Model building method, device, equipment and storage medium

Abstract

The present application proposes a model building method, device, equipment and storage medium, the method comprising: determining the vibration equation and design parameters of the target gearbox, the vibration equation is used to fit the vibration of the internal components of the target gearbox in motion; based on the vibration equation, design parameters and applied torque to the target gearbox, determining the predicted vibration signal at the target position of the target gearbox; obtaining the measured vibration signal at the target position under the applied torque; optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain the target vibration equation; and establishing a digital twin model corresponding to the target gearbox based on the target vibration equation. The embodiment of the present application can realize the dynamic transmission of the internal load of the gearbox by constructing the vibration equation corresponding to the gearbox, thereby making the output vibration signal more accurate, so that it can be determined whether the target gearbox has a fault based on the vibration signal.

Description

Technical Field

[0001] The present application belongs to the field of artificial intelligence technology, and specifically relates to a model building method, device, equipment and storage medium. Background Art

[0002] As an essential key component for connecting and transmitting power in mechanical equipment, gearboxes are widely used in many industries such as electric power, engineering machinery, warehousing and logistics, lifting and transportation. The working environment of gearboxes has the characteristics of variable load, variable speed, and variable external excitation. The uncertain factors in the operation of gearboxes lead to complex gearbox failure mechanisms.

[0003] Gearbox fault diagnosis based on vibration signals is a common fault diagnosis technology. Generally, gearbox fault diagnosis is assisted by simulating the internal structure of the gearbox. However, the vibration signal output by the current simulation method is not accurate enough, resulting in inaccurate fault diagnosis. Summary of the invention

[0004] The present application proposes a model building method, device, equipment and storage medium, which can solve the current technical problem of inaccurate gearbox fault diagnosis.

[0005] The first embodiment of the present application proposes a model building method, including:

[0006] Determine a vibration equation and design parameters of a target gearbox, wherein the vibration equation is used to fit the vibration of internal components of the target gearbox in motion;

[0007] determining a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and an applied torque for the target gearbox;

[0008] Acquiring a measured vibration signal of the target position under the applied torque;

[0009] Optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation;

[0010] A digital twin model corresponding to the target gearbox is established based on the target vibration equation.

[0011] The embodiment of the second aspect of the present application provides a model building device, including:

[0012] A determination module, used to determine the vibration equation and design parameters of the target gearbox, wherein the vibration equation is used to fit the vibration of the internal components of the target gearbox in a moving state;

[0013] The determination module is further configured to determine a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and an applied torque to the target gearbox;

[0014] An acquisition module, used for acquiring a measured vibration signal of the target position under the applied torque;

[0015] An optimization module, configured to optimize the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation;

[0016] A module is established to establish a digital twin model corresponding to the target gearbox based on the target vibration equation.

[0017] An embodiment of the third aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the first aspect above.

[0018] An embodiment of the fourth aspect of the present application provides a computer-readable storage medium on which a computer program is stored, and the program is executed by a processor to implement the method described in the first aspect above.

[0019] The technical solution provided in the embodiments of the present application has at least the following technical effects or advantages:

[0020] The present application proposes a model building method, device, equipment and storage medium, the method comprising: determining the vibration equation and design parameters of the target gearbox, the vibration equation is used to fit the vibration of the internal components of the target gearbox in motion; based on the vibration equation, design parameters and applied torque to the target gearbox, determining the predicted vibration signal at the target position of the target gearbox; obtaining the measured vibration signal at the target position under the applied torque; optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain the target vibration equation; and establishing a digital twin model corresponding to the target gearbox based on the target vibration equation. The embodiment of the present application can realize the dynamic transmission of the internal load of the gearbox by constructing the vibration equation corresponding to the gearbox, thereby making the output vibration signal more accurate, so that it can be determined whether the target gearbox has a fault based on the vibration signal.

[0021] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0023] Figure 1 A flow chart of a model building method provided by an embodiment of the present application is shown;

[0024] Figure 2 A framework diagram of a dynamic simulation system for a gearbox vibration state provided by an embodiment of the present application is shown;

[0025] Figure 3 A flowchart of a gearbox vibration state digital twin modeling method provided by an embodiment of the present application is shown;

[0026] Figure 4 A schematic diagram of the structure of a model building device provided in one embodiment of the present application is shown;

[0027] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present application is shown;

[0028] Figure 6 A schematic diagram of a storage medium provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0029] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0030] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by technicians in the field to which this application belongs.

[0031] Fault diagnosis technology based on vibration signals includes model-driven and data-driven fault diagnosis technology. The model-driven method solidifies expert experience in the mechanism model through theoretical modeling, and the data-driven method solidifies expert experience in the machine learning model through labeling. In the actual application of fault diagnosis technology, there is a large gap between the theoretical model and the actual gearbox, the data model is not extensible enough, and the static vibration theoretical model or data model cannot characterize the dynamic characteristics of gearbox vibration under complex working conditions. Therefore, the existing theoretical models and data models cannot meet the simulation of the gearbox vibration state under variable working conditions and the fault diagnosis technology requirements.

[0032] Digital twins are dynamic virtual models of physical entities that are built in a digital way with multiple dimensions, multiple time and space scales, multiple disciplines, and multiple physical quantities to simulate and characterize the properties, behaviors, rules, etc. of physical entities in real environments. Fault diagnosis technology based on digital twin models forms a pair of twins of the virtual space and real space of the gearbox through synchronous iterative optimization of data connections of virtual and real space models. In the process of digital twins, data is accumulated to form a dynamic knowledge base. Vibration state samples in new scenarios can be dynamically generated through digital twin model parameter settings to assist fault diagnosis.

