Vibration measurement method based on acceleration sensor compensation

Abstract

The present invention discloses a vibration measurement method based on acceleration sensor compensation, comprising: collecting vibration acceleration at various frequency points preset within a frequency measurement range by a vibration measuring product equipped with a MEMS acceleration sensor, comparing the collected vibration accelerations with preset vibration acceleration values, determining an attenuation coefficient at each frequency point, and establishing a compensation model for the vibration measuring product; collecting the measured vibration signal at an adaptive sampling rate by the vibration measuring product to obtain an acceleration sequence thereof, compensating the acceleration sequence according to the compensation model; calculating an acceleration peak value; performing time domain integration processing and detrending processing on the acceleration sequence to obtain a velocity sequence, and calculating an effective velocity value; performing time domain integration processing and detrending processing on the velocity sequence to obtain a displacement sequence, and calculating a peak-to-peak displacement value. The present invention improves detection accuracy by compensating the acceleration collected by the MEMS inertial sensor.

Description

Technical Field

[0001] The present invention relates to the technical field of vibration signal measurement and analysis, in particular to a vibration measurement method based on acceleration sensor compensation. Background Art

[0002] In recent years, with the rapid development of industrial automation, especially in large-scale heavy industries such as coal mining, power generation, and chemical engineering, unplanned downtime can lead to huge economic losses. To reduce the potential losses caused by unplanned downtime, regular and real-time monitoring of certain industrial parameters in the system to predict failures in advance has become particularly important. In particular, the development of MEMS technology has made various monitoring devices smaller, more portable, and more responsive.

[0003] MEMS inertial sensors offer advantages over existing piezoelectric ceramic inertial sensing methods, such as faster response and greater portability. However, because vibration signal transmission requires a rigid connection, MEMS inertial sensors are directly attached to the PCB, resulting in insufficient rigidity and more severe signal attenuation than with traditional piezoelectric ceramic vibration measurement equipment. Experiments have shown that the higher the vibration frequency of the measured object, the greater the signal attenuation. Summary of the Invention

[0004] In order to solve the problems existing in the prior art, the purpose of the present invention is to provide a vibration measurement method based on acceleration sensor compensation. The present invention uses compensation for the acceleration collected by the MEMS inertial sensor to achieve improved detection accuracy.

[0005] To achieve the above object, the present invention adopts a technical solution: a vibration measurement method based on acceleration sensor compensation, comprising the following steps:

[0006] Step 1: Use a vibration measurement product equipped with a MEMS accelerometer to collect vibration acceleration at each frequency point within a preset frequency measurement range, compare it with the preset vibration acceleration value, determine the attenuation coefficient at each frequency point, and establish a compensation model for this vibration measurement product;

[0007] Step 2: The vibration measuring product collects the measured vibration signal at an adaptive sampling rate to obtain its acceleration sequence, and compensates the acceleration sequence according to the compensation model;

[0008] Step 3: Calculate the peak acceleration after adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets;

[0009] Step 4: Perform time domain integration and detrending processing on the acceleration sequence to obtain the velocity sequence. After adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets, the effective value of the velocity is calculated.

[0010] Step 5: Perform time domain integration and detrending processing on the velocity series to obtain the displacement series. After adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets, the peak-to-peak value of the displacement is calculated.

[0011] As a further improvement of the present invention, the step 1 is specifically as follows:

[0012] According to the frequency measurement range of the vibration measurement product equipped with MEMS accelerometer, points are evenly divided and the acceleration at each frequency point is collected to establish the frequency-compensation coefficient curve:

[0013] coef f =A f ÷A0 f

[0014] where coef f is the compensation coefficient when the frequency is fHz, A f is the peak value of vibration acceleration at frequency fHz, A0 f is the peak acceleration value set by the vibration source at a frequency of fHz;

[0015] Establish polynomial fitting model and piecewise polynomial fitting model based on frequency-compensation coefficient curve:

[0016] coef f =c n f n +c n-1 f n-1 +…+c1f+c0

[0017]

[0018] Where c1 to c7 are the fitting curve coefficients, and f1 to f4 are the frequency segment ranges.

