CN109633269B - Wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum value - Google Patents
Wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum value Download PDFInfo
- Publication number
- CN109633269B CN109633269B CN201910001135.2A CN201910001135A CN109633269B CN 109633269 B CN109633269 B CN 109633269B CN 201910001135 A CN201910001135 A CN 201910001135A CN 109633269 B CN109633269 B CN 109633269B
- Authority
- CN
- China
- Prior art keywords
- wavelet decomposition
- time
- speed signal
- frequency power
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R23/00—Arrangements for measuring frequencies; Arrangements for analysing frequency spectra
- G01R23/16—Spectrum analysis; Fourier analysis
Landscapes
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum, which comprises the following steps: firstly, determining an optimal wavelet decomposition scale based on a time-frequency power spectral density maximum value; and secondly, enhancing the energy characteristics of the failure wave crest region of the propeller based on the optimal wavelet decomposition scale to obtain enhanced energy failure characteristics Eu and Ec of the speed signal wave crest region and the control signal wave crest region. The method can effectively identify the optimal wavelet decomposition scale of the speed signal of the underwater robot, and the determined optimal wavelet decomposition scales can be different due to different propeller fault degrees, so that the singular behavior signal-to-noise ratio of the speed signal is improved, and the propeller fault characteristic value is increased.
Description
Technical Field
The invention belongs to the underwater robot technology, is used for monitoring the state of a propeller of an underwater robot, and particularly relates to a wavelet decomposition scale determination and fault feature enhancement method based on a time-frequency power spectral density maximum value.
Background
The speed signal singular behavior caused by the propeller fault is weak generally, is influenced by noise of an underwater speed sensor, external interference noise such as ocean current and the like, has low signal-to-noise ratio and is difficult to identify, so that the fault characteristic directly extracted from the speed signal singular behavior is not obvious.
A typical method in the existing denoising method is a wavelet decomposition method, but the denoising effect of the method is affected by the decomposition scale. The known maximum eigenvalue method determines the wavelet decomposition scale directly from the maximum of the fault signature. However, in the wavelet decomposition results of all scales of the velocity signal of the underwater robot, the maximum value of the fault feature in some scales is not caused by the singular behavior of the fault, so that the direct use of the maximum value of the fault feature as the determination basis of the wavelet decomposition scale is not sufficient.
The known waveform priori knowledge method determines wavelet decomposition scale according to the priori knowledge of signal waveform, but singular behavior waveforms caused by faults of different degrees of a propeller in a speed signal are different, and when fault characteristics are extracted, the singular behavior waveforms are difficult to predict due to the fact that the fault degrees are difficult to predict, and the priori knowledge is difficult to obtain. In practice, the method generally determines a fixed wavelet decomposition scale according to the waveform prior knowledge of the historical data signals, and then decomposes all the detected signals by adopting the fixed wavelet decomposition scale. But this decomposition scale is only optimal for a certain degree of failure and not for all degrees of failure.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to solve the defects in the prior art and provides a wavelet decomposition scale determination and fault feature enhancement method based on a time-frequency power spectral density maximum value.
The technical scheme is as follows: the invention discloses a method for determining an optimal wavelet decomposition scale based on a time-frequency power spectral density maximum value, which comprises the following steps of:
first, the length of L is adopted1Intercepting the original data of the speed signal of the underwater robot by using the time window function;
second, for the length L1The speed signal data is processed by conventional smooth pseudo-Wigner-Willi distribution operation to obtain a smooth pseudo-Wigner-Willi spectrum SPWVD (n, m), n is a time beat sequence number, n is an integer, and n ∈ [1L ]1]M is a number of frequency bands, m is an integer, and m ∈ [1N ]3]，N3The maximum value of the frequency band serial number; obtaining the absolute value of the smooth pseudo-Vigrener-Weili spectrum SPWVD (n, m), obtaining the time-frequency power spectral density distribution SPWVDA (n, m) of the original speed signal, and determining the maximum value P in the time-frequency power spectral density distribution SPWVDA (n, m) of the original speed signal0；
Third, for the length L1The speed signal data is processed with N layers of conventional wavelet decomposition to obtain the approximate component of the N layer of wavelet, the smooth pseudo-Wigner-Weili spectrum of the approximate component of the wavelet is calculated, the absolute value of the smooth pseudo-Wigner-Weili spectrum is obtained to obtain the time frequency power density distribution, and the maximum value of the time frequency power density of the approximate component of the N layer of wavelet is determinedPN(ii) a N is a positive integer greater than 0;
the fourth step, judge if PN≤0.9PN-1If so, N-1 is the optimal wavelet decomposition scale and is recorded as M, namely M is N-1; if PN>0.9PN-1And if the N is equal to N +1, returning to the third step.
