the task list in sequence. Finally, infer the operations to be performed for each task based on the process view. Before each operation is executed, perform a statistical analysis of the manufacturing resources and allocate the resources to the factory of the production parameter calculator and the production action executor.

[0054] Step S53: Infer the action framework (represented by a set of sequential Action-type nodes) that should be executed for each completed operation (represented by a node of the Operation type).

[0055] Step S54: Infer the parameters required for each action (represented by a set of Parameter-type nodes).

[0056] Step S55: Produce a parameter calculator capable of calculating each parameter, and calculate the parameter values based on the production resources.

[0057] Step S56: Produce an action executor capable of executing each action. Determine the robot controlled by the action executor based on the production resources, and input a set of parameter key-value pairs related to the action to the action executor, and the action executor executes the action.

[0058] Step S57: The continuous actions executed by multiple action executors under multiple operations form the motion planning for completing the process.

[0059] Compared with the prior art, the present invention has the following advantages:

[0060] 1. The flexible manufacturing process mechanism model includes three parts: the flexible manufacturing process knowledge graph, the action executor, and the parameter calculator. The flexible manufacturing process knowledge graph can be structurally divided into an abstract concept layer, a concept instance layer, and an ability layer. From different perspectives of knowledge, it can be abstracted into a composition view, a machine resource view, a process view, and a mode view. The action executor is responsible for executing the robot-executable actions obtained by inference. The parameter calculator is responsible for calculating the parameter values required for various robot actions based on the process mechanism and the manufacturing resources in the inferred specific process manufacturing scenario. The above three parts cooperate with each other and complement each other, enabling the model to control the robot according to the manufacturing requirements to achieve intelligent path planning of the manufacturing process.

[0061] 2. The present invention can model production resources and production processes, express the relationships between process concepts, and the proposed machine resource view and process view can effectively express the connections between process knowledge, thus realizing the inference of process knowledge.

[0062] 3. The present invention can combine the flexible manufacturing production process knowledge graph with the robot control code to meet the requirements of flexible manufacturing of products under personalized customization, and can realize the planning of the robot's execution manufacturing motion trajectory from the manufacturing requirements to automatic.

[0063] 4. The present invention overcomes the disadvantages of low efficiency in traditional manual manufacturing and lack of flexibility in automated manufacturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is the construction process of the flexible manufacturing process mechanism model.

[0065] Figure 2 is the structure of the flexible manufacturing process knowledge graph.

[0066] Figure 3 is the machine resource view of the flexible assembly process knowledge graph.

[0067] Figure 4 is the process view of the flexible assembly process knowledge graph.

[0068] Figure 5 is the component view of the flexible assembly process knowledge graph.

[0069] Figure 6 is the schema view of the flexible assembly process knowledge graph.

[0070] Figure 7 is the implementation principle of the motion actuator and the parameter calculator. DETAILED DESCRIPTION OF THE INVENTION

[0071] The technical solution of the present invention will be further described below with reference to the accompanying drawings, but it is not limited thereto. Any modification or equivalent replacement of the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention shall be covered by the protection scope of the present invention.

[0072] The present invention provides a flexible manufacturing process mechanism model for realizing process execution, as Figure 1 shown. The model is composed of three parts: a flexible manufacturing process knowledge graph, a motion actuator, and a parameter calculator, wherein:

[0073] (1) Flexible manufacturing process knowledge graph: The flexible manufacturing process knowledge graph can express process knowledge from multiple levels and perspectives.

[0074] The flexible assembly process knowledge graph consists of triples, which are in the form of (head entity, relation, tail entity) composed of entities and relations. The head entity and tail entity correspond to Node, and the relation corresponds to Relation. The categories of nodes represented by Node include Abstraction, Instance, Capability, Goal, Operation, Action, and Parameter; the relations represented by Relation are the relationships between nodes, and the relation categories it contains are include, has_instance, has_capability, achieve_goal, operation, installed_on, has_action, has_parameter, and after.

