阅读说明：本技术 用于回转式多余物检测特征信号提取方法 (Method for extracting characteristic signal for rotary type redundancy detection ) 是由 袁静 赵倩 蒋会明 许冲 邵云飞 刘海江 于 2019-11-18 设计创作，主要内容包括：本发明提供了一种用于回转式多余物检测特征信号提取方法,因为根据初始回转动态信号获取初始IMF序列,再根据初始IMF序列获取基准待处理IMF和备选待处理IMF序列,接着获取待处理含噪IMF集合,然后,将待处理含噪IMF集合分解成多个局部子信号进行获取提纯样本后再进行重组后获取估计噪声分量,最后对估计噪声分量进行叠加获取回转估计信号,并通过回转估计信号获取多余物检测特征信号。所以,本发明的用于回转式多余物检测特征信号提取方法通过对信号进行精选重构及分解提纯再重构,有效地提高航天机电设备回转式多余物检测中噪声重构经验模式分解方法的关键噪声估计精度,从而为航天机电设备多余物智能检测提供了更为精确的特征信号获取方法。(The invention provides a method for extracting a characteristic signal for rotary type redundancy detection, which is characterized in that an initial IMF sequence is obtained according to an initial rotary dynamic signal, a reference IMF to be processed and an alternative IMF sequence to be processed are obtained according to the initial IMF sequence, then a noise-containing IMF set to be processed is obtained, then the noise-containing IMF set to be processed is decomposed into a plurality of local sub-signals, an estimation noise component is obtained after recombination is carried out after a purified sample is obtained, finally the estimation noise component is superposed to obtain a rotary estimation signal, and a redundancy detection characteristic signal is obtained through the rotary estimation signal. Therefore, the method for extracting the characteristic signal for the rotary type redundancy detection effectively improves the key noise estimation precision of the noise reconstruction empirical mode decomposition method in the rotary type redundancy detection of the aerospace electromechanical equipment by carrying out concentration reconstruction, decomposition, purification and reconstruction on the signal, thereby providing a more accurate characteristic signal acquisition method for the intelligent detection of the redundancy of the aerospace electromechanical equipment.)

1. A method for extracting a characteristic signal for rotary redundancy detection is used for acquiring a redundancy detection characteristic signal according to an initial rotary dynamic signal obtained by sampling according to a preset frequency in the rotary redundancy detection of aerospace electromechanical equipment, and is characterized by comprising the following steps of,

step S1, decomposing the initial rotation dynamic signal by an EMD (empirical mode decomposition) method to obtain an initial IMF sequence composed of a plurality of IMFs (eigenmode components);

step S2, acquiring a reference IMF to be processed and an alternative IMF sequence to be processed according to the initial IMF sequence;

step S3, acquiring a to-be-processed noisy IMF set composed of a plurality of to-be-processed noisy IMFs according to the reference to-be-processed IMF and the alternative to-be-processed IMF sequence;

step S4, obtaining each to-be-processed noisy IMF of the to-be-processed noisy IMF set to obtain a plurality of corresponding local sub-signal windows_l(i) The local sub-signal window_l(i) Of (2) a sampleThe point expression is shown in formula (1),

window_l(i)＝{c_l(t),t＝1+iw_l,…,(i+1)w_l},i＝0,…,nn/w_l-1 (1)

w_l＝f_s/f_l，w_lrounding off, f_lIs a gyration characteristic frequency, f, corresponding to the noisy IMF to be processed_sFor said predetermined frequency, c_l(t) is the sample point of the noisy IMF to be processed;

step S5, obtaining a plurality of purified noise samples corresponding to the local sub-signals

Tm_l,ithe expression of (b) is shown in formula (3),

E_l,iis the energy value of the local sub-signal,

step S6, recombining the plurality of purified noise samples in the order of the corresponding sample points to obtain a plurality of estimated noise components corresponding to the noise-containing IMF to be processed;

step S7, superimposing the plurality of estimated noise components to obtain an estimated noise signal of the slewing dynamic signal as a slewing noise estimation signal;

step S8, reconstructing a redundancy detection estimation signal according to the rotation noise estimation signal, and obtaining the redundancy detection feature signal.

