TWI407026B - Diagnosis method of ball screw preload loss via hilbert-huang transform and apparatus therefor - Google Patents

Abstract

A diagnosis method of ball screw preload loss via Hilbert-Huang Transform and the apparatus therefore is disclosed. The method is using for diagnosing a ball screw preload loss, comprises a process of sensing the motor torque current or motor voice when the ball screw is operation, filtering by means of Empirical Mode Decomposition (EMD), obtaining a Hilbert-Huang Spectrum(HHS) by Hilbert-Huang Transform (HHT), then abstracting a multiscale entropy(MSE) complexity model, comparing the signal of original status with the signal of preload loss, can diagnose successfully. Moreover, an apparatus comprises a sensor, a signal pre-processing unit, a signal post-processing unit, MSE abstracting unit and MSE complexity model output unit, for realizing the diagnosis method to achieve the purposes of prognostic effectiveness and utilizing convenience.

Description

Ball lead screw pre-pressure failure diagnosis method and device thereof

The invention relates to a ball lead screw pre-pressure failure diagnosis method and device thereof, in particular to a method and a device for generating multi-scale entropy (MSE) complexity mode using Hilbert yellow transformation (HHT) .

The ball lead screw is an important component of the precision linear drive and positioning system. The contact pressure (pre-pressure) generated by the geometric interference between the components during assembly is closely related. In order to achieve high positioning accuracy, the general method has the elimination of the ball guide. The gap of the screw is reduced to zero, or the rigidity of the ball lead screw is increased to reduce the amount of elastic deformation when subjected to the axial load. Both methods can be achieved by applying a preload to the ball lead screw. There are two kinds of pre-pressing methods. (1) Pre-pressing method of double-nut ball screw, that is, pre-pressing is applied in the middle of two nuts to pre-press, and the relative thickness is selected according to the pre-pressure. The pressing piece is placed between the nuts, the pre-pressure is applied, the two nuts generate the tensile load to form the stretching pre-pressure; (2) the pre-pressing mode of the single nut ball guiding screw, that is, the groove is built into the groove of the ball groove The steel ball with a slightly larger diameter makes the ball and the groove preload with four points of contact. Too small pre-pressure will lose the effect of pre-compression, while excessive pre-pressure will have adverse effects on life and heat dissipation; in general, according to different designs, the maximum pre-pressure of the ball lead screw is set to 10% of the dynamic load. Moderate preload is 6% to 8%, and mild preload is below 4%.

After a period of use, the ball lead screw is reduced due to movement wear, rigidity, or pre-stress due to various factors, which will result in lower positioning accuracy. In order to prevent the pre-stress failure, the factory often uses the working hours as the calculation. In order to avoid reducing the precision of the finished product, the ball lead screw exceeding the set working time is completely replaced, which is not economical; many designs have been developed, such as Taiwan patent disclosure. No. TW200940852, TW201002963, U.S. Patent Publication No. US20070068292, etc. proposes a mechanism for adjusting the pre-pressure to extend the service life of the ball lead screw, but it takes a lot of time to correct after the machine is stopped.

In order to be able to detect whether the ball lead screw is normal, Japanese Patent Publication No. JP2004361247 proposes a voltage at which the ball lead screw is oscillated to detect the use of a voltage difference to detect whether the ball lead screw is damaged; US Patent Publication No. US20090171594 discloses a position The difference signal estimates driving force, elastic deformation, etc., compared with a set threshold value to diagnose whether the machine is damaged.

Since the pre-pressure of the ball lead screw affects the positioning accuracy, and the mechanical size of the ball lead screw cannot be directly measured, how to diagnose the pre-pressure failure of the ball lead screw early is very important. However, the signal generated by the movement of the ball lead screw is an unsteady signal. If Fourier transform filtering is used in the prior art, the filtered result cannot be analyzed.

For the processing of non-steady-state signals, Ruqiang et al, in 2006 IEEE Transactions on Instrumentation and Measurement, vol. 55, proposed the use of Hilbert-Huang Transform for vibration mechanical mechanical inspection, however mechanical The vibration may have its periodicity. The Hilbert Huang transform can be used to convert the vibration signal into a time-frequency function of instantaneous amplitude and instantaneous frequency to exhibit the energy distribution of different types of signals; however, the motion of the ball lead screw is generated. The signal is much more complex than the vibration signal, and it is not possible to directly use the Hilbert Huang conversion to produce a quality simplification signal for diagnosis.

In view of the above problems of the prior art, one of the main objects of the present invention is to provide a ball lead screw pre-pressure failure diagnosis device, which generates a ball lead screw via the pre-pressure failure diagnosis device by using a signal generated when the ball lead screw moves. A multiscale Entropy Complexity model of the pre-stress signal is used to diagnose whether the ball lead screw pre-pressure is invalid, and provides the user with monitoring of the ball lead screw. If necessary, the pre-pressure can be adjusted or the ball guide can be replaced. Screw to maintain machining accuracy and maintain the life of the ball lead screw. The pre-stress failure diagnosis device comprises: a sensing unit, a signal pre-processing unit, a signal post-processing unit, a characteristic extraction unit and a complexity mode output unit.

