Servo motor and industrial robot

Abstract

This invention relates to a servo motor and an industrial robot, belonging to the field of intelligent drive technology for servo motors and industrial robots. The servo motor includes a motor body and further includes: a multimodal sensing module for forming a four-domain sensing network (electromagnetic, thermal, mechanical, and vibration) to integrate electromagnetic, thermal, mechanical, and vibration monitoring of the motor body; a multiphysics field construction module for 3D reconstruction of the motor based on CT scans to construct a multiphysics field coupling model of electromagnetic, thermal, and structural fields; a fault diagnosis module for identifying fault types through the multiphysics field construction module; and a maintenance strategy optimization module for tracing the vibration propagation path in a digital mirror when a fault is detected; simulating the impact of different maintenance schemes on the servo motor's operation through the digital mirror, and selecting the optimal scheme for output. The industrial robot includes a robot body and the aforementioned servo motor.

Description

Technical Field

[0001] This invention relates to a servo motor and an industrial robot, belonging to the field of intelligent drive technology for servo motors and industrial robots. Background Technology

[0002] The core components of industrial robots can be likened to the three major systems of the human body: servo motors act as the "power muscles," reducers as the "circulatory system," and controllers as the "nerve center." Among these, the servo motor system forms the foundation of the robot's motion execution. Its precision drive units are embedded in the joints of the robotic arm, directly determining the boundaries of the device's motion performance. Currently, industrial robot joint drives rely on servo systems for precise control. As application scenarios demand increasingly more compliant operation and positioning accuracy, the technical specifications of servo motors are showing an exponential growth trend. Specifically, this manifests in the need for instantaneous high output torque to cope with sudden load changes, millisecond-level dynamic response capabilities to ensure trajectory tracking accuracy, excellent torque-inertia matching characteristics for agile start-stop control, and an ultra-wide speed range to meet the demands of cross-condition operations from low-speed, high-precision to high-speed, high-efficiency. These technological breakthroughs collectively support the evolution of modern industrial robots towards greater intelligence and dexterity.

[0003] The following problems are commonly found in existing industrial robot servo motors:

[0004] Delay in fault diagnosis: Traditional detection relies on manual inspection or offline testing, which cannot capture hidden faults under dynamic operating conditions in real time;

[0005] Parameter configuration error: Core parameters such as the number of motor pole pairs and the number of encoder lines rely on manual calibration, which can easily lead to control inaccuracies due to installation deviations;

[0006] Insufficient environmental adaptability: The lack of a dynamic compensation mechanism for environmental factors such as vibration and temperature affects the robot's repeatability and positioning accuracy.

[0007] Complex structure assembly and disassembly: Traditional connection methods rely on welding or complex bolt fixing, resulting in high maintenance costs.

[0008] For example, Chinese Patent Publication No. CN115683235A discloses a method and device for detecting vibration faults in an industrial robot servo motor. The method includes: simulating and experimenting with vibration faults of the servo motor under different operating conditions and fault causes in both a simulation environment and a laboratory environment, acquiring corresponding multi-sensor measurement signals; establishing an SAE neural network model; using the multi-sensor measurement signals acquired in the simulation environment to form an unlabeled first multi-dimensional signal sample set, and initializing and training the SAE neural network model; using the multi-sensor measurement signals acquired in the laboratory environment to form a labeled second multi-dimensional signal sample set, and fine-tuning the SAE neural network model; controlling the servo motor under test to operate according to test examples corresponding to different operating conditions, and inputting the collected multi-dimensional signal samples into the SAE neural network model for vibration anomaly detection. This method can achieve automatic and accurate detection of servo motor vibration faults and analysis of fault causes. However, the fault detection is singular and still relies on manual inspection or offline testing, failing to "capture latent faults under dynamic operating conditions in real time." Summary of the Invention

[0009] The purpose of this invention is to provide a servo motor and industrial robot that solves the problem of delayed fault diagnosis and inability to capture latent faults under dynamic working conditions in real time.

[0010] The servo motor of the present invention includes a motor body and further includes;

[0011] Multimodal sensing module: used to form a four-domain sensing network of electromagnetic, thermal, mechanical and vibration, integrating electromagnetic, thermal and mechanical and vibration monitoring of the motor body;

[0012] Multiphysics construction module: used for 3D reconstruction of motors based on CT scans, constructing a multiphysics coupled model of electromagnetic field, thermal field, and structural field;

[0013] Fault diagnosis module: used to identify fault types by constructing modules using multiphysics.

[0014] Maintenance strategy optimization module: When a fault is detected, it traces the vibration propagation path in a digital mirror; it simulates the impact of different maintenance schemes on the operation of the servo motor through a digital mirror, and selects the optimal scheme for output.

