Processing process fault diagnosis method based on digital twinning

Abstract

The invention belongs to the technical field of fault diagnosis, and discloses a processing process fault diagnosis method based on digital twinning. Aiming at the problems of real-time monitoring and error traceability of existing fault diagnosis, the invention establishes a machining process fault diagnosis system based on digital twinning by analyzing the motion state and error transmission mode of each motion part of a machine tool in the machining process, and makes full use of an ultra-precise mechanical structure, a high-precision grating ruler and a high-performance motor servo system of the machine tool. The synchronization of the digital twin model and the operation process is realized through the data of the digital twin system of the machine tool, the real-time monitoring of the processing state and the diagnosis and traceability of the fault are realized, and any external sensor is not needed. The method has the advantages of non-contact, real-time monitoring, low cost, high precision and the like, error traceability can be realized through signal analysis of the digital twin system based on real-time data, the generation reason of the workpiece error can be determined, and a theoretical basis is provided for further improving the product precision and the product quality.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault diagnosis and relates to a machining process fault diagnosis method based on digital twin. Background Art

[0002] "Intelligence" is the inevitable trend of the development of manufacturing technology in the future, and intelligent manufacturing is its core. In automated manufacturing systems, continuous monitoring of processing status is particularly important. High-quality, efficient and online processing status monitoring systems have received more and more attention in the research of industry and manufacturing in recent years. As modern manufacturing systems develop in the direction of large-scale and complex, research and development of fault diagnosis technology to ensure the reliability, maintainability and safety of manufacturing system operation has become one of the hot topics closely watched at home and abroad. Production systems that run for a long time or under high load will inevitably produce various problems, even system crashes, which in turn cause safety problems and cause economic losses. Fault diagnosis technology can help discover potential hidden dangers in advance and avoid the above problems.

[0003] However, the specific manifestations of signals of different mechanical faults are different, and the methods of fault diagnosis are also diverse. There are certain difficulties in signal processing and information acquisition. There are still deficiencies in the diagnosis of early faults and weak and complex faults. However, the combination of processing systems and digital twins is a potential solution to improve the degree of automation and move towards intelligent manufacturing. However, there is currently a lack of in-depth modeling methods for digital twins. Many studies focus on the implementation and testing of artificial intelligence components. Studies have found that due to the opacity and unexplainability of artificial intelligence, the reasoning mechanism of digital twin models built using data-based methods has not been fully understood or explained by physics. Digital twin models are regarded as "black boxes" due to their lack of interpretability. Their interpretability is often poor when performing fault diagnosis, their generalization ability is questionable, and it is difficult to trace the source of the fault. Summary of the invention

[0004] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a method for diagnosing machining process faults based on digital twins. Aiming at the real-time monitoring and error tracing problems of existing fault diagnosis, the present invention establishes a digital twin system framework by analyzing the motion state and error transmission mode of each moving part of the machine tool during the machining process, and proposes a method for diagnosing machining process faults based on digital twins. The present invention makes full use of the ultra-precision mechanical structure, high-precision grating scale, and high-performance motor servo system of the machine tool, and realizes the synchronization of the digital twin model and the operation process through the data of the machine tool digital twin system, realizing real-time monitoring of the machining state and diagnosis and tracing of faults without any external sensors.

[0005] The technical solution of the present invention:

[0006] A method for fault diagnosis of machining processes based on digital twins has established a real-time visual detection framework of machining status driven by digital twins based on the real-time signals inside the machine tool. By extracting the internal control and feedback signals of the machine tool, a digital twin system of the CNC grinder is established to reflect the real-time working conditions without the need for any external sensors. It has the advantages of non-contact, real-time monitoring, low cost, and high precision, and realizes real-time and accurate visual monitoring of the machine tool status. The location analysis of product errors is realized. The digital twin system based on real-time signals provides decision-making assistance for the actual machining system by combining the simulation model with the operation data. By analyzing the signals of the digital twin system based on real-time data, error tracing can be achieved, and the causes of workpiece errors can be clarified, which provides a theoretical basis for further improving product precision and improving product quality. The specific steps are as follows:

[0007] Step 1: Kinematic modeling of each moving part of the machine tool

[0008] The moving parts of the machine tool are the main motion actuators that are closely related to the machining process. They are divided into motion systems based on linear motors and motion systems based on rotary motors. The motion systems based on linear motors include the machine tool X-axis and the machine tool Z-axis, and the motion systems based on rotary motors include the workpiece spindle or C-axis and the grinding axis. The kinematic model of the moving parts of the machine tool includes the control and feedback parameters of the main feed mechanism of the machine tool, and is mainly composed of three sub-models, namely the motor, the motor controller, and the load model. The electrical signal provided by the controller drives the motor to overcome the load movement. The electromechanical model of the linear motor motion system under load can be expressed as:

[0009]

[0010] Among them, F load is the real-time load, x(t) is the real-time position of the moving part, (t) is the first-order derivative of the real-time position of the moving part, M is the total mass of the moving part, B is the viscosity coefficient of the hydraulic guide, K i is the thrust coefficient of PMLSM, i q is the output current of the servo drive, and the real-time position of the motor is expressed as:

[0011]

[0012] Where C is the DC component, N(t) is the random noise, f0, A0, φ0 are the frequency, amplitude and initial phase of the motor fundamental vibration, respectively, and f i , A i ,φ i are the frequency, amplitude and phase of the non-fundamental frequency vibration of the motor respectively, and n is the order of the non-fundamental frequency vibration of the motor.

[0013] The dynamic model of the rotating motor motion system under load can be expressed as follows:

[0014]

[0015] Among them, the motor inertia J1 and the damping coefficient B1 are constants; θ is the real-time rotation angle of the motor, and the load torque is regarded as a disturbance that reduces the spindle speed; but the motor controller can adjust the real-time torque T1 of the motor by adjusting the current i(t) supplied to the motor, and then respond to the load, and the motor output torque T under load out It can be expressed as:

[0016] T out ( t ) = T 0 ( t ) + ∑ i = 1 n T n ( t )cos[ ∫ 2 π f n ( t ) dt + ϕ n ( t )]

[0017] Based on this, the motion model of each moving part of the machine tool can be established.

[0018] Step 2: Digital twin system of machining process based on kinematic model

[0019] According to the kinematic model of each moving part of the machine tool in step one, a digital twin system of the machining process can be established: in addition to external noise, the workpiece is installed on the C-axis, and the C-axis speed is relatively low, so the main component of the C-axis frequency domain is the low-frequency component; the C-axis is installed on the Z-axis guide, so the main component of the Z-axis frequency domain is the external noise and the low-frequency component of the workpiece; the grinding axis speed is much higher than the C-axis speed, and the grinding axis is installed on the X-axis guide, so the main components of the grinding axis and X-axis frequency domain are high-frequency components.

[0020] The five-dimensional architecture of the digital twin system includes physical entity PE, virtual entity VM, service SS, twin data DD and the links CN between the components, which can be expressed as:

[0021]

[0022] Among them, CNCGM is a CNC grinding machine, DT is a digital twin system; the virtual entity VM contains the kinematic models of the moving parts of the machine tool, which can be expressed as follows according to the load model in equations (1) and (3):

[0023]

[0024] Among them, F x Indicates the real-time load of the X-axis linear motion system, F z Indicates the real-time load of the Z-axis linear motion system, T CIndicates the real-time load of the workpiece spindle\C-axis rotation motion system, T G Represents the real-time load of the grinding axis rotary motion system; based on this, the moving parts are integrated into the digital twin system:

[0025]

[0026] Among them, P x Indicates the real-time power of the X-axis linear motion system, P z Indicates the real-time power of the Z-axis linear motion system, P C Indicates the real-time power of the workpiece spindle\C-axis rotary motion system, P G Indicates the real-time power of the grinding axis rotary motion system; v x Indicates the real-time speed of the X axis, v z Indicates the real-time speed of the Z axis, w c Indicates the real-time angular velocity of the workpiece spindle or C-axis, w g Indicates the real-time angular velocity of the grinding axis.