[0075] Considering the composition structure of the knowledge graph, the flexible assembly process knowledge graph can be analyzed from three levels: abstract concepts, concept instances, and capabilities.

[0076] Abstract concept layer: It represents the concepts abstracted from the flexible manufacturing field and also represents the relationships between concepts. In the flexible manufacturing field, production resources usually include robots, and robots are generally composed of robotic arms and fixtures. The installed_on relation is used to represent the installation relationship between the fixture and the robotic arm. The representation of the abstract concept layer is as follows:

[0077] T ACL ={(a i ,r1,a j )|a i ,a j ∈Abstraction,r1 = include}∪

[0078] {(a i ,r6,a j )|a i ,a j ∈Abstraction,r6 = installed_on}∪

[0079] {(c i ,r1,c j )|c i ,c j ∈Capability,r1 = include}∪

[0080] {(a,r3,c)|a∈Abstraction,c∈Capability,r3 = has_capability}

[0081] Concept Instance Layer: It is composed of instances of abstract concepts and abstract capabilities in the Abstract Concept Layer and their relationships. It represents the specific production and manufacturing resources in the flexible manufacturing field and the capabilities they possess. The formula is as follows:

[0082] T CIL ={(a, r2, i)|a ∈ Abstraction, i ∈ Instance, r2 = has_instance} ∪

[0083] {(i i , r1, i j )|i i , i j ∈ Instance, r1 = include}

[0084] Capability Layer: It describes the production goals in the product production process. The production goals will be decomposed into production steps. Entities of the Operation type represent the production steps in the production process, entities of the Action type represent the actions of the robot, and an action is an indivisible basic action. Entities of the Parameter type represent the control parameters related to the robot in the production process. Completing a production step requires the robot to execute multiple actions, and the has_action relationship represents the relationship between the operation and the robot actions. Completing a robot action requires several parameter values, and the has_parameter relationship represents the connection between the action and the parameters. Both the operations and actions need to be executed in the correct order, and the after relationship represents the execution order of the operations and actions. The representation of the Capability Layer is as follows:

[0085] T CL ={(i, r3, c)|i ∈ Instance, c ∈ Capability, r3 = has_capability} ∪

[0086] {(i, r4, g)|i ∈ Instance, g ∈ Goal, r4 = achieve_goal} ∪

[0087] {(g, r5, o)|g ∈ Goal, o ∈ Operation, r5 = operation} ∪

[0088] {(o, r7, n)|o ∈ Operation, n ∈ Action, r7 = has_action} ∪

[0089] {(n, r8, p)|n ∈ Action, p ∈ Parameter, r8 = has_parameter} ∪

[0090] {(o i , r9, oj )|o i ,o j ∈Operation, r9 = after} ∪

[0091] {(n i , r9, n j )|n i , n j ∈Action, r9 = after}

[0092] From the perspectives of the composition of the knowledge graph, machine resources, processes, and modes, the flexible manufacturing process knowledge graph includes a composition view, a machine resource view, a process view, and a mode view.

[0093] Machine resource view: A robot is a unique concept in the field of flexible manufacturing. A robot includes an end effector and a robotic arm. One type of end effector can be installed on one or more robotic arms, and more than one type of end effector can also be matched to the end of one robotic arm. In the actual application in a factory, one end effector corresponds to one robotic arm, and the situation of infrequent replacement is more common. The machine resource view stores information about robots and reflects the matching relationship between end effectors and robotic arms. It is a resource library for flexible manufacturing robots. As Figure 3 shown, i1 and i3 represent end effector instances, and i2, i4, and i5 represent robotic arm instances. End effector i1 can be installed at the end of robotic arm i2, and end effector i3 can be installed on either robotic arm i4 or robotic arm i5. The main functions of the machine resource view are: retrieving the robotic arms that can be matched by a known end effector; retrieving the end effectors that can be installed at the end of a known robotic arm.

