Data processing apparatus and method thereof

Abstract

The invention provides a data processing device and method for extracting resonance frequency from time domain signals, which can achieve high frequency resolution even for time domain signals with short sampling time, optimize data processing amount and rapidly extract resonance frequency. This data processing device is provided with: a modulation frequency-shift unit that processes a collected time-domain signal and obtains a frequency-shifted time-domain signal containing a frequency-reduced component obtained by shifting the entire frequency of the signal to a low frequency; a down-sampling unit that performs down-sampling processing on the frequency-shifted time-domain signal obtained by the modulation frequency-shifting unit to obtain a down-sampled frequency-shifted time-domain signal; a zero padding unit that adds zero data to the end of the data of the down-sampled frequency-shifted time-domain signal obtained by the down-sampling unit and performs a zero padding process; the Fourier transform part is used for performing Fourier transform on the data subjected to zero filling by the zero filling part to obtain a frequency domain signal; and a resonance frequency extraction unit that acquires a resonance frequency on the basis of the frequency domain signal obtained by the Fourier transform unit.

Description

Technical Field

[0001] The present invention belongs to the field of resonance frequency signal measurement, and specifically relates to a data processing method for measuring the resonance frequency of a mechanical oscillator, a magnetic resonance signal, and a resonant circuit. More specifically, it relates to a device and a method for extracting the resonance frequency from a time domain signal in real time, quickly, and stably. Background Art

[0002] The resonance frequency is an inherent property of a two-level system, and the measurement of the resonance frequency is crucial for analyzing and studying a two-level system. For example, the measurement of the resonance frequency of a mechanical oscillator can obtain information related to its mechanical properties, and the measurement of the resonance frequency of a nuclear spin can obtain information about the internal energy levels of an atom.

[0003] There are two general methods for measuring resonant frequency: 1. Locking the phase of the resonant system through phase-locked measurement. For example, the resonance of a mechanical oscillator is reflected in the vibration intensity as a resonance peak, and the phase at the resonance peak also changes greatly. Therefore, by locking the phase at the resonance peak through the resonance frequency, the measurement of the resonance frequency can be maintained; 2. Obtaining the resonance frequency by collecting time domain signals such as free induction decay signals and performing Fourier transform. These two methods have their own characteristics. The first method can quickly measure the resonance frequency, but due to the existence of PID feedback locked phase, there is a risk of unstable measurement. Although the second method has better measurement stability, it has the problem of long measurement time.

[0004] Specifically, through the second method, after collecting time domain signals at a certain sampling rate and sampling time, the discrete Fourier transform of this group of signals is used to obtain frequency domain signals, and the position of the resonance peak can be distinguished from the frequency domain signal. From the discrete Fourier transform, it can be seen that for a signal with a sampling rate of s points / second and a sampling time of t seconds, the bandwidth of the time domain signal obtained after Fourier transform is s / 2Hz, and the interval between each point in the frequency domain is 1 / t Hz. Therefore, in order to obtain a higher bandwidth, the sampling rate needs to be increased, and in order to obtain a better frequency resolution, a longer sampling time is required. However, a long sampling time will make the interval between each measurement longer, which will become undesirable in some applications.

[0005] In the prior art, a time domain signal with a short sampling time is subjected to Fourier transformation and then data fitting is performed to increase the frequency resolution. Figure 1a and Figure 1b is a schematic diagram showing the results of increasing frequency resolution through data fitting and the problems that may be encountered. Figure 1a As shown in the figure, the measured data can be fitted with Lorentz linear fitting. However, with this method, the measurement results of the resonance frequency will be different due to different fitting parameters. Moreover, for some more complex tests, such as those with more noise, the fitting may fail. Figure 1b As shown in the figure, a noise peak appears near the resonance peak, which makes curve fitting more difficult and makes it impossible to accurately measure the resonance frequency.

[0006] To this end, the present invention proposes a device and method for performing data processing on a collected time domain signal containing a resonance frequency, which can reduce the amount of data calculation and quickly and accurately extract the resonance frequency from the time domain signal. Summary of the invention

[0007] Technical problem to be solved by the invention

[0008] As mentioned above, there is a contradiction between the measurement time and frequency resolution, as well as a risk of fitting failure, in the processing of time domain signals. To facilitate understanding of the present invention, the main problems are specifically described below based on the basic principles.

[0009] The sampling theorem, also known as the Nyquist theorem, stipulates the relationship between the signal frequency and the sampling rate when a continuous signal is sampled into a discrete signal: when sampling a continuous signal X(t), the minimum frequency p(t) of the sampled signal must be greater than twice the maximum frequency of X(t), so that the sampled signal x[n] can restore the original signal X(t). In other words, the bandwidth of the sampled signal depends on the sampling rate fs, and the bandwidth is DC-fs / 2.

