An overload vibration data separation method based on a MEMS vibration sensor

Abstract

The application provides an overload vibration data separation method based on a MEMS vibration sensor, and the method achieves the purpose of simultaneously satisfying the precision of low-frequency overload signals and medium-high-frequency vibration signals by establishing a comprehensive filter evaluation index of filtered low-frequency overload signals and medium-high-frequency vibration signals, reasonably selecting filter parameters and obtaining an optimal filter function.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of environmental adaptability, and particularly relates to an overload vibration data separation method of a MEMS vibration sensor. BACKGROUND

[0002] Traditional aircraft respectively use different types of sensors to achieve the purpose of overload and vibration measurement. Overload sensors generally use charge sensors, and the measurement frequency requirement is as low as 0Hz. Vibration sensors generally use ICP (Integrated Circuits Piezoelectric) sensors, and the measurement minimum low frequency is greater than 0Hz. Overload sensors are generally large in size and have high requirements for installation space. Traditional aircraft simultaneously install overload sensors and vibration sensors to achieve the purpose of simultaneously measuring the overload and vibration of the aircraft.

[0003] With the development of MEMS (Micro Electro Mechanical System) vibration sensors, their application on aircraft has increased. In addition to being smaller in size and saving installation space, MEMS vibration sensors have a measurement frequency as low as 0Hz, and have the functions of overload and vibration measurement. However, MEMS vibration sensors also cause certain difficulties in data analysis. How to separate the mixed overload signal and medium-high frequency vibration signal in the MEMS vibration sensor has become a difficulty in MEMS vibration data analysis.

[0004] Traditional Butterworth filtering can ensure the authenticity of signals within the bandpass range, but may bring large errors to signals outside the bandpass range. For aircraft applications, it is necessary to simultaneously ensure the authenticity of overload signals and medium-high frequency vibration signals. The application of traditional Butterworth filtering method cannot meet the actual application requirements. SUMMARY

[0005] The present application aims to at least solve one of the technical problems existing in the prior art.

[0006] To this end, the present application provides an overload vibration data separation method based on a MEMS vibration sensor.

[0007] The technical solution of the present application is as follows:

[0008] According to an aspect, an overload vibration data separation method based on a MEMS vibration sensor is provided, which comprises:

[0009] Step one, using a MEMS vibration sensor, obtaining original vibration data x0(j) of a target to be measured, j=1, 2, 3, … r*fs; wherein fs is the sampling rate of the original vibration data, r is the sampling time, and r*fs is the data length of the x0 array.

[0010] Step two, let the order of SG filtering d = 1;

[0011] Step three, get the SG filtering window w(i), where w(i) is an odd number;

[0012] Step four, for each filtering window w(i), i = 1, 2,..., repeat the following steps five to step eight;

[0013] Step five, using the filtering order and filtering window size w(i) obtained in steps two and three as filtering parameters, SG filtering is performed on the original vibration data x0 to obtain the filtered low-frequency overload signal:

[0014] x1(j, i), j = 1, 2, 3,..., r*fs, i represents the i-th window;

[0015] Step six, based on the low-frequency overload signal x1(j, i) obtained in step five, calculate the medium-high frequency vibration signal x2(j, i);

[0016] Step seven, respectively, power spectral density analysis is performed on the low-frequency overload signal x1 and the medium-high frequency vibration signal x2, as well as the original vibration signal x0, to obtain the corresponding power spectral density: p1(k, i), p2(k, i), and p0(k), k = 0, 1, 2,..., N / 2; wherein N is the data length required for power spectral density analysis;

[0017] Step eight, taking 1Hz as the frequency demarcation point of the filtering precision evaluation of the low-frequency overload signal and the medium-high frequency vibration signal, calculate the filtering evaluation function v(d, i), as shown in the following formula:

[0018]

[0019] Wherein, when k = 1,..., m, it corresponds to the power spectral density below 1Hz in the f0 array;