[0019] As a further improvement of the present invention, the step 2 is specifically as follows:

[0020] When sampling, first collect the measured vibration signal at the highest sampling frequency and perform spectrum analysis to obtain the approximate frequency range of the measured signal:

[0021]

[0022] Where Y(k) is the result of the acceleration sequence after frequency domain transformation, n is the number of sampling points, and F(k) when Y(k) is the maximum value is the main frequency f of the measured signal signal ;

[0023] Calculate the appropriate sampling frequency and resample at this frequency to improve sampling accuracy. The appropriate sampling frequency is as follows:

[0024]

[0025] Where F is the configurable sampling rate of the MEMS accelerometer, ρ is the sampling magnification, and Fs is the selected frequency of the next sampling;

[0026] Use Fs to sample the measured signal to obtain a new acceleration sequence A1, and then compensate the acceleration sequence A1 to obtain the acceleration sequence A2:

[0027] A2(k)=A1(k)÷coef.

[0028] As a further improvement of the present invention, in step 3, step 4 and step 5, the adaptive high-pass filtering process is specifically as follows:

[0029] According to the sampling rate Fs selected by adaptive sampling, the filter cutoff frequencies Fls and Flp are calculated. The signal in the frequency range less than Fls is cut off, the signal in the frequency range from Fls to Flp is attenuated, and the signal in the frequency range greater than Flp remains unchanged:

[0030]

[0031] Wherein, r1 is the conversion coefficient of the cutoff frequency Flp, and r2 is the conversion coefficient of the cutoff frequency Fls;

[0032] Calculate the order N of the high-pass filter:

[0033]

[0034] Where b0 is the attenuation coefficient of the amplitude-frequency response signal attenuation interval, N0 is the initial order, if N0 is an even number, then N and N0 are equal, if N0 is an odd number, then N = N0 + 1. ceil(.) means rounding up, and floor(.) means rounding down.

[0035] Then, the filter coefficients [b, a] are approximately calculated using the least squares method, and the impulse response is smoothed using a window to complete the design of a high-pass filter with [b, a] as the coefficients.

[0036] As a further improvement of the present invention, in steps 4 and 5, the time domain integration adopts the trapezoidal formula or the Simpson formula to reduce the integration error, and then the trend caused by the integration is removed by removing the best straight line fitting line, and the root mean square value of the processed velocity sequence or displacement sequence is calculated to obtain the effective value of the velocity or displacement.

[0037] As a further improvement of the present invention, in steps 3 and 5, the calculation method of the peak acceleration and the peak-to-peak displacement is as follows:

[0038] According to the main frequency f of the signal signalAnd the sampling rate Fs estimates the number of sampling points N in one signal cycle s :

[0039] N s =Fs÷f signal

[0040] Press N in sequence A s The length is truncated to each subsequence A i , count the maximum and minimum values ​​of the subsequence:

[0041]

[0042] Where m is the number of subsequences that have been intercepted, max With A min Sort by large to small and small to large respectively, remove the first 3 values, and take the remaining maximum value a max and a min Calculate the peak value a peak =(a max -a min ) / 2, or peak-to-peak s p-p =s max -s min .

[0043] The vibration source in step 1 must be a calibrated device with stable output and programmable vibration frequency and acceleration, such as a vibration test bench, to ensure the reliability of the sampled data. The frequency points set for the vibration source must cover the product's frequency measurement range and be equally spaced. The acceleration setting for the vibration source should be as high as possible within a reasonable range. The compensation model can be a frequency-compensation coefficient parameter table or a frequency-compensation coefficient curve based on polynomial fitting, and can be a polynomial fitting curve segmented by frequency.

[0044] In step 2, sampling is performed by first acquiring the measured vibration signal at the highest sampling frequency and performing spectral analysis to obtain the approximate frequency range of the measured signal. Calculating an appropriate sampling frequency and re-sampling at that frequency can improve sampling accuracy. Before compensation, spectral analysis is performed on the sampled acceleration sequence to obtain the signal frequency or frequency distribution. Based on the frequency, a compensation model is used to calculate the compensation coefficient for acceleration compensation, effectively offsetting the attenuation of the vibration signal caused by conduction.

[0045] High-pass filtering is implemented using a software algorithm. Adaptive selection of filter parameters such as cutoff frequency and bandwidth based on the sampling frequency optimizes filtering performance. Because vibration signals are periodic, any unstable data that may appear after high-pass filtering should be discarded. Non-zero baseline offsets are removed to keep the mean of the periodic vibration signal close to zero. Peak acceleration is calculated by counting the maximum and minimum values ​​of the acceleration sequence.

