KR102419817B1 - Method for assessing preload degradation of ball screw - Google Patents

Abstract

A method for evaluating a preload deterioration of a ball screw (4) is implemented by a computing apparatus (1). The method includes the following steps of: acquiring a time domain vibration data entry based on a vibration signal related with vibrations of a ball (44) of the ball screw (4); acquiring a frequency domain vibration inducement data entry based on the time domain vibration data; acquiring a vibration eigenvector based on the frequency domain vibration inducement data entry; performing a preload evaluation based on the vibration eigenvector, a reference vibration vector and a preload evaluation range to acquire a result of the preload evaluation; and determining whether a preload deterioration occurs in the ball screw (4) based on the result of the preload evaluation.

Description

Method for evaluating preload deterioration of a ball screw

The present invention relates to a method for evaluating preload degradation, and more particularly, to a method for evaluating preload degradation of a ball screw.

Because of the advantages of high-precision motion, ball screws are widely used as transmission components in tool machines requiring relatively high positioning precision. In general, a ball screw includes a plurality of balls, a nut and a screw shaft. The nut is engaged with the screw shaft through a ball to perform a linear movement with respect to the screw shaft.

Most ball screws are preloaded to eliminate backlash between input (eg, rotation) and output (eg, linear movement). Insufficient preload can cause unwanted vibrations when the ball screw is moving. In addition, the preload of the ball screw gradually decreases with the period of use, eventually resulting in backlash. Due to undesirable vibration and backlash, the life of the ball screw can be shortened as a result and precision can be reduced.

Taiwan Invention Patent No. I653410B discloses a method of determining whether backlash is formed in a ball screw based on a result of determining whether a preload of the ball screw is present. If it is determined that the preload of the ball screw no longer exists, it is determined that backlash has been formed. However, this method is not suitable for determining whether preload deterioration has occurred in a state where backlash has not yet been formed.

Accordingly, it is an object of the present invention to provide a method for evaluating preload deterioration of a ball screw that can at least compensate for the drawbacks of the prior art.

According to the invention, the method is configured to be implemented by a computer device. The ball screw includes a nut, a plurality of balls and a return mechanism for recirculation of the balls. The computer device is in signal connection with a first sensor, mounted on the nut, adjacent the return mechanism, which periodically transmits a vibration signal related to the vibration of the ball to the computer device. This way,

A) obtaining an entry of time-domain vibration data based on the vibration signal received from the first sensor;

B) obtaining at least one entry of frequency-domain vibration derived data based on the time domain vibration data entry;

C) for each of the at least one frequency domain vibration guidance data entry, obtaining a vibration eigenvector based on the frequency domain vibration guidance data entry;

D) performing a preload evaluation based on the vibration eigenvector obtained for the plurality of reference vibration vectors, the preload evaluation range, and the at least one frequency domain vibration induction data entry to obtain a result of the preload evaluation; and

E) determining whether preload deterioration has occurred in the ball screw based on the result of the preload evaluation.

Other features and advantages of the present invention will become apparent from the embodiments of the following detailed description with reference to the accompanying drawings.
1 is a block diagram illustrating an embodiment of a system used to implement a method for evaluating preload deterioration of a ball screw according to the present invention;
2 is a perspective view showing an embodiment in which the first sensor and the second sensor of the system are mounted on a ball screw;
3 is a flowchart illustrating an embodiment of a training procedure for detecting preload degradation in a method according to the present invention;
4 is a flowchart illustrating an embodiment of a training procedure for backlash detection in a method according to the present invention.
5 and 6 are flowcharts together showing an embodiment of an evaluation procedure for preload deterioration and backlash in the method according to the present invention.
7 is a flowchart illustrating an embodiment of a sub-step for obtaining frequency-domain vibration derived data in a method according to the present invention.
8 is a flowchart illustrating an embodiment of a sub-step for obtaining an inertial-force eigenvector in a method according to the present invention.

1 and 2 , an embodiment of a method for evaluating preload deterioration of a ball screw 4 according to the present invention is configured to be implemented by the system 100 shown in FIG. 1 . The system 100 comprises a computer device 1 , a first sensor 2 and a second sensor 3 . The ball screw 4 includes a nut 41 , a screw shaft 42 , a plurality of balls 44 , and a return mechanism 43 for recirculating the balls 44 . The computer device 1 is in signal connection with the first sensor 2 and the second sensor 3 .