[0015] A multimodal sensing network is employed, fusing sensors from electromagnetic, thermal, mechanical, and vibration domains to achieve comprehensive monitoring of motor operating status, avoiding blind spots in single-parameter monitoring and improving the timeliness of fault early warning. A multiphysics coupling model is constructed, based on 3D reconstruction technology from CT scans, to build an electromagnetic-thermal-structural field coupling model, accurately simulating the complex physical field interactions within the motor, providing theoretical support for design optimization and fault prediction. A fault diagnosis closed-loop system, combining the multiphysics model with real-time sensor data, enables intelligent identification of fault types, reducing manual intervention and improving diagnostic accuracy. Digital mirroring technology is used to trace fault paths and simulate maintenance plans, selecting the optimal strategy to significantly reduce unplanned downtime and improve production efficiency.

[0016] Preferably, the multimodal sensing module includes:

[0017] Electromagnetic field sensor: installed at the end of the stator winding, used to measure the air gap magnetic flux density distribution in real time;

[0018] Temperature sensor: installed on the stator windings to measure temperature data in real time;

[0019] Stress sensor: installed in the yoke of the stator core for real-time measurement of stress distribution;

[0020] Vibration sensor: installed at the end of the output shaft of the motor body, used to measure vibration characteristics in real time.

[0021] Electromagnetic field sensors monitor the air gap magnetic flux density distribution in real time, optimizing electromagnetic design parameters, reducing the risk of magnetic circuit saturation, and improving motor efficiency. Temperature sensors accurately measure stator winding temperature, preventing insulation failure due to overheating and extending motor life. Stress sensors monitor stator core stress distribution, preventing mechanical fatigue and avoiding performance degradation caused by structural deformation. Vibration sensors capture output shaft vibration characteristics, detecting bearing wear or imbalance problems early and reducing the probability of sudden failures.

[0022] Preferably, the multiphysics construction module includes:

[0023] Electromagnetic field model building unit: used for building a three-dimensional electromagnetic field mathematical model of a motor based on CT scan;

[0024] Temperature model building unit: used to build a three-dimensional thermal field mathematical model of the motor based on CT scan;

[0025] Stress model building unit: used for building a three-dimensional mathematical model of the stress field of a motor based on CT scans;

[0026] Vibration model building unit: used to build a three-dimensional mathematical model of the vibration field of a motor based on CT scan;

[0027] Multiphysics building blocks: used to couple electromagnetic field mathematical models, thermal field mathematical models, stress field mathematical models, and vibration field mathematical models, and to define boundary conditions and initial conditions;

[0028] Model validation unit: used to calibrate model parameters using physical test data;

[0029] Dynamic optimization unit: used to iteratively update the model through real-time sensor data to achieve virtual-real mutual control.

[0030] The electromagnetic field model, based on a 3D model obtained from CT scans, accurately calculates the magnetic field distribution, optimizes coil layout, and reduces eddy current losses. The thermal field model predicts temperature gradient distribution, guiding the optimization of heat dissipation structures and preventing localized overheating. The stress field model analyzes areas of mechanical stress concentration, optimizing material selection and structural design to improve fatigue resistance. The vibration field model simulates vibration propagation paths, optimizing vibration reduction design and reducing noise and mechanical wear. Model validation and dynamic optimization calibrate model parameters using physical test data and iteratively update them using real-time sensor data, achieving virtual-real interoperability and adapting to dynamic operating conditions. Fault diagnosis is rapid, capable of capturing latent faults under dynamic operating conditions in real time.

[0031] Preferably, the multiphysics construction unit specifically includes:

[0032] Coupled variable definition sub-unit: used to define coupled variables to achieve data transfer between different physical fields. Coupled variables include temperature, stress, displacement, electromagnetic force, and vibration acceleration.

[0033] Boundary Condition and Load Setting Sub-element: Used to set boundary conditions and loads, which are configured based on the interactions and influences between different physical fields.

[0034] (1) Coupling of electromagnetic field and temperature field: The heat generated by electromagnetic field serves as the heat source of temperature field;

[0035] (2) Coupling of temperature field and stress field: thermal expansion or contraction caused by temperature change acts as a load on the stress field.

[0036] (3) Coupling of stress field and vibration field: Structural deformation caused by stress serves as the initial or boundary condition of the vibration field;

[0037] (4) Coupling of vibration field and electromagnetic field: dynamic inductance change caused by vibration;

[0038] Mesh generation and solver configuration sub-element: used to mesh the fully coupled model and configure the solver;

[0039] Solution and Result Analysis Sub-unit: Used to obtain the results of the fully coupled model, and to analyze and evaluate the accuracy and reliability of the model.

