CN116414085A

CN116414085A - Method and system for detecting dynamic characteristic deviation of feeding system

Info

Publication number: CN116414085A
Application number: CN202111651719.8A
Authority: CN
Inventors: 陈贤佑; 邱昱阩; 郑志钧; 程文男; 刘济铭
Original assignee: Hiwin Technologies Corp
Current assignee: Hiwin Technologies Corp
Priority date: 2021-12-30
Filing date: 2021-12-30
Publication date: 2023-07-11

Abstract

The invention provides a method for detecting dynamic characteristic deviation of a feeding system, which comprises the following steps: exciting the feeding system in a monitoring mode, and detecting the vibration of a sub-component of a component to be detected of the feeding system to generate a monitoring excitation signal; calculating a group of monitoring eigenvalues and a group of monitoring eigenvectors of the monitoring excitation signal by using a modal analysis method; judging the similarity between the group of monitoring eigenvalues and the group of monitoring eigenvectors and a group of standard eigenvalues and a group of standard eigenvectors of a digital twin model respectively by a mode verification method; and when the group of monitoring feature values and the group of monitoring feature vectors are judged to be dissimilar to the group of standard feature values and the group of standard feature vectors respectively, judging that the dynamic characteristics of the sub-component deviate. The sub-components of the feed system, which deviate in their dynamic behavior, can thereby be detected remotely and precisely.

Description

Method and system for detecting dynamic characteristic deviation of feeding system

Technical Field

The present invention relates to a status detecting system, and more particularly, to a method and a system for detecting dynamic characteristic deviation of a feeding system.

Background

In precision machinery manufacturing, it is particularly important to optimize a production machine, so for example, WO2020053083A1 provides a technique for optimizing a machine by using internal data of a digital twin model regulation machine controller of the machine. However, this technique can only read the equipment processing related data, and cannot diagnose whether the equipment element is abnormal.

The publication CN112446104a provides a method for identifying the deviation between an automation device and its digital twinning, which is to monitor parameters such as temperature, speed, acceleration, etc. by using sensors mounted on a body of processing material, and then compare the sensed result with digital twinning data simulated in advance, and when the sensor result is different from the analog value, identify the abnormal station of the workpiece in the automation device. However, this method cannot grasp the cause of occurrence of the abnormality and whether the dynamic characteristics of the transmission element system are degraded.

The patent US20210123830 discloses a method for monitoring the health status of a machine tool, which uses the data collected by a sensor to establish a health feature cluster (including displacement transmissibility, natural frequency, etc.) while exciting production equipment, and uses the cluster as a basis for judging the health status of the machine tool. However, this method still cannot accurately know which component on the machine is abnormal, and has the problem of poor sensitivity.

Since the state change of each element on the machine cannot be monitored in real time, the machine is usually detected to be abnormal when the finished product size is bad or abnormal noise is generated by the machine, and then the abnormal part on the machine can be tested by using the operation mode analysis (Operational Modal Analysis, OMA) technology. Not only is time consuming and laborious, but also the experimental data are not versatile.

Disclosure of Invention

The present invention is directed to a method and system for detecting deviations in the dynamic characteristics of a feed system that allows a monitor to remotely monitor the dynamic characteristics of key components (i.e., sub-components) in the feed system.

Another object of the present invention is to provide a method and a system for detecting deviation of dynamic characteristics of a feeding system, which can make a monitor instantly know whether the dynamic characteristics of each critical component deviate or are abnormal, so as to instantly process the abnormal critical component.

It is still another object of the present invention to provide a method and system for detecting a deviation of dynamic characteristics of a feed system, which can rapidly detect abnormal key components, thereby shortening a time for eliminating an obstacle to the stop of a production line.

It is still another object of the present invention to provide a method and system for detecting dynamic characteristic deviations of a feeding system, which can enable standard digital twin models corresponding to components to be detected to be used for feeding systems of different specifications.

