Servo feeding system fault diagnosis method and device and electronic equipment

Abstract

The invention provides a servo feeding system fault diagnosis method and device and electronic equipment, and the method comprises the steps: constructing a high-fidelity dynamic model for a target feeding system; vibration signals of different faults under different working conditions are collected, and a multi-working-condition fault database is constructed; performing a fault simulation experiment by using the high-fidelity dynamic model to obtain a simulation signal, and performing comparison in combination with the vibration signal to optimize and verify the high-fidelity dynamic model; and acquiring actual vibration data of the target feeding system, and comparing the actual vibration data with twin data to determine the fault type of the target feeding system. According to the invention, accurate fault diagnosis and prediction are carried out on the ball screw feeding system, and the problems of poor accuracy and real-time performance in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of numerical control machine tools, and particularly relates to a method, device and electronic equipment for servo feed system fault diagnosis. Background Art

[0002] A numerical control machine tool, abbreviated as a CNC machine tool, is an automated machine tool equipped with a program control system. This control system can logically process a program with control codes or other symbolic instructions, decode it, represent it in coded numbers, and input it into the numerical control device through an information carrier. After arithmetic processing, various control signals are sent out by the numerical control device to control the actions of the machine tool, and the parts are automatically processed according to the shape and size required by the drawing. The feed system is an important part of a numerical control machine tool, and the movement of the load is completed through the feed system.

[0003] With the development of intelligent manufacturing, the requirements for the stability and accuracy of the servo feed system of numerical control machine tools are increasing day by day. As the core transmission component, the health state of the ball screw directly affects the machining quality. However, the existing ball screw fault diagnosis methods mainly rely on manual experience or simple signal processing techniques, and there are problems such as low diagnostic accuracy, response lag, and difficulty in predictive maintenance.

[0004] For the problems of poor accuracy and real-time performance in the existing related technologies, no effective solutions have been proposed yet. Summary of the Invention

[0005] The present invention provides a method, device and electronic equipment for servo feed system fault diagnosis to solve the defects of poor accuracy and real-time performance in the existing related technologies and realize intelligent health management.

[0006] In the first aspect, the present invention provides a method for servo feed system fault diagnosis, including: Construct a high-fidelity dynamic model for the target feed system; the high-fidelity dynamic model includes a ball screw pair dynamic model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; Collect vibration signals of different faults under different working conditions and construct a multi-condition fault database; Use the high-fidelity dynamic model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamic model; Collect the actual vibration data of the target feed system, compare the actual vibration data with the twin data, and determine the fault type of the target feed system; the twin data is composed of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamic model.

[0007] A servo feed system fault diagnosis method provided by the present invention constructs a dynamic model of a ball screw pair for the target feed system, including: Analyze the mechanical characteristics of the balls in the raceway and construct the dynamic equation of the target feed system; Based on the dynamic equation of the target feed system, determine the coupling relationship between the rotation and linear motion of the ball screw; Combine the coupling relationship between the rotation and linear motion of the ball screw to construct a differential equation for the dynamic behavior of the ball screw pair in the target feed system, and generate the dynamic model of the ball screw pair.

[0008] A servo feed system fault diagnosis method provided by the present invention analyzes the mechanical characteristics of the balls in the raceway and constructs the dynamic equation of the target feed system, including: Analyze the contact forces between the balls and the raceway and between the two raceways, and construct the Lagrangian equation; Based on the displacement of the nut on the screw, operate on the Lagrangian equation to obtain the linear motion equation of the nut and determine the linear kinetic energy of the nut; Based on the rotation angle of the ball screw, operate on the Lagrangian equation to obtain the rotational motion equation of the ball screw and determine the rotational kinetic energy of the ball screw.

[0009] A servo feed system fault diagnosis method provided by the present invention constructs a five-degree-of-freedom rolling bearing vibration model for the target feed system, including: Based on the angular position of each rolling element and the angular velocity of the cage, determine the total deformation of the corresponding rolling element; Combine the Hertz contact theory to determine the contact force between each rolling element and the raceway, and construct the five-degree-of-freedom rolling bearing vibration model.

[0010] A servo feed system fault diagnosis method provided by the present invention collects vibration signals of different faults under different working conditions and constructs a multi-condition fault database, including: Drive the ball screw pair sample to move through a pre-built experimental test platform and collect the vibration signal of the ball screw pair sample; Simulate the working conditions of the normal state and different fault states, and sort and preprocess the vibration signals collected under different working conditions to construct a multi-condition fault database.

[0011] A servo feed system fault diagnosis method provided by the present invention combines the vibration signals for comparison and optimizes the high-fidelity dynamic model, including: Based on Kalman filtering and Bayesian estimation, dynamically match the vibration signal with the simulation signal; Quantify the similarity between the simulation signal and the vibration signal through probability density distribution analysis and frequency-domain feature comparison, and establish a transferability evaluation index; Adopt a domain adversarial network to optimize fault feature transfer, aiming to minimize the distribution difference between the simulation signal and the vibration signal, and adjust the model parameters of the high-fidelity dynamics model.