[0010] The frequency interval of each point after Fourier transform can be obtained through discrete Fourier transform. A time domain signal with N points can be transformed into a frequency domain signal with N points by discrete Fourier transform. The following formula (1) represents the discrete Fourier transform formula, where k is a period of N. And X[-k] = X[k]. Since the frequency domain image is symmetrical, generally speaking, the area that truly represents the frequency domain information is only in the positive half, and the number of points in the frequency domain is N / 2. Corresponding to the sampling rate fs and the sampling time t seconds, the interval between frequency points is 1 / t Hz.

[0011]

[0012] For a signal with a very high resonance frequency, in order to obtain a high resonance frequency measurement resolution, the sampling time needs to be extended as much as possible. This not only takes a long time to measure each resonance frequency signal, but also the amount of data is very large, and the data processing time will also be longer.

[0013] In the prior art, the above-mentioned method is used for frequency domain signals with short sampling time and large frequency interval (low frequency resolution). Figure 1a The resonance frequency can be obtained by fitting the data shown in FIG. However, as mentioned above, the resonance frequency obtained in this way depends on the fitting parameters and is not suitable for the case where the signal is disturbed, such as Figure 1b As shown, fitting the frequency domain signal will bring difficulties.

[0014] In this regard, the present invention proposes a method for processing data of a collected time domain signal containing a resonance frequency, which can achieve high frequency resolution even for a time domain signal with a short sampling time, optimize the data processing amount, and quickly extract the resonance frequency.

[0015] Technical means of solving problems

[0016] The first technical solution of the present invention is a data processing device for extracting the resonant frequency from the data of the collected time domain signal containing the resonant frequency, comprising: a modulation frequency shifting unit, which processes the collected time domain signal to obtain a frequency-shifted time domain signal, wherein the frequency-shifted time domain signal contains a down-frequency component obtained by shifting the frequency of the signal as a whole to a low frequency; a down-sampling unit, which down-samples the frequency-shifted time domain signal obtained by the modulation frequency shifting unit to obtain a down-sampled frequency-shifted time domain signal; a zero-padding unit, which adds a specified amount of zero data at the end of the data of the down-sampled frequency-shifted time domain signal obtained by the down-sampling unit for zero-padding; a Fourier transform unit, which Fourier transforms the data after zero-padding by the zero-padding unit to obtain a frequency domain signal; and a resonant frequency extraction unit, which obtains the resonant frequency based on the frequency domain signal obtained by the Fourier transform unit.

[0017] The second technical solution of the present invention provides a data processing device, wherein, based on the data processing device of the first technical solution, the modulation frequency shifting unit generates a sinusoidal generated signal with the same sampling rate and sampling time as the collected time domain signal, and multiplies the sinusoidal generated signal with the collected time domain signal point by point to obtain the frequency-shifted time domain signal containing the difference frequency component and the sum frequency component of the two.

[0018] The third technical solution of the present invention provides a data processing device, wherein, based on the data processing device of the second technical solution, an estimated value of the resonant frequency of the collected time domain signal is obtained in advance, and the difference between the frequency of the sinusoidal generated signal and the estimated value of the resonant frequency is smaller than the estimated value of the resonant frequency and greater than the estimated value of the broadening of the resonant peak of the frequency domain signal.

[0019] The fourth technical solution of the present invention provides a data processing device, wherein, based on the data processing device of the second technical solution, an estimated value of the resonant frequency of the collected time domain signal is obtained in advance, and the downsampling unit performs the downsampling processing by extracting one data at intervals of a specified number of points, and determines the specified number of values ​​based on the sampling theorem according to the difference between the frequency of the sinusoidally generated signal and the estimated value of the resonant frequency.

[0020] The fifth technical scheme of the present invention provides a data processing device, wherein, based on the data processing device of the first technical scheme, the ratio of the sampling time after the zero-padding process to the sampling time before the zero-padding process is less than or equal to the ratio of the number of data before the downsampling process to the number of data after the downsampling process.

[0021] A sixth technical solution of the present invention provides a data processing device, wherein, based on the data processing device of the first technical solution, the amount of zero data added by the zero padding unit is determined according to the required frequency resolution of the frequency domain signal.

[0022] The seventh technical scheme of the present invention provides a data processing device, wherein, based on the data processing device of the second technical scheme, the resonant frequency extraction unit identifies the maximum amplitude point of the frequency domain signal within a specified frequency range, and calculates the resonant frequency based on the corresponding frequency of the maximum amplitude point and the frequency of the sinusoidal generated signal.

[0023] The eighth technical scheme of the present invention provides a data processing device, wherein, based on the data processing device of the seventh technical scheme, for the frequency domain signal, the resonance frequency extraction unit identifies the maximum amplitude point of the frequency domain signal within the specified frequency range near the difference between the frequency of the sinusoidally generated signal and the estimated value of the resonance frequency.

[0024] The ninth technical solution of the present invention provides a data processing device, wherein, based on the data processing device of the first technical solution, the collected time domain signal is a free decay vibration signal of the free end of the cantilever beam obtained by driving the cantilever beam to vibrate and then stopping the driving signal in a cantilever beam force measurement device.