[0046] The acceleration sequence is integrated and detrended in the time domain as described in step 4 to obtain a velocity sequence. The RMS velocity is then calculated after adaptive high-pass filtering, discarding unstable data, and removing any non-zero baseline offsets. The time domain integration uses the trapezoidal or Simpson formula to minimize integration errors. Detrending caused by integration is then removed by removing the best straight-line fit. The RMS velocity is then calculated from the processed velocity sequence.

[0047] The velocity series is integrated and detrended in the time domain as described in step 5 to obtain a displacement series. After adaptive high-pass filtering, unstable data is discarded, and non-zero baseline offsets are removed, the peak-to-peak displacement is calculated. The peak-to-peak displacement is calculated by counting the maximum and minimum values ​​of the displacement series to reduce the impact of the accumulated error after two integrations on the peak-to-peak displacement calculation.

[0048] The beneficial effects of the present invention are:

[0049] 1. This invention uses MEMS inertial sensors to monitor the vibration of load motors in heavy industry. By analyzing the acceleration and other data obtained from the monitoring, it can provide early warning of possible motor failures, eliminating the economic losses caused by unplanned downtime.

[0050] 2. The present invention improves sampling accuracy and processing accuracy through adaptive sampling rate and adaptive filtering parameters, and compensates for the acceleration of the product at different frequency points. It can effectively solve the signal attenuation problem caused by insufficient rigidity in the vibration conduction path of the product, especially the signal attenuation problem of the increasingly used MEMS acceleration sensors under high-frequency vibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a partial structural diagram of the mining multi-parameter vibration speed sensor GZZ15 / 1500 in an embodiment of the present invention;

[0052] Figure 2 A flowchart of an embodiment of the present invention;

[0053] Figure 3 A frequency-compensation coefficient curve diagram in an embodiment of the present invention;

[0054] Figure 4 Schematic diagram of piecewise polynomial fitting of a frequency-compensation coefficient curve in an embodiment of the present invention;

[0055] Figure 5 Schematic diagram of the amplitude-frequency response curve of the high-pass filter in an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0057] Example

[0058] In this embodiment, a mining multi-parameter vibration speed sensor GZZ15 / 1500 equipped with a MEMS acceleration sensor model ism330dhc is used to collect vibration acceleration. Figure 1 The diagram below shows a partial structure. The MEMS accelerometer is soldered to a small PCB via a patch. The PCB is fixed to a rigid metal component on the base. The vibration transmission path is: external vibration source → base metal component → PCB → MEMS accelerometer.

[0059] Figure 2 Zhong 201 uses a vibration measurement product equipped with a MEMS accelerometer to collect vibration acceleration at each frequency point preset within the frequency measurement range, compares it with the preset vibration acceleration value, determines the attenuation coefficient of each frequency point, and establishes a compensation model for this vibration measurement product. The vibration source must be a calibrated device with stable output and the ability to set vibration frequency and acceleration, such as a vibration test bench, to ensure the reliability of the sampled data. The frequency points set by the vibration source cover the frequency measurement range of the product and are equally divided. The acceleration setting value of the vibration source should be as large as possible within a reasonable range. The compensation model can be a frequency-compensation coefficient parameter table or a frequency-compensation coefficient curve based on polynomial fitting, and can be a polynomial fitting curve segmented by frequency.

[0060] In this embodiment, the vibration source is a small, high-precision vibration test bench that has been verified and can cover the measurement range of GZZ15 / 1500: frequency 10Hz~1KHz, acceleration 0~15g. The measurement frequency points are evenly spaced at 10Hz intervals within the range of 10Hz~1KHz. In order to avoid destructive vibration, the acceleration is 0.5g below 50Hz, 2g below 100Hz, and 15g for the rest. GZZ15 / 1500 is used to collect the acceleration sequence of each frequency point respectively. The collection method is the same as Figure 2 202 before compensation, the data processing method is the same as Figure 2 The acceleration value at each frequency point is obtained by dividing the acceleration measured at each frequency point by the vibration source set value to obtain the compensation coefficient:

[0061] coef f =A f ÷A0 f ,f=10,20,…,1000Hz (1)