To achieve the above and other objects, according to one embodiment of the present invention, a method for detecting a dynamic deviation of a feeding system, the feeding system includes at least one part to be detected, each part to be detected includes at least one sub-part, the method for detecting a dynamic deviation of the feeding system is performed by a processor and includes the following steps: (A) Providing a sensor on the sub-assembly, the sensor in communication with the processor; (B) Exciting the feeding system in a monitoring mode, detecting the vibration of the corresponding sub-component through the sensor, and generating a monitoring excitation signal; (C) Calculating a group of monitoring eigenvalues and a group of monitoring eigenvectors of the monitoring excitation signal by a modal analysis method; (D) Judging the similarity of the group of monitoring eigenvalues and the group of monitoring eigenvectors corresponding to the sensor with a group of standard eigenvalues and a group of standard eigenvectors respectively by a mode verification method, wherein the group of standard eigenvalues and the group of standard eigenvectors are standard dynamic characteristics of a digital twin model, and the digital twin model is established for at least one part to be detected; and (E) when the group of monitoring eigenvalues and the group of monitoring eigenvectors corresponding to the sensor are judged to be dissimilar to the group of standard eigenvalues and the group of standard eigenvectors respectively, judging that the dynamic characteristics of the sub-component corresponding to the sensor deviate.

In some embodiments, the digital twin model corresponds to a set of first initial feature values and a set of first initial feature vectors that are generated when the digital twin model is built for the at least one part to be detected that is not mounted to the feed system, and the set of standard feature values and the set of standard feature vectors of the digital twin model are generated by: (F) Exciting the feeding system in an initial mode, detecting the vibration of the corresponding sub-component through the sensor and generating an initial excitation signal, wherein the initial mode refers to the stage of the feeding system before the at least one to-be-detected component just leaves the factory is assembled in the monitoring mode; (G) Calculating a set of second initial eigenvalues and a set of second initial eigenvectors of the initial excitation signal by using the modal analysis method; and (H) estimating the set of standard characteristic values and the set of standard characteristic vectors corresponding to the sensor according to the set of first initial characteristic values, the set of first initial characteristic vectors, the set of second initial characteristic values and the set of second initial characteristic vectors corresponding to the sensor through an optimization method.

In some embodiments, this step (H) is performed by the following formula:

wherein, the liquid crystal display device comprises a liquid crystal display device,

an nth first initial feature value of the set of first initial feature values; omega _n An nth second initial feature value of the set of second initial feature values; Δmac _n The similarity between the nth first initial feature vector in the set of first initial feature vectors and the nth second initial feature value in the set of second initial feature vectors is calculated according to a mode reliability criterion (Modal Assurance Criterion, MAC), and n is a positive integer.

In some embodiments, the method for detecting dynamic characteristic deviation of a feeding system further comprises the steps of: (I) Judging whether the group of monitoring characteristic values is smaller than a monitoring threshold value or not; and (J) when the group of monitoring characteristic values is smaller than the monitoring threshold value, judging that the sub-component corresponding to the sensor is abnormal.

In some embodiments, when the set of monitoring feature values and the set of monitoring feature vectors are determined to be dissimilar to the set of standard feature values and the set of standard feature vectors, respectively, the method for detecting a dynamic characteristic deviation of the feed system further comprises the steps of: (K) And updating the set of standard eigenvalues and the set of standard eigenvectors according to the set of monitoring eigenvalues and the set of monitoring eigenvectors by an optimization method, and defining the updated set of standard eigenvalues and the updated set of standard eigenvectors as the updated dynamic characteristics of the digital twin model.

In some embodiments, the modal analysis is an experimental modal analysis or an operational modal analysis.

In some embodiments, the feed system is excited in a tapping or motor-driven manner.

In some embodiments, the dynamic characteristics of the sub-component include mass, damping, or stiffness.

In some embodiments, the component to be detected is a linear slide or a ball screw, and the sub-component is a slide or a slide block when the component to be detected is the linear slide, or a screw or a nut when the component to be detected is the ball screw.

The present invention further provides a system for detecting dynamic property deviation of a feeding system according to an embodiment, the system comprises a processor configured to execute the method for detecting dynamic property deviation of a feeding system.