[0040] The coupling variables are defined to clarify key coupling variables such as temperature, stress, and displacement, enabling multi-physics data interaction and improving simulation accuracy. Coupling relationships between electromagnetic fields and temperature fields, and between temperature fields and stress fields, are defined to simulate physical interactions under actual working conditions. Mesh density and solver algorithms are optimized to balance computational efficiency and accuracy, accelerating design iterations. Solving and result analysis sub-elements evaluate model accuracy, identify potential design flaws, and guide parameter optimization.

[0041] Preferably, the electromagnetic field model establishment unit specifically includes:

[0042] Electromagnetic field geometric modeling subunit: used to build a 3D model of a motor based on CT scans, set up analysis modules, and create analysis geometric models;

[0043] Electromagnetic field meshing sub-elements: used to discretize a geometric 3D model into finite elements or meshes;

[0044] Electromagnetic field solver settings sub-element: used to select the solver type, set the excitation source and boundary conditions;

[0045] Electromagnetic field solution and post-processing sub-unit: used to solve for electromagnetic field distribution and visualize the results;

[0046] Electromagnetic field verification and optimization subunit: used to verify the accuracy of the model by comparing experimental data or analytical solutions, and to optimize parameters to improve performance.

[0047] Preferably, the fault diagnosis module diagnoses motor faults based on deep learning.

[0048] By automatically learning fault characteristics through deep learning algorithms, manual intervention is reduced, and diagnostic robustness is improved. It adapts to different operating conditions and fault types, reducing reliance on expert experience and enabling cross-scenario applications.

[0049] Preferably, the fault diagnosis module adopts an improved one-dimensional convolutional neural network, with the input layer consisting of electromagnetic-thermal-mechanical-vibration four-domain sensing signals, the intermediate layer consisting of 3 convolutional layers and 2 LSTM layers, and the output layer consisting of a Softmax classifier.

[0050] The input layer integrates electromagnetic, thermal, mechanical, and vibration signals for four-domain signal fusion, providing comprehensive diagnostic information and avoiding information silos. Three convolutional layers extract local features, and two LSTM layers capture temporal dependencies, improving fault classification accuracy through spatiotemporal feature capture. The output layer's Softmax classifier clarifies the probability of fault types, providing quantitative support for maintenance decisions.

[0051] Preferably, the maintenance strategy optimization module includes:

[0052] Fault tracing unit: used to trace the vibration propagation path in a digital mirror when a motor fault is detected;

[0053] Strategy optimization unit: used to simulate the impact of different maintenance plans on the production line through digital mirroring;

[0054] Output unit: Used to select the optimal solution for output;

[0055] Adaptive control unit: Used to automatically reduce speed or stop the machine when a motor fault is detected.

[0056] By tracing the vibration propagation path in a digital mirror, the source of the fault can be quickly located, shortening troubleshooting time. The impact of different maintenance plans on the production line can be simulated using digital mirrors to select the strategy with the lowest cost and highest efficiency. An executable maintenance plan is generated by combining the fault type with production needs, minimizing downtime losses. Automatic speed reduction or shutdown is implemented in the event of a fault to prevent the fault from escalating and ensure the safety of personnel and equipment.

[0057] The industrial robot described in this invention includes a robot body and a servo motor as described above.

[0058] By integrating the aforementioned servo motors, industrial robots can achieve real-time monitoring and multi-physics field coupling analysis across four domains: electromagnetic, thermal, mechanical, and vibration. This enables precise control of joint motion parameters (such as torque, speed, and position), significantly improving operational accuracy and dynamic response speed.

[0059] Compared with existing technologies, the advantages of the servo motor and industrial robot of the present invention are as follows:

[0060] The beneficial effects of the servo motor of the present invention are as follows:

[0061] 1. Leap in technical performance

[0062] Multi-dimensional perception and precise control achieve comprehensive monitoring of motor operating status through a four-domain perception network of electromagnetic, thermal, force and vibration. Combined with a multi-physics coupling model, it can accurately predict electromagnetic force distribution, temperature gradient, stress concentration and vibration propagation path, thereby improving motor control accuracy and dynamic response speed, which is significantly better than traditional servo motors.

[0063] The electromagnetic field-thermal field-structural field coupling model is constructed based on CT scan three-dimensional reconstruction technology. It can simulate the motor behavior under extreme conditions (such as overload and high temperature), guide the selection of materials and structural optimization (such as the optimization of the thickness of silicon steel sheets in the stator core and the improvement of the heat dissipation fin layout), so as to improve the power density of the motor and reduce torque fluctuation.