[0012] According to a servo feed system fault diagnosis method provided by the present invention, verify the high-fidelity dynamics model, including: Extract features and perform envelope spectrum analysis on the simulation signal and the vibration signal to obtain key features; Compare the key features with the feature theoretical values, and verify the effectiveness of the high-fidelity dynamics model according to the comparison results.

[0013] According to a servo feed system fault diagnosis method provided by the present invention, collect the actual vibration data of the target feed system, and compare the actual vibration data with the twin data to determine the fault type of the target feed system, including: Collect data of the target feed system through multiple pre-arranged sensors, and fuse the acquisition results based on a support vector machine or a deep learning model to generate actual vibration data; Compare the actual vibration data with the twin data to determine the fault type of the target feed system.

[0014] In a second aspect, the present invention also provides a servo feed system fault diagnosis device, including: A construction module for constructing a high-fidelity dynamics model for a target feed system; the high-fidelity dynamics model includes a ball screw pair dynamics model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; An acquisition module for acquiring vibration signals of different faults under different working conditions and constructing a multi-condition fault database; An optimization module for using the high-fidelity dynamics model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamics model; A diagnosis module for collecting the actual vibration data of the target feed system, and comparing the actual vibration data with the twin data to determine the fault type of the target feed system; the twin data is composed of the acquired vibration signals and the simulation signals of the optimized high-fidelity dynamics model.

[0015] In a third aspect, the present invention further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the servo feed system fault diagnosis method described in the first aspect above is implemented.

[0016] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the servo feed system fault diagnosis method described in the first aspect above is implemented.

[0017] In a fifth aspect, the present invention further provides a computer program product, including a computer program. When the computer program is executed by a processor, the servo feed system fault diagnosis method described in the first aspect above is implemented.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The servo feed system fault diagnosis method provided by the present invention constructs a high-fidelity dynamic model of the ball screw feed system based on digital twin, and combines multi-sensor data fusion to integrate high-precision modeling and data-driven, realizes real-time mapping and intelligent optimization of the physical system, and accurately diagnoses and predicts faults of the ball screw feed system, solving the problems of poor accuracy and real-time performance existing in the existing related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 is a flowchart of the servo feed system fault diagnosis method provided by the present invention; Figure 2 is a schematic diagram of the dynamic model of the ball screw pair in an embodiment of the present invention; Figure 3 is a schematic diagram of the five-degree-of-freedom rolling bearing vibration model in an embodiment of the present invention; Figure 4 is a schematic diagram of the simulation test bench in an embodiment of the present invention; Figure 5 is a schematic diagram of the envelope spectrum analysis under the pitting fault of the lead screw in an embodiment of the present invention; Figure 6 is a schematic diagram of the envelope spectrum analysis under the outer ring fault of the bearing in an embodiment of the present invention; Figure 7It is the probability density distribution diagram of the ball screw pitting failure in the embodiment of the present invention; Figure 8 It is the probability density distribution diagram of the fault signal of the outer ring of the support bearing in the embodiment of the present invention; Figure 9 It is the structural block diagram of the servo feed system fault diagnosis device provided by the present invention; Figure 10 It is the structural schematic diagram of the electronic device provided by the present invention. Specific implementation manners

[0021] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] The present invention provides a servo feed system fault diagnosis method, Figure 1 It is the flow chart of the servo feed system fault diagnosis method provided by the present invention. As Figure 1 shown, the method includes the following steps: Step S101, construct a high-fidelity dynamics model for the target feed system; the high-fidelity dynamics model includes a ball screw pair dynamics model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; Step S102, collect vibration signals of different faults under different working conditions, and construct a multi-condition fault database; Step S103, use the high-fidelity dynamics model to conduct a fault simulation experiment, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamics model; Step S104, collect the actual vibration data of the target feed system, and compare the actual vibration data with the twin data to determine the fault type of the target feed system; the twin data is composed of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamics model.

[0023] In this method, first, a high-fidelity dynamic model is constructed. The high-fidelity dynamic model can serve as an alternative model for the target feed system during the fault diagnosis process, simulating the operating states of the target feed system under different parameters or working conditions. Then, vibration signals under different working conditions and different faults are collected through experiments. Next, simulation signals during faults of the high-fidelity dynamic model are collected through simulation, and the simulation signals are dynamically compared with the experimentally collected signals to further optimize the high-fidelity dynamic model and improve its accuracy. Finally, during the actual working process of the target feed system, real-time comparison is made between the vibration signal data detected by multiple sensors and the twin data, thereby diagnosing faults in the target feed system. In the above process, a high-fidelity dynamic model of the ball screw feed system is constructed based on digital twin, combined with multi-sensor data fusion, integrating high-precision modeling and data-driven, realizing real-time mapping and intelligent optimization of the physical system, accurately diagnosing and predicting faults in the ball screw feed system, and solving the problems of poor accuracy and real-time performance existing in the existing related technologies.