[0033] Most of the current digital twin models only display vibration monitoring data, but do not reflect the load transfer process inside the gearbox, which results in inaccurate output vibration signals and inaccurate fault diagnosis for the gearbox.

[0034] In order to solve the above technical problems, the present application determines the vibration equation and design parameters of the target gearbox, and the vibration equation is used to fit the vibration of the internal components of the target gearbox in motion; based on the vibration equation, design parameters and applied torque to the target gearbox, the predicted vibration signal at the target position of the target gearbox is determined; the measured vibration signal at the target position under the applied torque is obtained; the vibration equation is optimized based on the predicted vibration signal and the measured vibration signal to obtain the target vibration equation; and a digital twin model corresponding to the target gearbox is established based on the target vibration equation. The embodiment of the present application can realize the dynamic transmission of the internal load of the gearbox by constructing the vibration equation corresponding to the gearbox, thereby making the output vibration signal more accurate, so that it can be determined whether the target gearbox has a fault based on the vibration signal.

[0035] The following describes a model building method, device, equipment and storage medium proposed according to an embodiment of the present application in conjunction with the accompanying drawings.

[0036] See also Figure 1 , the method specifically comprises the following steps:

[0037] S101. Determine the vibration equation and design parameters of the target gearbox.

[0038] The vibration equation is used to fit the vibration of the internal components of the target gearbox in motion.

[0039] Among them, the gearbox includes a housing, an input shaft, an output shaft, gears, bearings and other structures. The gearbox can rotate through the connection between the internal structures to provide power to the outside.

[0040] Among them, the vibration equation can be evolved based on the dynamic equation, which can represent the vibration conditions between the internal structures of the target gearbox in motion. The design parameters are the simplified dynamic parameters of each structure of the target gearbox, which may include mass, stiffness, damping, etc.

[0041] Each gearbox corresponds to a gearbox identifier, and each gearbox identifier corresponds to a set of vibration equations, each gearbox identifier corresponds to a structure included in the gearbox, and each structure corresponds to a set of design parameters. Therefore, the vibration equations and design parameters of the target gearbox can be determined based on the target gearbox identifier.

[0042] S102 : Determine a predicted vibration signal at a target position of the target gearbox based on a vibration equation, design parameters, and an applied torque to the target gearbox.

[0043] The target position is the output position of the predicted vibration signal.

[0044] In some embodiments, the vibration equation includes design parameters, load parameters, multiple load forces and unknown quantities, the position quantities include displacement quantities, velocity quantities and acceleration quantities, wherein the multiple load forces can be calculated based on the applied torque, the design parameters can be determined based on the gearbox structure, and the initial quantities of the load parameters can be set manually.

[0045] Furthermore, the vibration equation can be solved based on the design parameters, the load parameters, and the multiple load forces to obtain unknown quantities, and the predicted vibration signal of the target position can be further determined based on the unknown quantities.

[0046] S103, obtaining a measured vibration signal of the target position under applied torque.

[0047] Since the load parameters in the above vibration equation may be inaccurate, resulting in inaccurate output vibration signals, the load parameters of the vibration equation need to be optimized. Therefore, a sensor can be set at the target position to obtain the measured vibration signal of the target position under the applied torque.

[0048] S104: Optimize the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation.

[0049] Furthermore, the load parameters can be adjusted by predicting the vibration signal and the measured vibration signal to obtain the target vibration equation.

[0050] In some embodiments, the predicted vibration signal and the measured vibration signal may be analyzed to determine the load parameter values ​​corresponding to the predicted vibration signal and the measured vibration signal, and further the load parameter difference between the predicted vibration signal and the measured vibration signal.

[0051] The load parameter value in the vibration equation is optimized based on the load parameter difference, so that the load parameter value is substituted into the above vibration equation to obtain the target vibration equation.

[0052] S105. Establish a digital twin model corresponding to the target gearbox based on the target vibration equation.

[0053] After obtaining the target vibration equation, the corresponding digital twin model can be established in the following way:

[0054] Select the time step: Determine the discretization time step, which usually needs to be selected based on the characteristics of the system and the requirements of the simulation.

[0055] Numerical integration method: Choose an appropriate numerical integration method, such as the Euler method, trapezoidal method, or fourth-order Runge-Kutta method, to transform the continuous-time dynamic equations into discrete-time update equations.

[0056] Programming implementation: Use programming languages ​​(such as Python, Matlab, etc.) to convert the discretized dynamic equations into executable simulation codes.

[0057] Adding noise or uncertainty: Considering the noise or uncertainty that may exist in the actual system, corresponding items can be added to the model to better reflect the actual situation.

[0058] Verification and Validation: Validate the digital twin model and compare the model’s output with the actual system’s behavior to ensure the model’s accuracy and reliability.

[0059] The present application proposes a model building method, which includes: determining the vibration equation and design parameters of the target gearbox, the vibration equation is used to fit the vibration of the internal components of the target gearbox in motion; based on the vibration equation, design parameters and applied torque to the target gearbox, determining the predicted vibration signal at the target position of the target gearbox; obtaining the measured vibration signal at the target position under the applied torque; optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain the target vibration equation; and establishing a digital twin model corresponding to the target gearbox based on the target vibration equation. The embodiment of the present application can realize the dynamic transmission of the internal load of the gearbox by constructing the vibration equation corresponding to the gearbox, thereby making the output vibration signal more accurate, so that it can be determined whether the target gearbox has a fault based on the vibration signal.