2. The rotary type redundancy detection feature signal extraction method according to claim 1,

wherein, step S2 includes the following substeps:

step S2-1, taking the first IMF in the initial IMF sequence as the reference to-be-processed IMF, and taking the rest of the initial IMF sequence as to-be-determined IMF sequence;

step S2-2, sequentially obtaining the energy value corresponding to each IMF in the IMF sequence to be judged, and according to the energy value E_kA noise energy value estimation range of the corresponding confidence level of 99% is obtained,

lower limit of the noise energy value estimation range

upper limit of the noise energy value estimation range

Z_0.5for parameters that match 99% confidence levels, nn is the initial slew dynamic signal data length,

step S2-3, acquiring a corresponding noise energy value correction range according to the noise energy value estimation range;

and step S2-4, sequentially judging whether the energy values are in the corresponding correction noise range, and sequentially forming the alternative IMF sequences to be processed by the IMFs corresponding to the judgment result of yes judgment.

3. The rotary type redundancy detection feature signal extraction method according to claim 2,

wherein, in step S2-3, the upper limit of the noise energy value estimation range is added with one as the upper limit of the corresponding noise energy value correction range,

and reducing the lower limit of the noise energy value estimation range by one to be used as the lower limit of the corresponding noise energy value correction range.

4. The rotary type redundancy detection feature signal extraction method according to claim 2,

wherein, step S3 includes the following substeps:

step S3-1, selecting all IMFs with continuous sequence numbers in the initial IMF sequence with the reference IMF to be processed from the candidate IMF sequence to be processed, forming intermediate noise components with the reference IMF to be processed according to the sequence order of the initial IMF sequence, and taking the rest parts with discontinuous sequence numbers in the candidate IMFs to be processed as candidate noise-containing IMF sequences to be judged;

step S3-2, taking the difference between the initial rotation dynamic signal and the intermediate noise component as an intermediate characteristic component;

step S3-3, obtaining a correlation coefficient of the intermediate characteristic component and the intermediate noise component as a first correlation coefficient;

step S3-4, taking the first IMF of the candidate to-be-judged noisy IMF sequence as a current IMF, and moving the current IMF from the candidate to-be-judged noisy IMF sequence into the intermediate noise component to obtain a new intermediate noise component and a new intermediate characteristic component;

step S3-5, acquiring a correlation coefficient of the intermediate noise component and the intermediate characteristic component as a second correlation coefficient;

step S3-6, judging whether the second correlation number is less than or equal to the first correlation coefficient, if so, taking the second correlation number as a new first correlation coefficient, and entering step S3-7, if not, deleting the current IMF from the alternative to-be-judged noisy IMF sequence, and entering step S3-7;

and S3-7, judging whether the alternative to-be-judged noisy IMF sequence is empty, if not, entering the step S3-4, if so, taking each IMF of the intermediate noise component as the to-be-processed noisy IMF to form the to-be-processed noisy IMF set, and entering the step S4.

5. The rotary type redundancy detection feature signal extraction method according to claim 1,

in step S4, demodulating each to-be-processed noisy IMF in the to-be-processed noisy IMF set to obtain a corresponding slewing characteristic frequency.

6. The rotary type redundancy detection feature signal extraction method according to claim 1,

wherein, in step S8, the method includes the following substeps:

step S8-1, resampling the gyration noise estimation signal, and acquiring the redundancy detection estimation signal;

step S8-2, decomposing the redundancy detection estimation signal by an EMD method to obtain a residual signal and a plurality of decomposition characteristic signals;

step S8-3, summing the energy value of the residual signal and the energy values of the decomposition characteristic signals to obtain the characteristic energy sum, and taking the difference value between the energy value of the initial revolution dynamic signal and the characteristic energy sum as a characteristic energy difference value;

step S8-4, judging whether the relative ratio of the characteristic energy difference value and the energy value of the gyration noise estimation signal is smaller than a preset allowable error, if so, entering step S8-5, and if not, entering step S8-1;

and step S8-5, acquiring the redundancy detection characteristic signal according to the plurality of decomposition characteristic signals.

7. The rotary type redundancy detection feature signal extraction method according to claim 6,

in step S8-1, the slewing noise estimation signal is resampled in a manner of data random arrangement.

8. The rotary type redundancy detection feature signal extraction method according to claim 6,

wherein, in step S8-5, the redundancy detection feature signal is obtained by summing and averaging a plurality of the decomposed feature signals.

Technical Field

The invention belongs to the field of aerospace equipment detection, and particularly relates to a method for extracting a characteristic signal for rotary type redundancy detection.

