A digital twin model construction method for a collaborative robot joint

Abstract

The application discloses a kind of digital twin model construction methods for collaborative robot joint, comprising: geometry model construction module, mainly according to the geometric feature of robot joint, realize from part level, component level and product level etc. Dimensional construction geometry model;Rule model construction module, mainly realize the control rule extraction of robot joint control program and the construction of key part running damage rule;Physical model construction module, mainly realize the establishment of robot joint dynamics equation and key part temperature field;Behavior model construction module, mainly realize the construction of robot joint behavior tree and the solution of behavior corresponding parameter change curve;Model fusion module, mainly through data lake and asynchronous communication interface technology realizes the fusion between four-dimensional models.The present application can realize from geometry-physical-behavior-rule four-dimensional construction robot joint digital twin model, provide twin model support for robot joint fault diagnosis based on digital twin.

Description

Technical Field

[0001] This invention belongs to the fields of mechanical engineering and computer science, and specifically relates to a method for constructing digital twin models of collaborative robot joints. Background Technology

[0002] Currently, collaborative robots are widely used in various fields. However, their operation is affected by factors such as harsh working environments, sudden environmental changes, and component aging, leading to performance degradation, threatening the safe and stable operation of collaborative robots, and consequently causing production interruptions, increased maintenance costs, and safety risks. In industrial scenarios, signal-based deep learning has become the mainstream method for robot fault diagnosis. However, deep learning methods face challenges such as the diversity of fault modes and the difficulty in obtaining fault data. This invention discloses a method for constructing a full-dimensional digital twin model for collaborative robot joints. By incorporating fault knowledge into the digital twin modeling process, it provides digital twin model support to address the difficulty in obtaining robot joint fault data to a certain extent. Summary of the Invention

[0003] The technical problem to be solved by this invention is to provide a method for constructing digital twin models for collaborative robot joints, providing model support for solving the problem of difficulty in obtaining robot joint fault data using digital twin technology.

[0004] The technical problem solved by this invention is achieved by the following technical solution: a method for constructing digital twin models of collaborative robot joints, characterized in that:

[0005] Step (1): Construct the geometric model of the collaborative robot joints, the specific implementation of which is as follows:

[0006] Step (1.1): Construct the robot joint part-level geometric model: Based on the role of the parts in mechanical motion, they are divided into key parts (CP). i and auxiliary parts AP i The key components are precisely modeled according to their shape, material properties, and surface features, while the auxiliary components are modeled according to their shape and their mating relationship with the key components, resulting in the set of all components {CP}. i AP i The key components refer to those that directly participate in the mechanical motion process and have a direct impact on the motion, including the frameless motor bearing CP1 and several reducer gears CP. 21 CP 22 ...CP 2nThe auxiliary parts refer to components in the robot joint structure that play an auxiliary role to key components and have no direct impact on the mechanical movement process, including the frameless motor housing AP1, the frameless motor support AP2, the reducer support AP3, the controller AP4, and several sensors AP1. 51 AP 52 ...AP 5m Motor heat sink AP6, joint housing AP7, and several wiring harnesses AP 81 AP 82 ...AP 8k ;

[0007] Step (1.2): Construct the robot joint component-level geometric model: Based on the functional partitions of the robot joints, construct the part set {CP} built in step (1.1). i AP i The components are divided into a motor component set MC = {CP1, AP1, AP2, AP6} and a reducer component set RC = {AP3, CP6}. 21 CP 22 ...CP 2n}、Controller component set CC = {AP4, AP 81 ...AP 8i}, Sensor component set SC = {AP 51 ...AP 5m AP 8i+1 ...AP 8k Within each set, the part-level models are assembled into component-level models according to their mating and positional relationships.

[0008] Step (1.3): Constructing the product-level geometric model of the robot joint: Based on the position of each component in the joint shell, the mechanical transmission relationship between the components and the wiring harness connection relationship, assemble the four component-level geometric models constructed in step (1.2) into a product-level geometric model.