[0060] In some embodiments, determining vibration equations and design parameters of a target gearbox identification includes:

[0061] Determine a target gearbox corresponding to a target gearbox identifier; obtain a connection relationship of at least one gear corresponding to the target gearbox in the target gearbox and a shaft system type corresponding to the at least one gear; determine a vibration equation based on a connection relationship of the at least one gear in the target gearbox and a shaft system type corresponding to the at least one gear; and determine a design parameter corresponding to the shaft system type corresponding to the at least one gear from a correspondence between the shaft system type and the design parameter.

[0062] In some embodiments, a gearbox vibration model library and a gearbox vibration model parameter library may be pre-set, wherein the gearbox vibration model library includes a vibration equation framework of a gearbox with different gear pair combinations, wherein the different gear pair combinations are at least one gear included in the gearbox, the connection relationship between each gear and the structure in the gearbox, and the shaft system type corresponding to at least one gear, wherein the shaft system type includes parallel shaft spur gear pairs, planetary gear trains, bevel gear pairs, etc., and a combined shaft system type formed by a combination of various types. The gearbox vibration model parameter library includes design parameters and load parameters in the vibration equations of different shaft system types, wherein the design parameters include mass, damping, stiffness, etc., and the load parameters may be the amplitude of the waveform finally formed and the frequency of the waveform, etc.

[0063] In some embodiments, each gearbox identifier corresponds to a gearbox, and each gearbox includes at least one gear, a connection relationship between each gear and a structure in the gearbox, and a shaft system type corresponding to at least one gear to form a corresponding gear pair combination. The vibration equation of the gearbox can be determined based on the gear pair combination, and further, the design parameters and load parameters can be determined based on the shaft system type corresponding to at least one gear.

[0064] Furthermore, the design parameters and load parameters can be substituted into the vibration equation to solve the vibration equation.

[0065] In an embodiment of the present application, through a predefined gearbox vibration model library and a gearbox vibration model parameter library, the vibration equation corresponding to the target gearbox and the design parameters and load parameters corresponding to the vibration equation are determined based on the obtained identification of the target gearbox, so as to prepare to solve the vibration equation.

[0066] In some embodiments, determining a predicted vibration signal at a target position of a target gearbox based on a vibration equation, design parameters, and an applied torque for the target gearbox includes:

[0067] The vibration equation is solved using the design parameters and applied torque to obtain the predicted vibration acceleration, predicted velocity and predicted displacement of the target position;

[0068] The predicted vibration acceleration, predicted velocity and predicted displacement are substituted into a preset sub-vibration equation of the vibration equation to obtain a predicted vibration signal of the target position, wherein the predicted vibration signal includes a plurality of predicted acceleration component waveforms.

[0069] It is understandable that each gearbox needs to have a certain torque applied to it so that the gearbox can rotate to generate a vibration signal. Therefore, the applied torque corresponding to the target gearbox can be obtained, and the applied torque can be manually input or pre-set.

[0070] In some embodiments, assuming that the target gearbox is a parallel axis gear pair transmission gearbox, taking the X / Y direction vibration equation as an example, the vibration equation model is established as follows:

[0071] (1)

[0072] (2)

[0073] (3)

[0074] (4)

[0075] (5)

[0076] (6)

[0077] in, , , is the quality parameter, , , is the stiffness coefficient, , , is the damping coefficient, , , is the displacement of the mass block, , , is the velocity of the mass, , , is the acceleration of the mass block, , is the eccentric load, , is the bearing load, , is the gear meshing load, , For non-centering load.

[0078] Among them, the mass parameter, stiffness coefficient and damping coefficient can be determined based on the corresponding gear pair combination, and each load includes load parameters and displacement, velocity and acceleration of the mass block, among which the load parameter can be determined based on the gear pair combination, and each load parameter can be determined based on the applied torque, and the applied torque can be pre-set, among which the eccentric load can be calculated by the geometric relationship between the torque and the bearing, and the bearing load is the force and moment acting on the bearing, including radial load and axial load. According to the structure and stress conditions of the gearbox, the bearing load can be calculated by the torque and the geometric structure of the gearbox transmission system and the stress conditions inside the gearbox, the meshing load can be calculated by the gear module, number of teeth, pressure angle and other parameters, as well as the action of torque, and the misalignment load can be calculated by the geometric parameters of gear meshing, the gear meshing angle and the action of torque.

[0079] Therefore, the unknown quantities in the above vibration equation are only the displacement, velocity and acceleration of the mass block. The above vibration equation is solved based on the Runke-Kutta method to obtain the displacement, velocity and acceleration of the mass block.

[0080] The preset sub-vibration equation can be formula (1), where: To predict the vibration signal, Predict the acceleration component waveform for the bearing load, Predict the acceleration component waveform for misaligned loads, Predict the acceleration component waveform for eccentric loads, Predict the acceleration component waveform for the gear mesh load.

[0081] In some embodiments, determining a predicted vibration signal at a target position of a target gearbox based on a vibration equation, design parameters, and an applied torque for the target gearbox includes:

[0082] Using the preset learning model to learn the solution process and vibration equation, the target learning model is obtained;

[0083] The design parameters are input into the target learning model to obtain multiple predicted acceleration component waveforms of the target position of the target gearbox.