The invention also discloses a fault characteristic enhancement method based on the optimal wavelet decomposition scale, which comprises the following steps:
acquiring dynamic signal data of a speed signal and a propeller control signal of an underwater robot, and intercepting the dynamic signal by adopting a sliding time window, wherein the intercepting length L of the speed signal is1Control signal intercept length L2＝L1+1；
Secondly, determining the optimal wavelet decomposition scale M for the speed signal data based on a time-frequency power density maximum method; for the propeller control signal, the control signal change rate is obtained based on a discrete difference method, and after difference, the data length of the control signal change rate in a time window is L1The difference formula is shown as formula (1);
wherein c (n) is the propeller control signal, n is the time beat number, n ∈ [1L ]1]And Δ t is the sampling period,
the control signal change rate of the nth beat;
step three, controlling the rate of change of the speed signal u (n) in the time window
Respectively carrying out wavelet decomposition with the decomposition layer number being the optimal wavelet decomposition scale M to obtain the M-th layer wavelet approximate component u of the velocity signalMA(n), and control signal rate of change Mth layer wavelet approximation component
Step four, based on the conventional modified Bayes algorithm, the wavelet approximate component u of the velocity signal obtained in the step three is processedMA(n), and control signal rate of change wavelet approximation component
Processing to increase the amplitude of the singular behavior in the signal, the result of the processing being duMA(n) and dcMA(n)；
Step five, comparing the result d obtained in the step fouruMA(n) and dcMA(n) performing a convolution operation to concentrate the energy in the signal to obtain a convolution calculation result uconv(n) and cconv(n)；uconv(n) and cconv(n) the calculation is the same, here in uconv(n) for example, the convolution calculation is shown in equation (2):
uconv(n)＝duMA(n)*duMA(n) (2)
in the formula uconv(n) is duMA(n) the convolution operation result;
step six, the convolution calculation result u of the speed signal obtained in the step fiveconv(n), determining the positions of all minimum values, namely the positions of wave troughs, and determining u between two adjacent minimum valuesconv(n) accumulating and summing the amplitude values, taking the obtained result as the energy contained in the wave crest between two adjacent wave troughs, namely the energy of the wave crest region, and then selecting the maximum value of the energy of the wave crest region as the fault characteristic value of the propeller so as to obtain the energy fault characteristic value Eu of the wave crest region of the speed signal. Then c is processed in the same wayconvAnd (n) obtaining the energy fault characteristic value Ec of the peak region of the control signal.
Has the advantages that: the method can effectively identify the optimal wavelet decomposition scale of the speed signal of the underwater robot, and the determined optimal wavelet decomposition scales can be different due to different propeller fault degrees, so that the singular behavior signal-to-noise ratio of the speed signal is improved, and the propeller fault characteristic value is increased.
Drawings
Fig. 1 is a flowchart of an optimal wavelet decomposition scale determining method based on time-frequency power spectral density maxima according to the present invention.
FIG. 2 is a flow chart of the peak region energy feature enhancement method based on the optimal wavelet decomposition scale according to the present invention.
FIG. 3 is a flowchart of a wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maxima according to the present invention.