[0064] 2. Enhanced maintenance reliability

[0065] By using a deep learning-based fault diagnosis module to analyze four-domain sensing signals in real time, faults such as bearing wear and insulation aging can be predicted in advance. Combined with digital mirroring technology to trace the vibration propagation path, the fault location time is shortened and the rate of unplanned downtime is reduced.

[0066] 3. Energy efficiency and environmental benefits

[0067] Multiphysics coupling models guide the optimization of servo motor parameters (such as current harmonic suppression and heat dissipation structure improvement), thereby improving the overall energy efficiency of the motor, reducing energy consumption and carbon emissions at the same power.

[0068] 4. Adaptive safety control

[0069] When a motor fault is detected, the adaptive control unit can automatically trigger a speed reduction (speed reduced to a safe threshold) or a shutdown protection to avoid safety accidents such as robot arm collisions and workpiece damage. At the same time, through the coupling analysis of electromagnetic field and vibration field, the vibration suppression strategy is optimized to reduce robot operating noise and improve the working environment.

[0070] The industrial robot described in this invention has the following beneficial effects:

[0071] 1. Enhancement of Fault Prediction and Health Management (PHM) Capabilities

[0072] Through the deep learning fault diagnosis module of servo motors, industrial robots can analyze four-domain perception signals in real time and predict typical faults such as bearing wear and insulation aging 15-30 days in advance. Combined with digital mirroring technology, the vibration propagation path can be traced back to locate the fault source, improving efficiency and significantly reducing the risk of unplanned downtime.

[0073] 2. Maintenance cost and cycle optimization

[0074] The maintenance strategy optimization module supports simulating the impact of different maintenance plans on the production line (such as maintenance time and defect rate fluctuations) in a digital mirror. After selecting the optimal plan, maintenance time is shortened, spare parts inventory costs are reduced, and the total life cycle maintenance cost is reduced.

[0075] 3. Adaptive control and enhanced safety

[0076] When a motor fault is detected, the adaptive control unit can automatically trigger a speed reduction (speed reduced to a safe threshold) or a shutdown protection to avoid safety accidents such as robot arm collisions and workpiece damage. At the same time, through the coupling analysis of electromagnetic field and vibration field, the vibration suppression strategy is optimized to reduce robot operating noise.

[0077] 4. Improved energy efficiency and reliability

[0078] Multiphysics coupling models guide the optimization of servo motor parameters (such as current harmonic suppression and heat dissipation structure improvement), thereby improving the overall energy efficiency of industrial robots, extending MTBF (mean time between failures), and adapting to high-load, continuous production scenarios.

[0079] 5. Enhanced design iteration and customization capabilities

[0080] Based on CT scan-based 3D reconstruction and dynamic optimization units, industrial robot manufacturers can quickly verify the impact of new joint designs (such as lightweight materials and modular structures) on overall performance, shorten product development cycles, and support the rapid customization needs of flexible production lines for robot functions.

[0081] 6. Intelligent production network integration

[0082] Digital mirroring technology for servo motors can be seamlessly integrated with factory MES / ERP systems to enable cloud-based monitoring and collaborative scheduling of robot health status and maintenance plans, providing key equipment-level data support for industrial internet platforms and promoting the transformation to intelligent manufacturing. Attached Figure Description

[0083] Figure 1 This is a structural block diagram of a servo motor according to the present invention. Detailed Implementation

[0084] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0085] Example 1

[0086] like Figure 1 As shown, this embodiment discloses a servo motor, including a motor body, and also including;

[0087] Multimodal sensing module: used to form a four-domain sensing network of electromagnetic, thermal, mechanical and vibration, integrating electromagnetic, thermal and mechanical and vibration monitoring of the motor body;

[0088] Multiphysics construction module: used for 3D reconstruction of motors based on CT scans, constructing a multiphysics coupled model of electromagnetic field, thermal field, and structural field;

[0089] Fault diagnosis module: used to identify fault types by constructing modules using multiphysics.

[0090] Maintenance strategy optimization module: When a fault is detected, it traces the vibration propagation path in a digital mirror; it simulates the impact of different maintenance schemes on the operation of the servo motor through a digital mirror, and selects the optimal scheme for output.

[0091] Fault diagnosis has evolved from "passive response" to "proactive prevention"; control performance has evolved from "single operating condition adaptation" to "multi-process intelligent switching"; maintenance strategies have evolved from "regular replacement" to "on-demand maintenance"; and system reliability has evolved from "single-machine redundancy" to "end-to-end fault tolerance".