In this embodiment, each of the first sensor 2 and the second sensor 3 can be implemented as an accelerometer, but such an implementation is not limited to that disclosed herein, and in other embodiments each sensor is a displacement meter. meter) or a velocimeter. More specifically, the effective bandwidth of the first sensor 2 covers a frequency in the range of 0.1 Hz to 5 Hz, and the effective bandwidth of the second sensor 3 is within ten times the bandwidth of the spin frequency of the screw shaft 42 . Covering a frequency range (eg 0.1 Hz to 250 Hz), the second sensor 3 features a 20-bit digital resolution.

In this embodiment, the ball screw 4 is an external circulation ball screw, as shown in FIG. 2 , and the return mechanism 43 includes a return tube.

The first sensor 2 is mounted on a nut 41 and abuts the return mechanism 43 . The first sensor 2 periodically transmits to the computer device 1 a vibration signal related to the vibration of the ball 44 in the return mechanism 43 . The second sensor 3 is mounted on the nut 41 and transmits an inertial force signal related to the inertial force applied to the nut 41 along the direction in which the nut 41 is moving relative to the screw shaft 42 , to the computer device. (1) is transmitted periodically.

In the case of a ball screw having a structure different from that of the external circulation ball screw 4 described herein, the first sensor 2 and the second sensor 3 are located at appropriate positions on the ball screw to obtain a vibration signal and an inertial force signal, respectively. It should be noted that each can be mounted in the same location or in different locations.

In this embodiment, the computer device 1 may be implemented as a personal computer, a laptop computer, a notebook computer, a tablet computer, a computing server or a cloud server, but such implementation is not limited to that disclosed herein, and in other embodiments can be varied.

The computer device 1 includes a storage module 12 , a display module 13 , a communication module 11 signally connected to the first sensor 2 and the second sensor 3 , and a communication module 11 , a storage a processing module 14 electrically connected to the module 12 and the display module 13 .

The communication module 11 is implemented as a network interface controller or a wireless transceiver supporting a wired communication standard and/or a wireless communication standard (eg, a Bluetooth technology standard or a cellular network technology standard, etc.), but is not limited thereto.

The storage module 12 may be implemented as a flash memory, a hard disk drive (HDD), a solid state disk (SSD), an electrically-errasable programmable read-only memory (EEPROM), or other non-volatile memory device, but is not limited thereto. it is not

The display module 13 may be a liquid-crystal display (LCD), a light-emitting diode (LED) display, a plasma display panel, a projection display, or the like. However, the implementation of the display module 13 is not limited to the one disclosed herein and may be changed in other embodiments.

The processing module 14 is a processor, a central processing unit (CPU), a microprocessor, a micro control unit (MCU), a system on a chip (SoC), or a software method and/or hardware for implementing the functions discussed in the present invention. may be implemented by any circuitry configurable/programmable in a manner.

The storage module 12 of the computer device 1 stores a plurality of first training oscillation eigenvectors, a plurality of second training oscillation eigenvectors, a plurality of first training inertia force eigenvectors and a plurality of second training inertia force eigenvectors. .

Each of the first training oscillation eigenvectors (or second training oscillation eigenvectors) is a training kurtosis eigenvector in which the first training oscillation eigenvector (or second training oscillation eigenvector) represents the kurtosis of the corresponding frequency domain data entry. a vector, a training maximum peak value eigenvector indicating the maximum peak value of the frequency domain data entry, a training total energy eigenvector indicating the total energy of the frequency domain data entry, and any combination thereof.

Each of the first training inertia force eigenvectors (or second training inertia force eigenvectors) is a training peak to peak in which the first training inertia force eigenvector (or second training inertia force eigenvector) represents the peak-to-peak value of the corresponding time domain data entry. peak eigenvector, training representing the maximum peak value of a time domain data entry Maximum peak value eigenvector, training representing an average peak value that is the average of the absolute values of the absolute maximum and minimum positive peak values of the time domain data entry mean peak value eigenvectors, and any combination thereof.

It should be noted that implementations of the first and second training oscillation eigenvectors, and the first and second training inertia force eigenvectors are not limited to those disclosed herein and may be modified in other embodiments.

The method for evaluating the preload deterioration of the ball screw 4 according to the present invention includes a training procedure for preload deterioration detection (see Fig. 3), a training procedure for backlash detection (see Fig. 4), and preload deterioration and backlash evaluation procedure for (see FIGS. 5 and 6).