Drawings

Other aspects of the invention and their advantages will be found after studying the detailed description in conjunction with the following drawings:

FIG. 1 is a functional block diagram of a system for detecting dynamic bias of a feed system according to an embodiment of the invention.

FIG. 2 is a flow chart of a method of establishing standard dynamics of a digital twin model in an initial mode according to an embodiment of the present invention.

FIG. 3 is a flow chart of a method for monitoring dynamic bias and anomalies in a monitoring mode according to one embodiment of the present invention.

FIG. 4 is a schematic diagram of a sensor installed in a feed system at a viewing angle according to one embodiment of the present invention.

FIG. 5 is a schematic diagram of a sensor installed in a feed system at another view angle according to an embodiment of the present invention.

FIG. 6 is a graph of natural frequency versus slider stiffness for a work platform according to an embodiment of the present invention.

Description of the reference numerals

1 System for detecting dynamic characteristic deviations of a feed system

10 Server

11,11A,11B,11C,11D: sensor

12 vibration excitation device

13 processor(s)

131 control unit

132 modal analysis unit

133 Standard building Unit

134 similarity judging unit

135 abnormality determination unit

136 model updating unit

14 reservoir(s)

141 database

2 feed system

21 Linear module

22 working platform

23 ball screw

231 screw rod

232 screw cap

24:linear slide rail

241 slide rail

242,242A,242B,242C,242D, slider

M monitoring threshold value

F1, F2 natural frequency

R1 and R2, rigidity value.

Detailed Description

In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be understood by those of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures and/or components have not been described in detail so as not to obscure the present invention.

Referring to fig. 1 to 5, a system 1 for detecting dynamic characteristic deviation of a feeding system (hereinafter referred to as system 1) according to an embodiment of the present invention may be applied to perform a method for detecting dynamic characteristic deviation of a feeding system 2. The feeding system 2 comprises at least one linear module 21, each linear module 21 comprising a plurality of components, such as, but not limited to, a work platform 22 and at least one linear actuator. In the embodiment shown in fig. 4, the number of the linear driving devices is 3, which are a ball screw 23 and two linear slides 24 respectively; the ball screw 23 comprises a plurality of sub-components, screw 231 and nut 232, respectively; each linear slide rail 24 includes a plurality of sub-components, including a slide rail 241 and two slide blocks 242 mounted on the working platform 22 and movably sleeved on the slide rail 241. Each component has its dynamic characteristics such as, but not limited to, mass, damping, and stiffness. In the present embodiment, an example will be described in which the rigidity deviation of the four sliders 242 is confirmed with the two linear slides 24 as members to be detected.

The system 1 may include, for example and without limitation, a server 10, at least one sensor 11, and an excitation device 12. The server 10 includes a processor 13 and a memory 14 electrically connected to the processor 13.

The server 10 is installed with a plurality of software, so that the processor 13 and the storage 14 can be configured to include a control unit 131, a modality analysis unit 132, a standard establishment unit 133, a similarity determination unit 134, an anomaly determination unit 135, a model update unit 136 and a database 141 in the operation of the software. The control unit 131 may communicate with the modality analysis unit 132, the modality analysis unit 132 may communicate with the standard establishment unit 133, the similarity determination unit 134 may communicate with the abnormality determination unit 135, the abnormality determination unit 135 may communicate with the model update unit 136, and these units may communicate with the database 141 for accessing the database 141. The control unit 131 also communicates with the sensor 11 and the excitation device 12 to control the operation of the sensor 11 and the excitation device 12.