[0025] The tenth technical scheme of the present invention provides a data processing method for extracting the resonant frequency from the data of the collected time domain signal containing the resonant frequency, comprising: a modulation frequency shifting step, processing the collected time domain signal to obtain a frequency-shifted time domain signal, wherein the frequency-shifted time domain signal contains a down-frequency component obtained by moving the frequency of the signal as a whole to a low frequency; a downsampling step, downsampling the frequency-shifted time domain signal obtained by the modulation frequency shifting step to obtain a downsampled frequency-shifted time domain signal; a zero-padding step, adding a specified amount of zero data at the end of the data of the downsampled frequency-shifted time domain signal obtained by the downsampling step for zero-padding; a Fourier transform step, Fourier transforming the data after zero-padding in the zero-padding step to obtain a frequency domain signal; and a resonant frequency extraction step, obtaining the resonant frequency based on the frequency domain signal obtained by the Fourier transform step.

[0026] Effects of the Invention

[0027] The invention is used to process the collected time domain signal to extract the resonance frequency, which can reduce or at least not increase the data volume while increasing the frequency resolution, improve the data processing speed, and achieve extremely stable measurement results. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1a It is a schematic diagram showing an example of the result of increasing the frequency resolution by performing data fitting after Fourier transform on a time domain signal in the prior art.

[0029] Figure 1b It means adopting Figure 1a A schematic diagram of an example of a problem that may be encountered in the prior art.

[0030] Figure 2 It is a schematic diagram showing the main parts of the cantilever beam force measurement device.

[0031] Figure 3 It is a flowchart showing the overall process of measuring the resonance frequency of a cantilever beam.

[0032] Figure 4 It is a general flow chart of the data processing method of the present invention.

[0033] Figure 5 This is a schematic block diagram of a data processing device according to the present invention.

[0034] Figure 6 It is a flowchart of an embodiment of the data processing method of the present invention.

[0035] Figure 7 It is a schematic diagram of the corresponding spectrum of the time domain signal during the processing of an embodiment of the present invention.

[0036] Figure 8 It is a schematic diagram of the corresponding spectrum of the time domain signal during the processing of an embodiment of the present invention.

[0037] Fig. 9 It is a schematic diagram of the corresponding spectrum of the time domain signal during the processing of an embodiment of the present invention.

[0038] Fig.10 It is a frequency domain signal diagram after Fourier transformation in the processing process of an embodiment of the present invention.

[0039] Fig.11 It is the result of scanning field test on magnetic sample CrSBr using cantilever beam magnetic torque test method. DETAILED DESCRIPTION

[0040] The specific implementation of the present invention will be described below in conjunction with the accompanying drawings.

[0041] In the following embodiments, when referring to numbers of elements (including numbers, values, amounts, ranges, etc.), except for cases where it is particularly clearly stated and cases where it is obviously limited to specific numbers in principle, it is not limited to the specific numbers and can be above or below the specific numbers.

[0042] The implementation method described in this specification is only an example of a complete description and does not limit the scope of protection of the present invention. Based on the implementation method of the present invention, all other implementation methods that can be obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0043] The data processing method of the present invention can be implemented by a computer executing a program stored in a storage medium, or by implementing the various steps of the method through dedicated hardware. The storage medium can be any readable medium capable of storing data and programs, such as a hard disk, an optical disk, a flash disk, a chip, etc. The program can be any form of computer program, such as binary code, source code, instruction set, etc. The program can be executed by a processor on a computer, a server, a cloud computing device, etc., to implement the data processing method of the present invention.

[0044] The data processing method of the present invention is used to extract the resonance frequency from the collected discrete time domain signal containing the resonance frequency, and can be applied to measure the resonance frequency of the cantilever beam in a cantilever beam force measurement device in a dynamic working mode, for example. First, the cantilever beam force measurement device is briefly described.

[0045] [Cantilever beam force measurement device and drive device]

[0046] Figure 2 2 is a schematic diagram showing a main part of the cantilever beam force measuring device 200 .

[0047] like Figure 2 As shown, the cantilever beam force measurement device 200 includes a nano cantilever beam (hereinafter referred to as "cantilever beam") 201, a driving unit 202, an interference optical system 203, a laser light source 204, a detection unit 205, a collection unit 206 and an operation control system 207. The solid lines between the parts in the figure represent electrical connections, and the dot-dashed lines represent optical connections (optical paths).

[0048] The cantilever beam 201 can be, for example, a silicon-based cantilever beam prepared by an etching process, and can be in any common shape, such as a rectangular, T-shaped, U-shaped, triangular beam, etc. In the case of using optical interference to measure the vibration displacement of the free end, it is preferably configured in a T shape in order to increase the reflection area. The fixed end (right end in the figure) of the cantilever beam 201 is fixed to a driving part 202 composed of, for example, piezoelectric ceramics, and a magnetic sample 210 is transferred to the free end (left end in the figure), and a light reflecting surface is formed at the free end for reflecting incident light from an optical system (described later).

[0049] The cantilever beam 201 and the driving part 202 are located in a vacuum chamber 209 . Various magnetic fields are applied to the chamber 209 during operation. The magnetic field changes the resonance frequency f0 of the cantilever beam 201 on the magnetic sample 210 .