[0009] Step (2): Construct a rule model for the joints of the collaborative robot, the specific implementation of which is as follows:

[0010] Step (2.1): Constructing the control rule model for robot joints: Based on the control strategy of the robot joint program on the range and rate of change of robot joint parameters under external disturbances and its own faults, construct n "IF-THEN" rules R. i (i = 1, 2, ..., n).

[0011] Step (2.2): Constructing the output torque attenuation rule model of the robot joint: Based on the Archard wear equation (V = kFvd), construct the output torque attenuation rule model of the robot joint, where V is the wear volume of the part, k is the wear coefficient of the key part material, F is the surface contact pressure between the parts; v is the relative motion speed between the parts, d is the sliding distance between the parts; and the torque attenuation τ weak =μV, where μ is the torque attenuation coefficient.

[0012] Step (3): Construct the physical model of the collaborative robot joints, the specific implementation of which is as follows:

[0013] Step (3.1): Construct the torque dynamics equation of the robot joint: based on the Lagrange equation (L=T-U+τ) total Construct the torque dynamics equations for the robot joints, where T is the kinetic energy term, U is the potential energy term, and τ is the torque dynamics equation. total For output torque, τ total =τ gear -τ weak , τ gear The output torque of the reducer; τ weak The effect of wear on output torque is calculated using the torque attenuation rule model established in step (2.2).

[0014] Step (3.2): Constructing the temperature field of key components of the robot joint: Calculating the temperature field T on the surface of the key components based on the thermal network method. CP The calculation basis of the heat network method is the heat conduction equation, initial conditions, and heat exchange boundary conditions. The heat conduction equation is: Where x is the spatial coordinate, k(x) is the thermal conductivity, T(x, t) is the temperature distribution, ρ(x) is the density, C(x) is the specific heat capacity, and t is time; the initial condition is T(x, t=0)=T sensor T sensor The temperature is collected by the sensor; the heat exchange boundary conditions between the joints and key components are... Where h conv T is the thermal conductivity coefficient. CP Temperature of critical components;

[0015] Step (4): Construct the behavior model of the collaborative robot joints, the specific implementation of which is as follows:

[0016] Step (4.1): Constructing robot joint behavior tree nodes: Based on the behavior of robot joints under various conditions such as external disturbances and internal malfunctions, construct robot joint behavior tree nodes. The tree nodes include trigger nodes Ti, behavior nodes Bi, and execution nodes Pi. A trigger node represents a condition that can trigger a behavior, a behavior node represents a specific behavior, and an execution node represents a part within a robot joint.

[0017] Step (4.2): Connect robot joint behavior tree nodes: Using behavior node Bi as the intermediate layer, connect it upwards to trigger node Ti and downwards to execution node Pi. The connection between the trigger node and the behavior node indicates that the condition represented by the trigger node will trigger the behavior represented by the behavior node. The connection between the behavior node and the execution node indicates that the behavior represented by the behavior node is participated in by the part represented by the execution node.

[0018] Step (5): Integrate the geometry-physics-behavior-rules multidimensional model, the specific implementation of which is as follows:

[0019] Based on data lake technology, the data of the four models (geometric, physical, behavioral, and rule-based) constructed in steps (1) to (4) are classified and stored. An event-driven asynchronous communication interface is constructed between the multidimensional models and the data lake. The event types of the event-driven communication are data update events and data retrieval events. Data update events refer to the data updates performed by the four-dimensional models and the calculation of new data by the data processing program using the data managed in the current data lake. Data retrieval events refer to the requests made by the four-dimensional models to retrieve data managed in the data lake.

[0020] Furthermore, in step (5), based on data lake technology, the data of the four models constructed in steps (1) to (4)—geometric, physical, behavioral, and rule-based—are classified and stored. Specifically, the data of the geometric model includes information related to the shape, structure, and spatial position of each part of the robot joint. The shape of the model is stored using a 3D mesh, the spatial structure of the part-component-product level model is stored using point cloud, and the entity of the model is stored using CAD model format. The data of the physical model includes the material properties, mechanical characteristics, physical parameters, torque kinematic equations, and temperature field equations of each part of the robot joint. The tabular data of material properties and mechanical parameters is stored using tables, and XML is used to store the data. The data storage includes hierarchical physical model information, torque kinematic equations, and temperature field equations. A graph database is used to store the relationships between the physical models. The data of the behavior model includes descriptions of the motion, actions, and reactions of the robot joints. Event logs are used to store the behavioral events and state changes of the equipment. A time-series database is used to store time-related behavioral data. A graph database is used to store the behavior tree nodes and connections constructed in step (4). The data of the rule model includes descriptions of robot joint operating procedures and control rules. A text file is used to store the rule documents and operating procedures. A relational database is used to store the n rules established in step (2.1). A graph database is used to store the relationships between the rules.