The database 141 may store data such as, but not limited to, algorithms, thresholds, and various correspondences. The various corresponding relationships may be, for example, but not limited to, a corresponding relationship between a stiffness value and a natural frequency, a corresponding relationship between a material and a density, a corresponding relationship between a material and the young system , and a corresponding relationship between the sensor 11 and a sub-component of an object to be detected. The database 141 may also store related data of each component, such as, but not limited to, size data, texture data, position data, preset rigidity values, and corresponding relationships thereof. The dimensional data may be created or set, for example, but not limited to, by drawing software (e.g., without limitation, autoCAD) installed in the system 1 when drawing a three-dimensional part image of the part. The position data may be obtained by sampling pixel coordinates from the three-dimensional part image, for example, but not limited to, using a finite element method (Finite Element Method, FEM) or a continuum method (Continuum Mechanics). The position data is also related to the relative position of this component to other components in the feed system 2.

In the present embodiment, in order to detect the rigidity deviation of the sliders 242A to 242D, the number of the sensors 11 may be set to 4, that is, the sensors 11A to 11D, and the sliders 242A to 242D are respectively provided on the top surface of the work platform 22, as shown in fig. 4. However, the present invention is not limited to this embodiment. The sensor 11 is configured to detect vibrations of the work platform 22, and the sensor 11 may be, for example, but not limited to, an accelerometer. The excitation device 12 is configured to excite the feed system 2 to vibrate the table 22 by an external force. The excitation device 12 may excite the feed system 2, for example, but not limited to, in a tapping or motor-driven manner.

The following illustrates a method of detecting a rigid deviation of the two linear slides 24. In this detection method, the processor 13 first enters an initial mode to establish a standard of virtual dynamic characteristics, and then enters a monitoring mode to periodically or aperiodically monitor whether the dynamic characteristics of the sliders 242A-242D after starting operation deviate or are abnormal according to the standard. The initial mode refers to a stage in which the two-way linear guide 24 (i.e., the component to be inspected) just before shipment is assembled to the feed system 2 before entering the monitor mode. The monitoring mode refers to the stage at which the assembled feed system 2 has started to operate and needs to be monitored.

Referring to fig. 1 and 2 together, in the initial mode, the method for establishing the criteria of the virtual dynamic characteristics may include, for example, but not limited to, the following steps.

First, in step S11, since the sliders 242A-242D of the two linear slides 24 are fixed on the bottom surface of the working platform 22, and the slider pre-compression affects the vibration modes of the feeding structure (the two linear slides 24 together with the working platform 22) to different degrees, the mode analysis unit 132 may select the working platform 22 distributed with the sliders 242A-242D as a reference target for establishing a digital twin model, obtain the dimension data (such as but not limited to the length, the width and the height) of the working platform 22, the material data and the position data of the working platform 22, and obtain the rigidity value range of the slider 242 from the database 141, and based on these data, establish a digital twin model of the working platform 22 for the two linear slides 24 that have not been installed to the feeding system 2 by the software (such as but not limited to computer aided engineering (Computer Aided Engineering, CAE) software (such as but not limited to ANSYS push analysis software), and calculate a set of first initial feature values and a set of first initial feature vectors as initial dynamic characteristics of the twin model. The first initial eigenvalue is the initial natural frequency of the digital twin model, and the first initial eigenvector is the initial modality of the digital twin model. The digital twin model at this time is an initial digital twin model that has not been calibrated, and thus its initial dynamic characteristics slightly differ from the true dynamic characteristics of the two linear slides 24 that have been mounted to the feed system 2. The initial digital twin model and its initial dynamics are stored in the database 141 for later lookup.

On the other hand, in step S12, the sensors 11A to 11D may be mounted on the sliders 242A to 242D, respectively, as shown in fig. 4 and 5.

Next, in step S13, in the initial mode, the control unit 131 controls the excitation device 12 to excite the stationary feed system 2, and the sliders 242A to 242D vibrate. Meanwhile, the control unit 131 also controls the four sensors 11 to detect the vibration of the sliders 242A-242D, and the four sensors 11 correspondingly generate four initial excitation signals and transmit the four initial excitation signals back to the control unit 131.