[0050] The interference optical system 203 measures the displacement of the free end of the cantilever beam 201. It uses the laser emitted by the laser light source 204 as a light source, and guides one light in the interference optical path to the free end of the cantilever beam 201. After the light is reflected at the free end, it is coupled back to the interference optical system 203 and interferes with the other light in the interference optical path to obtain interference light. The interference light is detected by the detection unit 205 composed of a photoelectric detector. The signal detected by the detection unit 205 passes through the acquisition unit 206 composed of a data acquisition card, is sampled according to a specified sampling rate, and is input into the operation control system 207.

[0051] The operation control system 207 has a signal operation / processing function, a signal generation function, and a data output / storage function. For example, it can receive the user's settings and output a voltage signal of a specified frequency to the driving unit through a built-in signal generator, so that the driving unit 202 vibrates and drives the cantilever beam 201 to vibrate, and it can perform calculations based on the input signal and the set parameters, output / store the calculated various parameters, or further organize the parameters and output a visualized table, curve, etc. As a variation, the function of the acquisition unit 206 can also be integrated into the operation control system 207. The operation control system 207 can be implemented by a general-purpose computer connected to a signal generator, but the present invention is not limited to this.

[0052] Figure 3 It is a flowchart showing the overall process of measuring the resonance frequency of a cantilever beam.

[0053] In order to measure the resonant frequency f0 of the cantilever beam, three steps can be roughly included: a cantilever beam driving step S301, in which the cantilever beam is driven to vibrate with a large amplitude and the driving is stopped after a period of time; a free decay signal acquisition step S302, in which the free decay vibration signal of the free end of the cantilever beam is acquired; and a data processing step S303, in which the acquired free decay vibration signal is processed to obtain the resonant frequency of the cantilever beam.

[0054] Specifically, in the cantilever beam driving step S301 , a signal generator in the operation control system 207 outputs a driving signal of a specified frequency to the driving unit 202 based on a specified trigger signal to vibrate the driving unit 202 (ie, piezoelectric ceramics) and drive the cantilever beam 201 to vibrate.

[0055] The driving signal may be, for example, but not limited to, a voltage signal, which satisfies the condition that the cantilever beam 201 can vibrate with a large amplitude for subsequent measurement. Therefore, the frequency of the driving signal is preferably near the current resonance frequency f0 of the cantilever beam, for example, within the resonance bandwidth Δf.

[0056] Since there is no information about the resonant frequency of the cantilever beam at this time, in the cantilever beam driving step S301, it is preferred to roughly sweep the frequency spectrum of the cantilever beam to determine the approximate position of its current resonant frequency before officially driving the cantilever beam. The result of the frequency sweep depends on the resolution of the frequency sweep, etc., and is a relatively rough result. The frequency at this approximate position can be called the "sweep resonant frequency".

[0057] Next, by using the swept resonance frequency as the frequency of the drive signal during actual driving, the cantilever beam can be easily driven and vibrated with a large amplitude.

[0058] Then, after a certain period of vibration (to ensure that the cantilever beam has vibrated), the output of the driving signal is stopped.

[0059] In the free decay signal acquisition step S302 , in response to the cessation of the driving signal, a free decay vibration signal is acquired when the cantilever beam is no longer driven and performs free decay vibration.

[0060] After the output stops, the driving signal decreases to zero immediately, and the driving part 202 (piezoelectric ceramic) stops vibrating. However, the vibration of the cantilever beam 201 does not stop immediately, but the vibration gradually decays to zero by free decay.

[0061] After the output of the driving signal stops, the signal collected by the collecting unit 206 and obtained by the detecting unit 205 detecting the interference light output by the interference optical system 203 is obtained, that is, the collected free decay vibration signal of the specified sampling rate and sampling time is obtained, and waits for the next step of processing in the operation control system 207. Here, the collected signal is a signal of a certain period of time, preferably from the beginning of the cessation of the output of the driving signal until the cantilever beam 201 stops vibrating, but it is not limited to this, and the sampling time can also be shorter. Of course, the signal can also be collected continuously and uninterruptedly, and the operation control system 207 selects a signal that can characterize the attenuation process of the cantilever beam 201 for an appropriate period (for example, a certain period after the drive is stopped) as a free decay vibration signal.

[0062] Then, in the data processing step S303, the free decay vibration signal collected in step S302 is processed in the operation control system 207 to obtain the resonance frequency f0 of the cantilever beam.

[0063] In the data processing step S303, the data processing method of the present invention is applied to extract the resonant frequency f0. That is, the free decay vibration signal collected in the above step S302 corresponds to the "collected time domain signal containing the resonant frequency" (also referred to as "collected time domain signal") targeted by the present invention, which has a specified sampling rate and sampling time. Of course, the signal used as the processing object in the present invention is not limited to the free decay vibration signal, and can be any discrete time domain signal containing the resonant frequency. Moreover, the process of obtaining the time domain signal is not limited to the above steps S301-S302, and any method that a person skilled in the art can think of can be used.

