TWI436047B - System and method for evaluating mechanical vibration signal using mse - Google Patents

Description

Mechanical vibration signal evaluation system and method using multi-scale entropy

The present invention relates to a mechanical vibration signal evaluation system and method, and more particularly to a system and method for utilizing multi-scale entropy to evaluate mechanical vibration signals.

Machinery is a fairly common system and actuation principle in industry and life. In the field of machinery, as processing demands become more stringent, the requirements for processing quality tend to be high-precision and high standards.

As product complexity increases, it is often necessary to optimize multiple process characteristics while improving process quality. Therefore, good cutting frequency, tool wear, surface roughness and cutting force are the goals pursued by the machining. Since the mechanical working type is a dynamic behavior, it is composed of several operating modes, and when one of the parameters or the operating factor is mutated, it is likely to cause damage and danger to the overall structure and productivity. These subtle phenomena often use Fast Fourier Transform (FFT) and Root Mean Square (RMS) algorithms as analytical tools in traditional techniques, but often cannot be directly analyzed efficiently. The cause of the problem. In terms of FFT detection, only the most basic rough decision can be made for the machine. The RMS algorithm is characterized by fast and simple calculations and the ability to monitor any vibration changes in real time, but does not recognize the characteristics of the problem and the frequency distribution.

From this point of view, if the processing quality inspection of the machine is carried out by the conventional detection and evaluation technology, the effective analytical judgment effect cannot be achieved at present. In addition, the theory and judgment results of the conventional detection and evaluation techniques are often too Difficult, often require highly qualified personnel to be able to distinguish. Therefore, in the machinery market that continuously improves efficiency and pursues processing quality, it is necessary to have a more efficient and accurate detection and evaluation technology.

In view of this, the present invention is directed to a mechanical vibration signal evaluation system and method, and more particularly to a system and method for utilizing multi-scale entropy to evaluate mechanical vibration signals, thereby solving the above-described conventional problems.

One aspect of the present invention is a mechanical vibration signal evaluation system that uses a multi-scale entropy (MSE) algorithm to evaluate mechanical vibration signals, in addition to being able to quickly assess mechanical vibration conditions, which can produce more accurate critical result.

According to an embodiment of the present invention, the mechanical vibration signal evaluation system of the present invention comprises a filter module, a signal splitting module, a processor and a classifier. The filter module filters the timing signal by a frequency range and outputs the filtered timing signal. The signal splitting module is coupled to the filter module, receives the filtered timing signal, and divides the filtered timing signal by a clock interval to generate a plurality of signal segments. The processor is coupled to the signal segmentation module, receives the plurality of signal segments, and calculates a plurality of entropy values of each signal segment. a classifier coupled to the processor, having an entropy value comparison table, the classifier receiving the plurality of entropy values of each signal segment, and comparing the plurality of entropy values to an entropy value table to generate a mechanical vibration signal evaluation result.

In a practical application, the mechanical vibration signal evaluation system of the present invention further includes a sampling module, and the sampling module is coupled to the filtering module to sample the original signal at a preset frequency to generate a timing signal. In addition, the filter module can include a band pass filter. Furthermore, each entropy is the result of performing an operation on one of a plurality of operational scales in a multi-scale entropy algorithm. In addition, classification The entropy value comparison table can be established by inputting a plurality of standard entropy values in advance. The entropy value comparison table includes a plurality of entropy value intervals, and each entropy value interval includes at least one of the plurality of standard entropy values.

One aspect of the present invention resides in a method for evaluating mechanical vibration signals, which uses a multi-scale entropy algorithm to evaluate mechanical vibration signals. In addition to being able to quickly assess mechanical vibration conditions, a more accurate judgment result can be produced.

According to an embodiment of the present invention, the mechanical vibration signal evaluation method of the present invention comprises the steps of: filtering a timing signal by a frequency range to generate a filtered timing signal; and dividing the filtered timing signal by a clock interval, Generating a plurality of signal segments; computing a plurality of entropy values for each of the signal segments; comparing the plurality of entropy values to an entropy value table to generate a mechanical vibration signal evaluation result.

In practical applications, each entropy is the result of one of a plurality of operational scales in a multi-scale entropy algorithm. In addition, the mechanical vibration signal evaluation method further includes the following steps: inputting a plurality of standard entropy values in advance, and establishing an entropy value comparison table. The entropy value comparison table includes a plurality of entropy value intervals, and each entropy value interval includes at least one of the plurality of standard entropy values.