Fig. 4 is a time domain waveform diagram of a speed signal and a propeller control voltage signal of the underwater robot.
FIG. 5 is a time domain waveform diagram of velocity signals at different wavelet decomposition scales.
FIG. 6 is a graph of energy fault characteristic values of peak regions of velocity signals at different wavelet decomposition scales.
FIG. 7 is a velocity signal time-frequency power spectral density distribution diagram under different wavelet decomposition scales.
FIG. 8 is a time-frequency power spectral density maximum plot of velocity signals at different wavelet decomposition scales.
FIG. 9 is a graph of the energy of the known peak region and the extracted fault eigenvalues of the method of the present invention.
Fig. 10 is a plot of the time-frequency power spectral density maxima of a speed signal at a 10% fault level.
Fig. 11 is a fault feature distribution diagram corresponding to different fault degrees after enhancement based on a known maximum feature value method.
FIG. 12 is a graph of a fault signature distribution corresponding to different fault levels after enhancement based on the method of the present invention.
Detailed Description
The technical solution of the present invention is described in detail below, but the scope of the present invention is not limited to the embodiments.
As shown in fig. 1, the method for determining an optimal wavelet decomposition scale based on the time-frequency power spectral density maximum of the present invention specifically includes the following steps:
first, the length of L is adopted1Intercepting the original data of the speed signal of the underwater robot by a time window function of 400;
in the second step, the first step is that,for the length L1The velocity signal data of 400 is subjected to a conventional smooth pseudo-wigner-willi distribution operation to obtain a smooth pseudo-wigner-willi spectrum SPWVD (n, m), n being a time beat number, n being an integer, and n ∈ [1L1]M is a number of frequency bands, m is an integer, and m ∈ [1N ]3]，N3Is the maximum value of the band number, N3512; obtaining the absolute value of the smooth pseudo-Vigrener-Weili spectrum SPWVD (n, m), obtaining the time-frequency power spectral density distribution SPWVDA (n, m) of the original speed signal, and determining the maximum value P in the time-frequency power spectral density distribution SPWVDA (n, m) of the original speed signal0；
Third, for the length L1Performing N-layer conventional wavelet decomposition on the speed signal data of 400 (when the step is executed for the first time, N is 1) to obtain an N-th layer wavelet approximate component, calculating a smooth pseudo-Wigner-Weiley spectrum of the wavelet approximate component, calculating an absolute value of the smooth pseudo-Wigner-Weiley spectrum to obtain an instantaneous power density distribution, and determining an instantaneous power density maximum value PN。
The fourth step, judge if PN≤0.9PN-1If M is equal to N-1, the optimal wavelet decomposition scale is obtained; if PN>0.9PN-1And if the N is equal to N +1, returning to the third step.
As shown in fig. 2, the method for enhancing fault characteristics based on the optimal wavelet decomposition scale of the present invention specifically includes the following steps:
the method comprises the steps of firstly, acquiring dynamic signal data such as a speed signal of the underwater robot, a propeller control signal and the like, and intercepting the dynamic signal by adopting a sliding time window, wherein the intercepting length L of the speed signal 1400, the control signal is intercepted for a length L2＝401；
Secondly, determining the optimal wavelet decomposition scale M for the speed signal data based on the time-frequency power density maximum method provided by the invention; for a propeller control signal, obtaining a control signal change rate based on a discrete difference method, wherein after difference, the data length of the control signal change rate in a time window is 400, and a difference formula is shown as a formula (1);
where c (n) is the propeller control signal, n is the number of the time beat, n is 1,2,3, …, 400, Δ t is the sampling period, in this embodiment Δ t is 0.2s,
the control signal change rate of the nth beat;
thirdly, controlling the rate of change of the speed signal u (n) in the time window
Respectively performing conventional wavelet decomposition with the decomposition layer number being the optimal wavelet decomposition scale M to obtain the M-th layer wavelet approximate component u of the velocity signalMA(n), and control signal rate of change Mth layer wavelet approximation component
Fourthly, based on the conventional modified Bayes algorithm, the velocity signal wavelet approximate component u obtained in the third step is processedMA(n), and control signal rate of change wavelet approximation component
Processing to increase the amplitude of the singular behavior in the signal, the result of the processing being duMA(n) and dcMA(n)；
Step five, obtaining the result d of the step fouruMA(n) and dcMA(n) performing a convolution operation to concentrate the energy in the signal to obtain a convolution calculation result uconv(n) and cconv(n)。uconv(n) and cconv(n) the calculation is the same, here in uconv(n) illustrates the calculation process by way of example. The convolution calculation is shown in equation (2).