Then, in step S14, the control unit 131 provides the four initial excitation signals to the modal analysis unit 132, and the modal analysis unit 132 converts each initial excitation signal from the time domain signal to the frequency domain signal via the fast fourier transform (Fast Fourier Transform, FFT) by a modal analysis method through software (such as but not limited to CAE software) stored in the memory 14, so as to calculate a set of second initial eigenvalues and a set of second initial eigenvectors of each initial excitation signal. The second initial eigenvalue and the second initial eigenvector are the actual natural frequency and the actual modality, respectively, of the working platform 22 that has been mounted to the feed system 2. The modal analysis may be, for example, but not limited to, an experimental modal analysis or an operational modal analysis.

Finally, in step S15, the standard creating unit 133 obtains the set of second initial feature values and the set of second initial feature vectors of the initial excitation signal of each sensor 11 from the modal analyzing unit 132, obtains the set of first initial feature values and the set of first initial feature vectors from the database 141, and then uses software stored in the storage 14 (for example, but not limited to CAE software) to estimate a set of standard feature values and a set of standard feature vectors corresponding to each sensor 11 as standard dynamic characteristics of the digital twin model according to the set of first initial feature values, the set of first initial feature vectors, the set of second initial feature values and the set of second initial feature vectors corresponding to each sensor 11 by an optimization method. The digital twin model at this time is a calibrated standard digital twin model whose standard dynamics conform to the true dynamics of the two linear slides 24 that have been mounted to the feed system 2. The standard digital twin model and its standard dynamic characteristics are recorded in the database 141 by the standard establishing unit 133, and the corresponding relations between the standard dynamic characteristics and the four sensors 11 are also recorded in the database 141 by the standard establishing unit 133 for subsequent searching.

The optimization method described above can be performed, for example and without limitation, by the following formulaIn (a)

An nth first initial feature value of a set of first initial feature values; omega _n An nth second initial feature value of the set of second initial feature values; Δmac _n The similarity between the nth first initial feature vector in the group of first initial feature vectors and the nth second initial feature value in the group of second initial feature vectors is calculated according to a mode reliability criterion, and n is a positive integer.

After establishing the standard dynamics of the digital twin model, the processor 13 may enter a monitoring mode to further monitor the dynamic bias and anomalies of the sliders 242A-242D. As shown in fig. 1 and 3, in the monitoring mode, the method of monitoring the dynamic characteristic deviation and abnormality of the sliders 242A to 242D may include, for example, but not limited to, the following steps.

First, in step S21, the control unit 131 controls the excitation device 12 to excite the stationary feed system 2 in the monitor mode, and vibrates the sliders 242A to 242D. Meanwhile, the control unit 131 also controls the four sensors 11 to detect the vibration of the sliders 242A-242D, and the four sensors 11 correspondingly generate four monitoring excitation signals and transmit the four monitoring excitation signals back to the control unit 131.

Next, in step S22, the control unit 131 may provide the four monitoring excitation signals to the modal analysis unit 132, and the modal analysis unit 132 may convert the monitoring excitation signals corresponding to the sensors 11 from the time domain signals to the frequency domain signals via the fast fourier transform by using software (such as, but not limited to, CAE software) stored in the memory 14, so as to calculate a set of monitoring eigenvalues and a set of monitoring eigenvectors of the monitoring excitation signals of each sensor 11. And, the modality analysis unit 132 may further record the set of monitoring feature values and the set of monitoring feature vectors corresponding to each sensor 11 to the database 141. The modal analysis may be, for example, but not limited to, an experimental modal analysis or an operational modal analysis.

Then, in step S23, the similarity determining unit 134 may obtain, from the database 141, the standard dynamic characteristics of the digital twin model (i.e., the set of monitoring feature values and the set of monitoring feature vectors corresponding to the sensors 11), and obtain, from the database 141, the set of monitoring feature values and the set of monitoring feature vectors corresponding to the sensors 11 obtained by the calculation in step S22. Furthermore, the similarity determination unit 134 may further determine the similarity between the set of standard feature vectors and the set of monitoring feature vectors corresponding to the same sensor 11 and the set of standard feature values and the set of monitoring feature values corresponding to the same sensor 11 by using a mode verification method through software (such as, but not limited to, CAE software) stored in the storage 14. The modality validation method may be, for example, but not limited to, a modality reliability criterion.