[0021] The advantages of this invention compared to the prior art are:

[0022] This invention proposes a method for constructing digital twin models of collaborative robot joints. In the modeling process, it considers the control rules of parameters and the response behavior of the equipment when the joint is subjected to external disturbances and its own failure. It constructs a rule model for torque loss caused by wear of transmission parts during operation and establishes a four-dimensional digital twin model of collaborative robot joints that includes fault knowledge. It can simultaneously support the simulation generation of highly accurate data of collaborative robot joints under multiple conditions such as normal operation, disturbance, and failure. Attached Figure Description

[0023] Figure 1 This is a structural block diagram of the present invention;

[0024] Figure 2 This is a schematic diagram of the connected behavior tree. Detailed Implementation

[0025] This invention relates to a method for constructing digital twin models of collaborative robot joints, such as... Figure 1 As shown, its implementation method is as follows:

[0026] A method for constructing a digital twin model of a collaborative robot joint, comprising the following steps:

[0027] Step (1): Construct the geometric model of the collaborative robot's joints:

[0028] Step (1.1): Construct the robot joint part-level geometric model: Based on the role of the parts in mechanical motion, they are divided into key parts (CP). i and auxiliary parts AP i The key components are precisely modeled according to their shape, material properties, and surface features, resulting in the final product. Auxiliary components are modeled according to their shape and their mating relationship with the key components, resulting in the complete set of components {CP}. i AP i}

[0029] The key components refer to those that directly participate in the mechanical motion process and have a direct impact on the motion, including the frameless motor bearing CP1 and several reducer gears CP. 21 CP 22 ...CP 2n The auxiliary parts refer to components in the robot joint structure that play an auxiliary role to key components and have no direct impact on the mechanical movement process, including the frameless motor housing AP1, the frameless motor support AP2, the reducer support AP3, the controller AP4, and several sensors AP1. 51 AP 52 ...AP 5m Motor heat sink AP6, joint housing AP7, and several wiring harnesses AP 81 AP 82 ...AP 8k ;

[0030] Step (1.2): Construct the robot joint component-level geometric model: Based on the functional partitions of the robot joints, construct the part set {CP} built in step (1.1). i AP i The components are divided into a set of motor components MC = (CP1, AP1, AP2, AP6) and a set of reducer components RC = {AP3, CP6}. 21 CP 22 ...CP 2n}、Controller component set CC = {AP4, AP 81 ...AP 8i}, Sensor component set SC = {AP 51 ...AP 5m AP 8i+1 ...AP 8k Within each set, the part-level models are assembled into component-level models according to their mating and positional relationships.

[0031] The motor component is responsible for providing power and is assembled according to the contact surface relationship between the heat sink and the motor housing, the rigidity relationship between the support and the motor housing, and the axial alignment of the motor bearing and the motor housing; the reducer component is responsible for torque conversion and is assembled according to the meshing clearance between gears, axial alignment, radial alignment, helical fit, and gear center distance; the controller component is responsible for realizing the control function and is assembled according to the wiring harness routing and connector position; the sensor component is responsible for realizing the detection function and is assembled according to the sensor orientation angle and the rigidity relationship between the support structure and the sensor.

