TWI465938B - Speed ​​estimation method and module, power estimation unit and method - Google Patents

Description

Speed estimation method and module, power estimation unit and method

The invention relates to a module, a unit and a method, in particular to a method and module for estimating the speed, a power estimating unit and a method.

The tachometer for measuring the motor speed has been developed to date, and has stroboscopic speed measurements, frequency-type speed measurements, voltage-type speed measurements, etc., which can accurately measure the rotational speed of the motor. However, it has the following disadvantages:

First, the limit, frequency measurement and voltage measurement must be coupled to the rotating machine to measure the motor speed. However, based on design considerations, the shaft is usually enclosed in the motor. It is also necessary to disassemble the motor for measurement. In the case of a motor with a sealed casing, the tachometer described above is completely incapable of measuring its rotational speed.

Second, the complexity, photoelectric tachometer in order to measure the speed, and some of the motor mounted on the disc, increase the equipment of the motor, resulting in reduced practical value. Some reflection-tape is on the rotating shaft. Because the motor runs for many years, the return belt is easily contaminated, and the running motor will not stop to clean the return belt for measurement reasons. As the contamination of the return zone increases, the measured rotational speed error increases.

Third, the availability is limited, and the fault condition in which the measured motor speed is used for analysis does not cover a complex motor fault pattern.

For the above problems, the detection of the sound signal to estimate the speed of the motor is restrictive. Any rotating machine will emit an audible signal, which can be directly obtained from the air without disturbing the original signal form. Inconvenience is still easy. For the complexity, since the sound signal from the rotating machine is originally available, it is no longer necessary to install any equipment for the motor to measure directly. For availability, the sound signal has been considered as an effective signal for the diagnosis of motor fault diagnosis in many reports, and it can recognize many different types of faults through Fast Fourier Transform (FFT). Therefore, the sound detection can not only be used as the detection of the rotation speed, but also can integrate the fault diagnosis, improve the availability of the signal, and integrate the motor condition monitoring on the sound signal. In addition, the frequency of the frequency band in the sound spectrum can gently reflect the rotational speed, so that the voice detection has quite excellent conditions in the integrated fault condition and the operating condition, as shown in Fig. 1, the motor is at different speeds (1550 rpm, 1600 rpm, respectively). , 1650 rpm), the resulting sound band shift phenomenon, from Figure 1 can be seen in the spectrum of the frequency band to the speed of the smooth and continuous change.

As shown in Figure 2, in the literature "Wu Rongqing, Qin Chun, Cao Dapeng," audio sampling to detect the speed of the induction machine", The 17th Symposium on Electrical Power Engineering" or "R.-C. Wu, J.- I. Tsai, C.-T. Chiang, and C.-S. Ouyang, "Detection of Induction Motor Operation Condition by Acoustic Signal", INDIN 2010, Osaka, Japan, July 2010, pp. 792-797. A conventional method for measuring the rotational speed is suitable for estimating the rotational speed of a motor, and the rotational speed measuring method includes a table lookup procedure A1 and an estimated rotational speed program A2.

The establishment of the table lookup procedure A1 comprises the following steps:

Step A11: receiving a plurality of rotational speed values and a plurality of sound signals respectively corresponding to the equal rotational speed values, wherein the rotational speed values are measured when the motor is operated at different rotational speeds.

Step A12: Perform spectrum analysis on each sound signal by fast Fourier transform.

Step A13: taking out a frequency band in the spectrum for analysis, and calculating the frequency band according to the quadratic curve method to obtain a peak frequency of each sound signal, and the detailed implementation method of the quadratic curve method can refer to this document, so Repeat again.

Step A14: Establish a peak frequency and speed correspondence table according to the rotation speed values and the corresponding peak frequencies.

The estimated speed program A2 includes the following steps:

Step A21: receiving a sound signal generated when the motor is operated at a rotation speed, and performing an analysis operation on the sound signal to calculate a corresponding estimated peak frequency, wherein the detailed operation of the analysis operation is the same as steps A11 to A13. Therefore, it will not be repeated.

Step A22: Find the approximate data from the peak frequency and the rotational speed correspondence table according to the estimated peak frequency, and estimate the approximate data by a Lagrangian polynomial to obtain an estimated speed of the motor.

As shown in FIG. 3, a schematic diagram of the relationship between the estimated peak frequency and the actual frequency value obtained by using the quadratic curve method shows that the conventional method of measuring the rotational speed has the disadvantage that the estimated peak frequency has an actual frequency value. A fairly large error will make the speed estimate associated with the estimated peak frequency inaccurate.