[0024] In some of these embodiments, in step S101, a dynamic model of the ball screw pair for the target feed system is constructed, including: analyzing the mechanical characteristics of the balls in the raceway, and constructing the dynamic equation of the target feed system; based on the dynamic equation of the target feed system, determining the coupling relationship between the rotation and linear motion of the ball screw; combining the coupling relationship between the rotation and linear motion of the ball screw to construct a differential equation for the dynamic behavior of the ball screw pair in the target feed system, and generating a dynamic model of the ball screw pair.

[0025] In this embodiment, analyzing the mechanical characteristics of the balls in the raceway and constructing the dynamic equation of the target feed system includes: analyzing the contact forces between the balls and the raceway and between the two raceways, and constructing the Lagrangian equation; based on the displacement of the nut on the screw, operating on the Lagrangian equation to obtain the linear motion equation of the nut and determine the linear kinetic energy of the nut; based on the rotation angle of the ball screw, operating on the Lagrangian equation to obtain the rotational motion equation of the ball screw and determine the rotational kinetic energy of the ball screw.

[0026] Exemplarily, as Figure 2 shown, Figure 2 is a schematic diagram of the dynamic model of the ball screw pair in the embodiment of the present invention. The dynamic model of the ball screw pair comprehensively considers the conversion of rotational motion to linear motion, torsional vibration, and the translational and rotational motions of the nut in all directions, and constructs a differential equation describing the dynamic behavior of the ball screw pair based on Newton's laws of motion and the principles of contact mechanics.

[0027] Assume that the output torque of the servo motor is T, and the rotation angle of the shaft, the moment of inertia, rotational damping coefficient, and rotational stiffness of the shaft are respectively , , ; The moment of inertia, rotational damping coefficient, and rotational stiffness of the ball screw are , , ; The mass, damping coefficient, and stiffness of the workbench are , , ; The displacement of the workbench and the lead of the screw are , . Then the relationship between the displacement of the workbench and the rotation angle of the servo motor is:

[0028] where t represents time, J0 represents the equivalent moment of inertia, , B0 represents the equivalent damping coefficient, , K0 represents the equivalent stiffness, .

[0029] Due to the existence of the lead angle of the ball screw, the workbench and the screw will be affected by axial forces and generate axial vibrations. Therefore, when establishing the dynamic model of the ball screw pair, it is necessary to consider its axial movement degrees of freedom. At the same time, to reduce the computational complexity, it is assumed that the ball screw and the ball nut are rigid bodies, and their own elastic deformations are ignored. When performing precise dynamic modeling, first analyze the mechanical characteristics of the balls in the raceways. When the ball screw is driven by the rotation of the servo motor, under the action of the lead angle, the balls generate an axial thrust on the screw nut, causing contact forces between the balls and the screw raceway and the nut raceway. The contact force can be decomposed into component forces in the x, y, and z directions. Among them, the torque transmitted by the servo motor is T, the contact angle between the balls and the raceways is α, the lead angle of the screw is γ, and the contact forces between the balls and the two raceways and between the raceways are P. The specific expressions are as follows:

[0030]

[0031]

[0032] where P x represents the contact force in the x direction, P y represents the contact force in the y direction, and P z represents the contact force in the z direction. The basic form of Lagrange's equation is:

[0033] where is the Lagrangian, T represents kinetic energy, V represents potential energy, represents the generalized coordinates, denotes the corresponding generalized force. In a ball screw system, the generalized coordinates are and , and the generalized forces are and , respectively. F represents the external force received by the system, and T represents the external torque applied to the system.

[0034] For , the Lagrange equation yields the linear motion equation of the nut after calculation:

[0035]

[0036] where denotes the first derivative with respect to the displacement , i.e., the velocity, and denotes the second derivative with respect to the displacement , i.e., the acceleration.

[0037] For , the Lagrange equation yields the rotational motion equation of the screw after calculation:

[0038]

[0039] where denotes the first derivative with respect to the rotation angle , i.e., the angular velocity; denotes the second derivative with respect to the rotation angle , i.e., the angular acceleration. In the above expressions, the displacement of the nut on the screw is , the rotation angle of the ball screw is , and the external load on the system (axial load acting on the nut) is .