[0020] When k = m+1,..., N / 2+1, it corresponds to the power spectral density above 1Hz in the f0 array; d is the filtering order, and i represents the i-th window;

[0021] Step nine, find the optimal window size w opt and the minimum filtering evaluation function value v min (d, w opt );

[0022] Wherein, the minimum value of the filtering evaluation function is the minimum filtering evaluation function value, and the corresponding window size at this time is the optimal window size w opt ;

[0023] Step ten, increase the filter order by 1 as an interval value, and determine whether the filter order d is greater than 2, if yes, go to step eleven, if no, go to step four;

[0024] Step eleven, calculate the difference Δ of the adjacent two minimum filter evaluation functions, and the calculation formula is as follows:

[0025]

[0026] In the formula, w opt is the optimal filter window size corresponding to the order d-1, w' opt is the optimal filter window size corresponding to the order d-2;

[0027] Step twelve, determine whether the difference Δ of the minimum filter evaluation function is less than a preset threshold, if no, go to step four, if yes, the order d-1 is the optimal order d opt .

[0028] Step thirteen, take the optimal order d opt and the optimal window w opt corresponding thereto as the filter parameters, and perform SG filtering on the original signal x0 to obtain a low-frequency overload signal x1 opt .

[0029] Step fourteen, obtain a medium-high frequency vibration signal x2 opt according to the original signal x0 and the low-frequency overload signal x1 opt .

[0030] Further, the SG filter window w(i) is obtained in the following manner:

[0031] 1) Select a series of window sizes The selection principle is as follows:

[0032] Determine the starting window size: if 0.01*fs is even after taking the integer, directly take it as the starting window size, if 0.01*fs is odd after taking the integer, take its adjacent even number;

[0033] Determine the step window size dw: same as the starting window size;

[0034] Determine the termination window size: [(fs / dw) taking the integer +1]*dw.

[0035] 2) Obtain the SG filter window

[0036] Further, in step six, the medium-high frequency vibration signal is calculated in the following formula:

[0037] x2(j,i)=x0(j)-x1(j,i).

[0038] Further, the low frequency overload signal x1 is obtained from the original signal x0 by the following formula opt Obtaining the mid-high frequency vibration signal x2 opt :

[0039] x2 opt = x0-x1 opt .

[0040] Further, N is 2 n , n is a natural number, and N is not less than 2*fs.

[0041] Further, the preset threshold is 5%.

[0042] According to another aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the above-mentioned separation method when executing the computer program.

[0043] The above technical solution establishes a comprehensive filter evaluation index of the filtered low frequency overload signal and the mid-high frequency vibration signal, reasonably selects filter parameters, obtains an optimal filter function, and achieves the purpose of simultaneously satisfying the accuracy of the low frequency overload signal and the mid-high frequency vibration signal. The technical solution of the present application can at least achieve the following beneficial effects:

[0044] (1) effectively separating the overload information and vibration information in the MEMS vibration sensor measurement signal;

[0045] (2) proposing a filter evaluation function that takes into account the accuracy of the overload information and the accuracy of the vibration information;

[0046] (3) by reasonably selecting filter parameters, the overload information and vibration information in the original signal are maximally retained. BRIEF DESCRIPTION OF DRAWINGS

[0047] The accompanying drawings, which are included to provide a further understanding of the embodiments of the application and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0048] Figure 1 The measurement result of the MEMS vibration sensor is shown;

[0049] Figure 2 The measurement result of the traditional ICP sensor is shown;

[0050] Figure 3An overload vibration data separation flowchart for a MEMS vibration sensor is shown;

[0051] Figure 4 A low frequency power spectral density error versus window size is shown below 1 Hz (d = 1);

[0052] Figure 5 A power spectral density plot after filtering with different windows is shown below 1 Hz (d = 1);

[0053] Figure 6 A mid-high frequency power spectral density error versus window size is shown above 1 Hz (d = 1);

[0054] Figure 7 A power spectral density plot after filtering with different windows is shown above 1 Hz (d = 1);