uconv(n)＝duMA(n)*duMA(n) (2)
In the formula uconv(n) is duMA(n) convolution operation result.
The sixth stepConvolution calculation result u of velocity signal obtained in the fifth stepconv(n), determining the positions of all minimum values, namely the positions of wave troughs, and determining u between two adjacent minimum valuesconv(n) accumulating and summing the amplitudes, and taking the obtained result as the energy contained in the peak between two adjacent troughs, namely the energy of the peak area. In general, in a group uconvAnd (n) a plurality of minimum values exist, namely a plurality of peak region energies, and the maximum peak region energy is selected as a propeller fault characteristic value, so that a speed signal peak region energy fault characteristic value Eu is obtained. According to the same process, the energy fault characteristic value Ec of the control signal peak area can be obtained.
As shown in fig. 3, the present invention provides a wavelet decomposition scale determining and fault feature enhancing method based on time-frequency power spectral density maximum, which comprises the following specific steps:
firstly, determining an optimal wavelet decomposition scale based on a time-frequency power spectral density maximum value;
and secondly, enhancing the energy characteristics of the failure wave crest region of the propeller based on the optimal wavelet decomposition scale to obtain enhanced energy failure characteristics Eu and Ec of the speed signal wave crest region and the control signal wave crest region.
Example 1:
as shown in fig. 4, time domain waveforms of the underwater robot speed signal and the thruster control voltage signal in three operating states of the thruster are shown. And after the propeller breaks down, the actual output of the propeller is smaller than the theoretical output, and the output loss degree lambda of the propeller represents the failure degree of the propeller. In fig. 4, three sets of experimental result curves are counted, and each set of experimental process is: the target speed of the underwater robot is 0.3m/s, the underwater robot starts from a standstill, the speed is gradually increased, under the action of a closed-loop controller, the underwater robot tracks the upper target speed at a 100 th beat and stably operates at the target speed of 0.3m/s, and at a 250 th beat, the propellers respectively have propeller output loss faults with the fault degrees of lambda being 0%, lambda being 10% and lambda being 30% until the experiment is finished. In the experiment, the sampling frequency was 5 Hz.
And intercepting the speed signal data with the beat 101 to the beat 500 and the fault degree lambda being 30% in the figure 4 by adopting a time window, and performing wavelet decomposition on the data with the scales of 1-8 to obtain wavelet approximate components from the layer 1 to the layer 8, as shown in figure 5. From graphs (a) to (f), the singular behavior waveform caused by propeller failure in the speed signal is gradually apparent, and most pronounced in graph (f), as indicated by the dashed box in the graph. Starting from graph (g), the singular behavior waveform begins to weaken and diverge. In the graph (h) and the graph (i), the singular behavior waveform almost disappears. As can be seen from comparing fig. (a) to fig. (i), in the wavelet approximation component at layer 5 shown in fig. (f), the singular behavior waveform caused by propeller failure in the velocity signal is most significant, so the optimal wavelet decomposition scale should be 5.