[0040] By constructing the kinetic and potential energies of the system, the dynamic equations are obtained. The linear kinetic energy of the nut and the rotational kinetic energy of the ball screw are respectively:

[0041]

[0042] where T nut represents the linear kinetic energy of the nut, is the mass of the nut, is the linear velocity of the nut; T screw represents the rotational kinetic energy of the ball screw, is the moment of inertia of the ball screw, is the angular velocity of the ball screw. Combining the above formulas, we can obtain in terms of and the total kinetic energy equation dominated by:

[0043] where T represents the total kinetic energy of the system. Assume the linear stiffness of the ball screw is , and the potential energy of the system (the potential energy of the screw structure can be represented by a spring model) is:

[0044] where V represents the potential energy of the system. There is a coupling relationship between the rotation and linear motion of the ball screw, which is determined by the pitch of the screw. Taking the derivative of it, the coupling relationship between velocity and acceleration can be obtained, that is:

[0045]

[0046]

[0047] where represents displacement, represents the first derivative of displacement , that is, velocity; represents the second derivative of displacement , that is, acceleration; represents the angle of rotation, represents the first derivative of the angle of rotation , that is, angular velocity; represents the second derivative of the angle of rotation , that is, angular acceleration. A

[0048] When modeling the ball screw pair, consider its horizontal displacement , , vertical displacement , , axial displacement , , and simplify it to a six-degree-of-freedom system to construct its dynamic simulation model. At the same time, consider the centrifugal force caused by manufacturing or installation errors, and establish the dynamic differential equation of the ball screw pair:

[0049] where each parameter represents different physical meanings and system characteristics, represents the displacement of the screw in the x direction, represents the velocity of the screw in the x direction, represents the acceleration of the screw in the x direction, represents the displacement of the screw in the y direction, represents the velocity of the screw in the y direction, Represents the acceleration of the lead screw in the y direction, Represents the displacement of the lead screw in the z direction, Represents the velocity of the lead screw in the z direction, Represents the acceleration of the lead screw in the z direction, Represents the displacement of the nut in the x direction, Represents the velocity of the nut in the x direction, Represents the acceleration of the nut in the x direction, Represents the displacement of the nut in the y direction, Represents the velocity of the nut in the y direction, Represents the acceleration of the nut in the y direction, Represents the displacement of the nut in the z direction, Represents the velocity of the nut in the z direction, Represents the acceleration of the nut in the z direction; The total mass of the lead screw nut and the workbench and the mass of the lead screw part are respectively 、 , and their numerical values are directly related to the dynamic performance during system startup, stop, and speed change; the connection dampings between the lead screw and its fixed position and between the slider and the guide rail are respectively 、 , and appropriate damping settings can effectively reduce the resonance risk of the system and improve the system stability; the equivalent connection stiffnesses between the lead screw and its fixed position and between the slider and the guide rail are respectively 、 ; the components of the contact force in the x, y, and z directions are respectively 、 、 These contact force components are not only the key to driving the system motion but also reflect the stress distribution inside the system, which is of great significance for analyzing the wear, fatigue life, and failure mechanism of the ball screw. In addition, the frictional resistance between the slider and the guide rail is , the lead angle of the lead screw is , the motor torque is , the radius of the lead screw is , the eccentricity is , the angular frequency of the lead screw rotation is , the gravitational acceleration is g, and the time is t.

[0050] In this embodiment, the lead screw drive system applies a fault evolution and coupling response mechanism model based on Lagrange multi-body dynamics equations and Hertz contact theory, laying a physical foundation for the virtual entity modeling of the servo feed system and successfully establishing virtual entities for various failure faults of the servo feed system.

[0051] Correspondingly, a five-degree-of-freedom rolling bearing vibration model for the target feed system is constructed, including: determining the total deformation of the corresponding rolling element based on the angular position of each rolling element and the angular velocity of the cage; combining the Hertz contact theory to determine the contact force between each rolling element and the raceway, and constructing a five-degree-of-freedom rolling bearing vibration model.

[0052] The five-degree-of-freedom rolling bearing vibration model in this embodiment fully considers the vibration characteristics of the inner ring, outer ring, and rolling elements in the radial, axial, and tangential directions, as well as the contact nonlinearity between the balls and the raceways. On the basis of meeting the research accuracy and actual requirements, the dynamic simulation model of the support bearing is reasonably simplified to accurately describe the vibration response of the bearing under different working conditions, providing key support for the digital twin model of the servo feed system. Specifically, the model assumes that the motion is in the same plane, focuses on the horizontal and vertical vibrations of the inner and outer rings, regards the rolling elements as rigid bodies, ignores the stiffness, damping of elastohydrodynamic lubrication, and the influence of the moment of inertia of each component, uses the lumped mass method to estimate the mass of the components, and the elastic contact between the raceway and the rolling elements follows the Hertz contact theory. Thus, a system with 4 degrees of freedom for the horizontal and vertical directions of the inner and outer rings and 1 degree of freedom for the unit resonator is constructed. Among them, the unit resonator is used to simulate the high-frequency natural vibration of the excited bearing and other components, and can effectively capture the high-frequency vibration characteristics. Based on the above assumptions, the established dynamic simulation model of the support bearing is as Figure 3 shown, Figure 3 which is a schematic diagram of the five-degree-of-freedom rolling bearing vibration model in the embodiment of the present invention, and this model comprehensively integrates the key dynamic characteristics of the bearing.