[0064] [Cantilever beam resonance frequency extraction]

[0065] The data processing step S303 is described in detail below.

[0066] Figure 4 3 is a general flow chart of the above-mentioned data processing step S303, that is, the data processing method of the present invention. Figure 4 As shown, the data processing method (data processing step S303) of the present invention includes a modulation frequency shift step S401, a downsampling step S402, a zero padding step S403, a Fourier transform step S404 and a resonance frequency extraction step S405. Each of these steps will be described below.

[0067] (Modulation frequency shift step S401)

[0068] In the modulation frequency shifting step S401, the collected time domain signal is processed to obtain a frequency shifted time domain signal, wherein the frequency shifted time domain signal includes a frequency reduction component obtained by shifting the overall frequency of the signal to a low frequency.

[0069] The present invention assumes that a rough understanding of the resonant frequency value of the cantilever beam is already available. For example, it may be the "sweep resonance frequency" obtained by roughly sweeping the frequency as mentioned in the description of the above step S301, or it may be an estimated value obtained through other commonly used methods. Here, the estimated value is assumed to be f1.

[0070] Here, a new sinusoidal signal w1 (hereinafter referred to as "sinusoidal generated signal") is created with the same sampling rate and sampling time as the acquired time domain signal. g A value f2 lower or higher than the estimated value f1 of the resonance frequency is selected, and the amplitude is 1 (arbitrary unit). The value of f2 will be described later.

[0071] By multiplying the newly created sinusoidal signal w1 with the collected time domain signal containing the resonant frequency, we can get the signal containing its frequency f g The difference frequency with the resonant frequency f0 |f0-f g | and the sum frequency |f0+fg | signal. The collected time domain signal containing the resonant frequency can be regarded as the superposition of sine waves of various frequencies. In this way, the multiplication of the sine generated signal and the time domain signal is equivalent to the multiplication of these sine waves of various frequencies and the sine generated signal and then superposition. From the product-to-sum-difference formula shown in the following formula (2), it can be seen that the multiplication of the sine generated signal w1 and the sine wave of a certain frequency of the collected time domain signal can obtain two sine waves, one is the difference of their frequencies, and the other is the sum of their frequencies. This effect is equivalent to shifting the original frequency to low frequency and high frequency respectively.

[0072]

[0073] Where A is the amplitude, ω1, ω2, The frequency and phase of two sine waves are schematically shown.

[0074] We are only interested in the difference frequency component in the modulated frequency-shifted signal. As for the sum frequency component, since its frequency is very far away from the frequency of the difference frequency component, the subsequent Fourier transform will not affect the frequency identification of the difference frequency component, so there is no need to pay attention to it.

[0075] In the present invention, it is desirable that the frequency of the difference frequency component is relatively small. It should at least be smaller than the resonance frequency f0, so that it can become a "down-conversion signal" and the processing of the present invention is meaningful; and in order to enable the resonance peak of the difference frequency component to be fully reflected in the frequency domain in the subsequent Fourier transform, considering that the resonance peak has a certain width in the frequency domain (for example, see FIG. 1 ), its frequency should at least be greater than the resonance peak width (for example, the half-height full width).

[0076] Since the frequency of the difference frequency component |f0-f g |Close to the frequency f of the sinusoidal generated signal g The difference between the resonance peak width and the estimated resonance frequency value f1 is the aforementioned f2, so the above two conditions determine the value of f2, which satisfies: resonance peak width < f2 < resonance frequency estimated value f1, and preferably takes a smaller value as possible. For example, depending on the actual situation, it can be tens to hundreds of Hz. Here, the resonance peak width can be a pre-estimated value, for example, it can be obtained by the rough frequency sweep mentioned in the description of step S301 as with the resonance frequency estimated value.

[0077] It should be noted that the "low frequency" and "high frequency" mentioned in this specification are not absolute concepts but relative concepts. A frequency that is lower than the resonant frequency can be called a "low frequency", and a frequency that is higher than the resonant frequency can be called a "high frequency".

[0078] (Downsampling step S402)

[0079] In the down-sampling step S402, the frequency-shifted time-domain signal obtained by the modulation frequency-shifting step S401 is down-sampled to obtain a down-sampled frequency-shifted time-domain signal.

[0080] According to the sampling theorem, for low-frequency signals, the original signal can be restored without such a high sampling rate, so the signal can be downsampled. The downsampling method uses the method of point extraction, that is, extracting a data point at a specified interval of n points, thus reducing the bandwidth of the signal to 1 / (n+1) of the original. After downsampling, the number of data points is greatly reduced, which can speed up subsequent data processing such as Fourier transform.

[0081] Regarding the downsampling ratio, n here can be a natural number of n ≥ 1, and the specific value can be appropriately selected based on the sampling theorem according to the frequency of the above-mentioned difference frequency component of interest. Of course, the exact frequency of the difference frequency component is unknown, but it can be approximated as the frequency f of the sinusoidal generated signal g The difference f2 between the estimated resonance frequency value f1 and the estimated resonance frequency value f1.