The preload degradation detection training procedure includes steps 50 to 53 described below with reference to FIGS. 1 and 3 .

In step 50, the processing module 14 of the computer device 1 uses the first training vibration eigenvector as an input of the machine learning performed using an unsupervised learning algorithm, the first training vibration eigenvector Obtain a reference vibration vector in the data space spanned by the vector.

The unsupervised learning algorithm may include a clustering algorithm (eg, K-means clustering) and/or a self-organizing map (SOM) algorithm.

In a scenario in which the unsupervised learning algorithm is a clustering algorithm, the reference vibration vector thus obtained includes a center vector representing each of a plurality of vibration clusters obtained by performing the unsupervised learning algorithm.

In the scenario where the unsupervised learning algorithm is the SOM algorithm, the reference vibration vectors thus obtained include vectors respectively corresponding to neurons obtained by performing the SOM algorithm and updated more than a preset number of times.

In step 51, for each of the second training vibration eigenvectors, the processing module 14 of the computer device 1 configures a plurality of second vibration eigenvectors between the second training vibration eigenvector and each of the reference vibration eigenvectors obtained in step 50. 1 Calculate each candidate oscillation distance. In particular, each of the first candidate vibration distances calculated in this way is a Euclidean distance, but is not limited thereto.

In step 52 , for each of the second training vibration eigenvectors to which the calculated first candidate vibration distance corresponds, the processing module 14 of the computer device 1 determines which of the first candidate vibration distances to serve as the first target vibration distance Decide on the shortest one.

In step 53 , the processing module 14 of the computer device 1 determines, based on the first target vibration distance determined for the second training vibration eigenvector, whether preload deterioration has occurred in the ball screw 4 . Acquire the preload evaluation range used to More specifically, the distribution of the first target vibration distance is regarded as a normal distribution, and a 95% confidence interval (CI) of the distribution of the first target vibration distance is used as the preload evaluation range. However, the implementation of the preload evaluation range is not limited to the disclosure herein and may be changed in other embodiments.

The training procedure of backlash detection includes steps 60 to 63 described below with reference to FIGS. 1 and 4 .

In step 60, the processing module 14 of the computer device 1 uses the first training inertial force eigenvector as an input of machine learning performed using an unsupervised learning algorithm, and the data spanned by the first training inertial force eigenvector Obtain a reference inertia force vector in space.

In a scenario in which the unsupervised learning algorithm is a clustering algorithm, the reference inertial force vector thus obtained includes a center vector representing each of a plurality of inertial force clusters obtained by performing the unsupervised learning algorithm.

In the scenario where the unsupervised learning algorithm is the SOM algorithm, the reference inertia force vector thus obtained includes vectors respectively corresponding to neurons obtained by performing the SOM algorithm and updated more than another preset number of times.

In step 61 , for each of the second training inertia force eigenvectors, the processing module 14 of the computer device 1 configures a plurality of first eigenvectors between the second training inertia force eigenvector and each reference inertia force vector obtained in step 60 . Calculate each candidate inertia force distance. In particular, the calculated first candidate inertial force distances are respectively Euclidean distances, but are not limited thereto.

In step 62 , for each of the second training inertia force eigenvectors to which the calculated first candidate inertial force distance corresponds, the processing module 14 of the computer device 1 selects one of the first candidate inertial force distances to serve as the first target inertial force distance. Decide on the shortest one.

In step 63, the processing module 14 of the computer device 1 obtains a backlash evaluation range based on the first target inertial force distance determined with respect to the second training inertial force eigenvector. More specifically, the first target inertia force distance distribution is regarded as a normal distribution, and 95% CI of the first target inertial force distance distribution is used as the backlash evaluation range. However, the implementation of the backlash evaluation range is not limited to the disclosure of the present specification and may be changed in other embodiments.

1, 5 and 6 , the evaluation procedure for preload deterioration and backlash includes steps 70 to 85 described below.

In step 70 , the processing module 14 of the computer device 1 obtains a time domain vibration data entry based on the vibration signal received from the first sensor 2 .

In step 71, the processing module 14 of the computer device 1 obtains at least one frequency domain vibration induction data entry based on the time domain vibration data entry.

Specifically, step 71 includes substeps 710 to 714 as shown in FIG. 7 and described below.