For the example of judging the similarity of a set of standard feature vectors and a set of monitoring feature vectors with the modal reliability criterion, it can be calculated by the following formula, wherein MAC (r, q) represents the similarity;

a matrix representing the set of monitoring feature vectors;

a matrix representing the set of standard feature vectors; />

Is a transpose matrix representing the set of monitoring feature vectors; />

Is a transpose matrix representing the set of standard feature vectors.

Next, in step S24, the similarity determining unit 134 further compares the similarity obtained in step S23 with a similarity threshold (for example, but not limited to, 0.8) to determine whether the set of monitoring feature values and the set of monitoring feature vectors corresponding to the same sensor 11 are similar to the set of standard feature values and the set of standard feature vectors corresponding to the same sensor 11, respectively.

In step S24, when the similarity between the set of monitoring feature values corresponding to the same sensor 11 and the set of standard feature values is greater than or equal to the similarity threshold value, the similarity determination unit 134 determines that the set of monitoring feature values is similar to the set of standard feature values. Similarly, in step S24, when the similarity between the set of monitoring feature vectors corresponding to the same sensor 11 and the set of standard feature vectors is greater than or equal to the similarity threshold, the similarity determination unit 134 determines that the set of monitoring feature vectors is similar to the set of standard feature vectors. Since the database 141 records the correspondence between each sensor 11 and each slider 242, the similarity determination unit 134 may further determine that the current rigidity value of the slider 242 corresponding to the sensor 11 corresponding to the set of monitoring feature vectors similar to the set of standard feature vectors has no deviation according to the determination result of step S24 in step S25, and record the result in the database 141.

In contrast, in step S24, when the similarity between the set of monitoring feature values corresponding to the same sensor 11 and the set of standard feature values is smaller than the similarity threshold value, the similarity determination unit 134 determines that the set of monitoring feature values is not similar to the set of standard feature values. Similarly, in step S24, when the similarity between the set of monitoring feature vectors corresponding to the same sensor 11 and the set of standard feature vectors is smaller than the similarity threshold, the similarity determination unit 134 determines that the set of monitoring feature vectors is not similar to the set of standard feature vectors. At this time, the anomaly determination unit 135 may further obtain a monitoring threshold value M from the database 141 in step S26, and determine whether the set of monitoring feature values dissimilar to the set of standard feature values is smaller than the monitoring threshold value M.

If not, in step S26, the current dynamic characteristics of the slider 242 corresponding to the sensor 11 corresponding to the set of monitoring feature values are only deviated from the standard, but the deviation degree is still within the allowable range, so the abnormality determination unit 135 will determine that the current rigidity value of the slider 242 deviates in step S27, and record the determination result in the database 141. Then, in step S28, the model updating unit 136 may update the digital twin model and its virtual dynamic characteristics (i.e., the sets of standard feature values and the sets of standard feature vectors) recorded in the database 141 according to the set of monitoring feature values and the set of monitoring feature vectors by using software (for example, but not limited to, CAE software) stored in the storage 14 to optimize the method, and define updated sets of standard feature values and updated dynamic characteristics of the digital twin model according to the updated sets of standard feature values and the updated sets of standard feature vectors.

In contrast, in step S26, if yes, the current stiffness value of the slider 242 corresponding to the sensor 11 corresponding to the set of monitoring feature values not only deviates from the standard, but also exceeds the allowable range, so the anomaly determination unit 135 determines that the current stiffness value of the slider 242 has been abnormal in step S29, and correspondingly generates an anomaly signal and records the determination result in the database 141.