[0082] (Zero-filling step S403)

[0083] In the zero padding step S403, a predetermined amount of zero data is added to the end of the data of the down-sampled frequency-shifted time domain signal obtained in the down-sampling step S402 to perform zero padding processing.

[0084] As mentioned above, increasing the sampling time can increase the density of frequency domain points and increase the measurement resolution of the resonant frequency. Therefore, the existing time domain signal can be zero-filled to add zero points at the end of the data. For example, originally 1 second of data was collected (sampling time is 1 second), and the interval of frequency domain points is 1Hz. If 99 seconds of data are added after the existing data according to the original sampling rate, the interval of frequency domain points after Fourier transform is 0.01Hz, and the frequency resolution is increased by 100 times.

[0085] Therefore, in the zero-filling step S403, for the downsampled frequency-shifted time domain signal obtained in the downsampling step S402, zero data of a specified time is filled after the existing data at the same sampling rate (i.e., the sampling rate after downsampling). The amount of data after zero-filling will not be very large, because the downsampling process has been performed before, and the data points have been greatly reduced, which will not affect the speed of the subsequent Fourier transform.

[0086] In addition, the amount of zero data added to the end of the data can be appropriately set according to the actual situation. For example, if the measurement system is noisy and the frequency resolution needs to be improved to 0.01 Hz, then zero padding to 100 seconds is sufficient.

[0087] (Fourier transform step S404)

[0088] In the Fourier transform step S404, Fourier transform is performed on the data after zero padding in the zero padding step S403 to obtain a frequency domain signal.

[0089] The processing performed in the Fourier transform step S404 is no different from the prior art. The frequency domain signal is obtained by performing Fourier transform on the zero-filled data. Compared with directly performing Fourier transform on the original collected time domain data, the resolution of the frequency domain is increased and the bandwidth is reduced. In the present invention, the sampling rate is reduced through the above-mentioned downsampling step S402, and the sampling time is increased through the above-mentioned zero-filling step S403. By designing the downsampling ratio and the number of zero-fillings, the amount of data processed by the Fourier transform can be kept from increasing or even reduced, and the operation speed of the Fourier transform can be kept running quickly.

[0090] (Resonance Frequency Extraction Step S405)

[0091] In the resonance frequency extraction step S405 , the resonance frequency is obtained based on the frequency domain signal obtained in the Fourier transform step S404 .

[0092] As described above, since the data to be Fourier transformed has been subjected to the zero-filling step S403, the sampling time is increased and the frequency domain resolution is improved. Therefore, the frequency domain signal after the Fourier transform does not need to be zero-filled as in the prior art. Figure 1a , 1b In the resonance frequency extraction step S405, the frequency f corresponding to the point with the maximum amplitude can be directly identified from the frequency domain signal obtained in the Fourier transform step S404. r .

[0093] Specifically, after the modulation frequency shift in step S401, the time domain signal contains a difference frequency component and a sum frequency component, and the frequencies of both are reflected in the frequency domain signal. However, as mentioned above, the frequencies of the two are very different, so a part of the frequency range containing the difference frequency component can be intercepted from the frequency signal, and the rest can be discarded to identify the maximum amplitude. For example, considering that the frequency of the difference frequency component is close to |f1-f g |=f2, so the positive and negative specified range of frequency f2 can be intercepted, and the maximum amplitude point can be identified in it. The specified range here can be arbitrarily selected according to actual conditions, for example, the range of positive and negative f2 / 2 can be selected.

[0094] The frequency f identified here r It is not the resonant frequency itself, but the frequency of the difference frequency component that is moved to the low frequency after step S401, that is, |f0-f g |. Therefore, depending on the frequency f of the sinusoidal generated signal used in step S401 g , in f g >f1, the resonance frequency f0=fg -f r , in f g When f0 is less than f1, the resonance frequency f0 = f g +f r .

[0095] Thus, the resonance frequency of the collected time domain signal, that is, the resonance frequency f0 of the cantilever beam, is obtained.

[0096] Thus, in the data processing method of the present invention, the sampling time of the collected time domain signal is increased by zero padding, so the frequency resolution after Fourier transform can be improved, and the fitting process used in the prior art is not required, eliminating the problems encountered in the fitting process; and the amount of data is reduced by frequency shifting and downsampling before zero padding, so even if zero padding is performed, the amount of Fourier transform processing data will not be too large. Therefore, according to the data processing method of the present invention, a high frequency resolution can be achieved for a time domain signal with a short sampling time, and the amount of data processing is optimized, so that the resonant frequency can be quickly extracted.

[0097] Next, an overview of a data processing device to which the data processing method of the present invention is applied will be described.

[0098] As described above, the free decay vibration signal collected by the collecting unit 206 is input to the calculation control system 207 for calculation / processing. The data processing method of the present invention is applied to the calculation control system 207, so the calculation control system 207 substantially constitutes the data processing device of the present invention.

[0099] Figure 5 It is a schematic block diagram of the data processing device 500 of the present invention.