[0032] Step (1.3): Constructing the robot joint product-level geometric model: Based on the positions of each joint component within the joint housing, the mechanical transmission relationships between components, and the wiring harness connection relationships, assemble the four component-level geometric models constructed in step (1.2) into a product-level geometric model. Based on the absolute positions of the motor component, reducer component, controller component, and sensor component relative to the joint housing, their relative positions, and their orientations, fix the four components to the housing; based on the mechanical transmission relationships between the motor component and the reducer component, connect them according to axial and radial alignment to form a motor-reducer connector; based on the signal and power transmission relationships between the controller component, sensor component, and motor component, connect the signal lines and power lines.

[0033] Step (2): Construct a rule model for the joints of the collaborative robot, the specific implementation of which is as follows:

[0034] Step (2.1): Constructing the control rule model for robot joints: Based on the control strategy of the robot joint program for the range and rate of change of robot joint parameters under multiple conditions such as external disturbances and internal faults, construct n "IF-THEN" rules R. i (i = 1, 2, ..., n), f i It is a control function for the parameters. The external disturbances refer to sudden load changes, unexpected collisions, power outages, etc.; the internal faults refer to sensor malfunctions, abnormal motor output torque, abnormal reducer output torque, etc.

[0035] Step (2.2): Constructing the output torque attenuation rule model of the robot joint: Based on the Archard wear equation (V = kFvd), construct the output torque attenuation rule model of the robot joint, where V is the volume of wear of the part, k is the wear coefficient of the key part material, F is the surface contact pressure between the parts, obtained from the geometric model established in step (1); v is the relative motion speed between the parts, d is the sliding distance between the parts, calculated from the rotational speed relationship between the gear set and the output rotational speed detected by the encoder, and the torque attenuation τ weak =μV, where μ is the torque attenuation coefficient.

[0036] Step (3): Construct the physical model of the collaborative robot joints, the specific implementation of which is as follows:

[0037] Step (3.1): Construct the torque dynamics equation of the robot joint: based on the Lagrange equation (L=T-U+τ) total Construct the torque dynamics equations for the robot joints, where T is the kinetic energy term, calculated from the robot joint angular velocity and moment of inertia; u is the potential energy term, calculated from the joint mass; τ is the torque dynamics equation. total For output torque, τ total =τ gear -τ weak , τ gear The output torque of the reducer is the torque τ output by the motor. motor Calculate the transmission ratio ε and efficiency ∈ of the reducer gear set; τ weak The effect of wear on output torque is calculated using the torque attenuation rule model established in step (2.2).

[0038] Step (3.2): Constructing the temperature field of key components of the robot joint: Calculating the temperature field T on the surface of the key components based on the thermal network method. CP The calculation basis of the heat network method is the heat conduction equation, initial conditions, and heat exchange boundary conditions. The heat conduction equation is: Where x is the spatial coordinate, k(x) is the thermal conductivity, T(x, t) is the temperature distribution, ρ(x) is the density, C(x) is the specific heat capacity, and t is time; the initial condition is T(x, t=0)=T sensor T sensor The temperature is collected by the sensor; the heat exchange boundary conditions between the joints and key components are... Where h conv T is the thermal conductivity coefficient. CP Temperature of critical components;

[0039] Step (4): Construct the behavior model of the collaborative robot joints, the specific implementation of which is as follows:

[0040] Step (4.1): Constructing robot joint behavior tree nodes: Based on the behavior of robot joints under various conditions such as external disturbances and internal malfunctions, construct robot joint behavior tree nodes. The tree nodes include trigger nodes Ti, behavior nodes Bi, and execution nodes Pi. A trigger node represents a condition that can trigger a behavior, a behavior node represents a specific behavior, and an execution node represents a part within a robot joint.

[0041] Step (4.2): Connect the robot joint behavior tree nodes: Using behavior node Bi as the intermediate layer, connect it upwards to trigger node Ti and downwards to execution node Pi. The connection between a trigger node and a behavior node indicates that the condition represented by the trigger node will trigger the behavior represented by that behavior node. The connection between a behavior node and an execution node indicates that the behavior represented by the behavior node is participated in by the part represented by that execution node. The behavior tree after connection is as follows: Figure 2 As shown.