[0053] In Figure 3 it, k s 、k p 、k r represent the stiffness of the inner ring, outer ring, bearing housing, and unit resonator respectively; c s 、c p 、c r correspond to the damping of the inner ring, outer ring, bearing housing, and unit resonator respectively; m s 、m p 、m r are the masses of the inner ring, outer ring, bearing housing, and unit resonator respectively; x s 、x p 、y s 、y p 、y b represent the displacements of the inner and outer rings in the horizontal and vertical directions and the displacement of the unit resonator in the numerical direction respectively. When the bearing bears a pure radial load, load zones and non-load zones will appear on the bearing raceway. The rolling elements in the load zone are elastically deformed due to extrusion, which in turn causes relative displacement between the rings and triggers flexible vibration.

[0054] From the perspective of theoretical models, the relevant expression for the total deformation of the j-th rolling element is as follows:

[0055]

[0056]

[0057] Where, represents the total deformation of the rolling elements, c is the clearance, represents the angular position of the j-th rolling element, n b is the number of rolling elements, w c represents the angular velocity of the cage, is the initial angular position of the rolling element, w s is the angular velocity of the inner ring, a0 is the contact angle, D b is the diameter of the rolling element, D p is the pitch diameter of the bearing.

[0058] According to Hertz contact theory, the contact force (i.e., elastic restoring force) between the i th ball and the raceway is given by the following formula:

[0059] Where, represents the contact force between the i th ball and the raceway, δ z is the contact deformation between the rolling element and the inner and outer rings, z is the load deformation coefficient, usually 1.5 for ball bearings, k is the equivalent contact stiffness, and the expression for k is:

[0060] Where, k i is the contact stiffness between the inner ring and the rolling element, k o is the contact stiffness between the outer ring and the rolling element. In actual calculations, when the rolling element passes through the load zone, combining the above formulas, the contact force between a single rolling element and the raceway can be accurately calculated, and then the components f x and f y of the total non-linear contact force on the inner and outer rings in the x and y directions can be obtained through summation.

[0061] In some of these embodiments, in step S102, vibration signals of different faults under different working conditions are collected, and a multi-condition fault database is constructed, including: driving a ball screw pair sample to move through a pre-built experimental test platform, and collecting the vibration signals of the ball screw pair sample; simulating the working conditions of the normal state and different fault states, and sorting and preprocessing the vibration signals collected under different working conditions to construct a multi-condition fault database.

[0062] Exemplarily, Figure 4 is a schematic diagram of the simulation test bench in the embodiments of the present invention. As Figure 4 shown, an independently developed ball screw working condition simulation test bench is adopted. The test platform includes a platform main body, a drive control system, a signal acquisition system, etc. The platform main body is equipped with components such as a ball screw pair to be tested, a BKBF bearing seat, a lead screw nut seat, a coupling, a slider, a guide rail, and a load platform; the drive control system is equipped with equipment such as a servo motor, a driver, a servo controller, and a photoelectric gate sensor; the signal acquisition system mainly uses instruments such as a 3255A2 type piezoelectric acceleration sensor and a data acquisition instrument produced by DYTRAN Corporation of the United States. The sensor is installed on the lead screw nut seat to be close to the fault point and the force application point, reducing signal attenuation and interference, and ensuring that comprehensive vibration signals reflecting the performance state of the ball screw pair are collected.

[0063] For multiple fault types, multiple groups of vibration signal acquisition experiments of ball screw pairs are carried out in this embodiment. These experimental data will be used as the dataset basis for constructing a diagnostic model. To ensure the accuracy and reliability of the experimental data acquisition, the experimental process will be carried out according to the following steps: (1) Experimental preparation stage: Determine the ball screw pair samples required for the experiment, select a suitable acceleration sensor, prepare the installation fixture of the sensor and the supporting signal transmission cable to ensure the stability and accuracy of signal transmission; (2) Build an experimental test platform, install the lead screw and the bearing seat, and install the 3255A2 type and 3143M16 type acceleration sensors at appropriate positions on the ball screw nut and the bearing seat respectively, ensuring that the sensors are in close contact with the nut and firmly installed to avoid loosening or displacement during operation, thereby affecting the accuracy of signal acquisition; (3) Connect the experimental platform with the drive control system and the signal acquisition system, and set the parameters of the data acquisition equipment, including the adjustment of parameters such as the sampling frequency, sampling time, and gain; (4) According to the working characteristics of the ball screw and the expected fault characteristic frequency range, reasonably set the sampling frequency to ensure that the key information in the vibration signal can be accurately captured, and at the same time avoid excessive data volume and waste of computing resources caused by too high sampling frequency.

[0064] (5) Start the experimental test platform, set the motor speed to 300 rpm according to the experimental requirements, the uniform effective stroke is 600 mm, and the sampling frequency fs is set to 6400 Hz; (6) During the operation of the ball screw, synchronously start the data acquisition equipment, and collect the vibration signals output by the acceleration sensor in real time and store them in the local storage medium of the data acquisition equipment. Each time, 10 round-trip data are collected, and a total of 4 cycles are performed, obtaining a total of 40 groups of data; During the data acquisition process, closely monitor the operating status of the data acquisition equipment and the quality of the acquired signals to ensure the continuity and stability of data acquisition. If problems such as abnormal signals or data acquisition interruption are found, stop the experiment in a timely manner, troubleshoot the cause of the failure, and restart data acquisition after repair. (8) Conduct working condition simulation experiments in the normal state and the other five fault states in sequence, enabling the ball screw pair to operate under different working conditions, including different rotational speeds, loads, running directions, etc. for combined working conditions, and repeat the experimental procedures 2 - 7.