In a sub-step 710, the processing module 14 of the computer device 1 performs envelope processing on the time domain vibration data entry to derive a processed time domain vibration data entry. Implementations of envelope processing are well known to those skilled in the art, and detailed descriptions thereof are omitted for brevity.

In a sub-step 711 , the processing module 14 of the computer device 1 sends target time domain vibration data corresponding to a period during which the nut 41 of the ball screw 4 moves at a constant speed from the processed time domain vibration data entry. Search for entries. It should be noted that the period during which the nut 41 of the ball screw 4 moves at a constant speed can be determined from the preset rotational speed of the motor which induces the movement of the ball screw 4 .

In a substep 712 , the processing module 14 of the computer device 1 acquires at least one time domain vibration induction data entry based on the target time domain vibration data entry.

It should be noted that, when the at least one time domain vibration induction data entry is plural, the plurality of time domain vibration induction data entries each have a preset period corresponding to the same period.

When the length of the period corresponding to the target time domain vibration data entry is longer than the preset period, the processing module 14 is configured to: Divide the vibration data entry into a plurality of time domain vibration induction data entries.

On the other hand, if the length of the period corresponding to the target time domain vibration data entry is shorter than the preset period, the processing module 14 is configured to: The multiple entries of the target time domain vibration data are combined to form one of the plurality of time domain vibration guidance data entries. It should be noted that if the total length of the period corresponding to multiple entries of the target time domain vibration data to be combined is longer than the preset period, one of the thus formed plurality of time domain vibration induction data entries will be clipped to the preset period.

In substep 713, for each of the at least one time domain vibration induction data entry, the processing module 14 of the computer device 1 performs bandpass filtering processing on the time domain vibration induction data entry. The performance of the bandpass filtering process aims to attenuate the remainder of the time domain vibration guidance data entry while maintaining a portion of the time domain vibration guidance data entry that falls within a frequency range within ten times the spin frequency bandwidth of the screw shaft 42 . It should be noted that this is done with That is, when the frequency range that the first sensor 2 can detect is exactly 10 times the spin frequency bandwidth of the screw shaft 42 , the performance of the bandpass filtering process in step 713 may be omitted, the procedure of the method The flow proceeds directly to sub-step 714 after sub-step 712 .

In a sub-step 714, the processing module 14 of the computer device 1 performs a Fourier transform on each of the at least one time-domain vibration-induced data entry that has been subjected to the band-pass filtering process to derive a frequency-domain vibration-induced data entry.

In step 72, for each of the at least one frequency domain vibration induction data entry, the processing module 14 of the computer device 1 obtains a vibration eigenvector based on the frequency domain vibration induction data entry.

Each of the vibration eigenvector(s) includes a kurtosis eigenvector representing the kurtosis of the frequency domain vibration induction data entry, a maximum peak value eigenvector representing the maximum peak value of the frequency domain vibration induction data entry, and a frequency domain vibration induction data entry. total energy eigenvector representing the total energy, and any combination thereof. It should be noted that the component(s) of the vibration eigenvector(s) are not limited to the teachings herein and may be varied in other embodiments.

Next, the processing module 14 of the computer device 1 is configured to obtain the result of the preload evaluation, the vibration eigenvector(s) obtained in step 72, the reference vibration vectors obtained in step 50 and the reference vibration vectors obtained in step 53 Based on the obtained preload evaluation range, the preload evaluation described in steps 73 to 75 is performed.

In step 73, for each of the vibration eigenvector(s), the processing module 14 of the computer device 1 calculates a plurality of second candidate vibration distances, respectively, between each one of the vibration eigenvector and the reference vibration vector.

In step 74, for each of the vibration eigenvector(s) corresponding to the calculated second candidate vibration distance, the processing module 14 of the computer device 1 determines which of the second candidate vibration distances to serve as the second target vibration distance. Decide on the shortest one.

In step 75, the processing module 14 of the computer device 1 obtains an evaluation result of the preload based on the second target vibration distance(s) and the preload evaluation range determined for the vibration eigenvector(s).

In one embodiment, the processing module 14 of the computer device 1 calculates an average of the second target vibration distance(s) to obtain a vibration average, and then calculates an average of the vibration average to obtain a result of the preload evaluation. Determines whether it falls within the load evaluation range (eg, the vibration average falls within or outside the preload evaluation range).