The energy fault feature of the peak region is extracted from the wavelet approximate components of different decomposition scales shown in fig. 5, and the energy fault feature value of the peak region of the velocity signal under different wavelet decomposition scales is obtained, as shown in fig. 6. In fig. 6, the wavelet decomposition scale 0 represents the original velocity signal. As the wavelet decomposition scale increases, the peak region energy fault eigenvalue gradually increases and reaches a maximum at scale 8.
The known maximum characteristic value method determines the scale of the maximum value of the fault characteristic value as the optimal wavelet decomposition scale, so that the optimal wavelet decomposition scale is 8 according to the known maximum characteristic value method. However, as can be seen from fig. 5, in the wavelet approximation component of the layer 8 corresponding to the scale 8, the singular behavior waveform caused by the propeller fault in the velocity signal almost disappears, so the maximum value of the peak region energy corresponding to the scale 8 is not caused by the propeller fault, and it is not sufficient to use only the maximum value of the fault characteristic value as the basis for determining the wavelet decomposition scale.
The time-frequency power spectral density distributions of the wavelet approximation components of the different decomposition scales shown in fig. 5 were calculated, and the results are shown in fig. 7. From the graphs (a) to (f), the energy concentration region within the dashed box is substantially constant, while the high frequency portion in the time-frequency power spectral density distribution gradually decreases as the decomposition scale increases. Starting from graph (g), the energy concentration region within the dashed box starts to diverge, and this divergence phenomenon is more serious in graph (h) and graph (i).
Extracting the time-frequency power spectral density maximum value P corresponding to each decomposition scale in FIGS. 7(a) to 7(i)NAnd calculate PNAnd PN-1The results are shown in FIG. 8. This example is based on PN/PN-1And (5) less than or equal to 0.9, and determining M-N-1 as the optimal wavelet decomposition scale. In FIG. 8, P is from dimension 1 to dimension 5NAnd PN-1The ratios of (A) to (B) are all greater than 0.9. From dimension 5 to dimension 6, P6And P5The ratio of (a) to (b) is 0.878, which is less than 0.9, so the method determines that the optimal wavelet decomposition scale is 5. The best wavelet decomposition scale determined by the invention is consistent with the analysis result of fig. 5, which shows that the method for determining the best wavelet decomposition scale based on the time-frequency power spectral density maximum value is effective.
By length L1The time window of 400 intercepts the speed signal data of which the beat is 101 to 500 and the fault degree is lambda of 30% in fig. 4, and performs 5-scale wavelet decomposition on the data to obtain wavelet approximate components of the 1 st layer to the 5 th layer. And respectively extracting energy fault characteristic values of the peak region from the original speed signal and wavelet approximate components from the 1 st layer to the 5 th layer. And moving the time window backwards by one beat, extracting the speed signal data from beat 102 to beat 501, repeating the process, and extracting a group of new energy fault characteristic values of the peak area. And (4) continuously moving the time window backwards, and extracting a group of peak area energy fault characteristic values every time the time window is moved by a time beat. The time window is shifted backwards by 100 beats in total, and the extracted peak region energy fault characteristic value is shown in fig. 9.
As shown in fig. 9, the known peak region energy method extracts the peak region energy fault feature from the original velocity signal, and the present invention performs wavelet decomposition on the velocity signal based on the optimal wavelet decomposition scale, extracts the velocity signal wavelet approximate component, and then extracts the peak region energy fault feature from the wavelet approximate component.
As shown in fig. 9, the fault characteristic curve extracted by the known peak region energy method is at the bottom layer in the graph, while the fault characteristic curve extracted by the invention patent is always at the highest layer, i.e. is always larger than the fault characteristic extracted by the known peak region energy method and is larger than the peak region energy fault characteristic value extracted from the wavelet approximate components of the layers 1 to 4. The peak region energy characteristic enhancement method based on the optimal wavelet decomposition scale is effective.