[0065] After completing data acquisition, transfer the stored vibration data to a computer for subsequent preprocessing and analysis. In this experiment, sliding window overlapping sampling is carried out with M = 2048 and L = 1000. 240 samples are collected in each state, and the length of each sample is 2048 data points. The data is divided into dataset D, dataset E, and dataset F according to rotational speeds of 300, 600, and 1000 rpm respectively. See Table 1 for details: Table 1 Details of the working condition simulation dataset of the ball screw pair

[0066] In Table 1, WQF represents the screw bending fault, DSF represents the screw pitting fault, FHF represents the internal composite fault of the screw, NF represents the inner ring fault of the support bearing, and WF represents the outer ring fault of the support bearing. For each fault, there are three rotational speeds of 300r, 600r, and 1000r for experiments.

[0067] In some of these embodiments, in step S103, in combination with vibration signals for comparison, optimize the high-fidelity dynamic model, including: based on Kalman filtering and Bayesian estimation, dynamically match the vibration signals and simulation signals; through probability density distribution analysis and frequency domain feature comparison, quantify the similarity between the simulation signals and vibration signals, and establish a transferability evaluation index; adopt a domain adversarial network to optimize the transfer of fault features, aiming to minimize the distribution difference between the simulation signals and vibration signals, and adjust the model parameters of the high-fidelity dynamic model.

[0068] On this basis, verify the high-fidelity dynamic model, including: extract features and perform envelope spectrum analysis on the simulation signals and vibration signals to obtain key features; compare the key features with the theoretical feature values, and verify the effectiveness of the high-fidelity dynamic model according to the comparison results.

[0069] Exemplarily, based on Kalman filtering and Bayesian estimation, the experimental vibration signals (such as acceleration signals) are dynamically matched with the simulation signals, and the model parameters (such as wear amount, stiffness degradation coefficient) are adjusted. Specifically, through probability density distribution analysis and frequency domain feature comparison, the similarity between the simulation signal and the vibration signal (actual signal) is quantified, and a transferability evaluation index is established. As Figure 5 and Figure 6 shown Figure 5 is a schematic diagram of envelope spectrum analysis under the pitting fault of the lead screw in the embodiment of the present invention, Figure 6 is a schematic diagram of envelope spectrum analysis under the fault of the outer ring of the bearing in the embodiment of the present invention. From Figure 5 and Figure 6 time domain waveform diagrams, it can be observed that the impact waveforms of the simulation signal and the actual signal are very similar. The main frequency components of the simulation signal and the actual signal in the envelope spectrum of the lead screw pitting are 33.3315 Hz and 23.8322 Hz respectively, and the relative deviations from the theoretical characteristic frequency are 0.11% and 28.58% respectively. Obviously, there is unstable vibration of the lead screw in the initial stage of actual operation. Therefore, based on the difference between the latter two characteristic frequencies, the characteristic difference frequencies of the simulation signal and the actual signal are 33.3315 Hz and 33.6993 Hz respectively, and their relative deviations are 0.11% and 0.99% respectively; the main frequency components of the simulation signal and the actual signal in the envelope spectrum of the outer ring fault of the bearing are 13.8889 Hz and 13.8881 Hz respectively, and the relative deviations from the theoretical characteristic frequency are 0.46% and 0.47% respectively.

[0070] Since the probability distributions of the simulation signal and the actual signal are inconsistent, it may also lead to unsatisfactory transfer diagnosis effect. Therefore, before applying the simulation signal, it is necessary to analyze the transferability of the signal from the perspective of probability density distribution, and specifically estimate the probability density distributions of the two signals of the pitting fault of the lead screw raceway.

[0071] Figure 7 is the probability density distribution diagram of the pitting fault of the lead screw in the embodiment of the present invention. It can be seen from Figure 7 that since the influencing factors considered in the simulation signal are relatively single, the data distribution is relatively concentrated. There are more random influences in the actual signal, resulting in a relatively low probability density peak. Although there are certain differences between the simulation signal and the actual signal, the two probability density distributions are relatively close. Obviously, both signals contain important fault feature information, and the simulation signal of the lead screw pitting fault has the transferability for the actual lead screw pitting fault diagnosis.