In one embodiment, the processing module 14 of the computer device 1 determines a mode of the second target vibration distance(s) to obtain a vibration mode, and then sets the vibration mode to obtain a result of the preload evaluation. Determine whether or not it falls within the preload evaluation range (eg, the vibration mode falls within or outside the preload evaluation range).

It should be noted that the implementation of preload evaluation is not limited to the teachings herein and may be varied in other embodiments.

In step 76 , based on the result of the preload evaluation, the processing module 14 of the computer device 1 determines whether preload deterioration has occurred in the ball screw 4 . If it is determined that the preload deterioration has not occurred in the ball screw 4 (eg, the vibration average falls within the preload evaluation range), the procedure flow of the method proceeds to step 77 . Otherwise, that is, if it is determined that preload deterioration has occurred in the ball screw 4 (eg, the vibration average falls outside the preload evaluation range), the procedure flow of the method proceeds to step 78 .

In step 77, the processing module 14 of the computer device 1 generates a first message indicating that preload deterioration has not occurred in the ball screw 4, and the display module 13 for displaying the first message ) to control

In step 78 , if it is determined that preload deterioration has occurred in the ball screw 4 , the processing module 14 of the computer device 1 is configured to perform at least one time domain based on the inertial force signal received from the second sensor 3 . Obtain the inertial force data entry.

that each of the at least one time domain inertial force data entry corresponds to a reciprocating period in which the nut 41 of the ball screw 4 moves from the start position to the end position and from the end position back to the start position on the screw axis 42 . It is necessary to note. In one embodiment, each of the at least one time domain inertial force data entry only corresponds to an acceleration (or deceleration) sub-period of the reciprocating period, when the nut 41 is accelerated (or decelerated) during movement.

In step 79, for each of the at least one time domain inertial force data entry, the processing module 14 of the computer device 1 obtains an inertial force eigenvector based on the time domain inertial force data entry.

Each of the inertial force eigenvector(s) includes a peak-to-peak eigenvector representing the peak-to-peak value of the time domain inertial force data entry, a maximum peak value eigenvector representing the maximum peak value of the time domain inertial force data entry, and a time domain inertial force data an average peak value eigenvector representing an average peak value that is an average of the absolute value of the positive maximum peak value and the negative minimum peak value of the entry, and any combination thereof.

Specifically, step 79 includes sub-steps 790 to 792 described below with reference to FIG. 8 .

In sub-step 790, for each of the at least one time domain inertial force data entry, the processing module 14 of the computer device 1 performs envelope processing on the time domain inertial force data entry to derive a processed time domain inertial force data entry. do.

In a substep 791 , for each of the at least one processed time domain inertial force data entry respectively obtained from the at least one time domain inertial force data entry, the processing module 14 of the computer device 1 configures the processed time domain inertial force data entry Perform low-pass filtering processing on

It should be noted that the performance of the low-pass filtering processing aims to attenuate the remainder of the processed time domain inertial force data entry while retaining a portion of the processed time domain inertial force data entry that falls within the frequency range of 0.1 Hz to 5 Hz. do. That is, when the frequency range that can be detected by the second sensor 3 is exactly in the range of 0.1 Hz to 5 Hz, performing the low-pass filtering process in step 791 can be omitted, and the procedure flow of the method is a sub-step After 790, the flow proceeds directly to sub-step 792.

In a substep 792, for each of the at least one processed time domain inertial force data entry subjected to low-pass filtering processing, the processing module 14 of the computer device 1 obtains an inertial force eigenvector based on the time domain inertial force data entry. do.

Next, the processing module 14 of the computer device 1 obtains the inertial force eigenvector(s) obtained in step 79, the plurality of reference inertia force vectors obtained in step 60 and step 63 to obtain a backlash evaluation result. Based on the calculated backlash evaluation range, backlash evaluation to be described in steps 80 to 82 below is performed.

In step 80, for each of the inertial force eigenvector(s), the processing module 14 of the computer device 1 calculates, respectively, a plurality of second candidate inertial force distances between each one of the inertial force eigenvector and the reference inertial force vector. .

In step 81 , for each of the inertial force eigenvector(s) for which the corresponding second candidate inertial force distance is calculated, the processing module 14 of the computer device 1 determines which of the second candidate inertial force distances to serve as the second target inertial force distance. Decide on the shortest one.