[0094] Process view: The process view is a sub - graph of the process knowledge graph segmented from the perspective of the process. During the product production process, factory workers generally complete relevant operations for product production through a process manual. In a process manual, the main objective of the operation, that is, the process objective, is usually introduced first. Then, for each process objective, several steps required to achieve this objective, that is, work steps, are introduced. Within one work step, the object being operated on and the tools used remain unchanged. The content of the work steps is described in detail in the process manual. In the process view of the flexible assembly process knowledge graph, the process objective is decomposed into work steps, and the work steps are executed sequentially according to the after relationship between work steps. The implementation of work steps is accomplished by robots. After splitting the basic actions of the robot, each work step can be achieved by the robot executing several actions. According to the after relationship between actions, the robot actions related to this work step are executed sequentially. When each action is executed, control parameter values need to be passed to the robot to achieve accurate flexible manufacturing. An example of the process view is as Figure 4 .

[0095] Composition View: The composition view can hierarchically summarize process knowledge. It analyzes the process knowledge stored in the process knowledge graph from the perspective of concept composition. The composition view can also represent the composition of a product, describe in detail the composition of the manufactured product, and includes a description of the product manufacturing process. The structure of the composition view is as Figure 5 .

[0096] Mode View: The mode view represents the ontology of the flexible manufacturing field. It reflects the logical relationship between concepts, capabilities, and process goals in the flexible manufacturing field. It also shows the way of connection between process goals and capabilities. The structure of the mode view is as Figure 6 .

[0097] Matching process goals in the flexible manufacturing process knowledge graph is divided into two steps. First, retrieve all the process goals of the two operating objects separately, and then find the intersection of the process goals of the two operating objects as the common process goal.

[0098] (2) Parameter Calculator: The parameter calculator is used to provide a set of key-value pairs consisting of multiple parameters and parameter values for the actions to be performed by the action actuator. The implementation of the parameter calculator combines the simple factory design pattern and the configuration file. The factory for producing the parameter calculator initializes the production resources required for the parameter calculator to calculate parameters. When the factory produces the parameter calculator, the retrieval of the configuration file is based on the information described by the nodes of the Parameter type in the flexible manufacturing process knowledge graph. The parameters are divided into two categories (parameters related to the robot and parameters related to production resources such as operating objects). Each type of parameter has a corresponding parameter calculator, and each parameter calculator needs to inherit the corresponding parent class. The parameters used by the two types of parameter calculators are slightly different. The parameter calculator for calculating parameters related to the robot needs to use the robotic arm and gripper resources, and the parameter calculator for calculating parameters related to production resources such as operating objects needs operating object resources, reference object resources, etc. The process of calculating parameters is an embodiment of making full use of the process mechanism.

[0099] In the present invention, the function of the parameter calculator is to provide accurate parameters for the action actuator to perform specific actions, which is related to the specific process and manufacturing resources. The parameter calculator is temporarily created according to the information described in the nodes of the Parameter type. In order to obtain the objects of the action actuator and the parameter calculator more abstractly, the simple factory design pattern is used and combined with a configuration file to complete. The parameters are divided into two major categories, namely, robot-related parameters (such as the speed and acceleration of joint movement, etc.) and resource-related parameters (such as the rotation angle and end pose, etc.). The specific parameter calculator inherits from the RobotParameterCalculator class (an abstract class representing the parameter calculator related to robot-related parameters) or the ObjectParameterCalculator class (an abstract class representing the parameter calculator related to production resources such as the object to be operated) according to the category of the parameter. In this way, the factory that produces the parameter calculator can allocate different resources to the parameter calculator according to the parameter category. The information stored in the nodes of the Parameter type in the knowledge graph includes the parameter name represented by the node. In the configuration file, there is configuration information with the parameter name as the key and the full class name of the parameter calculator used to calculate the parameter as the value. The factory that produces the parameter calculator will produce the parameter calculator corresponding to the node according to the configuration file, so as to calculate the parameter values required for the action execution. The inheritance system of the parameter calculator is as Figure 7 shown.