Accordingly, a first object of the present invention is to provide a more accurate method of estimating the rotational speed of a rotational speed estimate.

The speed estimation method is adapted to estimate the rotation speed of a motor and is executed by a speed estimation module, and the speed estimation module includes a memory for storing a peak frequency and a speed correspondence table and is electrically connected to the memory. The processor of the body, the peak frequency and the speed correspondence table record a plurality of peak frequency values and a plurality of speed values respectively corresponding to the peak frequency values, and the speed estimation method comprises the following steps:

(A) using the processor to receive a sound signal generated when the motor operates at a rotational speed, and converting the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a frequency value And a corresponding amplitude value;

(B) using the processor to calculate an estimated peak frequency associated with the sound signal from the reference according to a centroid method;

(C) using the processor to perform a table lookup from the peak frequency and the rotational speed correspondence table of the memory according to the estimated peak frequency to obtain a plurality of related data, the related data having a plurality of peaks close to the estimated peak frequency a frequency value and a plurality of rotational speed values respectively associated with the peak frequency values; and

(D) using the processor to calculate the correlation data according to an interpolation method to obtain an estimated speed value associated with the sound signal.

A second object of the present invention is to provide a speed estimation module.

The speed estimation module is adapted to estimate the rotation speed of a motor, and comprises: a memory for storing a peak frequency and a speed correspondence table, wherein the peak frequency and the speed correspondence table record the peak frequency values and respectively correspond to the peaks a plurality of rotational speed values of the frequency value; and a processor receiving a sound signal generated by the motor operating at a rotational speed, and converting the sound signal to obtain a reference material having a plurality of frequency sampling points, each frequency The sampling point has a frequency value and a corresponding amplitude value; the processor calculates an estimated peak frequency associated with the sound signal according to a centroid method from the reference; the processor extracts the frequency from the memory according to the estimated peak frequency The peak frequency and the speed correspondence table are looked up to obtain a plurality of related data, and the related data has a plurality of peak frequency values close to the estimated peak frequency and a plurality of rotational speed values respectively related to the peak frequency values. The processor calculates the correlation data according to an interpolation method to obtain an estimated speed value associated with the sound signal.

A third object of the present invention is to provide a method for estimating the rotational speed.

The method for estimating the speed is applied to a speed estimation module. The speed estimation module includes a memory and a processor electrically connected to the memory, and the speed estimation method includes the following steps:

(A) using the processor to receive a plurality of rotational speed values respectively corresponding to a motor operating at different rotational speeds and a plurality of corresponding sound signals, and converting each sound signal into a reference material having a plurality of frequency sampling points, each The frequency sampling point has a frequency value and a corresponding amplitude value;

(B) using the processor to calculate a corresponding peak frequency from each reference material according to a centroid method; and

(C) using the processor to store the received rotational speed values and the peak frequencies corresponding to the equal rotational speed values in the memory to establish a peak frequency and rotational speed correspondence table.

A fourth object of the present invention is to provide a speed estimation module.

The speed estimation module comprises: a memory; and a processor, receiving a plurality of speed values and a plurality of corresponding sound signals, each sound signal being captured by a motor when operating at the corresponding speed value And the processor converts each sound signal into a reference material having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a corresponding amplitude value; the processor is from each reference material Calculating a corresponding peak frequency according to a centroid method; the processor stores the received speed values and the peak frequencies corresponding to the equal speed values in the memory to establish a peak frequency and speed correspondence table .

A fifth object of the present invention is to provide a power estimation method.

The power estimation method is adapted to estimate an input power of a motor and is performed by a power estimating unit, and the power estimating unit comprises a memory and a processor electrically connected to the memory, the memory storing a peak frequency And the input power correspondence table, the peak frequency and the speed correspondence table records a plurality of peak frequency values and a plurality of input power values respectively corresponding to the equal peak frequency values, and the power estimation method comprises the following steps:

(A) using the processor to receive a sound signal generated when the motor operates at a rotational speed, and converting the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a frequency value And a corresponding amplitude value;

(B) using the processor to calculate an estimated peak frequency associated with the sound signal from the reference according to a centroid method;

(C) using the processor to perform a table lookup from the peak frequency of the memory and the input power correspondence table according to the estimated peak frequency to obtain a plurality of related data, the related data having a plurality of correlation peak frequencies close to the estimated peak frequency a peak frequency value and a plurality of input power values respectively associated with the peak frequency values; and

(D) using the processor to calculate the correlation data according to an interpolation method to obtain an input power estimate associated with the sound signal.