[0042] Step (5): Integrate the geometry-physics-behavior-rules multidimensional model, the specific implementation of which is as follows:

[0043] Step (5.1) uses data lake technology to store the data of the four models constructed in steps (1) to (4): geometric, physical, behavioral, and rule-based models. The data of the geometric model includes information related to the shape, structure, and spatial position of each part of the robot joint. The shape of the model is stored using a 3D mesh, the spatial structure of the part-component-product level model is stored using point cloud, and the entities of the model are stored using CAD model formats (such as STL, OBJ, etc.). The data of the physical model includes the material properties, mechanical characteristics, physical parameters, torque kinematic equations, and temperature field equations of each part of the robot joint. The tabular data such as material properties and mechanical parameters are stored using tables, and the hierarchical structure is stored using XML. The physical model information, torque kinematic equations, and temperature field equations are stored in a graph database to store the relationships between the physical models; the behavioral model data includes descriptions of the motion, actions, and reactions of the robot joints, uses an event log to store the behavioral events and state changes of the equipment, uses a time series database to store time-related behavioral data (such as the motion trajectory of the equipment), and uses a graph database to store the behavioral tree nodes and connections constructed in step (4); the rule model data includes descriptions of robot joint operating procedures, control rules, etc., uses a text file to store rule documents and operating procedures, uses a relational database to store the n rules established in step (2.1), and uses a graph database to store the relationships between the rules.

[0044] Step (5.2) Construct an event-driven asynchronous communication interface between the multidimensional model and the data lake: Define the event triggering types as data update events and data call events. Data update events refer to the four-dimensional model updating data and the data processing program using the data managed in the current data lake to calculate new data. Data call events refer to the four-dimensional model making a call request to the data managed in the data lake.