The sensing unit is mounted on the machine device of the ball lead screw, and can transmit the signal generated by the change of the torque during the movement of the ball lead screw to the pre-processing unit of the signal. This signal is used in the embodiment of the present invention. The ball guide screw emits a sound signal generated by the sound or a current signal of the motor that drives the ball lead screw to rotate, but is not limited thereto.

The signal pre-processing unit has a Fast Fourier Transform (FFT) filtering function, which can convert the signal and generate an elegant and simple signal and transmit it to the signal post-processing unit.

The signal post-processing unit has a Hilbert-Huang Transform (HHT) function, which can convert the high-quality simplification signal into a time-frequency function of instantaneous amplitude and instantaneous frequency to present the high-quality simplification signal. The energy distribution forms a Hilbert-Huang Spectrum (HHS) and is sent to the characteristic extraction unit.

The characteristic extraction unit has a multiscale entropy (MSE) extraction function, and the Hilbert yellow map (HHS) can generate a multi-scale entropy complexity mode and transmit to the complexity mode output unit.

The complexity mode output unit may output the multi-scale entropy complexity mode in graphics or data or a combination thereof; the user may compare the original multi-scale entropy complexity mode with the current multi-scale entropy complexity mode, The change of the scale entropy is to understand the difference between the current ball lead screw pre-pressure and the original ball lead screw pre-pressure, to diagnose whether the ball lead screw pre-pressure is invalid, and to determine whether it is necessary to adjust the ball lead screw pre-pressure or replace the ball. Lead screw.

According to another object of the present invention, a ball lead screw pre-pressure failure diagnosing device is provided. When a fast Fourier transform cannot be used for a signal captured by a sensing unit, the signal pre-processing unit has an Empirical Mode Decomposition (Empirical Mode Decomposition, EMD) filtering function, which filters the high frequency noise to leave a low frequency signal with preload failure characteristics and generates an elegant and simple signal. The high quality simplification signal is a time frequency. The time-varying intrinsic mode function (IMF) is transmitted to the post-processing unit.

According to still another object of the present invention, a ball lead screw pre-pressure failure diagnosis device is provided, wherein the sensing unit further comprises a wireless transmission module, wherein the wireless transmission module can generate a ball lead screw due to a change in the rotational distance. The signal is transmitted to the pre-processing unit of the signal by wireless transmission; thus, one of the embodiments of the central monitoring system, the sensing unit with the wireless transmission module is respectively installed on the plurality of machine equipments, and each sensing The unit separately transmits the signal to the pre-signal processing unit, and a signal pre-processing unit and other units can separately process the signals transmitted by the sensing units to form a central monitoring implementation.

Another main object of the present invention is to provide a ball screw lead pre-pressure failure diagnosis method, which utilizes a signal generated when a ball lead screw moves, and through the pre-pressure failure diagnosis method, the director produces a ball guide screw pre-pressure signal. The scale entropy complexity mode is used to diagnose whether the ball lead screw pre-pressure is invalid, and provides the user with monitoring of the ball lead screw. If necessary, the pre-pressure can be adjusted or the ball lead screw can be replaced to maintain the machining accuracy and maintain the ball lead screw. The method of life. This includes the following steps:

S1: The signal presented when the ball lead screw is moved, the signal is generated by the change of the torque when the ball lead screw is moved. This signal is a voiceprint produced by the ball lead screw in the embodiment of the invention. The current signal generated by the signal or the current of the ball lead screw rotation, but not limited to this.

S2: The signal is processed by a signal pre-processing, wherein the signal pre-processing can be performed by fast Fourier transform (FFT) filtering, or by empirical mode decomposition (EMD) filtering, the signal can be filtered to remove high frequency noise and left. a low frequency signal having a preload failure characteristic to generate a high quality simplification signal;

S3: The high-quality simplification signal is processed by a signal, wherein the signal post-processing adopts Hilbert Yellow Transformation (HHT) to generate a time-frequency internal modal function (IMF) of instantaneous amplitude and instantaneous frequency. The time-frequency intrinsic mode function constitutes a Hilbert Yellow Map (HHS) of the signal;

S4: subjecting the Hilbert yellow spectrum of the signal to a characteristic extraction process, wherein the characteristic extraction process uses multi-scale entropy (MSE) extraction to generate a multi-scale entropy complexity mode of the signal;

S5: outputting the multi-scale entropy complexity mode of the signal, the user can compare the original multi-scale entropy complexity mode with the current multi-scale entropy complexity mode, and the multi-scale entropy (MSE) complexity change Knowing the difference between the current ball lead screw pre-pressure and the original ball lead screw pre-pressure, you can determine whether you need to adjust the ball lead screw pre-pressure or replace the ball lead screw.