[0100] (3) Action actuator: The action actuator is used to specifically implement the robot actions; the implementation of the action actuator adopts the method of combining the simple factory design pattern and the configuration file; the action actuator is initialized by the factory that produces the action actuator to control the manipulator, fixture, etc. used for the action execution; when the factory produces the action actuator, the retrieval of the configuration file is carried out according to the information described in the nodes of the Action type in the flexible manufacturing process knowledge graph; each robot action corresponds to an action actuator, and each action actuator has to inherit an action actuator parent class; the parameter values of the parameters required by the method for executing the robot action in the action actuator are calculated by the parameter calculator. The inference of which types of parameters are required for each action actuator is realized according to the process view of the flexible manufacturing process knowledge graph related to a certain process target.

[0101] In the present invention, an action executor is used to specifically implement the actions described by the nodes of the Action type in the flexible assembly process knowledge graph. The action executor is temporarily created according to the information described by the nodes of the Action type, and is completed by using the simple factory design pattern and combining with a configuration file. By creating a base class (ActionActuator) to represent the action executor of the abstract actions of the robot, the action executors of each action type of the robot are used as subclasses of this base class. For a robot with a gripper that can be opened and closed at the end, the actions of the robot are abstracted into six types, namely opening the gripper, closing the gripper, rotating the gripper (at the end), continuously rotating and moving the gripper, moving according to the end, and flipping the end face. These six actions are described by six classes respectively, and these classes all inherit the above base class. The information of the action type is provided in the nodes of the Action type. Through the configuration file, the class required to create the action executor representing a certain node can be retrieved. The configuration file serves as a connection bridge between the flexible manufacturing process knowledge graph and the action executor, and is used by the factory that produces the action executor to accurately produce the action executor for controlling the specified robot. The inheritance system of the action executor is as shown in Figure 7 shown.

[0102] According to the matched process objective, use the process view related to this process objective to plan the motion trajectory of the robot for executing the process. The steps are as follows:

[0103] Step S1: Retrieve the operations that need to be executed for this process according to the process objective, that is, retrieve the Operation nodes that have an operation relationship with this process objective, and use the topological sorting algorithm to sort according to the after relationship.

[0104] Step S2: Sequentially analyze each obtained Operation node to determine the production resources used in this operation, mainly including the controlled robot, the operation object, the capabilities of the operation object, etc.

[0105] Step S3: Allocate the resources matched in this operation to the factory of the project for producing the action executor and the production parameter calculator.

[0106] Step S4: Retrieve the robot actions related to the operation, that is, retrieve a set of Action nodes that have a has_action relationship with the Operation node, and use the topological sorting algorithm to sort according to the after relationship.

[0107] Step S5: Sequentially analyze the obtained actions, retrieve and calculate the parameters and parameter values required for each action, that is, retrieve a set of Parameter nodes that have a has_parameter relationship with this Action node.

[0108] Step S6: Traverse the obtained set of parameter nodes, and use the factory of the production parameter calculator to sequentially produce a parameter calculator for each parameter node that can calculate the parameter value expressed by the node.

[0109] Step S7: Use the factory of the production action executor to produce the action executor related to the Action node, input the obtained set of parameter key-value pairs related to the Action node into the action executor, and use the action executor to execute the robot action.

[0110] The continuity of the actions performed by multiple action executors is the action execution plan for achieving the process goal.

[0111] Through the coordinated cooperation of the flexible manufacturing process knowledge graph, action executors, and parameter calculators described above, flexible manufacturing of products for different manufacturing processes with a complete process can be achieved.