[0027] Step 3: Time-frequency domain analysis of the processing process

[0028] The digital twin system established based on the machining process can realize real-time monitoring of the machining process. By extracting the real-time position of each moving part of the machine tool in the digital twin system and comparing it with the command position of the moving parts of the machine tool during the machining process, the impact of the motion error during the machine tool movement on the workpiece machining process can be obtained:

[0029]

[0030] Among them, x instruction 、z instruction 、c instruction are the command positions of the XZC coordinates in the NC program after being discretized by the machine tool NC program, and x actual 、z actual 、c actual They are the actual positions of the XZC coordinates in the machine tool coordinate system during processing, and the difference between the two is x pe 、z pe 、c pe These are the effects on the three coordinates X, Z and C during the machining process.

[0031] Frequency domain analysis is an important technique in mechanical fault diagnosis, which reveals the frequency components and frequency response of signals by converting time domain signals into frequency domain signals. In machine tool fault diagnosis, frequency domain analysis can help identify and locate faults, because different fault types will show different characteristics in the frequency domain.

[0032] Frequency domain analysis can be achieved through Fourier transform, expressed as:

[0033]

[0034] Among them, k represents the corresponding moving part, F k (ω) is the transformed spectrum function, f k (t) is the original time domain signal, ω is the angular frequency, e -jωt is a complex exponential function. By multiplying and integrating the time domain signal with a series of complex exponential functions, we can obtain the spectrum distribution of the signal at different frequencies. Based on this, we can perform frequency domain analysis on the power signal during product processing, combine the processing parameters and the corresponding machine tool motion actuators, and determine the source of the processing error and the early manifestation of the fault. Through the frequency domain analysis results, the characteristic values ​​of the digital twin system signal during the processing process can be extracted, thereby realizing real-time monitoring of the processing process and making timely adjustments.

[0035] Step 4: Fault diagnosis and error location

[0036] While the digital twin system monitors the machining process, it can also realize fault diagnosis and error location of the machining process. Through frequency domain analysis, the machining error and characteristic value distribution law can be quickly obtained, and compared with the characteristic frequency signal of the main moving parts of the machine tool for fault diagnosis. System failure does not appear suddenly, but gradually occurs as the machining progresses and the parts and tools operate. These characteristics are often difficult to reflect in the time domain signal. Through frequency domain analysis, the early manifestation of the fault in the frequency domain can be obtained more sensitively, realizing early fault diagnosis and daily maintenance of equipment. In addition, the fault diagnosis results can also be compared with the physical structure of the main moving parts to realize error location, reduce the system error in the machining process, and ensure the stable operation of the product machining process.

[0037] Beneficial effects of the present invention: The method of the present invention does not require the installation of additional sensors and can be integrated into any open CNC system. In an open CNC system such as PMAC, by extracting CNC servo control signals such as current, voltage, power, position signals, etc., the same effect as the present invention can be achieved. The higher the accuracy of the CNC system, the higher the monitoring accuracy of the processing state. By selecting appropriate filtering, noise reduction and fitting algorithms, the monitoring accuracy of the processing process will be further improved, greatly broadening the scope of application of the method in industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of the machining process fault diagnosis method based on digital twin.

[0039] Figure 2It is a diagram of the error transmission relationship between the moving parts of the machine tool during the machining process.

[0040] Figure 3 It is a schematic diagram of the five-dimensional architecture of the digital twin machine tool.

[0041] Figure 4 It is the surface error distribution result of the cylindrical workpiece obtained based on the digital twin system.