According to the present invention, a ball lead screw pre-pressure failure diagnosis method and apparatus according to the present invention may have one or more of the following advantages:

(1) In the prior art or factory, whether the pre-pressure of the ball lead screw fails, and the direct and effective analysis cannot be directly performed, and the ball lead screw pre-pressure failure diagnosis method and device thereof can be directly and quickly compared through the original method. The multi-scale entropy complexity mode and the current multi-scale entropy complexity mode can effectively judge whether the ball lead screw pre-pressure is invalid, which will greatly facilitate the user, without stopping the inspection, and can improve the ball lead screw. life.

(2) In the practical application of the present invention, the wireless transmission type sensing unit of the ball lead screw pre-pressure failure diagnosis device of the present invention can be used to wirelessly transmit the signal of the ball lead screw to the remote pre-processing unit. Analysis and judgment, which can form the function of remote central monitoring, will give users greater convenience.

Please refer to FIG. 1 , which is a schematic diagram of the ball lead screw pre-pressure failure diagnosis device of the present invention. In the figure, the pre-stress failure diagnosis device 1 includes a sensing unit 11, a signal pre-processing unit 12, a signal post-processing unit 13, a characteristic extraction unit 14, and a complexity mode output unit 15. The sensing unit 11 is mounted on the machine device of the ball lead screw 2, and the sound or current signal generated by the change of the torque when the ball lead screw moves is transmitted to the signal pre-processing unit 12; the signal pre-processing unit 12 The function of fast Fourier transform (FFT) filtering or the function of empirical mode decomposition (EMD) filtering can convert the signal and generate an elegant and simple signal and transmit it to the signal post-processing unit 13 The signal post-processing unit 13 has a Hilbert Yellow Conversion (HHT) function, which can generate a Hilbert Yellow Map (HHS) and transmit it to the characteristic extraction unit 14; the characteristic extraction unit 14 has The multi-scale entropy (MSE) extraction function can generate a multi-scale entropy complexity model for the Hilbert Yellow spectrum (HHS) and transmit it to the complexity mode output unit 15.

Please refer to FIG. 2 , which is a step-by-step illustration of the method for pre-stress failure diagnosis of the ball lead screw of the present invention. The steps in the figure are as follows:

S1: Using the sensing unit 11 to extract the signal presented by the ball lead screw 2, the signal is generated by the change of the torque when the ball lead screw 2 moves; this signal is not limited, and the implementation is described later. In the first embodiment, the current rotated by the ball lead screw is a current signal, and the sound emitted by the ball lead screw in the second embodiment is a voiceprint signal.

S2: The signal is pre-processed by the signal pre-processing unit 12, wherein the signal pre-processing can adopt Fast Fourier Transform (FFT), or empirical mode decomposition (EMD), filtering out the low frequency and leaving the energy larger. The high frequency signal is used to generate a high quality simplification signal; the detailed processing is as follows: S21: setting the recursive condition of the intrinsic mode function, the recursive condition including an intrinsic mode function trend and an intrinsic mode function a constant; S22: the signal is subjected to a shifting process to generate a shifted signal; S23: the shift signal of the signal is subjected to an intrinsic mode function Processing to generate the signal quality simplification signal; S24: aligning whether the high quality simplification signal generated in step S23 satisfies the trend of the intrinsic mode function and the intrinsic mode function constant set in step S21, if not satisfied Then, the quality simplification signal is sent back to step S22 and then subjected to a shifting process; if it is satisfied, the quality simplification signal is sent to the next step.

S3: The quality simplification signal is post-processed by the signal post-processing unit 13, wherein the signal post-processing adopts Hilbert Yellow Conversion (HHT) to generate a time-frequency internal modulo of instantaneous amplitude and instantaneous frequency. The state function (IMF), the time-frequency intrinsic mode function constitutes a Hilbert Yellow Map (HHS) 131 of the signal (labeled in Figures 4 and 8).

S4: subjecting the Hilbert Yellow Spectrum (HHS) 131 to a characteristic extraction process by a characteristic extraction processing unit 14, wherein the characteristic extraction process uses multi-scale entropy (MSE) extraction to generate a multi-scale entropy complex of the signal. Degree mode.

S5: outputting the multi-scale entropy complexity mode by using the complexity mode output unit 15; the user can compare the original multi-scale entropy complexity mode with the current multi-scale entropy complexity mode to determine whether the ball lead screw 2 needs to be adjusted. Pre-pressure or replacement of the ball lead screw 2; for example, a multi-scale entropy complexity mode of the ball lead screw 2 at different rotational speeds (for example, 300 rpm, 1500 rpm, 3000 rpm of the subsequent embodiment), or a ball lead screw provided by a supplier 2 When the installation is completed, the multi-scale entropy complexity mode at different rotation speeds (for example, 300 rpm, 1500 rpm, 3000 rpm of the subsequent embodiment) is first performed; after the ball lead screw 2 is used for a period of time (or instant on-line), step S1 is performed. When the multi-scale entropy complexity mode at that time is obtained in S4, it can be judged whether the pre-pressure of the ball lead screw 2 is invalid.