[0055] Figure 8 A filter evaluation function versus window size is shown (d = 1);

[0056] Figure 9 A low frequency power spectral density error versus window size is shown below 1 Hz (d = 2);

[0057] Figure 10 A mid-high frequency power spectral density error versus window size is shown above 1 Hz (d = 2);

[0058] Figure 11 A filter evaluation function versus window size is shown (d = 2);

[0059] Figure 12 A low frequency power spectral density error versus window size is shown below 1 Hz (d = 3);

[0060] Figure 13 A mid-high frequency power spectral density error versus window size is shown above 1 Hz (d = 3);

[0061] Figure 14 A filter evaluation function versus window size is shown (d = 3);

[0062] Figure 15 A low frequency overload signal is shown versus the original signal;

[0063] Figure 16 A power spectral density of a low frequency overload signal versus the original signal is shown;

[0064] Figure 17 A mid-high frequency vibration signal is shown;

[0065] Figure 18 A power spectral density of a mid-high frequency vibration signal versus the original signal is shown. DETAILED DESCRIPTION

[0066] It should be noted that the embodiments and features of the embodiments in the present application can be combined with each other in the case of no conflict. The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The description of the at least one example embodiment is actually only illustrative, but not as any limitation on the present application and its application or use. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0067] It should be noted that the terms used herein are only intended to describe specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form, unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, there is a reference to the presence of a feature, step, operation, device, component and / or combinations thereof.

[0068] Unless specifically stated otherwise, the relative arrangement of components and steps, numerical expressions, and numerical values set forth in the various embodiments described herein are not limiting. It should be understood that the various parts shown in the drawings are not necessarily drawn to scale. Techniques, methods, and devices known to those of ordinary skill in the relevant art can not be discussed in detail, but should be considered as part of the description of the present application. In all examples shown and discussed herein, any specific value should be interpreted as merely an example, and not as a limitation. Therefore, other examples of the exemplary embodiments can have different values. It should be noted that similar reference numbers and letters represent similar items in the following drawings, and therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0069] As shown in FIG. 1, in one embodiment of the present application, a MEMS vibration sensor-based overload vibration data separation method is provided, which comprises: Figure 3

[0070] Step one, using a MEMS vibration sensor, obtaining original vibration data x0(j) of a target to be measured, j = 1, 2, 3, … r*fs; wherein fs is the sampling rate of the original vibration data, r is the sampling time, and r*fs is the data length of the x0 array;

[0071] Step two, setting the order d of the SG filter to 1;

[0072] ​Step three, obtain the SG filter window w(i), wherein w(i) is an odd number;

[0073] In this embodiment, a series of window sizes can be selected The selection principle is as follows:

[0074] Determination of the starting window size: take the even number after rounding 0.01*fs; if 0.01*fs is rounded to an odd number, take the adjacent even number (either +1 or -1); if 0.01*fs is rounded to an even number, the even number taking step can be omitted;

[0075] Determination of the step window size dw: same as the starting window size;

[0076] Determination of the ending window size: [(fs / dw) rounded +1]*dw;

[0077] Further, the SG filter window w(i) is obtained

[0078] Step four, for each filter window w(i), i = 1, 2,..., repeat steps five to eight as follows:

[0079] Step five, using the filter order and filter window size w(i) obtained in steps two and three as filter parameters, the original vibration data x0 is subjected to SG filtering to obtain the filtered low-frequency overload signal:

[0080] x1(j, i), j = 1, 2, 3,..., r*fs, i represents the i-th window;

[0081] In this embodiment, the Matlab calculation method is as follows:

[0082] x1 = sgolayfit(x0, d, w(i));

[0083] Step six, based on the low-frequency overload signal x1(j, i) obtained in step five, the medium-high frequency vibration signal x2(j, i) is calculated;

[0084] In this embodiment, the calculation formula of x2(j, i) is as follows:

[0085] x2(j, i) = x0(j) - x1(j, i);