Background

The surplus refers to all substances existing in the product, which are generated from the outside or from the inside and are not related to the specified state of the product. Common residues of aerospace electromechanical equipment comprise metals such as fallen standard parts, processed metal chips, missed assembly tools and welding tin chips, and non-metals such as wire skins, silica gel residues, adhesive tapes and aluminized films. The redundancy is introduced in various links such as production, processing, assembly, transportation and debugging of aerospace electromechanical equipment, and is finally sealed inside the product. The redundancy is a major quality accident hidden danger of the aerospace electromechanical equipment, and particularly, the active redundancy is always in an irregular free state, which may cause faults or failures such as device short circuit, circuit abnormity, mechanism blocking and the like, and even cause serious aerospace accidents. Therefore, the problem of preventing and controlling the redundant materials of the aerospace electromechanical devices is not very slow, and if the redundant materials in the aerospace electromechanical devices can be accurately and timely identified and detected, the method has great significance for improving the quality safety of products, ensuring the success rate of aerospace tasks, avoiding economic loss and catastrophic accidents and the like.

The existing redundant detection technology comprises means such as visual inspection, ear hearing, industrial endoscope detection, X-ray fluoroscopy, ultrasonic detection, particle collision noise detection and the like. The detection of the particle collision noise has a good detection effect on the movable redundant objects, and comprises rotary detection, random vibration detection and the like. Automatic detection equipment developed based on the technology is widely applied to various domestic spacecrafts. The particle collision noise-based redundancy detection process comprises the steps of signal acquisition, feature extraction, diagnosis and identification and the like, wherein redundancy signal feature extraction and processing are one of key links of the detection process, and feature signal extraction accuracy is directly related to the effectiveness of subsequent redundancy identification. However, the aerospace electromechanical devices are generally complex in structure and numerous in component parts, dynamic signals acquired in detection of redundancy such as rotation and the like are comprehensive reflection of response of each part, and factors such as complex and variable transmission paths, random vibration background noise interference, mutual coupling of multiple excitation sources and the like often cause weak characteristic signals generated by particle collision of redundancy (particularly tiny redundancy), low signal-to-noise ratio and difficult identification. Through the search and discovery of published patent applications, the aerospace electromechanical device redundancy detection patent application which is more relevant to the invention comprises the following steps: (1) the invention patent application CN201811344706.4 discloses a device and a method for detecting excess of a rotary radio device, which focuses on the design of the device and the excess detection flow based on the device; (2) the invention patent application CN201210528125.2 discloses a method for detecting and classifying and identifying the redundancy of an electronic device based on nonnegative tensor decomposition, which comprehensively utilizes multisource signals such as sound, acceleration and the like to carry out nonnegative tensor decomposition and classification and identification so as to improve the sensitivity and the precision of the redundancy detection; (3) the invention patent application CN201010123941.6 discloses a device and a method for detecting the sealed electronic component redundancy based on random vibration, which mainly focuses on the design of a detection device and an acceleration feedback signal processing technology thereof, and realizes the automatic detection of the sealed electronic component redundancy by means of Fourier transform and a time/frequency domain randomization technology. The research and development of the device for detecting the redundancy such as the multi-side re-rotation type and the random vibration type of the related patents for detecting the redundancy and the detection process itself, but the extraction and the processing of the weak characteristic signals in the detection of the redundancy (especially the rotation type redundancy) are not related and researched.

The invention utilizes an improved noise reconstruction empirical mode decomposition method to extract key weak signal characteristics in the rotary redundancy detection of the aerospace electromechanical equipment, and patents related to the invention on the basis theory comprise: (1) the invention patent ZL201310097502.6 discloses an integrated noise reconstruction empirical mode decomposition method for early mechanical and compound faults, which estimates a noise component in a signal to be detected by using a similar hard threshold processing mode, thereby realizing dynamic signal noise reduction and improving the mode confusion problem in the traditional empirical mode decomposition. The invention provides a basic noise reconstruction empirical mode decomposition method and concrete steps; (2) the invention patent application CN201811480857.2 discloses a radar precision stabilized platform operation reliability evaluation method, which introduces an adjacent coefficient threshold value technology into key noise estimation of noise reconstruction empirical mode decomposition, and calculates radar stabilized platform operation reliability evaluation indexes by normalized singular entropy. The invention patent application CN201811480857.2 makes two basic method improvements to the key noise estimation technique in the empirical mode decomposition method for noise reconstruction in patent ZL201310097502.6, that is: (1) forming a noise-containing eigenmode component by using a plurality of eigenmode components of a continuous sequence starting from the first eigenmode component; (2) the adjacent coefficient noise reduction technology is used for improving the original hard threshold processing mode. For the extra weak characteristic signals of the aerospace electromechanical devices, the key noise estimation technology in the original noise reconstruction empirical mode decomposition and improvement method in the two patents has the problem that the extra weak characteristic signals are mistakenly incorporated into noise components, which causes inaccurate extraction of important extra weak characteristic signals and leads to failure of detection and identification of the extra of the aerospace electromechanical devices.