[0084] In order to shorten the iterative calculation time, a preset learning model can be used to learn the above-mentioned solution process and the vibration equation, wherein the target learning model is a trained preset learning model.

[0085] Among them, the preset learning model can be trained in the following ways:

[0086] The design parameters and load parameters are input into the preset learning model, and the preset model outputs multiple training acceleration component waveforms. The multiple training acceleration component waveforms are compared with the corresponding predicted acceleration component waveforms to obtain the corresponding loss function values. The load parameters are adjusted based on the loss function value, and the training is continued until the loss function value is less than the preset loss function value or the number of training times is greater than the preset number of training times, so as to obtain a trained preset learning model, i.e., the target learning model.

[0087] The design parameters are input into the target learning model to obtain multiple predicted acceleration component waveforms of the target position of the target gearbox.

[0088] In an embodiment of the present application, a preset learning model can be used to learn the above-mentioned solution process and vibration equation, so that the learning model can be used to predict the acceleration component waveform and shorten the iterative calculation time.

[0089] In some embodiments, optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain the target vibration equation includes:

[0090] De-noise the measured vibration signal to obtain the target vibration signal;

[0091] Perform peak extraction on the target vibration signal to obtain multiple vibration signal peaks;

[0092] Multiple vibration signal peaks are filtered and decomposed using a multi-bandpass filter to obtain multiple measured acceleration component waveforms at the target position;

[0093] The vibration equation is optimized based on multiple predicted acceleration component waveforms output by the target learning model and multiple measured acceleration component waveforms to obtain the target vibration equation.

[0094] The process of acquiring the measured vibration signal may be implemented as follows: a sensor may be set to acquire the measured vibration signal at the target position.

[0095] The measured vibration signal generally includes noise, and the measured vibration signal can be subjected to denoising processing, wherein the denoising processing can be performed by using a filter, a wavelet transform, a mean filter, etc.

[0096] The peak value of the target vibration signal can reflect the abnormal vibration phenomenon in the mechanical system, so it can be used to detect and diagnose equipment failure. By monitoring the change of the peak value, problems such as gear damage and bearing failure can be discovered in time.

[0097] Generally, each vibration signal peak value in one cycle corresponds to an acceleration waveform component. Therefore, a plurality of vibration signal peak values ​​can be filtered and decomposed by a multi-bandpass filter to obtain a plurality of measured acceleration component waveforms at the target position.

[0098] Among them, the process of filtering and decomposing multiple vibration signal peaks by a multi-bandpass filter can be realized as follows:

[0099] According to the spectrum analysis results, the parameters required for designing multi-bandpass filters are determined, including center frequency, bandwidth, etc. The center frequency of the filter can be determined according to the position and amplitude of the peak, and the bandwidth can be determined according to the shape of the spectrum and signal characteristics.

[0100] Use the selected multi-bandpass filter design method to design the filter coefficients or frequency response. Common design methods include frequency response-based methods (such as frequency sampling method or window method) and time domain-based methods (such as impulse response design method).

[0101] The designed multi-bandpass filter is applied to the measured vibration signal, and the signal is decomposed into components of multiple frequency bands through the filter. Each component corresponds to a passband of the filter and contains the components of the original signal related to the frequency band.

[0102] It should be noted that in order to correspond to multiple predicted acceleration component waveforms, in the process of acquiring the measured vibration signal, it is necessary to measure the vibration signal at the corresponding target position and sample the vibration signal at a sampling frequency consistent with that in the simulation process.

[0103] Furthermore, the vibration equation can be optimized based on a plurality of predicted acceleration component waveforms output by the target learning model and a plurality of measured acceleration component waveforms to obtain a target vibration equation.

[0104] In the embodiment of the present application, the measured vibration signal can be processed so that the measured vibration signal can correspond one-to-one with the acceleration component waveform in the predicted vibration signal, so that the vibration equation can be optimized to obtain the target vibration equation.

[0105] In some embodiments, the vibration equation is optimized based on multiple predicted acceleration component waveforms and multiple measured acceleration component waveforms output by the target learning model to obtain the target vibration equation, including:

[0106] Selecting a target acceleration component waveform group, the target acceleration component waveform group comprising: a predicted acceleration component waveform and a measured acceleration component waveform;

[0107] Determine the gradient direction and parameter resolution according to the predicted acceleration component waveform and the measured acceleration component waveform;

[0108] Adjust the load parameters of the vibration equation based on the gradient direction and the parameter optimization value until the preset iteration conditions are met to obtain the target load parameters;

[0109] Based on the target load parameters, the target vibration equation is established.

[0110] After obtaining multiple predicted acceleration component waveforms and multiple measured acceleration component waveforms, the multiple predicted acceleration component waveforms and the multiple measured acceleration component waveforms can be grouped based on the frequency of the waveforms. As mentioned above, the multiple predicted acceleration component waveforms include the eccentric load predicted acceleration component waveform, the misaligned load predicted acceleration component waveform, the bearing load predicted acceleration component waveform, and the gear meshing load predicted acceleration component waveform. Then the multiple measured acceleration component waveforms include the eccentric load measured acceleration component waveform, the misaligned load measured acceleration component waveform, the bearing load measured acceleration component waveform, and the gear meshing load measured acceleration component waveform.