[0086] Step seven, the power spectral density analysis is performed on the low-frequency overload signal x1 and the medium-high frequency vibration signal x2, as well as the original vibration signal x0, to obtain the corresponding power spectral densities: p1(k, i), p2(k, i), and p0(k), k = 0, 1, 2,..., N / 2; wherein N is the data length required for power spectral density analysis;

[0087] In this embodiment, preferably, the size of N is 2 n , n is a natural number, and N is large enough (not less than 2*fs) to obtain the power spectrum of 0-1Hz low frequency overload;

[0088] In this embodiment, the power spectrum density analysis can use the pwelch function in matlab:

[0089] [p0,f0]=pwelch(x0,hanning(N),[],N,fs);

[0090] [p1(:,i),f0]=pwelch(x1(:,i),hanning(N),[],N,fs);

[0091] [p2(:,i),f0]=pwelch(x2(:,i),hanning(N),[],N,fs);

[0092] Wherein, p0, p1(:,i) and p2(:,i) are the power spectrum densities of signals x0, x1(:,i) and x2(:,i) respectively, f0 is the frequency corresponding to the power spectrum density, the unit is Hz, f0 array is {0, fs / N, 2*fs / N, …, N / 2*fs / N}, the array length is N / 2+1. x1(:,i), x2(:,i), p1(:,i) and p2(:,i) represent the whole elements of the i-th column of arrays x1, x2, p1 and p2 respectively;

[0093] Step eight, take 1Hz as the frequency demarcation point of low frequency overload signal and high frequency vibration signal filtering precision evaluation, calculate the filtering evaluation function v(d,i), as shown in the following formula:

[0094]

[0095] In the formula, when k=1,…,m, it corresponds to the power spectrum density below 1Hz in the f0 array;

[0096] When k=m+1,…,N / 2+1, it corresponds to the power spectrum density above 1Hz in the f0 array; d is the filter order, i represents the i-th window;

[0097] That is, the meaning of the filtering evaluation function of this embodiment: the power spectrum density error of the low frequency overload signal below 1Hz + the power spectrum density error of the high frequency vibration signal above 1Hz;

[0098] Because the MEMS vibration sensor needs to consider the accuracy of the low frequency part and the accuracy of the high frequency part at the same time, both parts must be considered at the same time;

[0099] Step nine, finding the optimal window size w opt and the minimum filter evaluation function value v min (d,w opt );

[0100] Wherein, the filter evaluation function takes the minimum value as the minimum filter evaluation function value, and the corresponding window size at this time is the optimal window size w opt ;

[0101] Step ten, increase the filter order by 1 as the interval value, and judge whether the filter order d is greater than 2, if yes, go to step eleven, if not, go to step four;

[0102] Step eleven, calculate the difference Δ of the adjacent two minimum filter evaluation functions, and the calculation formula is as follows:

[0103]

[0104] In the formula, w opt is the optimal filter window size corresponding to the order d-1, w' opt is the optimal filter window size corresponding to the order d-2;

[0105] Step twelve, judge whether the difference Δ of the minimum filter evaluation functions is less than the preset threshold value; if not, go to step four; if yes, the order d-1 is the optimal order d opt ;

[0106] Preferably, the preset threshold value is designed as 5%;

[0107] Step thirteen, using the optimal order d opt and the optimal window w opt corresponding thereto as the filter parameters, the SG filter is performed on the original signal x0 to obtain the low frequency overload signal x1 opt ;

[0108] Step fourteen, according to the original signal x0 and the low frequency overload signal x1 opt , the medium-high frequency vibration signal x2 opt is obtained;

[0109] In this embodiment, x2 opt =x0-x1 opt .