Disclosure of Invention

The invention is carried out in view of the above problems, and aims to provide a method for detecting weak characteristic signals of redundancy, which can realize signal adaptive decomposition, intelligent filtering and automatic noise reduction integrated processing in rotary redundancy detection of aerospace electromechanical equipment.

In order to achieve the purpose, the invention adopts the following technical scheme:

the invention provides a method for extracting a characteristic signal for rotary type redundancy detection, which has the characteristics that the method comprises the following steps: step S1, decomposing the initial rotation dynamic signal by EMD (empirical mode decomposition) method to obtain an initial IMF sequence composed of multiple IMFs (eigenmode components);

step S2, acquiring a reference IMF to be processed and an alternative IMF sequence to be processed according to the initial IMF sequence;

step S3, acquiring a to-be-processed noisy IMF set composed of a plurality of to-be-processed noisy IMFs according to the reference to-be-processed IMF and the alternative to-be-processed IMF sequence;

step S4, obtaining each to-be-processed noise-containing IMF of the to-be-processed noise-containing IMF set to obtain a plurality of corresponding local sub-signal windows_l(i) Local sub-signal window_l(i) The expression of (2) is shown in formula (1),

window_l(i)＝{c_l(t),t＝1+iw_l,…,(i+1)w_l},i＝0,…,nn/w_l-1 (1)

w_l＝f_s/f_l，w_lrounding off, f_lFor the characteristic frequency of revolution, f, corresponding to the noisy IMF to be processed_sAt a predetermined frequency, c_l(t) is a sample point of the noisy IMF to be processed;

step S5, obtaining a plurality of purified noise samples corresponding to the local sub-signals

Purifying noise samples

The expression of (A) is shown in formula (2),

Tm_l,ithe expression of (b) is shown in formula (3),

E_l,iis the energy value of the local sub-signal,

step S6, recombining a plurality of purified noise samples according to the sequence of corresponding sample points to obtain a plurality of estimated noise components corresponding to the IMF containing noise to be processed;

step S7, overlapping a plurality of estimation noise components to obtain an estimation noise signal of the rotation dynamic signal as a rotation noise estimation signal;

and step S8, reconstructing a redundancy detection estimation signal according to the rotation noise estimation signal, and acquiring a redundancy detection characteristic signal.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: wherein, step S2 includes the following substeps:

step S2-1, taking the first IMF in the initial IMF sequence as a reference IMF to be processed, and taking the rest of the initial IMF sequence as an IMF sequence to be judged;

step S2-2, sequentially acquiring the energy value corresponding to each IMF in the IMF sequence to be judged, and according to the energy value E_kA noise energy value estimation range of the corresponding confidence level of 99% is obtained,

lower limit of noise energy value estimation range

The expression of (A) is shown in formula (4),

upper limit of noise energy value estimation range

The expression of (A) is shown in formula (5),

Z_0.5for parameters that match the 99% confidence level, nn is the initial slew dynamic signal data length,

step S2-3, acquiring a corresponding noise energy value correction range according to the noise energy value estimation range;

and step S2-4, sequentially judging whether the energy value is in the corresponding correction noise range, and sequentially forming the IMF corresponding to the judgment result of yes into an alternative IMF sequence to be processed.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: in step S2-3, the upper limit of the noise energy value estimation range is increased by one to be the upper limit of the corresponding noise energy value correction range, and the lower limit of the noise energy value estimation range is decreased by one to be the lower limit of the corresponding noise energy value correction range.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: wherein, step S3 includes the following substeps:

step S3-1, selecting all IMFs with continuous sequence numbers in the initial IMF sequence with the reference IMF to be processed from the candidate IMF sequence to be processed, forming intermediate noise components with the reference IMF to be processed according to the sequence order of the initial IMF sequence, and taking the rest parts with discontinuous sequence numbers in the candidate IMFs to be processed as candidate noise-containing IMF sequences to be judged;

step S3-2, taking the difference between the initial rotation dynamic signal and the intermediate noise component as the intermediate characteristic component;

step S3-3, acquiring a correlation coefficient of the intermediate characteristic component and the intermediate noise component as a first correlation coefficient;

step S3-4, taking the first IMF of the alternative to-be-judged noisy IMF sequence as the current IMF, and moving the current IMF from the alternative to-be-judged noisy IMF sequence into the intermediate noise component to obtain a new intermediate noise component and a new intermediate characteristic component;

step S3-5, acquiring a correlation coefficient of the intermediate noise component and the intermediate characteristic component as a second correlation coefficient;

step S3-6, judging whether the second correlation number is less than or equal to the first correlation coefficient, if so, taking the second correlation number as a new first correlation coefficient, and entering step S3-7, if not, deleting the current IMF from the alternative noise-containing IMF sequence to be judged, and entering step S3-7;

and S3-7, judging whether the alternative to-be-judged noisy IMF sequence is empty, if not, entering the step S3-4, if so, taking each IMF of the intermediate noise component as a to-be-processed noisy IMF to form a to-be-processed noisy IMF set, and entering the step S4.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: in step S4, each to-be-processed noisy IMF in the to-be-processed noisy IMF set is demodulated, and a corresponding rotation characteristic frequency is obtained.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: wherein, in step S8, the method includes the following substeps:

step S8-1, resampling the rotary noise estimation signal, and acquiring a redundancy detection estimation signal;

step S8-2, decomposing the redundancy detection estimation signal by an EMD method to obtain a residual signal and a plurality of decomposition characteristic signals;

step S8-3, summing the energy value of the residual signal and the energy values of the plurality of decomposition characteristic signals to obtain a characteristic energy sum, and taking the difference value between the energy value of the initial rotation dynamic signal and the characteristic energy sum as a characteristic energy difference value;

step S8-4, judging whether the relative ratio of the characteristic energy difference value and the energy value of the gyration noise estimation signal is smaller than a preset allowable error, if so, entering step S8-5, and if not, entering step S8-1;

and step S8-5, acquiring a redundancy detection characteristic signal according to the plurality of decomposition characteristic signals.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: in step S8-1, the slewing noise estimation signal is resampled so that data is randomly arranged.

In the method for extracting the feature signal for detecting the rotary redundancy, the method can further have the following features: in step S8-5, a redundancy detection feature signal is obtained by summing and averaging the plurality of decomposed feature signals.

Action and Effect of the invention

According to the method for extracting the characteristic signal for rotary type redundancy detection, the initial IMF sequence is obtained according to the initial rotary dynamic signal, the reference IMF to be processed and the alternative IMF sequence to be processed are obtained according to the initial IMF sequence, then the noise-containing IMF set to be processed is obtained, then the noise-containing IMF set to be processed is decomposed into a plurality of local sub-signals, the noise samples are obtained and purified, then the noise components are recombined to obtain the estimated noise components, finally the estimated noise components are superposed to obtain the rotary noise estimated signal, and the redundancy detection characteristic signal is obtained through the rotary noise estimated signal. Therefore, the method for extracting the characteristic signal for the rotary type redundancy detection effectively improves the key noise estimation precision of the noise reconstruction empirical mode decomposition method in the rotary type redundancy detection of the aerospace electromechanical equipment by carrying out concentration reconstruction, decomposition, purification and reconstruction on the signal, thereby providing a more accurate characteristic signal acquisition method for the intelligent detection of the redundancy of the aerospace electromechanical equipment.

Drawings

FIG. 1 is a schematic diagram illustrating steps of a method for extracting a feature signal for detecting a revolving type redundancy in an embodiment of the present invention;

FIG. 2 is a characteristic composition diagram of an initial slew dynamic signal in an embodiment of the present invention;

FIG. 3 is a graph comparing an initial slewing dynamic noise signal to a slewing estimation noise signal in an embodiment of the invention;

FIG. 4 is a diagram of a redundancy detection feature signal in an embodiment of the present invention; and

fig. 5 is a diagram of a feature signal for detecting redundancy, which is obtained by a conventional technique.