[0111] Further, the multiple predicted acceleration component waveforms and the multiple measured acceleration component waveforms can be divided into multiple acceleration component waveform groups based on the frequency of the waveforms, and the multiple acceleration component waveform groups include: an eccentric load acceleration component waveform group, a misalignment load acceleration component waveform group, a bearing load acceleration component waveform group, and a gear meshing load predicted acceleration component waveform group. Each acceleration component waveform group includes a predicted acceleration component waveform and a measured acceleration component waveform. The target acceleration component waveform group can be any one of the multiple acceleration component waveform groups.

[0112] Among them, the gradient direction is the direction of parameter optimization. Assuming that the frequency of the predicted acceleration component waveform is greater than the amplitude of the measured acceleration component waveform based on the predicted acceleration component waveform and the measured acceleration component waveform, the gradient direction of the amplitude parameter is to reduce the amplitude. The parameter resolution is the step size of a parameter optimization. The parameter resolution can be flexibly set based on actual conditions. For example, if the amplitude of the predicted acceleration component waveform is 20 greater than the frequency of the measured acceleration component waveform, the parameter resolution can be 5.

[0113] The load parameters of the vibration equation are gradually adjusted until the preset iteration conditions are met to obtain the target load parameters, where the load parameters may include the amplitude and phase of each component waveform, etc., where the iteration condition may be that the difference between the load parameter values ​​of the predicted acceleration component waveform and the measured acceleration component waveform is less than the preset difference, or the number of iterations is less than the preset iteration threshold.

[0114] Furthermore, the target load parameters are substituted into the above vibration equation to obtain the target vibration equation.

[0115] In some embodiments, the above-mentioned solution process and the vibration equation are learned using a preset learning model, so the preset learning model can be trained to optimize the corresponding load parameters.

[0116] The design parameters and load parameters are input into the preset learning model, and the preset model outputs multiple training acceleration component waveforms. The multiple training acceleration component waveforms are compared with the corresponding measured acceleration component waveforms to obtain the corresponding loss function values. The load parameters are adjusted based on the loss function value, and training is continued until the loss function value is less than the preset loss function value or the number of training times is greater than the preset number of training times, so as to obtain the target load parameters.

[0117] In some embodiments, the method further comprises:

[0118] Input the design parameters and the current torque of the target gearbox into the digital twin model, and output multiple target acceleration component waveforms;

[0119] Multiple target acceleration component waveforms are input into a pre-trained fault identification model to output the fault cause and fault location.

[0120] Among them, the fault identification model can determine whether each target acceleration component waveform has a problem based on the waveform amplitude corresponding to each of the multiple target acceleration component waveforms.

[0121] For the acceleration component waveform of the eccentric load, if there is an obvious peak-to-peak value in the waveform, it may indicate that there is a serious eccentricity or imbalance problem inside the gearbox. This may be caused by damage, wear or poor assembly of the gear or shaft. Periodic waveforms usually indicate periodic vibration caused by gear meshing imbalance or bearing damage. This may be caused by gear damage, poor gear meshing or bearing wear. By performing frequency analysis on the vibration signal, the main frequency components of the vibration can be determined. These frequency components are usually associated with gear meshing frequency, bearing frequency and other system frequencies, which can help determine possible fault locations. Observing the phase difference of the vibration signal at different positions can help determine the propagation path of the vibration and the possible fault location. Changes in phase difference may indicate the reflection, propagation or interference of the vibration signal in the system.

[0122] Regarding the waveform of the acceleration component of the misaligned load, when fatigue, damage or fracture occurs on the tooth surface of a certain gear, the waveform of the misaligned load will change. Another possible reason is that the bearing inside the gearbox is faulty. Bearing failure usually causes the frequency component of the vibration signal to change, and it will cause specific waveform changes at the position of the misaligned load, such as the appearance of side frequencies or characteristic frequencies of the inner ring / outer ring. If the gears are not meshing well or the gears are damaged, the waveform of the misaligned load will also change. This situation may appear as periodic shocks or asynchronous vibration signals.

[0123] For the waveform of the acceleration component of the bearing load, inner ring damage or cracks: usually manifests as a higher frequency impact signal, which is related to the contact frequency of the ball or raceway. Outer ring damage or cracks: similar to inner ring damage, but the harmonic components in the spectrum may be different. Ball or roller damage: may cause periodic impact signals, and harmonic components related to the rolling frequency of rotation will appear in the spectrum. Raceway fatigue or cracks: usually produce low-frequency vibration signals, and side frequencies related to the rolling frequency of rotation may appear in the spectrum. Lubricant failure or deficiency: may cause increased friction and periodic high-frequency vibration signals. Lubricant contamination: a wider spectrum band may appear in the spectrum, reflecting unstable friction conditions. Improper installation or excessive bearing clearance will cause abnormal vibration signals, and irregular frequency components related to the speed may appear in the spectrum.

[0124] For the waveform of the acceleration component of the gear meshing load, tooth surface fatigue or fracture: usually manifested as a periodic impact signal, and harmonic components related to the gear rotation frequency will appear in the spectrum. Tooth surface wear: may cause the amplitude of the vibration signal to increase and the frequency component to change, manifested as an unstable vibration signal. Poor gear meshing or misalignment: may cause irregular frequency components to appear in the spectrum of the vibration signal, which is related to the gear rotation frequency and meshing frequency. Gear misalignment or offset: frequency components that do not match the meshing frequency may appear in the spectrum, manifested as asynchronous vibration signals. Bearing failure may cause the gear axis to shift or deform, thereby affecting the gear meshing quality, manifested as changes in the amplitude and frequency components of the vibration signal.