Claims

1. A method for constructing a flexible manufacturing process mechanism model for implementing process execution, characterized in that The model consists of three parts: a flexible manufacturing process knowledge graph, a parameter calculator, and an action executor, where: The flexible manufacturing process knowledge graph is composed of triples. The head entity and the tail entity correspond to Node, and the relationship corresponds to Relation. Among them, the categories of nodes represented by Node include Abstraction, Instance, Capability, Goal, Operation, Action, Parameter; the relationships represented by Relation are the relationships between nodes, and the relationship categories it contains are include, has_instance, has_capability, achieve_goal, operation, installed_on, has_action, has_parameter, after; among them, entities of the Operation type represent the production steps in the production process, entities of the Action type represent the actions of the robot, entities of the Parameter type represent the control parameters related to the robot in the production process, the include relationship represents the inclusion and composition relationship, the has_instance relationship represents the relationship between the abstract concept and the instance of this concept, the has_capability relationship indicates that a certain flexible manufacturing concept has a certain ability, the installed_on relationship indicates the installation relationship between the fixture and the robotic arm, the has_action relationship represents the relationship between the operation and the robot action, the has_parameter relationship represents the connection between the action and the parameter, and the after relationship represents the execution order of the operation and the action; From the perspective of the composition structure of the knowledge graph, the flexible manufacturing process knowledge graph includes an abstract concept layer, a concept instance layer, and a capability layer. From the perspectives of the composition, machine resources, process, and mode of the knowledge graph, it includes a composition view, a machine resource view, a process view, and a mode view; The parameter calculator is responsible for calculating the parameter values required for various robot actions according to the manufacturing resources and process mechanisms inferred from the flexible manufacturing process knowledge graph; The action executor is responsible for making specific implementations of the actions described by the nodes of the Action type in the flexible assembly process knowledge graph based on the knowledge inferred from the flexible manufacturing process knowledge graph and the parameter values calculated using the parameter calculator, obtaining the actions executable by the robot, and controlling the robot to execute the actions; The method includes the following steps: Step S1, construct the ontology of the flexible manufacturing process knowledge graph: Model industrial knowledge from the perspectives of the composition, process, resources, and mode of the process production equipment, and construct the ontology of the flexible manufacturing process knowledge graph according to the existing ontology construction method; Step S2, construct the flexible manufacturing process knowledge graph based on the ontology: Construct the flexible manufacturing process knowledge graph based on the AIC model, give specific instances to the concepts and relationships mentioned in the ontology, and write the entity classes corresponding to the flexible manufacturing process knowledge graph; Step S3. Use a parameter calculator to calculate control parameters: Define and configure a parameter calculator to calculate the parameters required for the inferred actions, and substitute a set of calculated parameters into the action executor to obtain an executable action. Step S4. Use an action executor to execute the robot actions: Define and configure an action executor to execute the inferred executable robot actions. Step S5. According to Steps S1 to S4, the flexible manufacturing process knowledge graph, parameter calculator, and action executor together form a flexible manufacturing process mechanism model. When a user inputs a manufacturing task, the flexible manufacturing process mechanism model is used to infer the motion trajectory of process execution. The specific steps are as follows: Step S51. Analyze the user input to obtain a task list required for the manufacturing task. Each task includes an operation object, a process name, and the robot to be used. Step S52. Process the task list in sequence. Finally, infer the operations to be performed for each task based on the process view. Before each operation is executed, perform a statistics of manufacturing resources and allocate the resources to the factory of the production parameter calculator and the production action executor. Step S53. Infer the action framework that should be executed for each completed operation. Step S54. Infer the parameters required for each action. Step S55. Use the factory of the production parameter calculator to produce a parameter calculator that can calculate each parameter, and calculate the parameter values based on the production resources. Step S56. Use the factory of the production action executor to produce an action executor that can execute each action. Determine the robot controlled by the action executor based on the production resources, and input a set of parameter key-value pairs related to the action to the action executor, and the action executor executes the action. Step S57. The continuous actions executed by multiple action executors under multiple operations form a motion plan for completing the process.

2. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, characterized in that The specific steps of Step S1 are as follows: Step S11. Determine that the domain of the flexible manufacturing process knowledge graph ontology is the flexible manufacturing domain, and use this ontology to perform process inference for flexible manufacturing. Step S12. Expand based on the AIC model: For the AIC model, retain the original entities and relationships of the model, and add entity types: Action representing the entity of an action, Parameter representing the entity of a parameter; add relationship types: installed_on, has_action, has_parameter, after, which represent the installation relationship of the robotic arm fixture, the association relationship between an operation and an action, the requirement relationship between an action and a parameter, and the sequential relationship between operations and between actions respectively. In addition, add a hidden relationship between a type of ability and the process mechanism of the process objective. Step S13. List the concepts of various entities in the flexible manufacturing domain, and define the classes representing the entities in the flexible manufacturing domain, the attributes of the classes, and the relationships between the classes.

3. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, characterized in that The specific steps of Step S3 are as follows: Step 31: Define the Parameter class for the parameter calculator of all types of parameters. Define the parameter calculator classes for robot-related parameters and resource-related parameters, both of which inherit from the Parameter class. Define the parameter calculator classes for specific different parameters and inherit the corresponding parent classes according to categories respectively, and write the implementation of calculating parameter values by the parameter calculator respectively. Step 32: Define a factory for the production parameter calculator. The defined factory retrieves the configuration file according to the input parameter type and produces a matching parameter calculator. Step 33: The parameter calculator calculates parameters based on the production resources. The parameters are divided into two major types, namely, the parameters related to the robot and the parameters related to the manufacturing resources.

4. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, wherein The specific steps of Step S4 are as follows: Step S41: Define the parent class of all action actuators, and create action actuators corresponding to different actions according to the categories of robot actions respectively. Step S42: Define a factory for the production of action actuators. The factory retrieves the configuration file according to the input action category and produces an action actuator that matches the action category. Step S43: The action actuator executes specific actions based on the parameter set calculated by the input parameter calculator.

5. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, characterized in that The abstract concept layer is represented as follows: Among them, the include relationship represents the inclusion and composition relationship. The abstract concepts are hierarchically divided through include, and the abstract capabilities are also represented by the include relationship to indicate more detailed capabilities. The has_capability relationship indicates that a certain concept of flexible manufacturing has a certain capability, and the installed_on relationship represents the installation relationship between the fixture and the robotic arm.

6. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, characterized in that The concept instance layer is represented as follows: Among them, the has_instance relationship represents the relationship between the abstract concept and the concept instance, and the include relationship is used to describe the composition relationship between instances.

7. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, characterized in that The capability layer is represented as follows: Among them, the entity of the Operation type represents the production step in the production process, the entity of the Action type represents the action of the robot, the entity of the Parameter type represents the control parameters related to the robot in the production process, the has_action relationship indicates the relationship between the operation and the robot action, the has_parameter relationship represents the connection between the action and the parameter, and the after relationship indicates the execution order of the operation and the action.

8. The method for constructing a flexible manufacturing process mechanism model for implementing process execution according to claim 1, characterized in that The machine resource view is used to describe the robot that executes the process path, obtain the fixture and the robotic arm at the end of the robot, store the information of the robot, and reflect the matching relationship between the fixture and the robotic arm; the process view is used to describe the process manual, from which process objectives and process execution steps, i.e., work steps, are obtained, and at the same time, the relationships between different types of actions of the robot and work steps, as well as the relationships between actions and parameters, are described; the mode view contains a mode overview among abstract concepts, instances, capabilities, and process objectives. There is a relationship r3 between the abstract concept and the capability, a relationship r4 between the abstract concept and the process objective, and there is an implicit association between the capability and the process objective through the process mechanism. The relationships among the abstract concept, the capability node, and between them map the relationships among the instances of the abstract concept and between the instance and the capability instance; The composition view is responsible for obtaining the data of the product's composition structure according to the product's composition structure.

Images