A sixth object of the present invention is to provide a power estimating unit.

The power estimating unit comprises: a memory, storing a peak frequency and an input power correspondence table, wherein the peak frequency and the speed correspondence table records a plurality of peak frequency values and a plurality of input power values corresponding to respective peak frequency values; and The processor receives a sound signal generated by a motor operating at a rotational speed, and converts the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a corresponding An amplitude value; the processor calculates an estimated peak frequency associated with the sound signal according to a centroid method in the reference; the processor calculates a peak frequency from the memory corresponding to the peak frequency and the input power Performing a lookup table to obtain a plurality of related data having a plurality of peak frequency values close to the estimated peak frequency and a plurality of input power values respectively associated with the peak frequency values; the processor correlates the correlations The data is calculated according to an interpolation method to obtain an input power estimate associated with the sound signal.

The above and other technical contents, features and advantages of the present invention will be apparent from the following detailed description of the preferred embodiments of the invention.

<Speed Estimation Unit>

As shown in FIG. 4, the preferred embodiment of the rotational speed estimating module 1 of the present invention is adapted to estimate the rotational speed of a motor 12, and includes a memory 10 and a processor 11 electrically coupled to the memory 10.

The speed estimation module 1 performs a speed estimation method. As shown in FIG. 5, the speed estimation method includes a table lookup procedure and a speed estimation program 5.

The processor 11 is electrically connected to a storage medium (not shown) for storing a pre-stored data, the pre-stored data includes a plurality of rotational speeds of the motor 12 and a plurality of corresponding audio signals, and the processor 11 The pre-stored data can be read from the storage medium and processed. Further, the plurality of rotational speeds in the pre-stored data can be obtained by measuring the motor 12 by a tachometer (not shown). Each sound signal is captured by the motor 12 when operating at the corresponding rotational speed.

<Create a table lookup procedure>

The establishment of the table lookup procedure comprises the following steps:

Step 2: The processor 11 receives a plurality of sound signals corresponding to the rotational speed values of the motor 12 operating at different rotational speeds and a plurality of corresponding equal rotational speed values, and converts each sound signal to obtain a sampling point having multiple frequencies. For reference, each frequency sampling point has a frequency value and a corresponding amplitude value.

Again, step 2 includes the following substeps:

Sub-step 21: The processor 11 is used to sample each sound signal according to a preset sampling rate to obtain a time domain sampling data.

Sub-step 22: Perform fast Fourier transform (FFT) on each time domain sample data by using the processor 11 to obtain a spectrum data, where the spectrum data includes a plurality of frequency sampling points.

Sub-step 23: The processor 11 is used to retrieve a plurality of frequency sampling points in each spectral data located in a target frequency band covering the maximum amplitude value as the reference material, as shown in FIG. 6.

Step 3: The processor 11 calculates a corresponding peak frequency from each reference data according to a centroid method, as shown in equations (1) and (2):

Where g is the number of frequency sampling points in the reference data, A _p+i is the amplitude value corresponding to the frequency sampling point P+i, and f _P+i is the frequency value corresponding to the frequency sampling point P+i, P Is the frequency sampling point with the largest amplitude, P + ε is the frequency sampling point where the second highest amplitude occurs. When the maximum amplitude in the reference data falls at the frequency sampling point P, the number of the reference data tends to concentrate in the center, so the number A larger amplitude value will also occur near the frequency sampling point P. The second highest amplitude occurs at P + ε, where when A _{P -1} > A _{P +1} , ε = -1, and g = 4, the reference data is shown in Fig. 7. When A _{P -1} < A _{P +1} , ε = 1, and the reference is shown in Fig. 8.

As shown in FIG. 9 , a schematic diagram of the relationship between the estimated peak frequency and the actual frequency value obtained by using the centroid method is compared with FIG. 3 , and it can be seen that when g=4, 6, and 8, the estimated peak frequency is close to the actual frequency value. , more accurate than the quadratic curve method (see Figure 3).

Step 4: The processor 11 outputs the received rotational speed values and the peak frequencies corresponding to the equal rotational speed values calculated in the step 3 to the memory 10 to establish a peak frequency and rotational speed correspondence table. The peak frequency and speed correspondence table is shown in FIG. 10, and the parameters n ₁ , n ₂ ~n _k-1 , n _k respectively represent multiple speed values, and the parameters f ₁ , f ₂ ~f _k-1 , f _k A plurality of peak frequency values corresponding to the equal speed values are respectively indicated.