[0100] like Figure 5 As shown, the data processing device 500 includes a modulation frequency shifting unit 501 , a down-sampling unit 502 , a zero padding unit 503 , a Fourier transform unit 504 and a resonance frequency extracting unit 505 .

[0101] The free decay vibration signal from the acquisition unit 206 is input to the modulation frequency shift unit 501 as the acquired time domain signal, and after being processed by various parts, the acquired resonance frequency is output from the resonance frequency extraction unit 505 .

[0102] Specifically, the modulation frequency shifting unit 501 processes the collected time domain signal to obtain a frequency-shifted time domain signal, wherein the frequency-shifted time domain signal includes a down-conversion component obtained by shifting the overall frequency of the signal to a low frequency.

[0103] The down-sampling unit 502 performs down-sampling processing on the frequency-shifted time-domain signal obtained by the modulation frequency-shifting unit 501 to obtain a down-sampled frequency-shifted time-domain signal.

[0104] The zero padding unit 503 adds a predetermined amount of zero data to the end of the data of the down-sampled frequency-shifted time domain signal obtained by the down-sampling unit 502 to perform zero padding processing.

[0105] The Fourier transform unit 504 performs Fourier transform on the data padded with zero by the zero-padding unit 503 to obtain a frequency domain signal.

[0106] The resonance frequency extraction unit 505 obtains the resonance frequency based on the frequency domain signal obtained by the Fourier transform unit 504 .

[0107] The specific processing details performed by the modulation frequency shifting unit 501, the down-sampling unit 502, the zero padding unit 503, the Fourier transform unit 504 and the resonance frequency extraction unit 505 substantially correspond to the above steps S401 to S405, and will not be repeated here.

[0108] Therefore, according to the data processing device of the present invention, a high frequency resolution can be achieved even for a time domain signal with a short sampling time, and the data processing amount can be optimized, so that the resonance frequency can be extracted quickly.

[0109] Next, examples will be described.

[0110] Example 1

[0111] In Example 1, the data processing method of the present invention is implemented by using LabVIEW program.

[0112] Figure 6 It is a program of the data processing method of the present invention written based on LabVIEW program. In order to explain the present invention more clearly, instead of the time domain signal, Figure 7 to Figure 9 A schematic diagram of the corresponding spectrum of the time domain signal during the processing of Example 1 is given. Fig.10 It is the frequency domain signal diagram obtained after Fourier transform.

[0113] Figure 7 It is the corresponding resonance peak spectrum of the original collected time domain signal. The sampling rate of the original collected time domain signal is 400k points / second, the sampling time is 1 second, and the estimated resonance frequency f1 is about 4040Hz.

[0114] First, the acquired time domain signal is modulated and frequency-shifted, and a discrete-point sinusoidal signal w1 is created using data processing software (LabVIEW). Its sampling rate and sampling time are the same as those of the acquired time domain signal, and the frequency f g is selected to be 100 Hz less than the estimated resonance frequency f1 (ie, f2 = 100 Hz). g =4040Hz-100Hz=3940Hz. The sine generated signal is multiplied point by point with the collected time domain signal to produce a frequency of f0+f gThe sum frequency component of the g ≈f2=100Hz difference frequency component, which is the aforementioned frequency-shifted time domain signal. Among them, we only focus on the difference frequency component, such as Figure 8 As shown in the corresponding spectrum diagram (difference frequency component only), the resonance frequency moves to around f2 = 100 Hz.

[0115] For the difference frequency component, the bandwidth of 500Hz with a sampling rate of 1000 points / second can completely include the resonance frequency. Therefore, the sampling rate of the original signal is reduced from 400,000 points / second to 1000 points / second for downsampling, that is, one point is extracted every 400 points (one point is extracted every 399 points) to obtain a downsampled frequency-shifted time domain signal. Fig. 9 The corresponding resonance peak spectrum (only the difference frequency component) of the down-sampled frequency-shifted time domain signal is shown.

[0116] Then, the downsampled frequency-shifted time domain signal is zero-filled to fill the original 1-second sampled data to 100 seconds. Since the sampling rate becomes 1000 points / second after downsampling, the total data volume of 100 seconds is 100,000 points. Compared with the original data with a sampling time of 1 second and a sampling rate of 400,000 points / second, the total data volume is reduced to 1 / 4, which can greatly speed up the subsequent Fourier transform processing speed.

[0117] Next, perform Fourier transform on the zero-filled data. Fig.10 It represents the frequency domain signal after Fourier transformation, and the interval of each point on the spectrum becomes 0.01Hz. No further fitting processing is required for this signal. The area around f2 = 100Hz plus or minus 50Hz is intercepted to identify the maximum amplitude point of the data in this area. The horizontal axis coordinate of the maximum amplitude point is f r The frequency f of the sinusoidal signal g The sum is the required resonant frequency f0.

[0118] It is clear from Example 1 that by adopting the present invention, a high frequency resolution can be achieved even for a time domain signal with a short sampling time, and the data processing amount is optimized, so that the resonance frequency can be extracted quickly.