[0042] Figure 5 It is the distribution of the workpiece surface error along the circumferential direction after the cylindrical surface is unfolded based on the digital twin system.

[0043] Figure 6 It is the frequency domain analysis result of the X-axis frequency signal based on the digital twin system frequency signal.

[0044] Figure 7 It is the frequency domain analysis result of the Z-axis frequency signal based on the digital twin system frequency signal.

[0045] Figure 8 It is the frequency domain analysis result of the C-axis frequency signal based on the digital twin system frequency signal.

[0046] Fig. 9 This is a schematic diagram of the internal structure of the C-axis.

[0047] Fig.10 This is a schematic diagram of the C-axis slotless brushless ring motor and windings.

[0048] In the figure: 1 brushless motor; 2 shaft core; 3 reading head; 4 grating ruler; 5 vacuum suction cup; 6 motor coil wire; 7 Hall sensor wire; 8 thermistor wire; 9 Hall sensor; 10 positioning threaded hole; 11 rotor assembly; 12 stator assembly. DETAILED DESCRIPTION

[0049] The specific implementation of the present invention is described in detail below in conjunction with the technical scheme and the accompanying drawings.

[0050] Figure 1 This is a flow chart of the fault diagnosis method for the machining process based on digital twins of the present invention. The specific steps of the fault diagnosis method flow are as follows:

[0051] Step 1: Kinematic modeling of each moving part of the machine tool

[0052] The moving parts of the machine tool are the main motion actuators that are closely related to the machining process. They are divided into motion systems based on linear motors and motion systems based on rotary motors. The motion system based on linear motors includes the machine tool X-axis and the machine tool Z-axis, and the motion system based on rotary motors includes the workpiece spindle or C-axis and the grinding axis. The motion model mainly includes the control and feedback parameters of the main feed mechanism of the machine tool. The feed mechanism consists of three sub-models, namely the motor, the motor controller, and the load model. The electrical signal provided by the controller drives the motor to overcome the load movement. The electromechanical model of the linear motor motion system under load can be expressed as:

[0053]

[0054] Among them, F load is the real-time load, x(t) is the real-time position of the moving part, (t) is the first-order derivative of the real-time position of the moving part, M is the total mass of the moving part, B is the viscosity coefficient of the hydraulic guide, K i is the thrust coefficient of PMLSM, i q is the output current of the servo drive, and the real-time position of the motor is expressed as:

[0055]

[0056] Where C is the DC component, N(t) is the random noise, f0, A0, φ0 are the frequency, amplitude and initial phase of the motor fundamental vibration, respectively, and f i , A i ,φ i are the frequency, amplitude and phase of the non-fundamental frequency vibration of the motor respectively, and n is the order of the non-fundamental frequency vibration of the motor.

[0057] The dynamic model of the rotating motor motion system under load can be expressed as follows:

[0058]

[0059] Among them, the motor inertia J1 and the damping coefficient B1 are constants; θ is the real-time rotation angle of the motor, and the load torque is regarded as a disturbance that reduces the spindle speed; but the motor controller can adjust the real-time torque T1 of the motor by adjusting the current i(t) supplied to the motor, and then respond to the load, and the motor output torque T under load out It can be expressed as:

[0060] T out ( t ) = T 0 ( t ) + ∑ i = 1 n T n ( t )cos[ ∫ 2 π f n ( t ) dt + ϕ n ( t )]

[0061] Based on this, the motion model of each moving part of the machine tool can be established.

[0062] Step 2: Digital twin system of machining process based on kinematic model

[0063] According to the kinematic model of each moving part of the machine tool in step one, a digital twin system of the machining process can be established: in addition to external noise, the workpiece is installed on the C-axis, and the C-axis speed is relatively low, so the main component of the C-axis frequency domain is the low-frequency component; the C-axis is installed on the Z-axis guide, so the main component of the Z-axis frequency domain is the external noise and the low-frequency component of the workpiece; the grinding axis speed is much higher than the C-axis speed, and the grinding axis is installed on the X-axis guide, so the main components of the grinding axis and X-axis frequency domain are high-frequency components.