Empirical Mode Decomposition (EMD) can process nonlinear and non-stationary signal data to solve the Intrinsic Mode Functions (IMF) by disassembling the original signal into a complex number of intrinsic modulus functions. (IMF), first the first empirical mode decomposition, first find the local maximum and local minimum of the entire signal (local maxima & minima), and then use the cubic spline line to create the envelope The interval interval (envelope), then averages every two envelope interval intervals, that is, the first component; and so on, until the average value approaches the horizontal line, the signal is the first inner 禀The modulus function (IMF) component; if the intrinsic modulus function (IMF) has not yet reached the ideal filtering (optimal mode), an empirical mode decomposition (EMD) iteration can be performed.

The sensing unit 11 picks up the signal sensed by the ball lead screw 2 during the movement of the ball lead screw 2, and the signal is composed of a real part and an imaginary part; a state vector of the real part (state vector) The state vector representing the symbol X ( q ) and the imaginary part represents Y ( q ), where q is represented as time and spatial coordinate; if the signal track is represented by a complex plane (complex trace) The symbol is Z( q ),

Z( q )=X(q)+iY(q) (1)

Where X ( q ) is the real part of Z( q ) and Y ( q ) is the imaginary part of Z( q ).

The sensing unit 11 captures the signal sensed when the ball lead screw 2 moves, and the amplitude portion X ( q ) is processed by signal pre-processing, such as fast Fourier transform (FFT) as the signal pre-processing, which can be expressed as:

For example, empirical mode decomposition (EMD) is used as signal pre-processing, which can be expressed as:

Where c _j ( q ) is the mode after empirical mode decomposition (EMD) decomposition, which is the sequence from the first mode to the last mode, and R _n is the residual value after the EMD decomposition (residue) ; c _j ( q ) contains the trend of X ( q ), which is the time-varying mean; thus the amplitude part X ( q ) is decomposed into n empirical modes ( Empirical modes) and a residual value R _n that can be regarded as the mean trend. When analyzing the empirical modal decomposition (EMD) data, no average value is required as the reference reference, and the local maximum or local minimum is used. The position of the value gives you an idea of the characteristics of the signal, so empirical mode decomposition (EMD) can handle both non-linear and non-steady-state data.

X (q) of the imaginary part Y (q), represents a plurality of types may be employed, for example, represents a fast Fourier transform of formula (FFT) or Hibbert conversion equation (Hilbert transform), when employed Hibbert conversion formula is as follows:

Where PV represents the principal value of the singular integral.

For the instant at q , the conjugate pair ( x ( q ), y ( q )) is expressed as a complex number:

z ( q )=α( q )‧ e ^{i θ( q )} (5)

Where α( q ) and θ( q ) represent the amplitude and phase of the instantaneous signal trajectory z ( q ) at q , respectively, and the instantaneous frequency ω( q ) is:

When the signal sensed by the ball lead screw 2 is filtered by the EMD decomposition of the signal pre-processing, that is, the equation (3) is processed by the Hilbert yellow conversion (HHT) signal post-processing, Hilbert The yellow transform (HHT) is processed for each intrinsic modulus function (IMF), and X ( q ) for each intrinsic modulus function (IMF) c _j ( q ) can be expressed as:

Where a _j ( q ) is the instantaneous amplitude and w _j ( q ) is the frequency.

The intrinsic modulus function (IMF) expresses the oscillation modes existing in the deep inner layers of the signal. Each intrinsic modulus function (IMF) contains only one oscillation mode, and there is no complicated carrier. Its instantaneous frequency ω( q) ) can be obtained by equation (6), so the characteristics of the signal can be easily analyzed using the intrinsic modulus function (IMF). When the intrinsic modulus function (IMF) is converted to Hilbert Yellow (HHT), the Hilbert Yellow Transform (HHT) can be used as a monotonic trend of long-period oscillations, and the amplitude and frequency can be clearly separated. A stereo-distribution map of amplitude-frequency-time is formed, which is Hilbert-Huang Spectrum.

Multi-scale entropy (MSE) is a complexity quantification method for testing sample entropy values at multiple scales; Hilbert yellow map (HHS) is subjected to characteristic extraction processing via characteristic extraction processing unit 14 to obtain coarse-grained time series The sampled entropy value varies with scale. The characteristic extraction of multi-scale entropy (MSE) is as follows: Hilbert Yellow Map (HHS) is a time series.