[0062] where coef f is the compensation coefficient when the frequency is fHz, A fis the peak value of vibration acceleration at frequency fHz, A0 f is the peak acceleration value of the vibration source at a frequency of fHz. f Comparison table, draw the frequency-compensation coefficient curve as shown Figure 3 When using, you can look up the table to get the corresponding compensation coefficient by the frequency of the signal. For GZZ15 / 1500, due to limited storage space, a polynomial fitting curve is used to establish f-coef f Corresponding relationship:

[0063] coef f =c n f n +c n-1 f n-1 +…+c1f+c0 (2)

[0064] In order to improve the compensation accuracy, the frequency-compensation coefficient curve is segmented fitted using formula (2), as shown in the following example: Figure 4 The effect shown is:

[0065]

[0066] Where c1 to c7 are the fitting curve coefficients, and f1 to f4 are the frequency segment ranges. The compensation coefficients are calculated using the fitting curve coefficients and the signal frequency. Different vibration transmission paths in different product designs can lead to differences in the frequency-compensation coefficient curve. Depending on the required compensation accuracy, the fitting curve will have different segmented representations.

[0067] Figure 2 The Zhong202 is a vibration measurement product that uses an adaptive sampling rate to acquire the measured vibration signal to obtain its acceleration sequence, which is then compensated according to a compensation model. During sampling, the measured vibration signal is first acquired at the highest sampling frequency and spectral analysis is performed to determine the approximate frequency range of the measured signal. The appropriate sampling frequency is then calculated and resampled at that frequency to improve sampling accuracy. Before compensation, the sampled acceleration sequence is spectrally analyzed to determine the signal frequency or frequency distribution. Based on this frequency, the compensation model is used to calculate the compensation coefficient for acceleration compensation calculation, effectively counteracting the attenuation of the vibration signal caused by conduction.

[0068] In this embodiment, the sampling rates that can be set for GZZ15 / 1500 are 6666Hz, 3333Hz, and 1666Hz. The signal set by the vibration table is used as the vibration source. GZZ15 / 1500 collects the measured signal at the highest sampling rate of 6666Hz to obtain the acceleration sequence A0. The number of sampling points is n, and n is an even number. A discrete Fourier transform or FFT is performed on A0 to obtain the spectrum signal of the measured signal:

[0069]

[0070] Where W n =e (-2πi) / n is one of the roots of unity of n. 0 to 6666 Hz is divided into points to obtain F(k), which is combined with Y(k). The F(k) at the maximum value of Y(k) is taken as the main frequency of the measured signal:

[0071]

[0072] where f signal is the main frequency of the measured signal. The sampling rate is selected using the following formula:

[0073]

[0074] Where F is the settable sampling rate, ρ = 20 is the sampling magnification, and Fs is the selected frequency for the next sampling. Use Fs to sample the measured signal to obtain a new acceleration sequence A1, and then calculate the main frequency of the measured signal using formulas (4) and (5). The signal frequency f calculated this time is signal Theoretically, the accuracy is higher than that of the initial sampling calculation. signal Substitute into formula (3) to calculate the compensation coefficient coef. Then compensate the acceleration sequence A1 to obtain A2:

[0075] A2(k)=A1(k)÷coef (7)

[0076] Figure 2 The peak acceleration is calculated in 203 after adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets. High-pass filtering is implemented using a software algorithm, adaptively selecting filter parameters such as cutoff frequency and bandwidth based on the sampling frequency to optimize filtering performance. Because the vibration signal is periodic, any unstable data that may appear after high-pass filtering should be discarded. Removing non-zero baseline offsets keeps the mean of the periodic vibration signal close to zero. Peak acceleration is calculated by statistically analyzing the maximum and minimum values ​​of the acceleration sequence.

[0077] In this embodiment, a FIR high-pass filter is selected for filtering, and the ideal amplitude-frequency response curve is as follows: Figure 5 As shown, Flp and Fls are the cutoff frequencies of the filter. The signal in the frequency range less than Fls is cut off, the signal in the frequency range from Fls to Flp is attenuated, and the signal in the frequency range greater than Flp remains unchanged. Fls and Flp need to be calculated through the sampling frequency Fs:

[0078]

[0079] Where r1 is the conversion coefficient of the cutoff frequency Flp, and r2 is the conversion coefficient of the cutoff frequency Fls. Under normal circumstances, r1 = 200 (when Fs = 6666Hz, r1 = 45), and r2 = 1.1. The order N of the high-pass filter is related to the value of the cutoff frequency, and after simplification, it is:

[0080]

[0081] Among them, b0 is the attenuation coefficient of the amplitude-frequency response signal attenuation interval, N0 is the initial order. If N0 is even, then N is equal to N0. If N0 is odd, then N = N0 + 1. ceil(.) rounds up, and floor(.) rounds down.