[0110] It can be seen that the key points of the embodiment scheme of the present application at least include:

[0111] (1) The embodiment of the present application is based on the SG filter algorithm, and according to the application purpose, the optimal data window size and the optimal fitting order are obtained by setting the filter evaluation function, so that the low frequency overload signal and the medium-high frequency vibration signal in the original MEMS vibration signal are effectively separated;

[0112] (2) The embodiment of the present application proposes a comprehensive filtering evaluation index for low-frequency overload signals and medium-high frequency vibration signals after filtering, taking into account the accuracy of the signals;

[0113] (3) The embodiment of the present application proposes that the frequency demarcation point for filtering accuracy evaluation of low-frequency overload signals and medium-high frequency vibration signals is 1Hz;

[0114] (4) The embodiment of the present application proposes a method of using the sum of absolute values of the difference between the power spectral densities of signals before and after filtering as a measure of the accuracy of the signals after filtering;

[0115] (5) The embodiment of the present application proposes a method of selecting the window size of the SG filter, i.e., a method of selecting the starting window, the step window, and the termination window.

[0116] That is, the embodiment of the present application establishes a comprehensive filtering evaluation index for low-frequency overload signals and medium-high frequency vibration signals after filtering, reasonably selects filtering parameters, obtains an optimal filtering function, and achieves the purpose of simultaneously satisfying the accuracy of low-frequency overload signals and medium-high frequency vibration signals. The technical solution of the present application can at least achieve the following beneficial effects:

[0117] (1) Effectively separates overload information and vibration information in the measurement signal of the MEMS vibration sensor;

[0118] (2) Proposes a filtering evaluation function that takes into account the accuracy of overload information and vibration information;

[0119] (3) By reasonably selecting filtering parameters, the overload information and vibration information in the original signal are maximally preserved.

[0120] According to another embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the above-mentioned separation method when executing the computer program.

[0121] The method of the present application is described in detail below with reference to a specific embodiment.

[0122] Taking the measurement results of a certain aircraft MEMS vibration sensor as an example, the process of separating overload signals and vibration signals is shown:

[0123] First, use the MEMS vibration sensor to obtain the original vibration data x0(j) of the target to be measured, j=1, 2, 3, … r*fs;

[0124] In the formula, fs is the sampling rate of the original vibration data, r is the sampling time, and r*fs is the data length of the x0 array;

[0125] In this embodiment, the obtained MEMS vibration signal is as shown in the figure Figure 1 As a comparison, Figure 2 The measurement result of the traditional ICP vibration sensor shows that the MEMS vibration sensor mixes low-frequency overload components, Figure 1 The signal is x0, the sampling time is 95 s, and the sampling frequency fs is 5120;

[0126] Second step, let the order d of SG filtering be 1;

[0127] Third step, select a series of window sizes The selection principle is as follows:

[0128] Determination of the starting window size: take the even number after 0.01*fs is rounded, if 0.01*fs is rounded to an odd number, take the adjacent even number (odd number+1 or-1); if 0.01*fs is rounded to an even number, the even number taking step can be omitted. In this embodiment, 0.01*fs is 51.2, which is rounded to 51. Under the even number taking principle, 50 can be taken, or 52 can be taken. In this embodiment, 50 is taken;

[0129] Determination of the step window size dw: same as the starting window size. In this embodiment, 50 is taken;

[0130] Determination of the termination window size: [(fs / dw) is rounded+1]*dw, in this embodiment, fs is 5120, fs / dw=102.4, which is rounded to 102, and after adding 1, it is 103, 103*50=5150;

[0131] Fourth step, get the SG filtering window

[0132] In this embodiment, the w(i) sequence is {51, 101, 151, …, 5151};

[0133] Fifth step, for each filtering window w(i), i=1, 2, …, repeat the following steps sixth to ninth steps;

[0134] Sixth step, take the order d and the window size w(i) as the filtering parameters, and perform SG filtering on the original vibration data x0 to obtain the filtered low-frequency overload signal x1. x1 is a matrix, the number of columns is the number of windows, and the number of rows is consistent with the length of x0. The Matlab calculation method is as follows:

[0135] x1=sgolayfit(x0,d,w(i))