[0092] The multimodal sensing module includes:

[0093] Electromagnetic field sensor: installed at the end of the stator winding to measure the air gap magnetic flux density distribution in real time; specifically, three miniature Hall sensors (range ±5T, resolution 1mT) are embedded at the end of the stator winding to measure the air gap magnetic flux density distribution in real time.

[0094] Temperature sensor: installed on the stator windings, used to measure temperature data in real time; specifically,

[0095] Stress sensor: installed in the stator core yoke for real-time measurement of stress distribution; specifically, fiber optic grating sensor (FBG, wavelength accuracy 1pm / ℃) is embedded in the stator core yoke, and stress distribution is inverted using the finite element inverse model; NTC thermistor (accuracy ±0.5℃) is embedded in the stator winding, and hot spot temperature is calculated using the heat conduction model.

[0096] Vibration sensor: installed at the end of the output shaft of the motor body, used to measure vibration characteristics in real time. Specifically, a triaxial MEMS accelerometer is integrated at the end of the motor output shaft, and the vibration characteristic frequency band is extracted by wavelet packet decomposition.

[0097] The multiphysics construction module includes:

[0098] Electromagnetic field model building unit: used for building a three-dimensional electromagnetic field mathematical model of a motor based on CT scan;

[0099] Temperature model building unit: used to build a three-dimensional thermal field mathematical model of the motor based on CT scan;

[0100] Stress model building unit: used for building a three-dimensional mathematical model of the stress field of a motor based on CT scans;

[0101] Vibration model building unit: used to build a three-dimensional mathematical model of the vibration field of a motor based on CT scan;

[0102] Multiphysics building blocks: used to couple electromagnetic field mathematical models, thermal field mathematical models, stress field mathematical models, and vibration field mathematical models, and to define boundary conditions and initial conditions;

[0103] Model validation unit: used to calibrate model parameters using physical test data;

[0104] Dynamic optimization unit: used to iteratively update the model through real-time sensor data to achieve virtual-real mutual control.

[0105] Electromagnetic field modeling steps

[0106] Model building: Select the solver and coordinate system according to the analysis object, and use the software's drawing function to build the model or import a CAD model. Taking a three-phase asynchronous motor as an example, it is necessary to construct the stator core, windings, rotor core, windings, motion domain (Band), and solution domain (Region).

[0107] Adding Materials: Select the component in the project tree, access the material library through the Properties interface, and assign the corresponding material to each component. If no suitable option is available in the material library, new materials must be added according to the rules.

[0108] Preprocessing settings:

[0109] Motion settings: Select the Band field, assign the Band in the Model and perform MotionSetup. Set the Type, Data, and Mechanical tabs according to your needs. When considering the startup process, check ConsiderMechanicalTransient.

[0110] Boundary condition settings: Select boundary conditions according to the characteristics of the model. For example, right-click the graph and select SelectEdges, then select the edges of the Region and select VetorPotential in Boundaries to set them.

[0111] Excitation Source Settings: Set the excitation source according to the solver type and excitation source type. Taking a winding as an example, select the winding in the project tree, right-click and select Coil to set up CoilExcitation. Add the winding and configure the power supply. Set the iron loss and eddy current loss solving options according to your needs.

[0112] Mesh partitioning settings: Select the component to be partitioned in the project tree and perform mesh partitioning.

[0113] Solver settings: In the project management panel, right-click on Analysis and select AddSolution Setup, then configure the settings according to the different solvers.

[0114] Post-processing: Review the electromagnetic field parameters and field quantity results, paying particular attention to the force results and field distribution diagrams. Analyze the accuracy of the simulation data using electromagnetic theory, and accumulate experience to solve potential problems.

[0115] Temperature field modeling steps

[0116] 3D modeling: Determine the work title and analysis heading, set up the analysis modules, and create the analysis geometry model. For example, when simulating the temperature field of the surrounding rock in a tunnel, create a rectangular model of the tunnel and the surrounding rock, and add monitoring points.

[0117] Element selection: Select the appropriate element type and options based on the analysis requirements.

[0118] Determining physical property parameters: Define the thermal properties of the material, such as thermal conductivity, specific heat capacity, and density.

[0119] Mesh generation: The model is meshed. A structured mesh can be used initially to ensure computational accuracy and speed, while paying attention to the number and quality of the meshes. To improve computational efficiency, the model can be simplified, such as by scaling down the full model proportionally.

[0120] Solver configuration: Select a suitable solver and configure relevant parameters, such as time unit, output time step, etc.

[0121] Heat source model loading: Set the heat source according to the actual situation, such as setting an internal forced convection boundary to simulate the influence of airflow on the surrounding rock, including parameters such as wind speed and wind temperature.