[0064] The five-dimensional architecture of the digital twin system includes physical entity PE, virtual entity VM, service SS, twin data DD and the links CN between the components, which can be expressed as:

[0065]

[0066] The virtual entity VM includes the motion model of each moving part of the machine tool established in step 1, which can be expressed as:

[0067]

[0068] Among them, F x Indicates the real-time load of the X-axis linear motion system, F z Indicates the real-time load of the Z-axis linear motion system, T C Indicates the real-time load of the workpiece spindle\C-axis rotation motion system, T G Represents the real-time load of the grinding axis rotary motion system; based on this, the moving parts are integrated into the digital twin system:

[0069]

[0070] Among them, P x Indicates the real-time power of the X-axis linear motion system, P z Indicates the real-time power of the Z-axis linear motion system, P C Indicates the real-time power of the workpiece spindle\C-axis rotary motion system, P G Indicates the real-time power of the grinding axis rotary motion system; v x Indicates the real-time speed of the X axis, v z Indicates the real-time speed of the Z axis, w c Indicates the real-time angular velocity of the workpiece spindle or C-axis, w g Indicates the real-time angular velocity of the grinding axis.

[0071] Step 3: Time-frequency domain analysis of the processing process

[0072] The digital twin system established based on the machining process can realize real-time monitoring of the machining process. By extracting the real-time position of each moving part of the machine tool in the digital twin system and comparing it with the command position of the moving parts of the machine tool during the machining process, the impact of the motion error during the machine tool movement on the workpiece machining process can be obtained:

[0073]

[0074] Among them, x instruction 、z instruction 、c instruction are the command positions of the XZC coordinates in the NC program after being discretized by the machine tool NC program, and x actual 、z actual 、c actual They are the actual positions of the XZC coordinates in the machine tool coordinate system during processing, and the difference between the two is x pe 、z pe 、c pe These are the effects on the three coordinates X, Z and C during the machining process.

[0075] Based on the established digital twin system, the error data of each axis can also be separated to obtain the influence of each motion axis on the overall machining error of the machine tool, such as Figure 4 shown. Figure 4 The figure shows the distribution of the error of the rear cylindrical workpiece in the XYZ coordinate axes during the machining process. It can be seen that the Z-axis error is the smallest, while the XY-axis error is the largest. This is because the Z-axis does not have interpolation motion during the cylindrical machining process, and it is a single-direction linear motion. The XY-axis coordinates are formed by interpolation of the X-axis and the C-axis, so the XY-axis error is relatively large. On the other hand, the Z-axis error represents the axial error, and the XY-axis represents the radial error. By projecting the three-axis error in the XZ direction, it can be found that there are 18 regular fluctuations in the x-direction, that is, the circumferential direction after expansion along the radius, such as Figure 5 As shown in the figure, it can be judged that the error shape in the processing process is not randomly distributed, and there is a system fault.

[0076] Frequency domain analysis is an important technique in mechanical fault diagnosis, which reveals the frequency components and frequency response of signals by converting time domain signals into frequency domain signals. In machine tool fault diagnosis, frequency domain analysis can help identify and locate faults, because different fault types will show different characteristics in the frequency domain.

[0077] Frequency domain analysis can be achieved through Fourier transform, expressed as:

[0078]

[0079] Among them, k represents the corresponding moving part, F k (ω) is the transformed spectrum function, f k (t) is the original time domain signal, ω is the angular frequency, e -jωt is a complex exponential function. By multiplying and integrating the time domain signal with a series of complex exponential functions, we can obtain the spectrum distribution of the signal at different frequencies. Based on this, we can perform frequency domain analysis on the power signal during product processing, combine the processing parameters and the corresponding machine tool motion actuators, and determine the source of the processing error and the early manifestation of the fault. Through the frequency domain analysis results, the characteristic values ​​of the digital twin system signal during the processing process can be extracted, thereby realizing real-time monitoring of the processing process and making timely adjustments.