In step 82, the processing module 14 of the computer device 1 obtains a backlash evaluation result based on the second target inertia force distance and the backlash evaluation range.

In one embodiment, the processing module 14 of the computer device 1 calculates an average of the second target inertial force distance(s) to obtain an inertial force average, and then determines whether the inertial force average falls within the backlash evaluation range. to obtain a backlash result (eg, the inertial force average falls within or outside the backlash evaluation range).

In one embodiment, the processing module 14 of the computer device 1 determines the mode of the second target inertial force distance(s) to obtain the inertial force mode, and then determines whether the inertial force mode falls within the backlash evaluation range. determine to obtain a backlash evaluation result (eg, the inertial force mode is within or outside the backlash evaluation range).

It should be noted that the implementation of the backlash evaluation performance is not limited to the disclosure herein and may be changed in other embodiments.

In step 83 , based on the backlash evaluation result, the processing module 14 of the computer device 1 determines whether backlash has occurred in the ball screw 4 . When it is determined that no backlash has occurred in the ball screw 4 (eg, the inertial force average falls within the backlash evaluation range), the procedure flow of the method proceeds to step 84 . Otherwise, that is, when it is determined that backlash has occurred in the ball screw 4 (eg, the inertial force average falls outside the backlash evaluation range), the procedure flow of the method proceeds to step 85 .

In step 84 , the processing module 14 of the computer device 1 generates a second message indicating that preload deterioration has occurred in the ball screw 4 but no backlash has occurred, and controls the display module 13 . to display the second message.

In step 85, the processing module 14 of the computer device 1 generates a third message indicating that backlash has occurred in the ball screw 4, and controls the display module 13 to display the third message.

In summary, the method for evaluating the preload deterioration of a ball screw according to the present invention uses a computer device 1 to perform a preload evaluation based at least on the vibration eigenvector(s), and the computer device 1 . to further perform a backlash evaluation based at least on the inertial force eigenvector(s). Each of the vibration eigenvector(s) is generated by a first sensor 2 mounted adjacent to the return mechanism 43 of the ball screw 4 and is associated with vibration of the ball 44 of the ball screw 4 . obtained from the signal, each of the inertial force eigenvector(s) is generated by the second sensor 3 mounted on the nut 41 of the ball screw 4 and is derived from the inertial force signal associated with the inertial force applied to the nut 41 . is obtained Then, the computer device 1 determines whether preload deterioration has occurred in the ball screw 4 based on the result of the preload evaluation and whether or not backlash has occurred in the ball screw 4 based on the backlash evaluation result do.

In the foregoing description, for purposes of explanation, numerous specific details have been set forth in order to provide a thorough understanding of the embodiments. It will be apparent, however, to one skilled in the art, that one or more different embodiments may be practiced without these few specific details. Also, throughout this specification reference to an embodiment marked with "one embodiment", "one embodiment", an ordinal number, etc. means that a particular feature, structure, or characteristic may be included in the embodiment of the invention. You have to understand that In the detailed description, various features are sometimes grouped together in a single embodiment, drawing, or description thereof, in order to streamline the invention and aid in understanding various inventive aspects, one or more features or specific details from one embodiment are presented where appropriate. Embodiments of the invention may be practiced with one or more features or specific details from other embodiments.

While the present invention has been described in connection with what is considered to be an exemplary embodiment, the invention is not limited to the disclosed embodiment, but is intended to cover all such modifications and equivalent arrangements, various arrangements included within the spirit and scope of the broadest interpretation. It is understood that it is intended to be inclusive.

Claims (10)