[0122] Results and data output: The calculation is run and the results, such as temperature field distribution and isotherms, are output. The results are analyzed to show the changes in surrounding rock temperature with time and location under different parameter conditions.

[0123] Stress field modeling steps

[0124] Establish a geological geometric model: Discretize the planar structure, characterize different structural zones, select key nodes and connect them to form the model outline.

[0125] Assign cell properties: Select the appropriate cell type, such as the 8node 183 cell, which has many advantages such as hyperelasticity and plasticity.

[0126] Define model material properties: Assign corresponding rock parameters to different regions, such as density, Young's modulus, and Poisson's ratio.

[0127] Unit mesh generation: Set different unit attributes according to the study area, specify the unit shape for mesh generation, and re-mesh the key areas to improve simulation accuracy.

[0128] Applying loads and constraints: Add reasonable constraint boundaries based on the actual geological background, and combine the plate movement trajectory and rate to transform displacement loads into stress loads, applying appropriate stress magnitude and direction.

[0129] Simulation Result Analysis: After setting the boundaries and the magnitude and direction of stress, the simulation results are calculated and the corresponding content is selected to generate a vector diagram, such as displacement field, stress field, deformation field, etc., to analyze the kinematic characteristics of the plate.

[0130] Vibration field modeling steps

[0131] Geometric model creation and simplification: Define the analysis objectives, create a geometric model, and simplify it according to the actual situation to improve computational efficiency.

[0132] Material properties and boundary conditions definition: Define material properties for each component in the model, such as elastic modulus and density; set reasonable boundary conditions, such as fixed supports and elastic supports, to simulate actual constraint conditions.

[0133] Meshing and quality control: Select an appropriate meshing strategy, mesh the model, and evaluate the mesh quality to ensure that the mesh can accurately describe the geometry and physical properties of the model.

[0134] Stimulus and response analysis:

[0135] Excitation input method: Set the excitation source according to the actual situation, such as harmonic force, impact load, etc., and determine the input position and method of the excitation.

[0136] Response extraction and analysis: Run the simulation, extract the vibration response of the model, such as displacement, velocity, acceleration, etc., and analyze the vibration characteristics.

[0137] Simulation verification of control effect:

[0138] The steps for simulation testing are as follows: conduct simulation testing and record relevant data.

[0139] Post-processing and evaluation of results: Post-process the simulation results, such as plotting frequency response curves and mode shapes, evaluate the control effect, and adjust and optimize the model based on the evaluation results.

[0140] The multiphysics building blocks are selected from simulation software interfaces or modules that support multiphysics coupling. Currently, many mature simulation software programs on the market support multiphysics coupling modeling, such as COMSOL Multiphysics, ANSYS, and ABAQUS. These software programs provide rich physics interfaces and coupling tools, facilitating data transfer and interaction between different physics fields. Specifically, this includes:

[0141] Coupled variable definition sub-unit: used to define coupled variables to achieve data transfer between different physical fields. Coupled variables include temperature, stress, displacement, electromagnetic force, and vibration acceleration.

[0142] The coupling variables are defined to clarify key coupling variables such as temperature, stress, and displacement, enabling multi-physics data interaction and improving simulation accuracy. Coupling relationships between electromagnetic fields and temperature fields, and between temperature fields and stress fields, are defined to simulate physical interactions under actual working conditions. Mesh density and solver algorithms are optimized to balance computational efficiency and accuracy, accelerating design iterations. Solving and result analysis sub-elements evaluate model accuracy, identify potential design flaws, and guide parameter optimization.

[0143] Boundary Condition and Load Setting Sub-element: Used to set boundary conditions and loads, which are configured based on the interactions and influences between different physical fields.

[0144] (1) Coupling of electromagnetic field and temperature field: The heat generated by electromagnetic field serves as the heat source of temperature field;

[0145] (2) Coupling of temperature field and stress field: thermal expansion or contraction caused by temperature change acts as a load on the stress field.

[0146] (3) Coupling of stress field and vibration field: Structural deformation caused by stress serves as the initial or boundary condition of the vibration field;

[0147] (4) Coupling of vibration field and electromagnetic field: dynamic inductance change caused by vibration;

[0148] In coupled models, boundary conditions and loads need to be set appropriately. Boundary conditions refer to the constraints on physical quantities or their derivatives on the model boundaries, such as fixed constraints, symmetric boundaries, periodic boundaries, etc. Loads refer to external forces or excitations acting on the model, such as electromagnetic forces, heat flux, pressure, etc. In coupled models, the setting of boundary conditions and loads needs to take into account the interactions and influences between different physical fields.