[0080] Based on this, the machine tool detection data is analyzed in the frequency domain, such as Figure 6-Figure 8 The figure shows the frequency domain analysis results based on the frequency signal of the digital twin system. It can be seen that the machine tool frequency distribution has a "ZC-WT-X" propagation path: the C and X axes are in direct contact with the tooling (WT) system, so Figure 6 The main frequency component of the X-axis is the low-frequency component of 200 Hz of the grinding axis amplitude, while the low-frequency component of the processing process is around 0-5 Hz, that is, the amplitude of the C-axis is 0.833 Hz and its frequency multiples are 1.66, 2.5 Hz, 3.33 Hz and 4.16 Hz, etc. The other axis that is in direct contact with the WT system is the C-axis, whose main frequency component is 0.83 Hz and its frequency multiples caused by the C-axis speed. In addition, there is a group of 7.5 Hz and its high-frequency components of 15 Hz and 22.5 Hz, such as Figure 8 As shown. The Z axis is connected to the WT system through the C axis and connected to the grinding wheel axis through the bed - X axis, so theoretically it should have the grinding wheel axis frequency and the workpiece spindle frequency components. Figure 7 As shown in the figure, the X-axis amplitude is the highest, the C-axis amplitude is the lowest, the low-frequency component is not significant, and the main response is the amplitude corresponding to the X-axis frequency. The amplitude corresponding to the C-axis is doubled to 18.285 Hz, 19.119 Hz, 19.952 Hz and 20.786 Hz near the X-axis frequency due to interaction. Figure 4 From the error distribution results of the workpiece, it can be determined that the frequency domain component of the surface error is 15 Hz. Based on this, the error can be clearly located on the C-axis, that is, the 7.5 Hz caused by the C-axis torque 18 winding and its high-frequency components 15 Hz and 22.5 Hz.

[0081] Step 4: Fault diagnosis and error location

[0082] While the digital twin system monitors the machining process, it can also realize fault diagnosis and error location of the machining process. Through frequency domain analysis, the machining error and characteristic value distribution law can be quickly obtained, and compared with the characteristic frequency signal of the main moving parts of the machine tool for fault diagnosis. System failure does not appear suddenly, but gradually occurs as the machining progresses and the parts and tools operate. These characteristics are often difficult to reflect in the time domain signal. Through frequency domain analysis, the early manifestation of the fault in the frequency domain can be obtained more sensitively, realizing early fault diagnosis and daily maintenance of equipment. In addition, the fault diagnosis results can also be compared with the physical structure of the main moving parts to realize error location, reduce the system error in the machining process, and ensure the stable operation of the product machining process.

[0083] Combined with the frequency domain results, the C-axis structure is analyzed. The C-axis is a high-rigidity, high-rotation accuracy static pressure air spindle that provides high-precision rotation motion and angle positioning functions for processing. The overall structure of the C-axis is as follows: Fig. 9 As shown in the figure, it mainly includes slotless brushless ring motor and other components. At the beginning of the design, in order to increase the C-axis load, the C-axis adopts the torque motor drive mode, and the spindle is driven by the PMSM motor direct drive mode. There are 18 iron core windings inside the spindle motor, such as Fig.10 As shown in the figure, in order to ensure the load and rigidity of the C-axis at the beginning of the machine tool design, the spindle motor torque was selected to be relatively large. However, in the working condition described in this example, the actual processing load is relatively small, far less than the machine tool load range, which leads to excessive torque of the C-axis motor. Finally, there are 18 torque fluctuations during the C-axis motor's rotation, which is finally reflected in the surface morphology of the workpiece.