In the following examples of the sensors 11A to 11D, since the wear of the sliders 242A to 242D of the two linear slides 24 is different after a period of operation, in step S24, it is determined whether the set of monitoring feature values corresponding to each sensor 11 is similar to the corresponding set of standard feature values, it can be primarily determined that the rigidity value of at least one of the sliders 242A to 242D is deviated, and in step S24, it is determined whether the set of monitoring feature vectors corresponding to each sensor 11 is similar to the corresponding set of standard feature vectors, and it can be more precisely determined which of the sliders 222 of the sliders 242A to 242D has the rigidity value deviated. As shown in fig. 6, the vertical axis represents different natural frequencies of the working platform 22, the horizontal axis represents different rigidity values of the slider, each natural frequency corresponds to a rigidity value, the rigidity value R1 corresponding to the natural frequency F1 is the rigidity value of the slider with high pre-compression just delivered, the rigidity value R2 corresponding to the natural frequency F2 (i.e. the monitor threshold value M) is the rigidity value of the slider with no pre-compression, and the correspondence between the natural frequency and the rigidity value of the slider is stored in the database 141 in advance.

In step S24 of this example, if the set of monitoring feature values and the set of monitoring feature vectors corresponding to each sensor 11 of the sensors 11A-11D are not similar to the set of standard feature values and the set of standard feature vectors corresponding to the sensor 11, the set of monitoring feature values corresponding to each sensor 11 is smaller than the natural frequency F1, and the rigidity values of the sliders 242A-242D corresponding to the sensors 11A-11D are also reduced and lower than the rigidity value R1.

In step S26 of this example, if only the set of monitoring characteristic values corresponding to the sensor 11D is smaller than the natural frequency F2, it indicates that the rigidity value of the sliders 242A-242C corresponding to the sensors 11A-11C is still greater than or equal to the rigidity value R2 although it is decreasing, and the rigidity value of the slider 242D corresponding to the sensor 11D is already decreasing to be lower than the rigidity value R2, the decreasing amplitude thereof is beyond the allowable range, and the slider 242D is in an abnormal state. At this time, the abnormality determination unit 135 will send out an abnormality signal corresponding to the slider 242D. Therefore, abnormal elements can be quickly found out.

On the other hand, if the monitoring information (i.e. each obtained set of monitoring feature values and each set of monitoring feature vectors and the detection results thereof) obtained at each time point can be presented in a user interface (not shown) in communication with the processor 13, the monitor can instantly control the states of the sliders 242A-242D at the time point, so that when an abnormality occurs in one slider 242 (e.g. the slider 242D), the abnormal slider 242D can be instantly replaced. In addition, by the steps S24 to S27 and S29, the monitor can not only know that at least one of the sliders 242A to 242D mounted on the working platform 22 is damaged and needs to be replaced, but also know which slider 242 needs to be replaced accurately. The user interface may be displayed on a display of a computer device electrically connected to the server 10 and a display of a computer device remotely connected to the server 10. Therefore, the purpose of monitoring at the near end and the far end can be achieved.

Although the above embodiments use two linear slides 24 as the members to be detected, the present invention is not limited thereto. In other embodiments, the ball screw 23 may also be used as the component to be inspected; in step S11, the screw 231 is selected as a reference target for establishing a digital twin model, and a digital twin model is established and a set of first initial feature values and a set of first initial feature vectors of the digital twin model are calculated by using the rigidity value range, screw size, screw material and screw position data of the screw cap 232 (i.e., subcomponent) stored in advance in the database 141; the change in the stiffness value of the nut 232 is then monitored in real time by steps S12-S15 and steps S21-28, wherein the sensor 11 is mounted on the nut 232.

In addition, the built digital twin models can be applied to feeding systems with different specifications, and the method for detecting dynamic characteristic deviation of the feeding systems provided by the invention can be applied to feeding systems with different specifications.

Although the present invention is disclosed in the above embodiments, the present invention is not limited thereto. The present invention is not limited to the above-described embodiments, but is capable of modification, variation and combination of embodiments without departing from the spirit and scope of the present invention. Reference is made to the appended claims for a review of the scope of the invention.