[0149] Mesh generation and solver configuration sub-element: used to mesh the fully coupled model and configure the solver;

[0150] Mesh the coupled model and configure the solver. Mesh generation is the process of discretizing the geometric model into finite element methods or meshes. The quality and density of the mesh directly affect computational accuracy and efficiency. In coupled models, the interactions between different physical fields may concentrate in certain regions, thus requiring mesh refinement in these regions to improve computational accuracy. Solver configuration refers to selecting appropriate solver types and parameters to ensure the stability and convergence of the coupled model.

[0151] Solution and Result Analysis Sub-unit: Used to obtain the results of the fully coupled model, and to analyze and evaluate the accuracy and reliability of the model.

[0152] Run simulations to obtain and analyze the results of the coupled model. The results may include information from multiple physical fields, such as electromagnetic field distribution, temperature distribution, stress distribution, and vibration characteristics. Analyzing these results allows us to understand the interactions and influences between different physical fields, assess the accuracy and reliability of the model, and provide a scientific basis for engineering design and optimization.

[0153] Preferably, the fault diagnosis module diagnoses motor faults based on deep learning.

[0154] By automatically learning fault characteristics through deep learning algorithms, manual intervention is reduced, and diagnostic robustness is improved. It adapts to different operating conditions and fault types, reducing reliance on expert experience and enabling cross-scenario applications.

[0155] Preferably, the fault diagnosis module adopts an improved one-dimensional convolutional neural network, with the input layer consisting of electromagnetic-thermal-mechanical-vibration four-domain sensing signals, the intermediate layer consisting of 3 convolutional layers and 2 LSTM layers, and the output layer consisting of a Softmax classifier.

[0156] The input layer integrates electromagnetic, thermal, mechanical, and vibration signals for four-domain signal fusion, providing comprehensive diagnostic information and avoiding information silos.

[0157] Three convolutional layers extract local features, and two LSTM layers capture temporal dependencies, thus improving fault classification accuracy through spatiotemporal feature capture. The output layer's Softmax classifier is used to define the probability of fault types, providing quantitative support for maintenance decisions.

[0158] The maintenance strategy optimization module includes:

[0159] Fault tracing unit: used to trace the vibration propagation path in a digital mirror when a motor fault is detected;

[0160] Strategy optimization unit: used to simulate the impact of different maintenance plans on the production line through digital mirroring;

[0161] Output unit: Used to select the optimal solution for output;

[0162] Adaptive control unit: Used to automatically reduce speed or stop the machine when a motor fault is detected.

[0163] By tracing the vibration propagation path in a digital mirror, the source of the fault can be quickly located, shortening troubleshooting time. The impact of different maintenance plans on the production line can be simulated using digital mirrors to select the strategy with the lowest cost and highest efficiency. An executable maintenance plan is generated by combining the fault type with production needs, minimizing downtime losses. Automatic speed reduction or shutdown is implemented in the event of a fault to prevent the fault from escalating and ensure the safety of personnel and equipment.

[0164] For example, bearing failure prediction and maintenance decision-making.

[0165] Fault evolution process:

[0166] T=0h: Pitting with a diameter of 0.1mm and a depth of 0.02mm appears on the outer ring of the bearing;

[0167] T=100h: The energy of the vibration signal in the 850Hz frequency band increases by 12dB, and the 1D-CNN model predicts a failure probability of 85%;

[0168] T=150h: The fiber optic grating sensor detected a core stress of 75MPa (close to the warning threshold of 80MPa).

[0169] Digital Twin Decision Making:

[0170] Simulate two scenarios in a digital mirror:

[0171] Option A: Replace the bearing immediately (downtime 2 hours, cost $500);

[0172] Option B: Continue operation until T=200h (downtime risk probability 30%, downtime loss $20,000).

[0173] Optimal decision: Choose option A, which reduces the production line OEE loss from 2.5% to 0.3%.

[0174] Example 2

[0175] The industrial robot described in this invention includes a robot body and a servo motor as described above.

[0176] By integrating the aforementioned servo motors, industrial robots can achieve real-time monitoring and multi-physics field coupling analysis across four domains: electromagnetic, thermal, mechanical, and vibration. This enables precise control of joint motion parameters (such as torque, speed, and position), significantly improving operational accuracy and dynamic response speed.

[0177] The above description is only a preferred embodiment of this practice, but the scope of protection of this embodiment is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the scope of the technology disclosed in this embodiment, based on the technical solution and inventive concept of this embodiment, should be covered within the scope of protection of this embodiment.