[0084] Based on the processing fault diagnosis system and method based on digital twin proposed by the present invention, high-precision processing status real-time monitoring and processing effect prediction over time can be realized, and the cost of using external sensors is reduced. The method meets the requirements of on-site real-time monitoring of the processing process, improves processing efficiency and workpiece accuracy, and can be integrated into any open CNC system. By extracting CNC servo control signals such as current, voltage, power, position signals, etc., the same effect as the present invention can be achieved, and online monitoring in an industrial environment can be easily realized. On this basis, the source of processing errors can be traced, and the causes of workpiece errors can be clarified, providing a theoretical basis for further improving product accuracy and improving product quality. The higher the accuracy of the CNC system, the higher the monitoring accuracy of the processing state. By selecting appropriate filtering, noise reduction and fitting algorithms, the monitoring accuracy of the processing process will be further improved, greatly broadening the scope of application of this method in industry.

Claims

1. A method for fault diagnosis of machining process based on digital twin, characterized in that: Here are the steps: Step 1: Kinematic modeling of the moving parts of the machine tool Construct a dynamic model of a linear motor-based motion system under load; Construct dynamic models of motion systems based on rotating motors under load; Build up the torque output through the motor under load; Step 2: Digital twin system of machining process based on kinematic model According to the kinematic model of the moving parts of the machine tool in step 1, a digital twin system of the machining process is established: Step 3: Time-frequency domain analysis of the processing process The digital twin system established based on the machining process realizes real-time monitoring of the machining process; Step 4: Fault diagnosis and error location While the digital twin system monitors the machining process, it can also perform fault diagnosis and error location in the machining process.

2. The method for fault diagnosis of a machining process based on digital twin according to claim 1, characterized in that: In step 1, the kinematic modeling process of the moving parts of the machine tool is as follows: The moving parts of the machine tool are motion actuators that are closely related to the machining process, and are divided into motion systems based on linear motors and motion systems based on rotary motors. The motion system based on linear motors includes the machine tool X-axis and the machine tool Z-axis, and the motion system based on rotary motors includes the workpiece spindle or C-axis and the grinding axis. The kinematic model of the moving parts of the machine tool includes the control and feedback parameters of the machine tool feed mechanism, which is mainly composed of three sub-models, namely the motor, controller and load model. The electrical signal provided by the controller drives the motor to overcome the load movement. The dynamic model of the motion system based on the linear motor under load is expressed as: Among them, F load is the real-time load, x(t) is the real-time position of the moving part, (t) is the first-order derivative of the real-time position of the moving part, M is the total mass of the moving part, B is the viscosity coefficient of the hydraulic guide, K i is the thrust coefficient of PMLSM, i q is the output current of the servo drive, and the real-time position of the motor is expressed as: Where C is the DC component, N(t) is the random noise, f0, A0, φ0 are the frequency, amplitude and initial phase of the motor fundamental vibration, respectively, and f i , A i ,φ i are the frequency, amplitude and phase of the non-fundamental frequency vibration of the motor, and n is the order of the non-fundamental frequency vibration of the motor; The dynamic model of the motion system based on the rotating motor under load is expressed as follows: Among them, the motor inertia J1 and the damping coefficient B1 are constants, θ is the real-time rotation angle of the motor, and the load torque is regarded as a disturbance that reduces the spindle speed; however, the motor controller can adjust the real-time torque T1 of the motor by adjusting the current i(t) supplied to the motor, and then respond to the load. Under load, the motor outputs the torque T out It is expressed as: Based on this, the kinematic model of the moving parts of the machine tool is established.