Claims

1. A method of detecting a deviation in a dynamic characteristic of a feed system, the feed system comprising at least one part to be detected, each part to be detected comprising at least one sub-part, characterized in that: the method for detecting dynamic characteristic deviation of the feeding system is executed by a processor and comprises the following steps:

(A) Providing a sensor on the sub-assembly, the sensor in communication with the processor;

(B) Exciting the feeding system in a monitoring mode, detecting the vibration of the corresponding sub-component through the sensor, and generating a monitoring excitation signal;

(C) Calculating a group of monitoring eigenvalues and a group of monitoring eigenvectors of the monitoring excitation signal by a modal analysis method;

(D) Judging the similarity of the group of monitoring eigenvalues and the group of monitoring eigenvectors corresponding to the sensor with a group of standard eigenvalues and a group of standard eigenvectors respectively by a mode verification method, wherein the group of standard eigenvalues and the group of standard eigenvectors are standard dynamic characteristics of a digital twin model, and the digital twin model is established for at least one part to be detected; and

(E) And when the group of monitoring eigenvalues and the group of monitoring eigenvectors corresponding to the sensor are judged to be dissimilar to the group of standard eigenvalues and the group of standard eigenvectors respectively, the dynamic characteristics of the sub-component corresponding to the sensor are judged to deviate.

2. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: said digital twin model corresponds to a set of first initial eigenvalues and a set of first initial eigenvectors, which are generated when the digital twin model is built for the at least one part to be detected that is not mounted to the feed system, and which are generated by the steps of:

(F) Exciting the feeding system in an initial mode, detecting the vibration of the corresponding sub-component through the sensor and generating an initial excitation signal, wherein the initial mode refers to the stage of the feeding system before the at least one to-be-detected component just leaves the factory is assembled in the monitoring mode;

(G) Calculating a set of second initial eigenvalues and a set of second initial eigenvectors of the initial excitation signal by using the modal analysis method; and

(H) And estimating the set of standard characteristic values and the set of standard characteristic vectors corresponding to the sensor according to the set of first initial characteristic values, the set of first initial characteristic vectors, the set of second initial characteristic values and the set of second initial characteristic vectors corresponding to the sensor through an optimization method.

3. A method of detecting a deviation in dynamic characteristics of a feed system as claimed in claim 2, wherein: the step (H) is performed by the following formula:

an nth first initial feature value of the set of first initial feature values; omega _n An nth second initial feature value of the set of second initial feature values; Δmac _n The similarity between the nth first initial feature vector in the group of first initial feature vectors and the nth second initial feature value in the group of second initial feature vectors is calculated according to a mode reliability criterion, and n is a positive integer.

4. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: further comprising the following steps:

(I) Judging whether the group of monitoring characteristic values is smaller than a monitoring threshold value or not; a kind of electronic device with high-pressure air-conditioning system

(J) And when the group of monitoring characteristic values is smaller than the monitoring threshold value, judging that the sub-component corresponding to the sensor is abnormal.

5. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: when the set of monitoring feature values and the set of monitoring feature vectors are not similar to the set of standard feature values and the set of standard feature vectors, respectively, the method for detecting the dynamic characteristic deviation of the feeding system further comprises the following steps:

(K) And updating the set of standard eigenvalues and the set of standard eigenvectors according to the set of monitoring eigenvalues and the set of monitoring eigenvectors by an optimization method, and defining the updated set of standard eigenvalues and the updated set of standard eigenvectors as the updated dynamic characteristics of the digital twin model.

6. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: the modal analysis method is an experimental modal analysis method or an operation modal analysis method.

7. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: the feed system is excited in a tapping or motor-driven manner.

8. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: this dynamic characteristic of the sub-assembly includes mass, damping or stiffness.

9. A method of detecting a deviation in dynamic characteristics of a feed system according to claim 1, wherein: the component to be detected is a linear slide rail or a ball screw, when the component to be detected is the linear slide rail, the sub-component is a slide rail or a slide block, and when the component to be detected is the ball screw, the sub-component is a screw or a screw cap.

10. A system for detecting a deviation in a dynamic characteristic of a feed system, characterized by: comprising a processor configured to perform the method of detecting a deviation in a dynamic characteristic of a feed system as claimed in claim 1.