[0125] In some embodiments, Figure 2 A framework diagram of a dynamic simulation system for the vibration state of a gearbox is shown.

[0126] like Figure 2 As shown, the simulation system includes a gearbox model analysis module, a gearbox vibration model library, a gearbox vibration model parameter library, a vibration equation solver, a vibration signal adaptive filter, a discrete signal parameterization module, a load model parameter optimization module, and a gearbox load display module.

[0127] The gearbox vibration model library defines the vibration equation framework of the gearbox with different gear pair combinations. The concentrated mass method is used to simplify the gear shaft into several mass blocks. The simplified vibration equations corresponding to different gear pair combinations are different.

[0128] The gearbox vibration model parameter library defines the mass m, damping c, stiffness k and load-related parameters in the vibration equations of different gear pairs. The vibration equation is composed of the gearbox vibration model and the gearbox vibration model parameters.

[0129] The gearbox model parsing module searches for the corresponding vibration model and model parameters in the gearbox vibration model library and the gearbox vibration model parameter library according to the gearbox model;

[0130] The vibration equation solver uses the Ronck-Kutta method to solve the generated vibration equations;

[0131] The vibration signal feature extraction module uses smoothing algorithm and AMPD peak extraction algorithm to identify the peak value of the measured vibration signal spectrum;

[0132] The discrete signal parameterization module establishes a parameterized signal model based on the peak recognition result, fits the simulated signal and the measured signal, and forms a parameterized AM / FM signal model;

[0133] The load model parameter optimization module uses the gradient descent method to perform multi-objective parameter optimization on the load model parameters. After optimization iterations, the iterative calculation is stopped when the criterion converges;

[0134] The gearbox load display module dynamically displays the optimized gearbox load in the form of load curves, part load field cloud maps, etc.

[0135] Figure 3 A flow chart of a gearbox vibration state digital twin modeling method provided in an embodiment of the present application is shown.

[0136] like Figure 3 As shown in the figure, the process includes a virtual space gearbox vibration method, a measured signal extraction method, and a digital twin model parameter feedback optimization method.

[0137] The virtual space gearbox vibration method includes:

[0138] S301. Select the industrial gearbox shafting type.

[0139] Industrial gearbox shaft system types include parallel shaft spur gear pairs, planetary gear trains, bevel gear pairs, etc., and combined shaft system types formed by combinations of various types.

[0140] S302. Generate a vibration equation based on the industrial gearbox shaft system type.

[0141] Each industrial gearbox shaft system type corresponds to a vibration equation, and the vibration equation corresponding to the gearbox shaft system type can be called based on the industrial gearbox shaft system type.

[0142] S303, substitute design parameters, load parameters and applied torque.

[0143] Each gearbox needs to have a certain torque applied to it so that it can rotate to generate a vibration signal. Therefore, the applied torque corresponding to the target gearbox can be obtained. Further, the design parameters and load parameters corresponding to the industrial gearbox shaft system type are substituted into the above vibration equation.

[0144] S304, solving the above vibration equation based on the Longke Kutta algorithm to obtain a predicted vibration signal.

[0145] The above vibration equations are solved based on the Runke-Kutta method to obtain the displacement, velocity and acceleration of the mass block.

[0146] S305: Predict parameterization of vibration signals.

[0147] The displacement, velocity and acceleration of the mass block are substituted into the preset sub-vibration equation to obtain a plurality of predicted acceleration component waveforms corresponding to the predicted vibration signal.

[0148] S306: Use the preset learning model to learn the above steps S304 and S303.

[0149] The preset learning model is used as a proxy model to replace the vibration differential equation in the iterative process, thus shortening the iterative calculation time.

[0150] The measured signal extraction methods include:

[0151] S307. Select the vibration measurement point of the industrial gearbox.

[0152] The vibration measurement point is consistent with the output position of the vibration signal of the above-mentioned virtual space gearbox vibration method.

[0153] S308, measuring the actual vibration signal of the detection point.

[0154] A sensor may be provided to acquire a measured vibration signal at a detection point.

[0155] S309: Decompose the measured vibration signal.

[0156] Design a multi-bandpass filter to filter and decompose the measured vibration signal.

[0157] S310, load characteristic identification.

[0158] The decomposed waveforms of the multiple measured acceleration components are characterized and the load corresponding to each measured acceleration component waveform is determined.

[0159] S311. Parameterize the measured vibration signal.

[0160] The plurality of measured acceleration component waveforms are converted into acceleration component waveforms corresponding to the plurality of predicted acceleration component waveforms.

[0161] The digital twin model parameter feedback optimization method includes:

[0162] S312. Optimize load parameters using gradient descent method.

[0163] The gradient direction and parameter resolution are determined according to the predicted acceleration component waveform and the measured acceleration component waveform; the load parameters of the vibration equation are adjusted based on the gradient direction and the parameter optimization value until the preset iteration conditions are met to obtain the target load parameters.

[0164] Further, the target load parameters are substituted into the above vibration equation.

[0165] S313. Dynamically update the display results.

[0166] The present application also provides a model building device, which is used to execute the above Figure 1 The model building method provided in the embodiment. Figure 4 As shown, the device includes: a determination module 401, an acquisition module 402, an optimization module 403 and an establishment module 404.