<Estimated speed program>

The difference between the estimated speed program 5 and the established table lookup program is that it includes the following steps:

Step 50: The processor 11 receives a sound signal generated when the motor 12 operates at a rotating speed, and converts the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a The frequency value and a corresponding amplitude value, and the detailed implementation of converting the sound signal is similar to sub-steps 21 to 23, and therefore will not be repeated.

Step 51: Using the processor 11 to calculate an estimated peak frequency of the sound signal related to the unknown rotational speed value from the reference data according to a centroid method, as shown in equations (1) and (2), respectively.

Step 52: The processor 11 performs a table lookup from the peak frequency and the rotational speed correspondence table of the memory 10 according to the estimated peak frequency to obtain a plurality of related data, and the related data has a plurality of approximate peak frequencies. The peak frequency value and a plurality of rotational speed values respectively associated with the peak frequency values.

Step 53: Using the processor 11 to perform Lagrangian polynomial interpolation according to the correlation data to calculate to obtain the speed estimation value.

As shown in FIG. 11, when the peak frequency of the unknown sound signal is f _y , three related data ( n _{m - 2} , f _{m -2} ) closest to f _y are found from the peak frequency and the rotational speed correspondence table, ( n) _{m -1} , f _{m -1} ) and ( n _m ,f _m ), then the Lagrange polynomial is as shown in equation (3):

Where f _y is the estimated peak frequency value, f _{m - i} , f _{m - j} are peak frequency values close to the estimated peak frequency, respectively, and n(f _{m - i} ) is related to the peak frequency value f _{m - i} Speed value.

As shown in FIG. 12, a correspondence relationship between a peak frequency and an estimated value of the rotational speed obtained by performing the experiment according to the above method is shown.

<Power Estimation Unit>

As shown in FIG. 13, the preferred embodiment of the power estimating unit 6 of the present invention is adapted to estimate the input power of a motor 12, and the difference between the power estimating unit 6 and the speed estimating module 1 is that the memory 10 A table of peak frequency and input power is stored, and the peak frequency and the input power correspondence table record multiple peak frequencies and corresponding input powers respectively, and the corresponding relationship between the speed and the input power is as shown in equation (4).

Where f _y is the estimated peak frequency value, f _{m - i} , f _{m - j} are peak frequency values close to the estimated peak frequency, respectively, p (f _{m - i} ) is related to the peak frequency value f _{m - i} Input power value.

As shown in FIG. 14, the correspondence between a peak frequency and the input power estimated value obtained by performing experiments according to the above method is shown.

As shown in FIG. 15, the power estimating unit 6 performs a power estimating method, which includes the following steps:

Step 2: The processor 11 receives a sound signal generated when the motor 12 operates at a rotational speed, and converts the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a The frequency value and a corresponding amplitude value.

Step 3: Using the processor 11 to calculate an estimated peak frequency associated with the sound signal from the reference material according to a centroid method.

Step 7: using the processor 11 to perform a table lookup from the peak frequency and the input power correspondence table of the memory 10 according to the estimated peak frequency to obtain a plurality of related data, where the related data has multiple close proximity to the estimated peak. The peak frequency value of the frequency and a plurality of input power values respectively associated with the peak frequency values.

Step 8: The processor 11 is used to calculate the correlation data according to an interpolation method to obtain an input power estimate associated with the sound signal.

In summary, the above embodiment has the following advantages:

1. The estimated peak frequency obtained by the centroid method is more accurate than the quadratic curve method, so the speed estimation value related to the estimated peak frequency is also more accurate.

2. Easy to detect, no need to touch the motor itself or add any equipment to the motor.

3. The relationship between input power and frequency can also be used as power detection. The extended detection function is quite simple, and the obtained input power estimation value is more accurate than the prior art because the speed estimation value is more accurate.

4. The monitoring of the rotational speed and input power can be integrated by the sound signal sampling, and the integration effect is better than the prior art due to the more accurate speed estimation value and the input power estimation value.

The above is only the preferred embodiment of the present invention, and the scope of the invention is not limited thereto, that is, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent.