Detailed Description

In order to make the technical means, the creation features, the achievement objects and the effects of the present invention easy to understand, the following embodiments describe the method for detecting the feature signal of the revolving type redundancy in detail with reference to the accompanying drawings.

Fig. 1 is a schematic step diagram of a method for extracting a feature signal for rotary redundancy detection in an embodiment of the present invention.

As shown in fig. 1, in the method S100 for extracting a characteristic signal for detecting a rotary redundancy, a redundancy detection characteristic signal is obtained according to an initial rotary dynamic signal obtained by sampling according to a predetermined frequency in the rotary redundancy detection of an aerospace electromechanical device.

Fig. 2 is a characteristic composition diagram of an initial slew dynamic signal in an embodiment of the present invention.

As shown in FIG. 2, in the present embodiment, the initial rotation dynamic signal x is an artificial mixed signal, which includes a sinusoidal rotation signal x in the detection of the rotation redundancy of the analog aerospace electromechanical device₁Simulating particle collision signal x in rotary redundancy detection of aerospace electromechanical equipment₂The method comprises the following steps of (namely, redundancy characteristic signals), a noise signal r in the rotary redundancy detection of the aerospace electromechanical equipment, and a simulated mixed signal x of the rotary redundancy detection of the aerospace electromechanical equipment. As can be seen from the simulated mixed signal, the redundancy characteristic signal x₂The signal is weak, and the effective identification is difficult to be directly carried out from the initial rotation dynamic signal x.

The method for extracting the characteristic signal for detecting the rotary redundancy comprises the following steps:

step S1, decomposing the initial rotation dynamic signal by an EMD (empirical mode decomposition) method to obtain an initial IMF sequence composed of a plurality of IMFs (eigenmode components).

In this embodiment, EMD decomposition is performed on the initial slewing dynamic signal x to obtain an initial IMF sequence { c }_k(t),k＝1,…,13}。

Step S2, acquiring a reference to-be-processed IMF and an alternative to-be-processed IMF sequence according to the initial IMF sequence, including the following sub-steps:

and step S2-1, taking the first IMF in the initial IMF sequence as a reference IMF to be processed, and taking the rest part of the initial IMF sequence as an IMF sequence to be judged.

In the present embodiment, i.e. c₁(t) as a reference IMF to be processed, { c_k(t), k is 2, …,13} as the IMF sequence to be judged.

Step S2-2, sequentially acquiring the energy value corresponding to each IMF in the IMF sequence to be judged, and according to the energy value E_kA noise energy value estimation range of the corresponding confidence level of 99% is obtained,

lower limit of noise energy value estimation range

The expression of (A) is shown in formula (4),

upper limit of noise energy value estimation range

The expression of (A) is shown in formula (5),

Z_0.5for parameters that match the 99% confidence level, nn is the initial slew dynamic signal data length.

In the present embodiment, the IMF sequence { c ] to be determined is calculated_k(t), k 2, …, 13) for each IMF energy value ln { E }_k K 2, …,13 and an estimated redundancy noise energy lower bound for each IMF within 99% of the confidence level

And upper limit of

And step S2-3, acquiring a corresponding noise energy value correction range according to the noise energy value estimation range.

And adding one to the upper limit of the noise energy value estimation range to serve as the upper limit of the corresponding noise energy value correction range, and subtracting one from the lower limit of the noise energy value estimation range to serve as the lower limit of the corresponding noise energy value correction range.

In the present embodiment, the upper limit of the noise energy value correction range is set to

The lower limit of the noise energy magnitude correction range is

And step S2-4, sequentially judging whether the energy value is in the corresponding correction noise range, and sequentially forming the IMF corresponding to the judgment result of yes into an alternative IMF sequence to be processed.

In this embodiment, the energy values lnE are determined one by one_kAnd if the IMF is judged to be in the closed interval range of the upper limit and the lower limit of the noise energy value correction range, sequentially forming the IMF corresponding to the judgment result of yes into an alternative IMF sequence to be processed, and discarding the rest IMF data.