[0167] A determination module 401 is used to determine a vibration equation and design parameters of a target gearbox, wherein the vibration equation is used to fit the vibration of an internal component of the target gearbox in motion;

[0168] The determination module 401 is further configured to determine a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters and the applied torque to the target gearbox;

[0169] An acquisition module 402 is used to acquire a measured vibration signal of the target position under the applied torque;

[0170] An optimization module 403 is used to optimize the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation;

[0171] The establishing module 404 is used to establish a digital twin model corresponding to the target gearbox based on the target vibration equation.

[0172] The present application proposes a model building device, which determines the vibration equation and design parameters of the target gearbox, wherein the vibration equation is used to fit the vibration of the internal components of the target gearbox in motion; based on the vibration equation, the design parameters and the applied torque to the target gearbox, determines the predicted vibration signal at the target position of the target gearbox; obtains the measured vibration signal at the target position under the applied torque; optimizes the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain the target vibration equation; and establishes a digital twin model corresponding to the target gearbox based on the target vibration equation. The embodiment of the present application can realize the dynamic transmission of the internal load of the gearbox by constructing the vibration equation corresponding to the gearbox, thereby making the output vibration signal more accurate, so that it can be determined whether the target gearbox has a fault based on the vibration signal.

[0173] In some embodiments, the determination module 401 is specifically configured to:

[0174] Determine a target gearbox corresponding to the target gearbox identifier;

[0175] Acquire a connection relationship of at least one gear corresponding to the target gearbox in the target gearbox and a shaft system type corresponding to the at least one gear;

[0176] Determining a vibration equation based on a connection relationship of the at least one gear in the target gearbox and a shaft system type corresponding to the at least one gear;

[0177] From the correspondence between the shaft system type and the design parameters, the design parameters corresponding to the shaft system type corresponding to the at least one gear are determined.

[0178] In some embodiments, the determination module 401 is further specifically configured to:

[0179] Solving the vibration equation using the design parameters and the applied torque to obtain a predicted vibration acceleration, a predicted velocity, and a predicted displacement of a target position;

[0180] The predicted vibration acceleration, the predicted velocity and the predicted displacement are substituted into a preset sub-vibration equation of the vibration equation to obtain a predicted vibration signal of the target position, wherein the predicted vibration signal includes a plurality of predicted acceleration component waveforms.

[0181] In some embodiments, the determination module 401 is further specifically configured to:

[0182] Using a preset learning model to learn the solution process and the vibration equation, to obtain a target learning model;

[0183] The design parameters are input into the target learning model to obtain a plurality of predicted acceleration component waveforms of a target gearbox target position.

[0184] In some embodiments, the determination module 401 is further specifically configured to:

[0185] De-noising the measured vibration signal to obtain a target vibration signal;

[0186] Performing peak extraction on the target vibration signal to obtain multiple vibration signal peaks;

[0187] Using a multi-bandpass filter to filter and decompose the multiple vibration signal peaks to obtain multiple measured acceleration component waveforms at the target position;

[0188] The vibration equation is optimized based on the multiple predicted acceleration component waveforms output by the target learning model and the multiple measured acceleration component waveforms to obtain a target vibration equation.

[0189] In some embodiments, the determination module 401 is further specifically configured to:

[0190] Selecting a target acceleration component waveform group, wherein the target acceleration component waveform group includes: a predicted acceleration component waveform and a measured acceleration component waveform;

[0191] Determining a gradient direction and a parameter optimization value according to the predicted acceleration component waveform and the measured acceleration component waveform;

[0192] Adjusting the load parameters of the vibration equation based on the gradient direction and the parameter optimization value until a preset iteration condition is met to obtain a target load parameter;

[0193] Based on the target load parameters, a target vibration equation is established.

[0194] The above device also includes an output module, which is used to:

[0195] Inputting design parameters and the current torque of the target gearbox into the digital twin model, and outputting a plurality of target acceleration component waveforms;

[0196] The multiple target acceleration component waveforms are input into a pre-trained fault identification model, and the fault cause and fault location are output.

[0197] The present application also provides an electronic device to execute the above model building method. Figure 5 It shows a schematic diagram of an electronic device provided by some embodiments of the present application. Figure 5As shown, the electronic device 7 includes: a processor 700, a memory 701, a bus 702 and a communication interface 703, and the processor 700, the communication interface 703 and the memory 701 are connected via the bus 702; the memory 701 stores a computer program that can be run on the processor 700, and when the processor 700 runs the computer program, it executes the model building method provided in any of the aforementioned embodiments of the present application.

[0198] The memory 701 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the device network element and at least one other network element is realized through at least one communication interface 703 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0199] The bus 702 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory 701 is used to store programs, and the processor 700 executes the programs after receiving the execution instructions. The model building method disclosed in any implementation of the aforementioned embodiment of the present application may be applied to the processor 700, or implemented by the processor 700.

[0200] The processor 700 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 700. The above processor 700 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a readily available programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 701, and the processor 700 reads the information in the memory 701 and completes the steps of the above method in combination with its hardware.

[0201] The electronic device provided in the embodiment of the present application and the model building method provided in the embodiment of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented therein.

[0202] The present application also provides a computer-readable storage medium corresponding to the model building method provided in the above embodiment. Figure 6 The computer-readable storage medium shown is a CD 30 on which a computer program (ie, a program product) is stored. When the computer program is run by a processor, the model building method provided by any of the aforementioned embodiments will be executed.

[0203] It should be noted that examples of computer-readable storage media may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0204] The computer-readable storage medium provided in the above-mentioned embodiments of the present application and the model building method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0205] It should be noted that:

[0206] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known structures and technologies are not shown in detail so as not to obscure the understanding of this description.