A method for evaluating preload deterioration of a ball screw, comprising:
The method is implemented by a computer device, wherein the ball screw comprises a nut, a plurality of balls, a screw shaft, and a return mechanism for recirculating the balls;
the computer device is in signal connection with a first sensor, the first sensor mounted on the nut and adjacent to the return mechanism, wherein the return mechanism periodically sends a vibration signal related to vibration of the ball to the computer device send to
The computer device is also in signal connection with a second sensor, the second sensor being mounted to the nut and configured to receive an inertial force signal related to an inertial force applied to the nut along a direction in which the nut travels with respect to the screw axis. periodically transmit to said computer device;
The method is
A) obtaining an entry of time-domain vibration data based on the vibration signal received from the first sensor;
B) obtaining at least one entry of frequency-domain vibration derived data based on the time domain vibration data entry;
C) for each of the at least one frequency domain vibration guidance data entry, obtaining a vibration eigenvector based on the frequency domain vibration guidance data entry;
D) performing a preload evaluation based on a plurality of reference vibration vectors, a preload evaluation range, and the vibration eigenvector obtained for the at least one frequency domain vibration induction data entry to obtain a result of the preload evaluation step and
E) determining whether preload deterioration has occurred in the ball screw based on the result of the preload evaluation
including,
The method, after step E),
H) when it is determined that preload deterioration has occurred in the ball screw, obtaining at least one time domain inertial force data entry based on the inertial force signal received from the second sensor;
I) for each of the at least one time domain inertial force data entry, obtaining an inertial force eigenvector based on the time domain inertial force data entry;
J) performing a backlash evaluation based on a plurality of reference inertial force vectors, a backlash evaluation range and the inertial force eigenvector obtained for the at least one time domain inertial force data entry to obtain a result of the backlash evaluation;
K) determining whether backlash occurs in the ball screw based on a result of the backlash evaluation
further comprising
Way. The method of claim 1,
The step B) is,
B-1) performing envelope processing on the time domain vibration data entry to derive a processed time domain vibration data entry;
B-2) a sub-step of retrieving, from the processed time domain vibration data entry, a target time domain vibration data entry corresponding to a period in which the nut of the ball screw moves at a constant speed;
B-3) a sub-step of obtaining the at least one frequency domain vibration induction data entry based on the target time domain vibration data entry;
Way. 3. The method of claim 2,
The sub-step B-3) is,
B-3-1) a sub-step of obtaining at least one time domain vibration induction data entry based on the target time domain vibration data entry;
B-3-2) for each of the at least one time domain vibration derivation data entry, performing a Fourier transform on the time domain vibration derivation data entry to derive the frequency domain vibration derivation data entry;
Way. 4. The method according to any one of claims 1 to 3,
In the step C), each of the vibration eigenvectors is a kurtosis eigenvector indicating the kurtosis of the frequency domain vibration induction data entry, and a maximum peak value eigenvector indicating the maximum peak value of the frequency domain vibration induction data entry. , and at least one of a total energy eigenvector representing the total energy of the frequency domain vibration induction data entry.
Way. 4. The method according to any one of claims 1 to 3,
the computer device stores a plurality of first training vibration eigenvectors and a plurality of second training vibration eigenvectors;
The method, prior to step D),
F) using the first training oscillation eigenvector as input in machine learning performed using an unsupervised learning algorithm, a reference oscillation vector in the data space specified by the first training oscillation eigenvector the steps of obtaining
G) obtaining the preload evaluation range based on the reference vibration vector and the second training vibration eigenvector
How to include more. 6. The method of claim 5,
In step F) above,
The unsupervised learning algorithm includes a clustering algorithm,
The reference vibration vector includes a center vector representing each of a plurality of vibration clusters obtained by performing the unsupervised learning algorithm,
Way. 6. The method of claim 5,
In step F) above,
The unsupervised learning algorithm includes a Self-Organizing Map (SOM) algorithm,
The reference vibration vector includes vectors respectively corresponding to neurons obtained by performing the SOM algorithm and updated more than a preset number of times,
Way. 6. The method of claim 5,
The step G) is,
G-1) for each of the second training vibration eigenvectors, respectively calculate a plurality of first candidate vibration distances between the second training vibration eigenvector and each one of the reference vibration vectors obtained in step F) sub-steps,
G-2) determining, for each of the second training vibration eigenvectors to which the calculated first candidate vibration distance corresponds, the shortest one of the first candidate vibration distances to serve as a first target vibration distance;
G-3) comprising the sub-step of obtaining the preload evaluation range based on the first target vibration distance determined for the second training vibration eigenvector,
Way. 4. The method according to any one of claims 1 to 3,
Step D) is,
D-1) for each of the vibration eigenvectors, a sub-step of respectively calculating a plurality of second candidate vibration distances between each one of the vibration eigenvector and the reference vibration vector;
D-2) a sub-step of determining, for each of the vibration eigenvectors to which the calculated second candidate vibration distance corresponds, the shortest one of the second candidate vibration distances to serve as a second target vibration distance;
D-3) comprising a sub-step of obtaining a result of the preload evaluation based on the second target vibration distance determined for the preload evaluation range and the vibration eigenvector,
Way. delete

Images