In summary, the mechanical vibration signal evaluation system and method of the present invention uses a multi-scale entropy algorithm to evaluate mechanical vibration signals. Because of the multi-scale entropy algorithm, multi-dimensional analysis can be performed according to different operational scales. Therefore, the present invention can compare a plurality of entropy values calculated by a plurality of operation scales, not just a single entropy value. Thereby, the mechanical vibration signal evaluation system and method of the present invention can greatly improve the accuracy of the judgment result in addition to the rapid evaluation of the mechanical vibration condition.

The advantages and spirit of the present invention will be further understood from the following detailed description of the invention.

Referring to FIG. 1, FIG. 1 is a block diagram showing a mechanical vibration signal evaluation system according to an embodiment of the present invention. As shown in FIG. 1 , the mechanical vibration signal evaluation system 1 of the present invention includes a filter module 10 , a signal splitting module 12 , a processor 14 , a classifier 16 , and a sampling module 18 . The filter module 10 is electrically connected between the signal splitting module 12 and the sampling module 18, and the processor 14 is electrically connected between the signal splitting module 12 and the classifier 16.

The filter module 10 filters the timing signals in a frequency range and outputs the filtered timing signals. In practice, the filter module 10 can include or can be a bandpass filter and the frequency range is from 0.5 Hz to 30 Hz. That is to say, the filter module 10 can filter out unwanted signals, leaving only signals with a frequency range of 0.5 Hz to 30 Hz. On the other hand, if the frequency of the mechanical vibration signal to be detected is high, the filter module 10 can be designed according to the frequency of the mechanical vibration signal to be detected, so the filter module 10 is not limited to only leaving the frequency range at 0.5. Signal from Hz to 30Hz.

The signal segmentation module 12 receives the filtered timing signals output by the filter module 10, and divides the filtered timing signals by a clock interval to generate a plurality of signal segments. In practice, the signal segmentation module 12 is configured to divide the signal into a plurality of segments to further process the segmented signal, and the clock interval is not limited herein. For example, the clock interval may be 30 seconds.

The processor 14 receives the plurality of signal segments and operates a plurality of entropy values for each of the signal segments. In practice, each entropy is the result of computing in one of multiple operational scales in a multi-scale entropy (MSE) algorithm. For example, the processor 14 may use 20 operational scales to perform the multi-scale entropy algorithm, and each of the (1st to 20th) operational scales may be divided by the multi-scale entropy algorithm. The entropy values of the respective (1st to 20th) are not generated and output to the classifier 16. Here, the present invention does not limit the number of operational scales, and the skilled person can select the appropriate number of operational scales.

It should be noted that the present invention does not limit its operation mode because the multi-scale entropy algorithm is an algorithm that can be understood by those skilled in the art, although it can be implemented in a number of different ways. The operation of multi-scale entropy, but as long as it is a multi-scale entropy operation and each of the operational scales can generate respective entropy values, which belongs to the scope of the present invention, the technician can determine the operation mode by himself.

The classifier 16 has an entropy value comparison table, and the classifier 16 receives the plurality of entropy values of each signal segment, and compares the plurality of entropy values with an entropy value table to generate a mechanical vibration signal evaluation result. . In practice, classifier 16 can be a linear or non-linear classifier. And, the classifier 16 may input a plurality of standard entropy values in advance to establish an entropy value comparison table. The entropy value comparison table includes a plurality of entropy value intervals, and each entropy value interval includes at least one of the plurality of standard entropy values. For example, before using the mechanical vibration signal evaluation system 1 to evaluate the mechanical vibration signal, the entropy value comparison table may be first adjusted by inputting a number of standard values or by a professional to enable the classifier 16 to accurately classify the processing. The entropy value entered by the device 14.

In addition, the plurality of entropy value sections respectively correspond to a plurality of operation scales, for example, the entropy values calculated by the first to the 20th operation scales may be compared by the first to the 20th entropy value sections, respectively. Wherein, each entropy interval should include a range for indicating different mechanical vibration signal evaluation parameters. In addition, the user can select which mechanical vibration signal evaluation result to be generated by which entropy interval.