3. The method for fault diagnosis of a machining process based on digital twin according to claim 2, characterized in that: In step 2, the process of the digital twin system of the machining process based on the kinematic model is as follows: According to the kinematic model of the moving parts of the machine tool in step 1, a digital twin system of the machining process is established: in addition to external noise, the workpiece is installed on the C-axis, and the main component of the C-axis frequency domain is the low-frequency component; the C-axis is installed on the Z-axis guide, so the main component of the Z-axis frequency domain is the external noise and the low-frequency component of the workpiece; the grinding axis speed is higher than the C-axis speed, and the grinding axis is installed on the X-axis guide, so the main components of the grinding axis and the X-axis frequency domain are high-frequency components; The five-dimensional architecture of the digital twin system includes physical entity PE, virtual entity VM, service SS, twin data DD and the links CN between the components, which can be expressed as: Among them, CNCGM is a CNC grinding machine, and DT is a digital twin system; The virtual entity VM contains the kinematic models of the moving parts of the machine tool. The load model in equations (1) and (3) is expressed as: Among them, F x Indicates the real-time load of the X-axis linear motion system, F z Indicates the real-time load of the Z-axis linear motion system, T C Indicates the real-time load of the workpiece spindle or C-axis rotary motion system, T G Represents the real-time load of the grinding axis rotary motion system; based on this, the moving parts of the machine tool are integrated into the digital twin system: Among them, P x Indicates the real-time power of the X-axis linear motion system, P z Indicates the real-time power of the Z-axis linear motion system, P C Indicates the real-time power of the workpiece spindle or C-axis rotary motion system, P G Indicates the real-time power of the grinding axis rotary motion system; v x Indicates the real-time speed of the X axis, v z Indicates the real-time speed of the Z axis, w c Indicates the real-time angular velocity of the workpiece spindle or C-axis, w g Indicates the real-time angular velocity of the grinding axis.

4. The method for fault diagnosis of a machining process based on digital twin according to claim 3 is characterized in that: In step 3, the time-frequency domain analysis process of the processing is as follows: The digital twin system established based on the machining process realizes real-time monitoring of the machining process. By extracting the real-time position of each moving part of the machine tool in the digital twin system and comparing it with the command position of the moving parts of the machine tool during the machining process, the influence of the motion error during the machine tool movement on the workpiece machining process is obtained: Among them, x instruction 、z instruction 、c instruction are the command positions of the XZC coordinates in the NC program after being discretized by the machine tool NC program, and x actual 、z actual 、c actual They are the actual position of the XZC coordinates in the machine tool coordinate system during processing, and the difference between the command position and the actual position x pe 、z pe 、c pe They are the influences on the three coordinates X, Z and C during the machining process; Frequency domain analysis reveals the frequency components and frequency response of a signal by converting the time domain signal into a frequency domain signal. In machine tool fault diagnosis, frequency domain analysis helps identify and locate faults, because different fault types will show different characteristics in the frequency domain. Frequency domain analysis can be achieved through Fourier transform, expressed as: Among them, k represents the corresponding moving part, F k (ω) is the transformed spectrum function, f k (t) is the original time domain signal, ω is the angular frequency, e -jωt It is a complex exponential function; by multiplying the time domain signal with a series of complex exponential functions and integrating them, the frequency spectrum distribution of the signal at different frequencies is obtained; based on this, the power signal in the product processing process is analyzed in the frequency domain, and the processing parameters and the corresponding machine tool motion system are combined to determine the source of the processing error and the early manifestation of the fault; through the frequency domain analysis results, the characteristic values ​​of the digital twin system signal in the processing process are extracted, thereby realizing real-time monitoring of the processing process and making timely adjustments.

5. The method for fault diagnosis of a machining process based on digital twin according to claim 4, characterized in that: In step 4, the process of fault diagnosis and error location is as follows: While the digital twin system monitors the machining process, it can also perform fault diagnosis and error location of the machining process. Through frequency domain analysis, the machining error and eigenvalue distribution law can be quickly obtained and compared with the characteristic frequency signals of the main moving parts of the machine tool to perform fault diagnosis. In addition, the fault diagnosis results can also be compared with the physical structure of the main moving parts to achieve error location, reduce system errors in the machining process, and ensure the stable operation of the product machining process.

Images