[0072] Figure 8 is the probability density distribution diagram of the fault signal of the outer ring of the support bearing in the embodiment of the present invention. As Figure 8As shown, the probability density distribution of the simulation signal is relatively concentrated, while for the actual signal, due to various random factors from other components, its probability density peak is lower and the probability distribution is relatively smooth. Although there are certain differences between the simulation signal and the actual signal, the probability density distributions of the two under the same fault state are relatively close. A fault feature migration strategy based on the domain adversarial mechanism can be used to reduce the difference in the fault feature distributions of the two signals under the same fault type, thereby realizing the migration from the simulation signal to the actual fault diagnosis.

[0073] Based on the above embodiments, in step S104, actual vibration data of the target feed system is collected, and the actual vibration data is compared with the twin data to determine the fault type of the target feed system, including: collecting data of the target feed system through a plurality of pre-arranged sensors, and fusing the acquisition results based on a support vector machine or a deep learning model to generate actual vibration data; comparing the actual vibration data with the twin data to determine the fault type of the target feed system.

[0074] Exemplarily, for pitting faults, the transient impact frequency (theoretical characteristic frequency ± 5% deviation) and the envelope spectrum energy distribution are extracted as key features; based on a support vector machine (SVM) or a deep learning model, multi-sensor data (acceleration, stress) is fused to achieve fault classification (normal, pitting, wear, bending, etc.).

[0075] The present invention also provides a servo feed system fault diagnosis device. The servo feed system fault diagnosis device provided by the present invention will be described below. The servo feed system fault diagnosis device described below can be correspondingly referred to the servo feed system fault diagnosis method described above. Figure 9 is the structural block diagram of the servo feed system fault diagnosis device provided by the present invention, as Figure 9 shown. The device includes: A construction module 901, configured to construct a high-fidelity dynamics model for the target feed system; the high-fidelity dynamics model includes a ball screw pair dynamics model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; An acquisition module 902, configured to acquire vibration signals of different faults under different working conditions and construct a multi-condition fault database; An optimization module 903, configured to perform fault simulation experiments using the high-fidelity dynamics model to obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamics model; The diagnostic module 904 is used to collect the actual vibration data of the target feed system, compare the actual vibration data with the twin data, and determine the fault type of the target feed system; the twin data is composed of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamic model.

[0076] When this device is in use, first, the construction module 901 constructs a high-fidelity dynamic model, which can be used as an alternative model of the target feed system during the fault diagnosis process to simulate the operating state of the target feed system under different parameters or working conditions. Then, the acquisition module 902 collects vibration signals under different working conditions and different faults through experiments. The optimization module 903 then collects the simulation signals of the high-fidelity dynamic model during faults through simulation, and dynamically compares the simulation signals with the signals collected by experiments to further optimize the high-fidelity dynamic model and improve its accuracy. Finally, the diagnostic module 904 diagnoses the faults of the target feed system by detecting the actual working process of the target feed system and comparing the vibration signal data detected by multiple sensors with the twin data in real time. In the above process, a high-fidelity dynamic model of the ball screw feed system is constructed based on digital twin, combined with multi-sensor data fusion, integrating high-precision modeling and data-driven, realizing the real-time mapping and intelligent optimization of the physical system, accurately diagnosing and predicting the faults of the ball screw feed system, and solving the problems of poor accuracy and real-time performance existing in the existing related technologies.

[0077] Figure 10 An example of the physical structure diagram of an electronic device is shown as Figure 10 shown. The electronic device may include: a processor 1001, a communication interface 1002, a memory 1003, and a communication bus 1004. Among them, the processor 1001, the communication interface 1002, and the memory 1003 complete mutual communication through the communication bus 1004. The processor 1001 can call the logical instructions in the memory 1003 to execute the fault diagnosis method for the servo feed system, and this method includes: Construct a high-fidelity dynamic model for the target feed system; the high-fidelity dynamic model includes a ball screw pair dynamic model based on a six-degree-of-freedom differential equation system and a five-degree-of-freedom rolling bearing vibration model; Collect vibration signals of different faults under different working conditions and construct a multi-condition fault database; Use the high-fidelity dynamic model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamic model; Collect the actual vibration data of the target feed system, compare the actual vibration data with the twin data, and determine the fault type of the target feed system; the twin data consists of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamic model.

[0078] In addition, when the logical instructions in the above-mentioned memory 1003 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs and other various media that can store program codes.

[0079] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the servo feed system fault diagnosis method provided by the above-mentioned various methods. The method includes: Construct a high-fidelity dynamic model for the target feed system; the high-fidelity dynamic model includes a ball screw pair dynamic model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; Collect vibration signals of different faults under different working conditions and construct a multi-condition fault database; Use the high-fidelity dynamic model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamic model; [[ID=?]] Collect the actual vibration data of the target feed system, compare the actual vibration data with the twin data, and determine the fault type of the target feed system; the twin data consists of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamic model.

[0080] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the servo feed system fault diagnosis method provided by the above-mentioned various methods. The method includes: Construct a high-fidelity dynamic model for the target feed system; the high-fidelity dynamic model includes a ball screw pair dynamic model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; Collect vibration signals of different faults under different working conditions and construct a multi-condition fault database; Use the high-fidelity dynamic model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamic model; Collect the actual vibration data of the target feed system, compare the actual vibration data with the twin data, and determine the fault type of the target feed system; the twin data is composed of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamic model.