[0136] Seventh step, based on the low-frequency overload signal x1(j,i) obtained in the sixth step, calculate the medium-high frequency vibration signal x2(j,i), and the calculation formula is as follows:

[0137] x2(j,i) = x0(j) - x1(j,i);

[0138] Step 8, power spectral density analysis is performed on the original vibration signal x0, the low frequency overload signal x1 and the medium-high frequency vibration signal x2, respectively, to obtain the corresponding power spectral density:

[0139] p0(k), p1(k,i), p2(k,i), k = 0, 1, 2, … N / 2;

[0140] wherein N is the data length required for power spectral density analysis, and the size thereof is required to be 2 n n is a natural number. Note that N should be large enough (not less than 2*fs) to obtain the power spectrum of the low frequency overload of 0-1 Hz;

[0141] The power spectral density analysis can use the pwelch function in matlab.

[0142] [p0,f0] = pwelch(x0,hanning(N),[],N,fs);

[0143] [p1(:,i),f0] = pwelch(x1(:,i),hanning(N),[],N,fs);

[0144] [p2(:,i),f0] = pwelch(x2(:,i),hanning(N),[],N,fs);

[0145] wherein p0, p1(:,i) and p2(:,i) are the power spectral densities obtained by signals x0, x1(:,i) and x2(:,i), respectively, f0 is the frequency corresponding to the power spectral density, with the unit of Hz, and the f0 array is {0, fs / N, 2*fs / N, …, N / 2*fs / N}, with the array length being N / 2+1. x1(:,i), x2(:,i), p1(:,i) and p2(:,i) represent the entire elements of the i-th column of the arrays x1, x2, p1 and p2, respectively;

[0146] In this embodiment, fs = 5120 Hz, and N = 2 14 The f0 array has the unit of Hz, and the specific values are {0, 0.3125, 0.6250, 0.9375, 1.250, …, 2560}, and the frequencies of 0-1 Hz in the f0 array are 0 Hz, 0.3125 Hz, 0.6250 Hz and 0.9375 Hz;

[0147] Step 9, taking 1 Hz as the frequency demarcation point for the filtering precision evaluation of the low frequency overload signal and the medium-high frequency vibration signal, the filtering evaluation function v(d,i) is calculated. The calculation formula is as follows:

[0148]

[0149] In this embodiment, m = 4, there are 4 frequencies below 1 Hz;

[0150] The first part on the right side of equation (1) represents the power spectrum density error of the part below 1 Hz. When it is plotted as the vertical axis and the window size, the larger the window, the greater the error below 1 Hz, so the smaller the window is better. See Figure 4 and Figure 5 ;

[0151] The second part on the right side of equation (1) represents the power spectrum density error of the part above 1 Hz. When it is plotted as the vertical axis and the window size, the smaller the window, the greater the error above 1 Hz, so the larger the window is better.

[0152] See Figure 6 and Figure 7 ;

[0153] Step 10, find the optimal window size w opt and the minimum filter evaluation function value v min (d, w opt ). The window size corresponding to the minimum value of the filter evaluation function is the optimal window size w opt , and the corresponding filter evaluation function value is the minimum filter evaluation function value v min (d, w opt );

[0154] In this embodiment, according to equation (1), the relationship between the filter evaluation function v(d, i) and the window size is as follows Figure 8 As can be seen from the figure, when the filter order is 1, the optimal window size w opt is 500, and the corresponding minimum filter evaluation function v min (1, 500) is 0.03622;

[0155] Step 11, increase the filter order, the filter order is increased by 1;

[0156] Step 12, determine whether the filter order is greater than 2;

[0157] Step 13, the filter order is less than or equal to 2, return to step 5-12, until the filter order is greater than 2;

[0158] In this embodiment, when the filter order is 2, the condition is not met, and it needs to return to step 5-12; the related image is as follows Figures 9-11 : when the filter order d = 2, the optimal filter window is 2950, and the minimum filter evaluation function is 0.01223;