X ={ x ₁ , x ₂ ,..., x _i ,..., x _L } (8)

Set a scale factor τ to establish a time series of samples:

Among them, 1 j L / τ, that is, the time series length of each sampling is N = L / τ, when the scale factor is 1, it is the original time series. When the time series is N, the time series is:

Wherein, if a time series of length m is cut,

in ( j ) and ( i ) The probability that the distance between them is less than γ is:

among them, (γ) for any time ( j ) and ( i ) The probability that the distance between them is less than γ. The sample entropy is defined as:

Due to τ≠0, the sampling entropy can be counted as follows:

Define multi-scale entropy (MSE) as a set of sampling entropy at multiple scales. Multi-scale entropy (MSE) can be calculated as follows:

Therefore, multi-scale entropy (MSE) can be used to describe the complexity of time series on different time scales; for different pre-stress of ball lead screw 2, different complexity can be presented (different multi-scale entropy complexity mode) . If the user compares the original multi-scale entropy complexity mode with the current multi-scale entropy complexity mode, it is determined whether the ball lead screw 2 is invalid by judging the difference between the two.

<First Embodiment>

4 to 7 are schematic views of a first embodiment of a ball lead screw pre-pressure failure diagnosis method and apparatus thereof, and in FIG. 4, the sensing unit 11 is in a current sense in this embodiment. In order to realize the measuring transmitter 111, the current sensing transmitter 111 is mounted on the rotary driving motor of the ball guiding screw 2 (not shown), and can output the motor rotation speed (rpm) of the rotary driving motor of the ball guiding screw 2 and The motor torque current (mA) is the (A) line and the (B) line of Fig. 5(a), respectively. The rotation of the ball lead screw 2 drives the motor to rotate in the forward direction at regular or irregular timing (motor speed). (rpm) and motor torque current (mA) are positive values or reverse rotation (motor speed (rpm) and motor torque current (mA) are negative); 5(a) and 5(b) The drawing is plotted at 300 rpm and can be processed and plotted in the same way for different speeds. Since the motor torque current (mA) is non-linear and unstable, it is difficult to analyze it by conventional techniques or to extract its characteristics for comparison.

The current sensing transmitter 111 transmits a motor torque current (mA) signal to the pre-signal processing unit 12, see the small image above the current sensing transmitter 111 in FIG. 4 (ie, Figure 5(a)), before the signal The processing unit 12 performs signal pre-processing; in this embodiment, empirical mode decomposition (EMD) is used to perform empirical mode decomposition (EMD) on the motor torque current (mA) signal, and the processing result is shown in FIG. 5(b). . From Fig. 5(b), the motor torque current (mA) signal is recursed multiple times through the aforementioned steps S21 to S24, and the non-linear and unstable signal can be processed into a high quality simplification signal after ten times of retransmission. The amplitude and time of the imf1~imf10 and the residual value are respectively shown; if the imd10 shows the mode of the EMD decomposition of the equation (3), the mode c _j ( q ) changes with time, and the res display formula (3) Residual value R _n of the EMD after decomposition.

The quality simplification signal is further processed by the signal post-processing unit 13 for signal post-processing. In this embodiment, Hilbert Yellow Conversion (HHT) is used, and the converted signal constitutes a Hilbert Yellow Map of the signal (HHS). 131, Hilbert Yellow Map (HHS) 131 as shown in Figure 6; wherein, Figure 6(a) shows the imf6 processed Hilbert Yellow Map (HHS) of Figure 5(b) 131 The Hilbert Yellow Map (HHS) 131 after imf7 of Figure 5(b) is shown by Figure 6(b); the Hilbert Yellow Map (HHS) 131 is also plotted in Figure 4. To illustrate; from Fig. 6, the intrinsic mode function of the motor torque current (mA) signal can be clearly distinguished.

In Fig. 7, the Hilbert Yellow Map (HHS) 131 of the motor torque current (mA) signal is subjected to characteristic extraction processing via the characteristic extraction processing unit 14, and in this embodiment, multi-scale entropy (MSE) extraction is used to A multi-scale entropy complexity mode that produces the signal. The multi-scale entropy complexity mode of the ball lead screw 2 at different rotational speeds is shown in Fig. 7. Figure 7(a) shows that at 300 rpm, the pre-pressure of the ball lead screw 2 is reduced from 6% to 4%, 2%. Different multi-scale entropy complexity mode, the pre-pressure of the ball lead screw 2 is 6%, and when the pre-pressure of the ball lead screw 2 is reduced to 4% or 2%, the pre-pressure of the ball lead screw 2 is invalid. The % or 2% multi-scale entropy complexity mode can be clearly and correctly diagnosed; the same 7(b) shows that the pre-pressure of the ball lead screw 2 is reduced from 6% to 4%, 2% at 1500 rpm. The multi-scale entropy complexity mode; Figure 7(c) shows the multi-scale entropy complexity mode in which the pre-pressure of the ball lead screw 2 is reduced from 6% to 4% and 2% at 3000 rpm.

After comparing the multi-scale entropy complexity mode, the user can effectively diagnose whether the ball lead screw 2 pre-pressure is invalid, which can greatly facilitate the user and does not need to stop the inspection. For the gradually failing ball lead screw 2, it can be immediately Maintenance and increase the life of the ball lead screw.