1. . . Speed estimation module

10. . . Memory

11. . . processor

12. . . motor

2~4. . . Established steps

21~23. . . Substeps of conversion

5. . . Speed estimation program

50~51. . . Estimated substep

6. . . Power estimation unit

7~8. . . Power estimation steps

Figure 1 is a schematic diagram showing a gentle and continuous change in the frequency spectrum versus the rotational speed;

2 is a flow chart of a conventional method for measuring a rotational speed;

3 is a schematic diagram illustrating the relationship between the estimated peak frequency and the actual frequency value obtained by the quadratic curve method;

4 is a block diagram of a preferred embodiment of the speed estimation module of the present invention;

Figure 5 is a flow chart of the method for estimating the rotational speed by the rotational speed estimating module;

Figure 6 is a schematic diagram illustrating a target frequency band covering a maximum amplitude value;

Figure 7 is a schematic diagram showing an example of a reference material located within a target frequency band;

Figure 8 is a schematic view showing another example of the reference material;

Figure 9 is a schematic diagram showing the relationship between the estimated peak frequency and the actual frequency value obtained by the centroid method;

Figure 10 is a schematic diagram showing a correspondence table of peak frequencies and rotational speeds;

Figure 11 is a schematic diagram showing three related data closest to the peak frequency;

Figure 12 is a schematic diagram showing the correspondence between a peak frequency and the estimated speed value;

Figure 13 is a block diagram of a preferred embodiment of the power estimation unit of the present invention;

Figure 14 is a schematic diagram showing the correspondence between a peak frequency and the estimated input power; and

Figure 15 is a flow chart of the power estimation unit performing a power estimation method.

1. . . Speed estimation module

10. . . Memory

11. . . processor

12. . . motor

Claims (12)