[0045] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0046] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for constructing a digital twin model of a collaborative robot joint, characterized in that, The method steps are as follows: Step (1): Construct the geometric model of the collaborative robot joints, the specific implementation of which is as follows: Step (1.1): Construct the robot joint part-level geometric model: Based on the role of the parts in mechanical motion, they are divided into key parts (CP). i and auxiliary parts AP i For key components CP i Modeling is performed based on shape, material properties, and surface features for auxiliary parts (AP). i Modeling based on its shape and its mating relationship with key components yields the complete set of parts {CP}. i AP i The key components refer to those that directly participate in the mechanical motion process and have a direct impact on the motion, including the frameless motor bearing CP1 and several reducer gears CP. 21 CP 22 ...CP 2n The auxiliary parts refer to components in the robot joint structure that play an auxiliary role to key components and have no direct impact on the mechanical movement process, including the frameless motor housing AP1, the frameless motor support AP2, the reducer support AP3, the controller AP4, and several sensors AP1. 51 AP 52 ...AP 5m Motor heat sink AP6, joint housing AP7, and several wiring harnesses AP 81 AP 82 ...AP 8k ; Step (1.2): Construct the robot joint component-level geometric model: Based on the functional partitions of the robot joints, construct the part set {CP} built in step (1.1). i AP i The components are divided into a motor component set MC = {CP1, AP1, AP2, AP6} and a reducer component set RC = {AP3, CP6}. 21 CP 22 ...CP 2n }、Controller component set CC = {AP4, AP 81 ...AP 8i }, Sensor component set SC = {AP 51 ...AP 5m AP 8i+1 ...AP 8k Within each set, the part-level models are assembled into component-level models according to their mating and positional relationships. Step (1.3): Construct a product-level geometric model of the robot joint: Based on the position of each component in the joint shell, the mechanical transmission relationship between the components and the wiring harness connection relationship, assemble the four component-level geometric models constructed in step (1.2) into a product-level geometric model; Step (2): Construct a rule model for the joints of the collaborative robot, the specific implementation of which is as follows: Step (2.1): Constructing the control rule model for robot joints: Based on the control strategy of the robot joint program on the range and rate of change of robot joint parameters under external disturbances and its own faults, construct n "IF-THEN" rules R. i (i = 1, 2...n); Step (2.2): Constructing the output torque attenuation rule model of the robot joint: Based on the Archard wear equation V = kFvd, construct the output torque attenuation rule model of the robot joint, where V is the volume of wear of the part, k is the wear coefficient of the key part material, F is the surface contact pressure between the parts; v is the relative motion speed between the parts, d is the sliding distance between the parts; and the torque attenuation τ weak =μV, where μ is the torque attenuation coefficient; Step (3): Construct the physical model of the collaborative robot joints, the specific implementation of which is as follows: Step (3.1): Construct the torque dynamics equation of the robot joint: based on the Lagrange equation L=T-U+τ total Construct the torque dynamics equations for the robot joints, where T is the kinetic energy term, U is the potential energy term, and τ is the torque dynamics equation. total For output torque, τ total =τ gear -τ weak , τ gear The output torque of the reducer; τ weak The effect of wear on output torque is calculated using the torque attenuation rule model established in step (2.2); Step (3.2): Constructing the temperature field of key components of the robot joint: Calculating the temperature field T on the surface of the key components based on the thermal network method. CP The calculation basis of the heat network method is the heat conduction equation, initial conditions, and heat exchange boundary conditions. The heat conduction equation is: Where x is the spatial coordinate, k(x) is the thermal conductivity, T(x, t) is the temperature distribution, ρ(x) is the density, C(x) is the specific heat capacity, and t is time; the initial condition is T(x, t=0)=T sensor T sensor The temperature is collected by the sensor; the heat exchange boundary conditions between the joints and key components are... Where h conv T is the thermal conductivity coefficient. CP Temperature of critical components; Step (4): Construct the behavior model of the collaborative robot joints, the specific implementation of which is as follows: Step (4.1) Constructing robot joint behavior tree nodes: Based on the behavior of robot joints under multiple conditions such as external disturbances and their own failures, construct robot joint behavior tree nodes; the tree nodes include trigger nodes Ti, behavior nodes Bi and execution nodes Pi. A trigger node represents a condition that can trigger a behavior, a behavior node represents a specific behavior, and an execution node represents a part within a robot joint. Step (4.2): Connect robot joint behavior tree nodes: With behavior node Bi as the intermediate layer, connect trigger node Ti upwards and execution node Pi downwards; the connection between trigger node and behavior node indicates that the condition represented by the trigger node will trigger the behavior represented by the behavior node, and the connection between behavior node and execution node indicates that the behavior represented by the behavior node is participated in by the part represented by the execution node. Step (5): Integrate the geometry-physics-behavior-rules multidimensional model, the specific implementation of which is as follows: Based on data lake technology, the data of the four models (geometric, physical, behavioral, and rule) constructed in steps (1) to (4) are classified and stored; an event-driven asynchronous communication interface is constructed between the multidimensional model and the data lake, wherein the event types of the event-driven communication are data update events and data call events; Data update events refer to data updates performed by the four-dimensional models, as well as data processing programs using the data managed in the current data lake to calculate new data. Data retrieval events refer to data retrieval requests made by the four-dimensional models to the data managed in the data lake.

2. The method according to claim 1, characterized in that, In step (5), based on data lake technology, the data of the four models constructed in steps (1) to (4) – geometric, physical, behavioral, and rule – are classified and stored. The specific storage method is as follows: the data of the geometric model includes information related to the shape, structure and spatial position of each part of the robot joint. The shape of the model is stored using a three-dimensional mesh, the spatial structure of the part-component-product level model is stored using point cloud, and the entity of the model is stored using CAD model format; the data of the physical model includes the material properties, mechanical properties, physical parameters, torque kinematic equations and temperature field equations of each part of the robot joint. The tabular data of material properties and mechanical parameters is stored using tables, the hierarchical physical model information, torque kinematic equations and temperature field equations are stored using XML, and the relationships between the physical models are stored using a graph database; the data of the behavior model includes the description of the motion, action and reaction of the robot joint. The behavior events and state changes of the equipment are stored using an event log, the time-related behavior data is stored using a time series database, and the behavior tree nodes and connections constructed in step (4) are stored using a graph database; the data of the rule model includes the description of the robot joint operation procedures and control rules. The rule documents and operation procedures are stored using text files, the n rules established in step (2.1) are stored using a relational database, and the relationships between the rules are stored using a graph database.

Images