[0159] Step 14: If the filter order is greater than 2, calculate the difference Δ between two adjacent minimum filter evaluation functions. The calculation formula is as follows:

[0160]

[0161] In the formula, w opt w' is the optimal filter window size corresponding to an order of d-1. opt It is the optimal filter window size corresponding to the order d-2;

[0162] In this embodiment, when the filter order d = 3, the difference between two adjacent minimum filter evaluation functions is:

[0163]

[0164] Step 15: Determine whether the difference Δ of the minimum filter evaluation function is less than 5%;

[0165] Step 16: If not satisfied, return to steps 5 through 15 until the difference of the minimum filter evaluation function is within 5%.

[0166] like Figures 12-14 As shown, in this embodiment, when the filter order d = 3, the requirement that the difference between the minimum filter evaluation functions is less than 5% is not met. Therefore, steps five through fifteen are returned to calculate the optimal window size and the minimum filter evaluation function when the filter order d = 3. The calculation results are basically consistent with those when the filter order d = 2. The optimal window size is 2950, ​​and the minimum filter evaluation function is 0.01223. After completing the above calculations, d = 4, and the difference Δ between adjacent minimum filter evaluation functions is calculated.

[0167]

[0168] Step 17: The difference Δ between the minimum filter evaluation functions satisfies the requirement of being less than 5%. At this point, the corresponding order d-1 is the optimal order d. opt ;

[0169] In this embodiment, when d=4, the difference Δ of the minimum filter evaluation function meets the requirement of being less than 5%, and increasing the filter order can no longer improve the filter accuracy. The optimal filter order is d-1, that is, the optimal filter order is 3.

[0170] Step 18, using the optimal order d opt and its corresponding optimal window w opt Using the filtering parameters, the original signal x0 is subjected to SG filtering to obtain the low-frequency overload signal x1. opt ;

[0171] In this embodiment, the original signal x0 is subjected to SG filtering using the optimal order 3 and its corresponding optimal window 2950 as filtering parameters to obtain the low-frequency overload signal x1.opt See Figure 15 , compared with the power spectrum density of the original signal, see Figure 16 ;

[0172] Nineteenth step, get the mid-high frequency vibration signal x2 opt = x0-x1 opt ;

[0173] In this embodiment, the obtained mid-high frequency overload signal x2 opt See Figure 17 , compared with the power spectrum density of the original signal, see Figure 18 .

[0174] Features described and / or illustrated above in relation to one embodiment can be used in one or more other embodiments in the same or similar way, and / or in combination with or in place of features in other embodiments.

[0175] It should be emphasized that the term "comprises / comprising" when used in this text, refers to the presence of the stated features, integers, steps or components but does not preclude the presence or addition of one or more other features, integers, steps, components or groups thereof.

[0176] The above method of the present application can be realized by hardware, or by hardware in combination with software. The present application relates to a computer readable program which, when executed by a logic component, enables the logic component to realize the above-described apparatus or constituent components, or to realize the above-described various methods or steps. The present application also relates to a storage medium for storing the above program, such as a hard disk, a magnetic disk, an optical disk, a DVD, a flash memory, etc.

[0177] Many of the features and advantages of the embodiments are apparent from the detailed description, and thus, it is intended that the appended claims cover all such features and advantages of the embodiments within their true spirit and scope. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the embodiments of the present application to the exact construction and operation described and illustrated, and accordingly, all suitable modifications and equivalents should be included within the scope of the embodiments of the present application.

[0178] The parts of the present application not described in detail are known to those skilled in the art.