In addition, when the entropy value comparison table is adjusted, a plurality of entropy value areas are usually set. Corresponding to multiple standard entropy values respectively, the fault tolerance rate of the mechanical vibration signal is improved, so that the judgment of the mechanical vibration signal can be protected from some undesired factors.

The sampling module 18 samples the original signal at a predetermined frequency to generate a timing signal. In practice, the original signal generated by detecting the state of mechanical vibration is often a long-term signal. If the signal is analyzed and processed from the original signal, the computing resources will be excessively consumed. Therefore, in order to reduce the amount of calculation, the amount of signal data entering the filter module 10 can be effectively reduced, and the sampling module 18 can reduce the signal frequency from the original signal. For example, the preset frequency can be preset to 256 Hz.

In addition, the mechanical vibration signal evaluation result generated by the mechanical vibration signal evaluation system 1 of the present invention is graphically presented, and the graphic may include a multi-scale entropy map. Please refer to FIG. 2A to FIG. 2C. FIG. 2A to FIG. 2C respectively show a timing chart, a time-frequency diagram and a multi-scale entropy map of the elevator ascending vibration. Taking the elevator ascending vibration as an example in practical application, Figure 2A shows the change of the amplitude of the elevator ascending vibration signal with time, and Figure 2B shows the change of the elevator's rising vibration signal with time. Figure 2C shows the The multi-scale entropy of the elevator ascending vibration signal corresponding to time. Compared with FIG. 2A and FIG. 2B, in the multi-scale entropy map (FIG. 2C) generated by the mechanical vibration signal evaluation system 1 of the present invention, it can be clearly determined that the elevator has abnormal jitter during the time period of 25 seconds to 65 seconds. Happening.

The following is a detailed description of the mechanical vibration signal evaluation method and the drawing of the present invention.

Referring to FIG. 1 and FIG. 3, FIG. 3 is a flow chart showing a method for evaluating a mechanical vibration signal according to an embodiment of the present invention. As shown in the figure, in step S20, the sampling module 18 samples an original signal at a predetermined frequency to generate a timing signal, and the sampling module 18 can further transmit the timing signal to the filtering module 10. In practice, the sampling module 18 can select other sampling conditions for the original signal. Sampling, for example, the sampling module 18 can also be used to eliminate signals with excessive or too small amplitudes, thereby reducing the amount of computation of signals entering the filtering module 10.

Next, in step S22, the filter module 10 filters the timing signals by a frequency range to generate the filtered timing signals.

Next, in step S24, the signal segmentation module 12 receives the filtered timing signal from the filter module 10, and divides the filtered timing signal by a clock interval to generate a plurality of signal segments.

Next, in step S26, the processor 14 receives the plurality of signal segments and operates a plurality of entropy values for each of the signal segments in a multi-scale entropy algorithm. In practice, the processor 14 uses a multi-scale entropy algorithm and operates each of the signal segments on multiple operational scales to correspondingly generate a plurality of entropy values.

Next, in step S28, the classifier 16 compares the plurality of entropy values with the entropy value table to generate a mechanical vibration signal evaluation result. For example, the processor 14 receives a certain signal segment, and then calculates 20 entropy values of the signal segment through a multi-scale entropy algorithm on 20 operational scales, and then inputs the 20 entropy values into the classifier 16. For example, when the entropy value calculated by the i-th operation scale falls within the first range of the i-th entropy value interval, the classifier 16 outputs the first processing quality evaluation parameter; when the entropy value falls Within the second range of the i-th entropy interval, the classifier 16 outputs a second processing quality evaluation parameter.

It is worth mentioning that if the timing signal is long, the mechanical vibration signal evaluation system 1 can further perform step S26 and step S28 repeatedly, and the mechanical vibration signal evaluation result in each signal segment can be classified, thereby obtaining a series of The mechanical vibration signal is evaluated to determine the state of mechanical vibration at each time point.

In summary, the mechanical vibration signal evaluation system and method of the present invention uses a multi-scale entropy algorithm to evaluate mechanical vibration signals. Further, the present invention The mechanical vibration signal evaluation system and method can be widely applied to the analysis of various mechanical vibration signals. Since the multi-scale entropy algorithm can perform multi-dimensional analysis according to different operation scales, the present invention is not only based on a single entropy. The value is judged. Thereby, the mechanical vibration signal evaluation system and method of the present invention can greatly improve the accuracy of the judgment result in addition to the rapid evaluation of the mechanical vibration condition of the subject. On the other hand, the classifier of the present invention can be adjusted in advance by a professional to make the evaluation result of the present invention more accurate and reliable.