Step S3, obtaining a to-be-processed noisy IMF set composed of multiple to-be-processed noisy IMFs according to the reference to-be-processed IMF and the candidate to-be-processed IMF sequence, including the following substeps:

and step S3-1, selecting all IMFs with continuous sequence numbers in the initial IMF sequence with the reference IMF to be processed from the candidate IMF sequence to be processed, forming intermediate noise components with the reference IMF to be processed according to the sequence order of the initial IMF sequence, and taking the rest parts with discontinuous sequence numbers in the candidate IMFs to be processed as candidate noise-containing IMF sequences to be judged.

In this embodiment, the candidate pending IMF and the reference pending IMF are { c }_l(t), l ═ 1,2,3,6,7,8,12}, where { c } is present₁(t),c₂(t),c₃(t) } consecutive in the initial IMF sequence, will { c_l(t), 1,2,3 is superimposed as the intermediate noise component np (t), and { c ═ c is superimposed_l(t), l ═ 6,7,8,12} as alternative noisy IMF sequences to be judged.

In step S3-2, the difference between the initial rotation dynamic signal and the intermediate noise component is used as an intermediate feature component.

The difference between the initial rotation dynamic signal x and the intermediate noise component np (t) is used as the intermediate characteristic component sp (t).

Step S3-3, a correlation coefficient of the intermediate feature component and the intermediate noise component is acquired as a first correlation coefficient cc.

And step S3-4, taking the first IMF of the alternative to-be-judged noisy IMF sequence as the current IMF, and shifting the current IMF from the alternative to-be-judged noisy IMF sequence into the intermediate noise component to obtain a new intermediate noise component and a new intermediate characteristic component.

In this embodiment, the first IMF of the candidate noisy IMF sequence to be determined obtained in step 3-1 is c₆(t) a new intermediate noise component np (t) consisting of { c }_l(t), l is obtained by superposition of 1,2,3,6, and the new intermediate feature component sp (t) is the difference between the initial rotation dynamic signal x and the new intermediate noise component np (t).

In step S3-5, the correlation coefficient between the intermediate noise component and the intermediate feature component is obtained as the second correlation coefficient cc'.

And S3-6, judging whether the second correlation number is less than or equal to the first correlation coefficient, if so, taking the second correlation number as a new first correlation coefficient, and entering step S3-7, otherwise, deleting the current IMF from the alternative noise-containing IMF sequence to be judged, and entering step S3-7.

In this embodiment, if it is determined in the determination in step 3-6 that the determination is yes, the first intermediate noise component is the IMF series { c }_l(t), l is the superposition of 1,2,3, 6), and the candidate noise-containing IMF sequence to be judged is { c_l(t), l ═ 7,8,12 }; if not, then the first intermediate noise component is still IMF sequence { c }_l(t), l is the superposition of 1,2, 3), and the candidate noise-containing IMF sequence to be judged is { c_l(t),l＝7,8,12}。

And S3-7, judging whether the alternative to-be-judged noisy IMF sequence is empty, if not, entering the step S3-4, if so, taking each IMF of the intermediate noise component as a to-be-processed noisy IMF to form a to-be-processed noisy IMF set, and entering the step S4.

In the present embodiment, after the determination of step S3-7, the noise-containing IMF set to be processed is { c }_l(t),l＝1,2,3,6,7,8}。

Step S4, obtaining each to-be-processed noise-containing IMF of the to-be-processed noise-containing IMF set to obtain a plurality of corresponding local sub-signal windows_l(i) Local sub-signal window_l(i) The sample point expression of (2) is shown in formula (1):

window_l(i)＝{c_l(t),t＝1+iw_l,…,(i+1)w_l},i＝0,…,nn/w_l-1 (1)

w_l＝f_s/f_land w is_lRounding off, f_sFor a predetermined frequency of sampling the initial slewing dynamic signal, f_lIs the gyration characteristic frequency corresponding to the noise-containing IMF to be processed.

Specifically, for the selected noisy IMF set { c) to be processed_l(t), l ═ 1,2,3,6,7,8} local sub-signal window under adaptive sliding window_l(i)，w_lFor adaptive sliding window width, if c is truncated at the adaptive sliding window_lAnd if the last local subsignal is not wide enough, zero padding is carried out on the samples at the insufficient points.

Step S5, obtaining a plurality of purified noise samples corresponding to the local sub-signals

Purifying noise samples

The expression of (A) is shown in formula (2),

Tm_l,ithe expression of (b) is shown in formula (3),

E_l,iis the energy value of the local sub-signal.