<Second embodiment>

8 to 11 are schematic views of a second embodiment of a ball lead screw pre-pressure failure diagnosis method and apparatus thereof, and in FIG. 8, the sensing unit 11 has a sense of sound in this embodiment. The measurement transmitter 112 is implemented. The sound sensing transmitter 112 includes a microphone (not shown) and a wireless transmission module 113. The microphone is mounted on the rotary drive motor of the ball lead screw 2 to output a ball guide. The rotation of the screw 2 drives the sound (db) of the motor and is converted into an electronic type of voiceprint signal; the microphone used in this embodiment can receive sound at a frequency of 60 Hz to 20,000 Hz, in this embodiment, when the rotational speed is 300 rpm, The sound frequency signal generated by the ball screw 2 and the nut and the groove of the screw collide with each other between about 300 and 600 Hz, so this is the main analysis area; when the speed is 1500 rpm, the sound frequency signal is about 100~ Between 2500HZ, this is the main analysis area; the above analysis area is only for explanation, but not limited to this.

The voiceprint signal is wirelessly transmitted to the signal pre-processing unit 12 of the central monitoring center by the wireless transmission module 113. The amplitude of the voiceprint signal is as shown in Fig. 9. The figure 9(a) shows the voiceprint signal of the ball screw 2 with a pre-pressure drop of 2% when the ball screw 2 is rotated at 300 rpm, and the figure 9(b) is When the ball lead screw 2 is rotated at 300 rpm, the ball lead screw 2 pre-stresses the original 6% voiceprint signal.

The wireless transmission module 113 wirelessly transmits the voiceprint signal to the signal pre-processing unit 12, referring to the small image on the left side of the sensing unit 11 in FIG. 8 (ie, FIG. 9), and the signal pre-processing unit 12 performs signal pre-processing. In this embodiment, the empirical mode decomposition (EMD) is used to perform the empirical mode decomposition (EMD) of the voiceprint signal into a high quality simplification signal; at 300 rpm, the voiceprint signal is reduced to 2% when the pre-pressure is reduced to 2%. The empirical mode decomposition (EMD) is a high quality simplification signal, which is the amplitude and time change of imf1~imf10 and the residual value, as shown in Fig. 10; the processing method and result are similar to the first embodiment, not here. Narration.

The quality simplification signal is further processed by the signal post-processing unit 13 for signal post-processing. In this embodiment, Hilbert Yellow Conversion (HHT) is also used, and the converted signal constitutes a Hilbert yellow map of the signal ( HHS) 131, for example, Fig. 11, Fig. 11 shows the processed Hilbert Yellow Map (HHS) 131 of Fig. 10; the time-frequency map processing method is similar to the first embodiment, and will not be described herein.

The Hilbert Yellow Map (HHS) 131 of the voiceprint signal is subjected to characteristic extraction processing via the characteristic extraction processing unit 14, and in this embodiment, multi-scale entropy (MSE) extraction is also used to generate a multi-scale entropy complex of the signal. The degree mode is plotted under the characteristic extraction processing unit 14 of FIG. 8 to facilitate the description. The multi-scale entropy complexity mode at different rotational speeds is detailed as shown in FIG. 12; FIG. 8 shows only the ball guide under the characteristic extraction processing unit 14 When the screw 2 is rotated at 300 rpm, the pre-pressure of the ball lead screw 2 is reduced from 6% to 2%, and the multi-scale entropy (MSE) complexity of the voiceprint signal is increased.

The multi-scale entropy complexity mode of the ball lead screw 2 at different rotational speeds is shown in Fig. 12(a), and Fig. 12(a) shows that the preload of the ball lead screw 2 is lowered by 6% (C line) at 300 rpm. To the multi-scale entropy complexity mode of 4% (B line) and 2% (A line), the pre-pressure of the ball lead screw 2 is 6%, when the pre-pressure of the ball lead screw 2 is reduced to 2%. The failure of the pre-pressure of the ball lead screw 2 can be clearly diagnosed correctly by the 2% multi-scale entropy complexity mode; the same 12(b) shows that the pre-pressure of the ball lead screw 2 is 6% at 1500 rpm. (C line) reduced to 4% (B line) and 2% (A line) different multi-scale entropy complexity mode; Figure 12 (C) shows that at 3000 rpm, the ball lead screw 2 pre-pressure is (C line) Reduces the multi-scale entropy complexity mode to 4% (B line) and 2% (A line).

This embodiment is a remote monitoring embodiment using a voiceprint signal. A sound sensing transmitter 112 is mounted on each equipment machine of the factory, and the microphone of the sound sensing transmitter 112 is used to collect the rotation of the ball lead screw 2 when the ball is turned. The voiceprint signal is transmitted to the remote signal pre-processing unit 12 by the wireless transmission module 113. Therefore, the central monitoring room can conveniently and effectively diagnose whether the pre-pressure of each ball lead screw 2 is invalid, and can be immediately used for the gradually failing ball lead screw 2 Maintenance and increase the efficiency of factory operations.