[0119] Example 2

[0120] Using the data processing method of Example 1, the magnetization of the CrSBr magnetic sample was measured, and the resonant frequency of the cantilever beam and its characteristics of changing with the magnetic field were obtained. The results are shown in Fig.11 middle.

[0121] like Fig.11As shown, the changes in the resonance frequency and the quality factor reflect the magnetization characteristics of the magnetic sample. As the external magnetic field increases, the resonance frequency reflects the magnetization state corresponding to the CrSBr sample. Using the data processing method of the present invention, even small changes in the resonance frequency can be effectively distinguished.

[0122] Therefore, by adopting the data processing method of the present invention to extract the resonance frequency, the resonance frequency of the cantilever beam can be simply and stably measured, and the characteristics of the magnetic properties of the sample can be more comprehensively and meticulously understood.

[0123] The above briefly describes the implementation mode and embodiments of the present invention. It should be appreciated that the present invention is not limited to the above description and may be modified in various ways.

Claims

1. A data processing device for extracting a resonance frequency from data of a time domain signal containing a resonance frequency, characterized in that: include: A modulation frequency shifting unit processes the collected time domain signal to obtain a frequency-shifted time domain signal, wherein the frequency-shifted time domain signal includes a frequency reduction component obtained by shifting the overall frequency of the signal to a low frequency; A down-sampling unit, which performs down-sampling processing on the frequency-shifted time-domain signal obtained by the modulation frequency-shifting unit to obtain a down-sampled frequency-shifted time-domain signal; a zero padding unit for adding a predetermined amount of zero data to the end of the data of the down-sampled frequency-shifted time domain signal obtained by the down-sampling unit to perform zero padding processing; A Fourier transform unit, which performs Fourier transform on the data after the zero-padding by the zero-padding unit to obtain a frequency domain signal; and A resonance frequency extraction unit is configured to obtain a resonance frequency based on the frequency domain signal obtained by the Fourier transform unit.

2. The data processing device according to claim 1, characterized in that: The modulation frequency shifting unit generates a sinusoidal generated signal having the same sampling rate and sampling time as the collected time domain signal, and multiplies the sinusoidal generated signal with the collected time domain signal point by point to obtain the frequency-shifted time domain signal containing the difference frequency component and the sum frequency component of the two.

3. The data processing device according to claim 2, characterized in that: obtaining in advance an estimated value of the resonance frequency of the collected time domain signal, A difference between the frequency of the sinusoidal generated signal and the estimated value of the resonance frequency is smaller than the estimated value of the resonance frequency and larger than an estimated value of a broadening of a resonance peak of the frequency domain signal.

4. The data processing device according to claim 2, characterized in that: obtaining in advance an estimated value of the resonance frequency of the collected time domain signal, The downsampling unit performs the downsampling process by extracting one data point at a predetermined number of intervals. Furthermore, the value of the predetermined number is determined based on the sampling theorem according to the difference between the frequency of the sinusoidal generation signal and the estimated value of the resonance frequency.

5. The data processing device according to claim 1, characterized in that: The ratio of the sampling time after the zero-padding process to the sampling time before the zero-padding process is less than or equal to the ratio of the number of data before the downsampling process to the number of data after the downsampling process.

6. The data processing device according to claim 1, characterized in that: The amount of zero data added by the zero padding unit is determined according to the required frequency resolution of the frequency domain signal.

7. The data processing device according to claim 2, characterized in that: The resonance frequency extraction unit identifies the maximum amplitude point of the frequency domain signal within a specified frequency range, and calculates the resonance frequency according to the corresponding frequency of the maximum amplitude point and the frequency of the sinusoidal generated signal.

8. The data processing device according to claim 7, characterized in that: The resonance frequency extraction unit identifies, for the frequency domain signal, a maximum amplitude point of the frequency domain signal within the prescribed frequency range around a difference between a frequency of the sinusoidal generation signal and an estimated value of the resonance frequency.

9. The data processing device according to claim 1, characterized in that: The collected time domain signal is a free decay vibration signal of the free end of the cantilever beam obtained by driving the cantilever beam to vibrate and then stopping the driving signal in the cantilever beam force measurement device.

10. A data processing method for extracting a resonance frequency from collected data of a time domain signal containing a resonance frequency, characterized in that: include: A frequency shifting step of processing the collected time domain signal to obtain a frequency shifted time domain signal, wherein the frequency shifted time domain signal includes a frequency reduction component obtained by shifting the overall frequency of the signal to a low frequency; A downsampling step of downsampling the frequency-shifted time domain signal obtained by the modulation frequency-shifting step to obtain a downsampled frequency-shifted time domain signal; a zero-padding step of adding a predetermined amount of zero data to the end of the data of the down-sampled frequency-shifted time domain signal obtained by the down-sampling step to perform zero-padding processing; A Fourier transform step, performing Fourier transform on the data after zero padding in the zero padding step to obtain a frequency domain signal; and The resonance frequency extraction step obtains the resonance frequency based on the frequency domain signal obtained in the Fourier transform step.

Images