Specifically, for a plurality of local sub-signals window_l(i) Local extreme value threshold processing (calculation formula is shown as (2)) is implemented to obtain self-adaptive local purified noise samples

And step S6, recombining the plurality of purified noise samples according to the sequence of the corresponding sample points to obtain a plurality of estimated noise components corresponding to the IMF containing noise to be processed.

Specifically, a plurality of obtained are

In step S7, the estimated noise components are superimposed to obtain an estimated noise signal of the slewing dynamic signal as a slewing noise estimation signal.

In the present embodiment, the slewing noise estimation signal is obtained by superposition

And acquiring a redundancy detection characteristic signal according to the rotation noise estimation signal.

Step S8, obtaining a redundancy detection feature signal according to the slewing noise estimation signal, including the following substeps:

and step S8-1, resampling the rotation noise estimation signal and obtaining a redundancy detection estimation signal.

Wherein the signal is estimated for the slewing noise

Is resampled in a manner that the data is randomly arranged.

Fig. 3 is a graph comparing an initial slewing dynamic noise signal with a slewing estimation noise signal in an embodiment of the invention.

As shown in FIG. 3, in this embodiment, the gyration noise estimation signal obtained by the present invention

The amplitudes of the plurality of sampling points are basically equivalent to the amplitudes of the corresponding points of the initial rotation dynamic noise signal, and the effectiveness of the noise estimation method for the invention is demonstrated.

Step S8-2: and decomposing the redundancy detection estimation signal by an EMD method to obtain a residual signal and a plurality of decomposition characteristic signals.

In the present embodiment, for

Performing EMD self-adaptive decomposition and integrated average processing on the residue detection estimation signal obtained by resampling to obtain a residual signal r_j(t) and a plurality of decomposition feature signals { c_j,k(t),k＝1,…,13}。

Step S8-3: and summing the energy value of the residual signal and the energy values of the plurality of decomposition characteristic signals to obtain a characteristic energy sum, and taking the difference value between the energy value of the initial rotation dynamic signal and the characteristic energy sum as a characteristic energy difference value.

Step S8-4: and judging whether the relative ratio of the characteristic energy difference value and the energy value of the gyration noise estimation signal is smaller than a preset allowable error, if so, entering the step S8-5, and if not, entering the step S8-1.

Specifically, the predetermined allowable error is 0.04.

Step S8-5: and acquiring a redundancy detection characteristic signal according to the plurality of decomposition characteristic signals.

And acquiring a redundancy detection characteristic signal by summing and averaging the plurality of decomposition characteristic signals.

Fig. 4 is a diagram of a redundancy detection feature signal in an embodiment of the present invention.

As shown in fig. 4, in this embodiment, the decomposition feature signal is obtained after 26 times of judgment, and then the average operation is performed on the decomposition result (i.e. the decomposition feature signal) to obtain the feature signal of the rotary type redundancy (i.e. the purified simulation feature signal of the redundancy of the aerospace electromechanical device)

Andas can be seen from FIG. 4, the signature signal of the revolving redundancy includes 6 kinds of simulation signatures

Implicit in the original signal x is the weak characteristic signal x of the redundancy₂Is successfully extracted from

And a sinusoidal revolution signal x₁Is clearly shown in

Therefore, the embodiment effectively realizes the integrated processing of the rotary redundancy detection simulation mixed signal self-adaptive decomposition, the intelligent filtering and the automatic noise reduction of the aerospace electromechanical equipment, and obtains the weak characteristic signal x of the redundancy of the aerospace electromechanical equipment₂The effective expression and intelligent extraction of the method.

Fig. 5 is a diagram of a feature signal for detecting redundancy, which is obtained by a conventional technique.

As shown in FIG. 5, the initial rotation dynamic signal x detected by the same rotation redundancy of the aerospace electromechanical device is analyzed by using a conventional EMD, and the initial rotation dynamic signal x comprises 6 simulation characteristics { c }_k(t), k is 1, …,6}, wherein, the weak characteristic signal x of the redundancy₂Is shown in c₂But the feature is contaminated by interference noise, is not easily recognized, and is a sinusoidal revolution signal x₁Is shown in c₄But this feature exhibits significant pattern aliasing. Therefore, compared with the embodiment, the analysis result of the traditional EMD on the simulation signal of the rotary redundancy detection of the aerospace electromechanical device is not ideal.