A method for estimating a rotational speed, which is suitable for estimating a rotational speed of a motor and executed by a rotational speed estimating module, wherein the rotational speed estimating module comprises a memory for storing a peak frequency and a rotational speed correspondence table, and an electrical connection is The processor of the memory, the peak frequency and the rotational speed correspondence table records a plurality of peak frequency values and a plurality of rotational speed values respectively corresponding to the peak frequency values, and the rotational speed estimating method comprises the following steps: (A) receiving the The motor operates a sound signal generated at a rotational speed, and converts the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a corresponding amplitude value; B) using the processor to calculate an estimated peak frequency associated with the sound signal from the reference according to a centroid method; (C) using the processor to correspond to the peak frequency of the memory from the peak frequency according to the estimated peak frequency A table lookup is performed to obtain a plurality of related data having a plurality of peak frequency values close to the estimated peak frequency and a plurality of correlations respectively associated with Value of the peak frequency value of the rotational speed and the like; and (D) the processor calculates the data and the like in accordance with an interpolation method to obtain an estimate of a speed of the sound signal associated with use. The method for estimating a rotational speed according to claim 1, wherein the step (A) comprises the following sub-step: sub-step (A1): using the processor to sample the sound signal according to a preset sampling rate to obtain a time domain sampling data; sub-step (A2): using the processor to perform fast Fourier transform on the time domain sample data to obtain a spectrum data, the spectrum data includes a plurality of frequency sampling points; and sub-step (A3): utilizing The processor extracts a frequency sampling point in each of the spectral data that is within a target frequency band covering the maximum amplitude value as the reference material. According to the method for estimating the rotational speed described in claim 1, wherein the estimated peak frequency value is calculated in the following manner in the step (B): Where g is the number of frequency sampling points in the reference, P is the frequency sampling point with the largest amplitude, P + ε is the frequency sampling point where the second highest amplitude occurs, and A _p+i is the frequency sampling point P +i Corresponding amplitude values, when A _{P -1} > A _{P +1} , ε = -1, and when A _{P -1} < A _{P +1} , ε = 1. According to the method for estimating the rotational speed described in claim 1, wherein the estimated speed is calculated in the following manner in the step (D): Where f _y is the estimated peak frequency value, f _{m - i} , f _{m - j} are peak frequency values close to the estimated peak frequency, respectively, and n ( f _{m - i} ) is related to the peak frequency value f _{m - i} Speed value. A speed estimation module is suitable for estimating the rotation speed of a motor, and comprises: a memory for storing a peak frequency and a speed correspondence table, wherein the peak frequency and the speed correspondence table record the peak frequency values and respectively correspond to the peaks a plurality of rotational speed values of the frequency value; and a processor receiving a sound signal generated by the motor operating at a rotational speed, and converting the sound signal to obtain a reference material having a plurality of frequency sampling points, each frequency The sampling point has a frequency value and a corresponding amplitude value; the processor calculates an estimated peak frequency associated with the sound signal according to a centroid method from the reference; the processor extracts the frequency from the memory according to the estimated peak frequency The peak frequency and the speed correspondence table are looked up to obtain a plurality of related data, and the related data has a plurality of peak frequency values close to the estimated peak frequency and a plurality of rotational speed values respectively related to the peak frequency values. The processor calculates the correlation data according to an interpolation method to obtain an estimated speed value associated with the sound signal. According to the speed estimation module of claim 5, the processor obtains the reference data by: the processor sampling the sound signal according to a preset sampling rate to obtain a time domain sampling data; The processor performs fast Fourier transform on the time domain sample data to obtain a spectrum data, the spectrum data includes a plurality of frequency sampling points; and the processor extracts each spectrum data in a target frequency band covering a maximum amplitude value. Multiple frequency sampling points are used as the reference material. According to the speed estimation module described in claim 5, the processor calculates the corresponding estimated peak frequency value according to the following manner: Where g is the number of frequency sampling points in the reference, P is the frequency sampling point with the largest amplitude, P + ε is the frequency sampling point where the second highest amplitude occurs, and A _p+i is the frequency sampling point P +i Corresponding amplitude values, when A _{P -1} > A _{P +1} , ε = -1, and when A _{P -1} < A _{P +1} , ε = 1. According to the speed estimation module described in claim 5, the processor calculates the speed estimation value according to the following manner: Where f _y is the estimated peak frequency value, f _{m - i} , f _{m - j} are peak frequency values close to the estimated peak frequency, respectively, and n(f _{m - i} ) is related to the peak frequency value f _{m - i} Speed value. A method for estimating a rotational speed is implemented by a rotational speed estimating module, the rotational speed estimating module includes a memory, and a processor electrically connected to the memory, and the rotational speed estimating method comprises the following steps: (A) utilizing The processor receives a plurality of rotational speed values corresponding to a motor operating at different rotational speeds and a plurality of corresponding sound signals, and converts each sound signal into a reference material having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a corresponding amplitude value; (B) using the processor to calculate a corresponding peak frequency from each reference material according to a centroid method; and (C) using the processor to receive the same The rotational speed values and the peak frequencies corresponding to the respective rotational speed values are stored in the memory to establish a peak frequency versus rotational speed correspondence table. A speed estimation module comprising: a memory; and a processor receiving a plurality of speed values and a plurality of corresponding sound signals, each sound signal being captured by a motor when operating at the corresponding speed value And the processor converts each sound signal into a reference material having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a corresponding amplitude value; the processor is from each reference material Calculating a corresponding peak frequency according to a centroid method; the processor stores the received speed values and the peak frequencies corresponding to the equal speed values in the memory to establish a peak frequency and speed correspondence table . A power estimation method for estimating an input power of a motor and being performed by a power estimating unit, wherein the power estimating unit comprises a memory, and a processor electrically connected to the memory, the memory storing a peak The frequency and input power correspondence table, the peak frequency and the speed correspondence table record a plurality of peak frequency values and a plurality of input power values respectively corresponding to the equal peak frequency values, and the power estimation method comprises the following steps: (A) utilizing the processor Receiving a sound signal generated by the motor operating at a rotating speed, and converting the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a corresponding amplitude value (B) using the processor to calculate an estimated peak frequency associated with the sound signal from the reference according to a centroid method; (C) utilizing the processor to obtain the peak frequency from the memory based on the estimated peak frequency Performing a look-up table in the input power correspondence table to obtain a plurality of related data, the related data having a plurality of peak frequencies close to the estimated peak frequency a value and a plurality of input power values respectively associated with the peak frequency values; and (D) using the processor to calculate the correlation data according to an interpolation method to obtain an input power estimate associated with the sound signal . A power estimating unit includes: a memory storing a peak frequency and an input power correspondence table, wherein the peak frequency and the input power correspondence table records a plurality of peak frequency values and a plurality of input power values corresponding to respective peak frequency values; a processor receives a sound signal generated by a motor operating at a rotational speed, and converts the sound signal to obtain a reference data having a plurality of frequency sampling points, each frequency sampling point having a frequency value and a phase Corresponding amplitude value; the processor calculates an estimated peak frequency associated with the sound signal according to a centroid method in the reference; the processor corresponds to the peak frequency and the input power from the memory according to the estimated peak frequency Performing a lookup table to obtain a plurality of related data having a plurality of peak frequency values close to the estimated peak frequency and a plurality of input power values respectively associated with the peak frequency values; the processor The correlation data is calculated according to an interpolation method to obtain an input power estimate associated with the sound signal.