[0081] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0082] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or equivalently replace some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fault diagnosis method for a servo feed system, characterized in that Including: Construct a high-fidelity dynamic model for the target feed system; the high-fidelity dynamic model includes a ball screw pair dynamic model based on a six-degree-of-freedom differential equation set and a five-degree-of-freedom rolling bearing vibration model; Collect vibration signals of different faults under different working conditions and construct a multi-condition fault database; Use the high-fidelity dynamic model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamic model; Collect the actual vibration data of the target feed system and compare the actual vibration data with the twin data to determine the fault type of the target feed system; the twin data is composed of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamic model.

2. The servo feed system fault diagnosis method according to claim 1, wherein Construct a ball screw pair dynamic model for the target feed system, including: Analyze the mechanical characteristics of the balls in the raceway and construct the dynamic equation of the target feed system; Based on the dynamic equation of the target feed system, determine the coupling relationship between the rotation and linear motion of the ball screw; Combine the coupling relationship between the rotation and linear motion of the ball screw to construct the differential equation of the dynamic behavior of the ball screw pair in the target feed system and generate the ball screw pair dynamic model.

3. The servo feed system fault diagnosis method according to claim 2, characterized in that Analyze the mechanical characteristics of the balls in the raceway and construct the dynamic equation of the target feed system, including: Analyze the contact forces between the balls and the raceways and between the two raceways, and construct the Lagrange equation; Based on the displacement of the nut on the screw, operate on the Lagrange equation to obtain the linear motion equation of the nut and determine the linear kinetic energy of the nut; Based on the rotation angle of the ball screw, operate on the Lagrange equation to obtain the rotation motion equation of the ball screw and determine the rotation kinetic energy of the ball screw.

4. The servo feed system fault diagnosis method according to claim 1, wherein, Construct a five-degree-of-freedom rolling bearing vibration model for the target feed system, including: Based on the angular position of each rolling element and the angular velocity of the cage, determine the total deformation of the corresponding rolling element; Combine the Hertz contact theory to determine the contact force between each rolling element and the raceway and construct the five-degree-of-freedom rolling bearing vibration model.

5. The servo feed system fault diagnosis method according to claim 1, wherein Collect vibration signals of different faults under different working conditions and construct a multi-condition fault database, including: Drive the ball screw pair sample to move through a pre-built experimental test platform and collect the vibration signals of the ball screw pair sample; Simulate the working conditions of the normal state and different fault states, and organize and preprocess the vibration signals collected under different working conditions to construct a multi-condition fault database.

6. The servo feed system fault diagnosis method according to claim 1, characterized in that Combine the vibration signals for comparison and optimize the high-fidelity dynamic model, including: Based on Kalman filtering and Bayesian estimation, dynamically match the vibration signals with the simulation signals; Through probability density distribution analysis and frequency domain feature comparison, quantify the similarity between the simulation signals and the vibration signals and establish a transferability evaluation index; Optimize the fault feature migration using a domain adversarial network, and adjust the model parameters of the high-fidelity dynamics model with the goal of minimizing the distribution difference between the simulation signal and the vibration signal.

7. The servo feed system fault diagnosis method according to claim 1, characterized in that Verify the high-fidelity dynamics model, including: Extract features and perform envelope spectrum analysis on the simulation signal and the vibration signal to obtain key features; Compare the key features with the theoretical feature values, and verify the effectiveness of the high-fidelity dynamics model according to the comparison results.

8. The servo feed system fault diagnosis method according to claim 1, characterized in that Collect the actual vibration data of the target feed system, and compare the actual vibration data with the twin data to determine the fault type of the target feed system, including: Collect data of the target feed system through multiple pre-arranged sensors, and fuse the collected results based on a support vector machine or a deep learning model to generate actual vibration data; Compare the actual vibration data with the twin data to determine the fault type of the target feed system.

9. A fault diagnosis device for a servo feed system, characterized in that, Including: A construction module for constructing a high-fidelity dynamics model for the target feed system; the high-fidelity dynamics model includes a ball screw pair dynamics model based on a six-degree-of-freedom differential equation system and a five-degree-of-freedom rolling bearing vibration model; A collection module for collecting vibration signals of different faults under different working conditions and constructing a multi-condition fault database; An optimization module for using the high-fidelity dynamics model to conduct fault simulation experiments, obtain simulation signals, and compare them with the vibration signals to optimize and verify the high-fidelity dynamics model; A diagnosis module for collecting the actual vibration data of the target feed system, and comparing the actual vibration data with the twin data to determine the fault type of the target feed system; the twin data is composed of the collected vibration signals and the simulation signals of the optimized high-fidelity dynamics model.

10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, When the processor executes the program, it implements the servo feed system fault diagnosis method according to any one of claims 1 to 8.

Images