[0082] The filter coefficients [b, a] are approximated by the least squares method using the parameters of formulas (8) and (9), and the impulse response is smoothed using a window. The rational transfer function of the filter is usually expressed as:

[0083]

[0084] where n a is the feedback filter order, n b is the order of the feedforward filter, X(z) is the z-domain transform of the sequence before filtering, and Y(z) is the z-domain transform of the sequence after filtering. Substituting the filter coefficients [b, a] of this embodiment into formula (10), the equivalent time domain difference equation is:

[0085]

[0086] Where A2 is the acceleration sequence before filtering, A3 is the acceleration sequence after filtering, b is a 1×N array, and a is a 1×1 array. Discarding unstable filtered data:

[0087] A4(k)=A3(k+len),len <N (12)

[0088] Where len is the length of the discarded sequence. Then remove the non-zero baseline offset:

[0089]

[0090] in is the acceleration mean. A5 is the processed acceleration sequence with a length of n1, based on which the acceleration peak value is calculated. According to the signal frequency f signal And the sampling rate Fs estimates the number of sampling points N in one signal cycle s :

[0091] N s =Fs÷f signal (14)

[0092] Press N in A5 s The length is truncated to each subsequence A i , take its maximum and minimum values:

[0093]

[0094] Where m is the number of subsequences that have been truncated. max With A min Sort by large to small and small to large respectively, remove the first 3 values, and take the remaining maximum value a max and a min Calculate the peak acceleration:

[0095] a peak =(a max -a min ) / 2 (16)

[0096] Figure 2 In Figure 204, the acceleration sequence is integrated and detrended in the time domain to obtain a velocity sequence. After adaptive high-pass filtering, unstable data is discarded, and non-zero baseline offsets are removed, the effective velocity value is calculated. The time domain integration uses the trapezoidal or Simpson formula to minimize integration errors. The trend caused by the integration is then removed by removing the best straight-line fit. The RMS value of the processed velocity sequence is then calculated.

[0097] In this embodiment, the time domain integration adopts the trapezoidal formula:

[0098]

[0099] Where Δt is the sampling interval, and V1(k) is the vibration velocity sequence. Perform a linear fit on V1(k) as y=c1x+c0, and then subtract the fitted straight line:

[0100] V2(k)=V2(k)-(c1k+c0) (18)

[0101] Consistent with the acceleration processing, V2(k) is processed by adaptive high-pass filtering, unstable data is discarded, and non-zero baseline offset is removed to obtain V3(k). Calculate the effective value of velocity:

[0102]

[0103] Figure 2In Figure 205, the velocity series is integrated and detrended in the time domain to obtain a displacement series. After adaptive high-pass filtering, unstable data is discarded, and non-zero baseline offsets are removed, the peak-to-peak displacement is calculated. The peak-to-peak displacement is calculated by counting the maximum and minimum values ​​of the displacement series, minimizing the impact of the accumulated error after two integrations on the peak-to-peak displacement calculation.

[0104] In this embodiment, the processing of displacement is consistent with that of velocity. V2(k) is integrated in the time domain and detrended to obtain S1(k). Then, after adaptive high-pass filtering, discarding unstable data and removing non-zero baseline offsets, the vibration displacement sequence S2(k) is obtained. Then, the displacement peak value s is obtained according to the method of formula (14), (15), and (16). peak , the peak-to-peak displacement is:

[0105] s p-p =s peak ×2 (20)