Claims

1. A servo motor comprising a motor body, characterized by, Also includes; Multimodal sensing module: used to form a four-domain sensing network of electromagnetic, thermal, mechanical, and vibration, integrating electromagnetic, thermal, mechanical, and vibration monitoring of the motor body; the multimodal sensing module includes: Electromagnetic field sensor: installed at the end of the stator winding, used to measure the air gap magnetic flux density distribution in real time; Temperature sensor: installed on the stator windings to measure temperature data in real time; Stress sensor: installed in the yoke of the stator core for real-time measurement of stress distribution; Vibration sensor: installed at the end of the output shaft of the motor body, used to measure vibration characteristics in real time; Multiphysics construction module: used for 3D reconstruction of motors based on CT scans, constructing a multiphysics coupled model of electromagnetic field, thermal field, and structural field; the multiphysics construction module includes: Electromagnetic field model building unit: used for building a three-dimensional electromagnetic field mathematical model of a motor based on CT scan; Temperature model building unit: used to build a three-dimensional thermal field mathematical model of the motor based on CT scan; Stress model building unit: used for building a three-dimensional mathematical model of the stress field of a motor based on CT scans; Vibration model building unit: used to build a three-dimensional mathematical model of the vibration field of a motor based on CT scan; Multiphysics building blocks: used to couple electromagnetic field mathematical models, thermal field mathematical models, stress field mathematical models, and vibration field mathematical models, and to define boundary conditions and initial conditions; Model validation unit: used to calibrate model parameters using physical test data; Dynamic optimization unit: used to iteratively update the model through real-time sensor data to achieve virtual-real mutual control; Fault diagnosis module: used to identify fault types by constructing modules using multiphysics. Maintenance strategy optimization module: When a fault is detected, it traces the vibration propagation path in a digital mirror; it simulates the impact of different maintenance schemes on the operation of the servo motor through a digital mirror, and selects the optimal scheme for output.

2. A servo motor according to claim 1, characterized in that, The multiphysics construction unit specifically includes: Coupled variable definition sub-unit: used to define coupled variables to achieve data transfer between different physical fields. Coupled variables include temperature, stress, displacement, electromagnetic force, and vibration acceleration. Boundary Condition and Load Setting Sub-element: Used to set boundary conditions and loads, which are configured based on the interactions and influences between different physical fields. (1) Coupling of electromagnetic field and temperature field: The heat generated by electromagnetic field serves as the heat source of temperature field; (2) Coupling of temperature field and stress field: thermal expansion or contraction caused by temperature change acts as a load on the stress field; (3) Coupling of stress field and vibration field: Structural deformation caused by stress serves as the initial or boundary condition of vibration field; (4) Coupling of vibration field and electromagnetic field: dynamic inductance change caused by vibration; Mesh generation and solver configuration sub-element: used to mesh the fully coupled model and configure the solver; Solution and Result Analysis Sub-unit: Used to obtain the results of the fully coupled model, and to analyze and evaluate the accuracy and reliability of the model.

3. A servo motor according to claim 1, characterized in that, The electromagnetic field model establishment unit specifically includes: Electromagnetic field geometric modeling subunit: used to build a 3D model of a motor based on CT scans, set up analysis modules, and create analysis geometric models; Electromagnetic field meshing sub-elements: used to discretize a geometric 3D model into finite elements or meshes; Electromagnetic field solver settings sub-element: used to select the solver type, set the excitation source and boundary conditions; Electromagnetic field solution and post-processing sub-unit: used to solve for electromagnetic field distribution and visualize the results; Electromagnetic field verification and optimization subunit: used to verify the accuracy of the model by comparing experimental data or analytical solutions, and to optimize parameters to improve performance.

4. A servo motor according to claim 1, characterized in that, The fault diagnosis module diagnoses motor faults based on deep learning.

5. A servo motor according to claim 1, characterized in that, The fault diagnosis module adopts an improved one-dimensional convolutional neural network. The input layer is a four-domain sensing signal of electromagnetic, thermal, force and vibration. The middle layer contains 3 convolutional layers and 2 LSTM layers. The output layer is a Softmax classifier.

6. A servo motor according to claim 1, characterized in that, The maintenance strategy optimization module includes: Fault tracing unit: used to trace the vibration propagation path in a digital mirror when a motor fault is detected; Strategy optimization unit: used to simulate the impact of different maintenance plans on the production line through digital mirroring; Output unit: Used to select the optimal solution for output; Adaptive control unit: Used to automatically reduce speed or stop the machine when a motor fault is detected.

7. An industrial robot, comprising a robot body, characterized in that, It also includes a servo motor as described in any one of claims 1-6.

Images