The above is intended to be illustrative only and not limiting. Any equivalent modifications or alterations to the spirit and scope of the invention are intended to be included in the scope of the appended claims.

1. . . Pre-pressure failure diagnostic device

11. . . Sensing unit

111. . . Current sensing transmitter

112. . . Sound sensing transmitter

113. . . Wireless transmission module

12. . . Signal pre-processing unit

13. . . Signal post processing unit

131. . . Hilbert Yellow Map (HHS)

14. . . Characteristic extraction unit

15. . . Complexity mode output unit

2. . . Ball lead screw

S1, S2, S3, S4, S5. . . Method step

as well as

S21, S22, S23, S24. . . Method step

1 is a schematic view of a pre-pressure failure diagnosis device for a ball lead screw of the present invention;

2 is an explanatory diagram of steps of a method for pre-stress failure diagnosis of a ball lead screw of the present invention;

Figure 3 is an explanatory diagram of the pre-processing steps of the empirical mode decomposition (EMD) signal for the pre-pressure failure diagnosis method of the ball lead screw of the present invention;

Figure 4 is a schematic view showing a first embodiment of a ball screw lead pre-pressure failure diagnosis method and apparatus thereof according to the present invention;

Figure 5 is a diagram showing changes in (a) signals of the first embodiment of the present invention; and (b) a variation diagram of an empirical mode decomposition (EMD) process and results;

Figure 6 is a first embodiment of the invention (a) a Hilbert yellow map (HHS) formed using Hilbert yellow transition (HHT) with imf6; and (b) using Hilbert with imf7 Hilbert yellow map (HHS) formed by special yellow transition (HHT); comparison map;

Figure 7 is a first embodiment of the present invention, the ball lead screw is (a) 300 rpm; (b) 1500 rpm; (c) 3000 rpm; multi-scale entropy complex with respect to 6%, 4%, and 2% of the pre-pressure remaining Comparison graph of degree mode;

Figure 8 is a schematic view showing a second embodiment of a ball screw lead pre-pressure failure diagnosis method and apparatus thereof according to the present invention;

Figure 9 is a second embodiment of the present invention, when the ball lead screw is at 2% of the (a) pre-pressure at 300 rpm; and (b) the remaining 4% of the pre-stress is an audio signal;

Figure 10 is a diagram showing the variation of the empirical mode decomposition (EMD) process and results of the ball guide screw at 300 rpm in the second embodiment of the present invention;

Figure 11 is a Hilbert yellow map (HHS) formed by using a Hilbert yellow transition (HHT) in a ball screw of 3000 rpm in a second embodiment of the present invention;

Figure 12 is a second embodiment of the ball lead screw of the present invention with (a) 300 rpm; (b) 1500 rpm; (c) 3000 rpm; multi-scale entropy complex with respect to 6%, 4% and 2% of the pre-pressure remaining Comparison chart of degree mode.

1. . . Pre-pressure failure diagnostic device

11. . . Sensing unit

12. . . Signal pre-processing unit

13. . . Signal post processing unit

14. . . Characteristic extraction unit

15. . . Complexity mode output unit

as well as

2. . . Ball lead screw

Claims (12)