Claims

1. A method for separating overload vibration data based on MEMS vibration sensors, characterized in that, The separation method includes: Step 1: Use a MEMS vibration sensor to acquire the raw vibration data x0(j) of the target under test, j = 1, 2, 3, ... r*fs; where fs is the sampling rate of the raw vibration data, r is the sampling time, and r*fs is the data length of the x0 array; Step 2: Set the order of the SG filter, d = 1; Step 3: Obtain the SG filter window w(i), where w(i) is an odd number; Step 4: For each filter window w(i), i = 1, 2, ..., repeat steps 5 to 8. Step 5: Using the filter order and filter window size w(i) obtained in Steps 2 and 3 respectively as filter parameters, perform SG filtering on the original vibration data x0 to obtain the filtered low-frequency overload signal: x1(j,i),j=1,2,3,…r*fs,i represents the i-th window; Step 6: Based on the low-frequency overload signal x1(j,i) obtained in Step 5, calculate the medium- and high-frequency vibration signal x2(j,i); Step 7: Perform power spectral density analysis on the low-frequency overload signal x1, the mid-to-high frequency vibration signal x2, and the original vibration signal x0 respectively to obtain the corresponding power spectral densities: p1(k,i), p2(k,i), and p0(k), k = 0, 1, 2, ... N / 2; where N is the data length required for power spectral density analysis. Step 8: Using 1Hz as the frequency boundary for evaluating the filtering accuracy of low-frequency overload signals and medium-to-high-frequency vibration signals, calculate the filtering evaluation function v(d,i), as shown in the following formula: In the formula, when k = 1, ..., m, it corresponds to the power spectral density below 1 Hz in the f0 array; When k = m+1,…,N / 2+1, the power spectral density above 1Hz in the f0 array corresponds to the following values; d is the filter order, and i represents the i-th window. Step 9: Find the optimal window size w opt and the minimum filter evaluation function value v min (d,w opt ); The minimum value of the filtering evaluation function is defined as the minimum value of the filtering evaluation function, and the corresponding window size is the optimal window size w. opt ; Step 10: Increase the filter order by 1 at intervals and determine whether the filter order d is greater than 2. If yes, proceed to step 11; otherwise, proceed to step 4. Step 11: Calculate the difference Δ between two adjacent minimum filter evaluation functions. The calculation formula is as follows: In the formula, w opt w' is the optimal filter window size corresponding to an order of d-1. opt It is the optimal filter window size corresponding to the order d-2; Step 12: Determine if the difference Δ between the minimum filter evaluation functions is less than the preset threshold; if not, proceed to Step 4; if yes, then the order d-1 is the optimal order d. opt ; Step 13: Using the optimal order d opt and its corresponding optimal window w opt Using the filtering parameters, the original signal x0 is subjected to SG filtering to obtain the low-frequency overload signal x1. opt ; Step Fourteen: Based on the original signal x0 and the low-frequency overload signal x1 opt Acquire mid-to-high frequency vibration signals x2 opt .

2. The overload vibration data separation method based on a MEMS vibration sensor according to claim 1, characterized in that... The SG filter window w(i) is obtained as follows: 1) Select a series of window sizes w ~ (i), and the selection principle is as follows; Determine the starting window size: If 0.01*fs is an even number after rounding, use it directly as the starting window size; if 0.01*fs is an odd number after rounding, use its adjacent even number. Determine the step window size (dw): same as the initial window size; Determine the termination window size: [(fs / dw) rounded up + 1] * dw. 2) Obtain the SG filter window w(i) = w~(i)+1.

3. A method for separating overload vibration data based on a MEMS vibration sensor according to claim 1 or 2, characterized in that, In step six, the mid-to-high frequency vibration signal is calculated using the following formula: x2(j,i)=x0(j)-x1(j,i).

4. The overload vibration data separation method based on a MEMS vibration sensor according to claim 3, characterized in that, The following formula is used based on the original signal x0 and the low-frequency overload signal x1. opt Acquire mid-to-high frequency vibration signals x2 opt : x2 opt =x0-x1 opt 。 5. The overload vibration data separation method based on a MEMS vibration sensor according to claim 1, characterized in that, N is of size 2 n n is a natural number, and N is not less than 2*fs.

6. The overload vibration data separation method based on a MEMS vibration sensor according to claim 1, characterized in that, The preset threshold is 5%.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the separation method according to claims 1-6.

Images