The features and spirit of the present invention will be more apparent from the detailed description of the preferred embodiments. On the contrary, the intention is to cover various modifications and equivalents within the scope of the invention as claimed.

1‧‧‧Mechanical vibration signal evaluation system

10‧‧‧Filter module

12‧‧‧Signal splitting module

14‧‧‧ Processor

16‧‧‧ classifier

18‧‧‧Sampling module

S20~S28‧‧‧Step process

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a block diagram showing a mechanical vibration signal evaluation system in accordance with an embodiment of the present invention.

Figure 2A to Figure 2C show the timing diagram, time-frequency diagram and multi-scale entropy diagram of the elevator ascending vibration.

FIG. 3 is a flow chart showing a method for evaluating a mechanical vibration signal according to an embodiment of the present invention.

1‧‧‧Mechanical vibration signal evaluation system

10‧‧‧Filter module

12‧‧‧Signal splitting module

14‧‧‧ Processor

16‧‧‧ classifier

18‧‧‧Sampling module

Claims (18)

A mechanical vibration signal evaluation system comprises: a filter module, filtering a timing signal in a frequency range and outputting a filtered timing signal; a signal dividing module coupled to the filtering module, receiving the filtered timing And dividing the filtered timing signal by a clock interval to generate a plurality of signal segments; a processor coupled to the signal segmentation module, the processor receiving the signal segments, and calculating each of the signal segments An entropy value; and a classifier coupled to the processor, having an entropy value comparison table, the classifier receiving the isentropic value of each of the signal segments, and comparing the entropy values to the entropy value table According to the results of a mechanical vibration signal evaluation. The mechanical vibration signal evaluation system of claim 1, wherein the mechanical vibration signal evaluation result is graphically presented. The mechanical vibration signal evaluation system of claim 2, wherein the graphic comprises a multi-scale entropy map. The mechanical vibration signal evaluation system of claim 1, further comprising: a sampling module coupled to the filtering module to sample an original signal at a predetermined frequency to generate the timing signal. The mechanical vibration signal evaluation system of claim 1, wherein the filter module comprises a band pass filter. The mechanical vibration signal evaluation system according to claim 1, wherein each of the entropy values is a result of performing operation on one of a plurality of operational scales in a multi-scale entropy algorithm. The mechanical vibration signal evaluation system according to claim 1, wherein the classifier establishes the entropy value comparison table by inputting a plurality of standard entropy values in advance. The mechanical vibration signal evaluation system of claim 7, wherein the entropy value comparison table includes a plurality of entropy value intervals, each of the entropy value intervals including at least one of the standard entropy values. The mechanical vibration signal evaluation system of claim 1, wherein the classifier is a linear classifier or a nonlinear classifier. The mechanical vibration signal evaluation system of claim 1, wherein the timing signal is a mechanical vibration signal. A method for evaluating a mechanical vibration signal, comprising the steps of: filtering a timing signal by a frequency range, thereby generating a filtered timing signal; dividing the filtered timing signal by a time interval to generate a plurality of signal segments; Computing a plurality of entropy values of each of the signal segments; and comparing the entropy values to an entropy value table to generate a mechanical vibration signal evaluation result. The method for evaluating a mechanical vibration signal according to claim 11, wherein the mechanical vibration signal evaluation result is graphically presented. The method for evaluating a mechanical vibration signal according to claim 12, wherein the graphic comprises a multi-scale entropy map. The method for evaluating a mechanical vibration signal according to claim 11, further comprising the step of sampling an original signal at a predetermined frequency to generate the timing signal. The method for evaluating a mechanical vibration signal according to claim 11, wherein each of the entropy values is a result of performing an operation on one of a plurality of operational scales in a multi-scale entropy algorithm. The method for evaluating the mechanical vibration signal described in claim 11 of the patent application further includes the following steps: A plurality of standard entropy values are input in advance, and the entropy value comparison table is established. The mechanical vibration signal evaluation method according to claim 16, wherein the entropy value comparison table includes a plurality of entropy value intervals, and each of the entropy value intervals includes at least one of the standard entropy values. The method for evaluating a mechanical vibration signal according to claim 11, wherein the timing signal is a mechanical vibration signal.