[0106] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A vibration measurement method based on acceleration sensor compensation, characterized in that: The following steps are involved: Step 1: Use a vibration measurement product equipped with a MEMS accelerometer to collect vibration acceleration at each frequency point within a preset frequency measurement range, compare it with the preset vibration acceleration value, determine the attenuation coefficient at each frequency point, and establish a compensation model for this vibration measurement product; The step 1 is specifically as follows: According to the frequency measurement range of the vibration measurement product equipped with MEMS accelerometer, points are evenly divided and the acceleration at each frequency point is collected to establish the frequency-compensation coefficient curve: coef f =A f ÷A0 f where coef f is the compensation coefficient when the frequency is fHz, A f is the peak value of vibration acceleration at frequency fHz, A0 f is the peak acceleration value set by the vibration source at a frequency of fHz; Establish polynomial fitting model and piecewise polynomial fitting model based on frequency-compensation coefficient curve: coef f =c n f n +c n-1 f n-1 +…+c1f+c0 Among them, c1 to c7 are fitting curve coefficients, and f1 to f4 are frequency segment ranges; Step 2: The vibration measuring product collects the measured vibration signal at an adaptive sampling rate to obtain its acceleration sequence, and compensates the acceleration sequence according to the compensation model; Step 3: Calculate the peak acceleration after adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets; Step 4: Perform time domain integration and detrending processing on the acceleration sequence to obtain the velocity sequence. After adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets, the effective value of the velocity is calculated. Step 5: Perform time domain integration and detrending processing on the velocity series to obtain the displacement series. After adaptive high-pass filtering, discarding unstable data, and removing non-zero baseline offsets, the peak-to-peak value of the displacement is calculated.

2. The vibration measurement method based on acceleration sensor compensation according to claim 1, characterized in that: The step 2 is specifically as follows: When sampling, first collect the measured vibration signal at the highest sampling frequency and perform spectrum analysis to obtain the approximate frequency range of the measured signal: Where Y(k) is the result of the acceleration sequence after frequency domain transformation, n is the number of sampling points, and F(k) when Y(k) is the maximum value is the main frequency f of the measured signal signal ; Calculate the appropriate sampling frequency and resample at this frequency to improve sampling accuracy. The appropriate sampling frequency is as follows: Where F is the configurable sampling rate of the MEMS accelerometer, ρ is the sampling magnification, and Fs is the selected frequency of the next sampling; Use Fs to sample the measured signal to obtain a new acceleration sequence A1, and then compensate the acceleration sequence A1 to obtain the acceleration sequence A2: A2(k)=A1(k)÷coef.

3. The vibration measurement method based on acceleration sensor compensation according to claim 2, characterized in that: In step 3, step 4 and step 5, the adaptive high-pass filtering process is specifically as follows: According to the sampling rate Fs selected by adaptive sampling, the filter cutoff frequencies Fls and Flp are calculated. The signal in the frequency range less than Fls is cut off, the signal in the frequency range from Fls to Flp is attenuated, and the signal in the frequency range greater than Flp remains unchanged: Among them, r1 is the conversion coefficient of the cutoff frequency Flp, and r2 is the conversion coefficient of the cutoff frequency Fls; Calculate the order N of the high-pass filter: Where b0 is the attenuation coefficient of the amplitude-frequency response signal attenuation interval, N0 is the initial order, if N0 is an even number, N and N0 are equal, if N0 is an odd number, then N = N0 + 1; ceil(.) means rounding up, and floor(.) means rounding down; Then, the filter coefficients [b, a] are approximately calculated using the least squares method, and the impulse response is smoothed using a window to complete the design of a high-pass filter with [b, a] as the coefficients.

4. The vibration measurement method based on acceleration sensor compensation according to claim 3, characterized in that: In steps 4 and 5, the time domain integration uses the trapezoidal formula or Simpson's formula to reduce the integration error. Then, the trend caused by the integration is removed by removing the best straight-line fitting line. The root mean square value of the processed velocity series or displacement series is calculated to obtain the effective value of the velocity or displacement.

5. The vibration measurement method based on acceleration sensor compensation according to claim 4, characterized in that: In steps 3 and 5, the peak acceleration and peak-to-peak displacement values ​​are calculated as follows: According to the main frequency f of the signal signal And the sampling rate Fs estimates the number of sampling points N in one signal cycle s : N s =Fs÷f signal Press N in sequence A s The length is truncated to each subsequence A i , count the maximum and minimum values ​​of the subsequence: Where m is the number of subsequences that have been intercepted, max With A min Sort by large to small and small to large respectively, remove the first 3 values, and take the remaining maximum value a max and a min Calculate the peak value a peak =(a max -a min ) / 2, or peak-to-peak s p-p =s max -s min ; Among them, s max and s min are the maximum and minimum values ​​of the displacement sequence respectively.

Images