[0207] Similarly, it should be understood that in order to streamline the present application and help understand one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present application, the various features of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following schematic diagram: the claimed application requires more features than the features clearly stated in each claim. More specifically, as reflected in the claims below, the inventive aspects are less than all the features of the single embodiment disclosed above. Therefore, the claims following the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself serves as a separate embodiment of the present application.

[0208] In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present application and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.

[0209] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A model building method, characterized in that: include: Determine a vibration equation and design parameters of a target gearbox, wherein the vibration equation is used to fit the vibration of internal components of the target gearbox in motion; determining a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and an applied torque for the target gearbox; Acquiring a measured vibration signal of the target position under the applied torque; Optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation; Establishing a digital twin model corresponding to a target gearbox based on the target vibration equation; The step of determining a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and the applied torque to the target gearbox comprises: Solving the vibration equation using the design parameters and the applied torque to obtain a predicted vibration acceleration, a predicted velocity, and a predicted displacement of a target position; Substituting the predicted vibration acceleration, the predicted velocity and the predicted displacement into a preset sub-vibration equation of the vibration equation to obtain a predicted vibration signal of the target position, wherein the predicted vibration signal includes a plurality of predicted acceleration component waveforms; The step of determining a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and the applied torque to the target gearbox comprises: Using a preset learning model to learn the solution process and the vibration equation, to obtain a target learning model; Inputting the design parameters into the target learning model to obtain a plurality of predicted acceleration component waveforms of a target position of a target gearbox; The step of optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation includes: De-noising the measured vibration signal to obtain a target vibration signal; Performing peak extraction on the target vibration signal to obtain multiple vibration signal peaks; Using a multi-bandpass filter to filter and decompose the multiple vibration signal peaks to obtain multiple measured acceleration component waveforms at the target position; The vibration equation is optimized based on the multiple predicted acceleration component waveforms output by the target learning model and the multiple measured acceleration component waveforms to obtain a target vibration equation.

2. The method according to claim 1, characterized in that The step of determining the vibration equation and design parameters of the target gearbox includes: Determine a target gearbox corresponding to the target gearbox identifier; Acquire a connection relationship of at least one gear corresponding to the target gearbox in the target gearbox and a shaft system type corresponding to the at least one gear; Determining a vibration equation based on a connection relationship of the at least one gear in the target gearbox and a shaft system type corresponding to the at least one gear; From the correspondence between the shaft system type and the design parameters, the design parameters corresponding to the shaft system type corresponding to the at least one gear are determined.

3. The method according to claim 1, characterized in that The step of optimizing the vibration equation based on the multiple predicted acceleration component waveforms output by the target learning model and the multiple measured acceleration component waveforms to obtain a target vibration equation includes: Selecting a target acceleration component waveform group, wherein the target acceleration component waveform group includes: a predicted acceleration component waveform and a measured acceleration component waveform; Determining a gradient direction and a parameter optimization value according to the predicted acceleration component waveform and the measured acceleration component waveform; Adjusting the load parameters of the vibration equation based on the gradient direction and the parameter optimization value until a preset iteration condition is met to obtain a target load parameter; Based on the target load parameters, a target vibration equation is established.

4. The method according to any one of claims 1 to 3, characterized in that: The method further comprises: Inputting design parameters and the current torque of the target gearbox into the digital twin model, and outputting a plurality of target acceleration component waveforms; The multiple target acceleration component waveforms are input into a pre-trained fault identification model, and the fault cause and fault location are output.

5. A model building device, characterized in that: include: A determination module, used to determine the vibration equation and design parameters of the target gearbox, wherein the vibration equation is used to fit the vibration of the internal components of the target gearbox in motion; The determination module is further configured to determine a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and an applied torque to the target gearbox; The step of determining a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and the applied torque to the target gearbox comprises: Solving the vibration equation using the design parameters and the applied torque to obtain a predicted vibration acceleration, a predicted velocity, and a predicted displacement of a target position; Substituting the predicted vibration acceleration, the predicted velocity and the predicted displacement into a preset sub-vibration equation of the vibration equation to obtain a predicted vibration signal of the target position, wherein the predicted vibration signal includes a plurality of predicted acceleration component waveforms; The step of determining a predicted vibration signal at a target position of a target gearbox based on the vibration equation, the design parameters, and the applied torque to the target gearbox comprises: Using a preset learning model to learn the solution process and the vibration equation, to obtain a target learning model; Inputting the design parameters into the target learning model to obtain a plurality of predicted acceleration component waveforms of a target position of a target gearbox; An acquisition module, used for acquiring a measured vibration signal of the target position under the applied torque; An optimization module, configured to optimize the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation; The step of optimizing the vibration equation based on the predicted vibration signal and the measured vibration signal to obtain a target vibration equation includes: De-noising the measured vibration signal to obtain a target vibration signal; Performing peak extraction on the target vibration signal to obtain multiple vibration signal peaks; Using a multi-bandpass filter to filter and decompose the multiple vibration signal peaks to obtain multiple measured acceleration component waveforms at the target position; Optimizing the vibration equation based on the multiple predicted acceleration component waveforms output by the target learning model and the multiple measured acceleration component waveforms to obtain a target vibration equation; A module is established to establish a digital twin model corresponding to the target gearbox based on the target vibration equation.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor runs the computer program to implement the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the method according to any one of claims 1 to 4.

Images