A pre-pressure failure diagnosis method for diagnosing a pre-pressure condition of a ball lead screw includes the following steps: S1: a signal presented when the ball lead screw is moved; S2': the signal is pre-processed by a signal, The signal pre-processing adopts fast Fourier transform (FFT) filtering to generate a high-quality simplification signal; S3: the high-quality simplification signal is processed by a signal, wherein the signal post-processing adopts Hilbert Huang conversion (HHT), a time-frequency intrinsic mode function (IMF) that produces instantaneous amplitude and instantaneous frequency. The time-frequency intrinsic mode function (IMF) forms a Hilbert yellow map (HHS) of the signal. S4: The Hilbert yellow spectrum (HHS) of the signal is subjected to a characteristic extraction process, wherein the characteristic extraction process uses multi-scale entropy (MSE) extraction to generate a multi-scale entropy complexity mode of the signal. S5: outputting the multi-scale entropy complexity mode of the signal for comparing the multi-scale entropy complexity mode of the ball lead screw normal pressure, thereby diagnosing whether the ball lead screw pre-pressure is invalid. The method for pre-stress failure diagnosis according to claim 1, wherein the signal of step S1 is a current signal of a current change caused by a change in the torque of the ball lead screw. A pre-pressure failure diagnosis method for diagnosing a pre-pressure condition of a ball lead screw includes the following steps: S1: a signal presented when the ball lead screw is moved; S2: the signal is pre-processed by a signal, wherein The signal pre-processing uses empirical mode decomposition (EMD) filtering to generate a high quality simplification signal, which is a time-frequency intrinsic mode function (IMF); S3: the quality simplification signal is passed A post-processing of the signal, wherein the post-processing of the signal uses a Hilbert Yellow Transform (HHT) to generate a time-frequency intrinsic mode function (IMF) of instantaneous amplitude and instantaneous frequency, the intrinsic mode of the time-frequency The function constitutes a Hilbert Yellow Map (HHS) of the signal; S4: the Hilbert Yellow Spectrum (HHS) of the signal is subjected to a characteristic extraction process, wherein the characteristic extraction process uses multi-scale entropy (MSE) Extracting to generate a multi-scale entropy complexity mode of the signal; S5: outputting the multi-scale entropy complexity mode of the signal for comparing the multi-scale entropy complexity mode of the ball lead screw normal pressure Diagnostic ball guide Pre pressure is invalid. The method for pre-stress failure diagnosis according to claim 3, wherein the signal of step S1 is selected from a current signal or a current change caused by a change in the torque when the ball lead screw is moved. One of the voiceprint signals or a combination thereof, which is caused by a change in the rotational speed of the ball lead screw. For example, the pre-stress failure diagnosis method described in claim 3, wherein the signal pre-processing of step S2 uses empirical mode decomposition (EMD), further comprising the following steps: S21: setting a recursive return of an intrinsic mode function Condition, the recursive condition includes an intrinsic mode function trend and an intrinsic mode function constant; S22: the signal is shifted by a program to generate a shift signal; S23: the shift of the signal The signal is processed by an intrinsic mode function to generate a quality simplification signal of the signal; S24: comparing whether the high quality simplification signal generated in step S23 satisfies the trend of the intrinsic mode function set in step S21 and The intrinsic mode function constant, if not satisfied, returns the high quality simplification signal to step S22 and then through a shifting process; if it is satisfied, the quality simplification signal is sent to step S3. A pre-pressure failure diagnosis device is a multi-scale entropy complexity mode for generating a ball lead screw pre-pressure signal, comprising: a sensing unit, a signal pre-processing unit, a signal post-processing unit, and a characteristic extraction unit And a complexity mode output unit; wherein the sensing unit is connected to the ball lead screw, and a signal generated by the change of the torque when the ball lead screw is moved is transmitted to the signal pre-processing unit; the signal pre-processing unit The function of fast Fourier transform (FFT) filtering can convert the signal and generate a high quality simplification signal and transmit it to the signal post processing unit; the signal post processing unit has Hilbert Yellow Conversion (HHT) function. The high quality simplification signal can be generated into a Hilbert yellow map (HHS) and sent to the characteristic extraction unit; the characteristic extraction unit has a multi-scale entropy (MSE) extraction function, and the Hilbert yellow spectrum can be (HHS) generating a multi-scale entropy complexity mode and transmitting to the complexity mode output unit; the complexity mode output unit may be the multi-scale entropy complexity mode, Output in one or a combination of graphics or data. The pre-stress failure diagnosis device according to claim 6, wherein the sensing unit is a current sensing transmitter that senses a current when the ball lead screw moves and samples the current to generate A current signal is sent to the pre-processing unit of the signal. The pre-stress failure diagnosis device according to claim 6, wherein the sensing unit further comprises a wireless transmission module, wherein the wireless transmission module can generate a signal caused by a change in the rotational distance of the ball lead screw. It is transmitted to the pre-processing unit of the signal by wireless transmission. A pre-pressure failure diagnosis device is a multi-scale entropy complexity mode for generating a ball lead screw pre-pressure signal, comprising: a sensing unit, a signal pre-processing unit, a signal post-processing unit, and a characteristic extraction unit And a complexity mode output unit; wherein the sensing unit is connected to the ball lead screw, and a signal generated by the change of the torque when the ball lead screw is moved is transmitted to the signal pre-processing unit; the signal pre-processing unit The function of empirical mode decomposition (EMD) filtering converts the signal and produces a high quality simplification signal, which is a time-frequency intrinsic mode function (IMF) and is transmitted to the signal for processing. The signal post-processing unit has a Hilbert Yellow (HHT) conversion function, which can generate a Hilbert Yellow Map (HHS) and transmit it to the characteristic extraction unit; the characteristic extraction unit The multi-scale entropy (MSE) extraction function can generate a multi-scale entropy complexity mode of the Hilbert yellow spectrum (HHS) and transmit it to the complexity mode output unit; The mode of the output unit may be a multi-scale entropy complexity mode, a graphic, or one or a combination of output data. The pre-stress failure diagnosis device according to claim 9, wherein the sensing unit is a sound sensing transmitter that senses a sound emitted by the ball lead screw and samples the sound. A tone signal is generated and transmitted to the pre-processing unit of the signal. The pre-stress failure diagnosis device according to claim 9, wherein the sensing unit is a current sensing transmitter that senses a current when the ball lead screw moves and samples the current to generate A current signal is sent to the pre-processing unit of the signal. The pre-stress failure diagnosis device according to claim 9, wherein the sensing unit further comprises a wireless transmission module, wherein the wireless transmission module can generate a signal caused by a change in the rotational distance of the ball lead screw , transmitted to the pre-processing unit of the signal by wireless transmission.