Intercepting speed signal data with the beat number of 101-500 and the fault degree of lambda being 10% in the graph 4 by adopting a time window, performing wavelet decomposition on the speed signal data with the scales of 1-8 to obtain wavelet approximate components of layers 1-8, calculating the time-frequency power spectral density distribution of the wavelet components with different scales, and extracting the maximum value P of the time-frequency power spectral density corresponding to each decomposition scaleNAnd calculate PNAnd PN-1The results are shown in FIG. 10. According to the inventionN/PN-1And (5) less than or equal to 0.9, and determining M-N-1 as the optimal wavelet decomposition scale. In FIG. 10, P is from dimension 1 to dimension 4NAnd PN-1The ratios of (A) to (B) are all greater than 0.9. From dimension 4 to dimension 5, P5And P4The ratio of (a) to (b) is 0.890, which is less than 0.9, so the method of the present invention determines the optimal wavelet decomposition scale to be 4. The results of joint analysis of fig. 8 and fig. 10 show that the optimal wavelet decomposition scales determined by the present invention may be different according to the different propeller failure degrees.
With a length L1Capturing speed signal data of which the beat is 101-500, the fault degree is lambda 10% and lambda 30% in the time window of 400 in the graph 4, determining a decomposition scale by adopting a known maximum characteristic value method, then extracting energy fault characteristics of a peak region, and extracting the energy fault characteristics of the peak region by adopting the wavelet decomposition scale determination and fault characteristic enhancement method based on the time-frequency power spectral density maximum value; with a length L2And (3) intercepting control signal data of which the beat is 101-501 in the time window of 401 in fig. 4, the fault degree is lambda of 10% and lambda of 30%, determining a decomposition scale by adopting a known maximum characteristic value method, extracting energy fault characteristics of a peak region, and extracting the energy fault characteristics of the peak region by adopting the method. And continuously moving the time window L1 and the time window L2 backwards, and extracting a group of speed signal fault characteristics and control signal fault characteristics every time the time is moved by one time beat. The time window is moved backwards by 100 beats in total, and the extracted fault characteristics are distributed as shown in figure 11,As shown in fig. 12.
As shown in fig. 11, based on the fault features extracted by the known maximum eigenvalue method, the fault features corresponding to the fault degree of 10% and the fault features corresponding to the fault degree of 30% are mixed together and are difficult to distinguish. As shown in fig. 12, in the fault features extracted in this embodiment, the fault features corresponding to the same fault degree are grouped together, and the fault features corresponding to different fault degrees are far apart from each other and are easy to distinguish. Comparing fig. 11 and fig. 12, it is demonstrated that compared with the known maximum eigenvalue method, the extracted fault characteristics of the present invention are more favorable for propeller state monitoring and fault degree identification, that is, the wavelet decomposition scale determination and fault characteristic enhancement method based on the time-frequency power density spectrum maximum value of the present invention is effective.
Claims (2)
1. A method for determining an optimal wavelet decomposition scale based on a time-frequency power spectral density maximum value is characterized by comprising the following steps: the method comprises the following steps:
first, the length of L is adopted1Intercepting the original data of the speed signal of the underwater robot by using the time window function;
second, for the length L1The speed signal data is processed by conventional smooth pseudo-Wigner-Willi distribution operation to obtain a smooth pseudo-Wigner-Willi spectrum SPWVD (n, m), n is a time beat sequence number, n is an integer, and n ∈ [1L ]1]M is a number of frequency bands, m is an integer, and m ∈ [1N ]3]，N3The maximum value of the frequency band serial number; obtaining the absolute value of the smooth pseudo-Vigrener-Weili spectrum SPWVD (n, m), obtaining the time-frequency power spectral density distribution SPWVDA (n, m) of the original speed signal, and determining the maximum value P in the time-frequency power spectral density distribution SPWVDA (n, m) of the original speed signal0；
Third, for the length L1Performing N-layer conventional wavelet decomposition on the speed signal data to obtain an N-layer wavelet approximate component, calculating a smooth pseudo-Wigner-Weiley spectrum of the wavelet approximate component, solving an absolute value of the smooth pseudo-Wigner-Weiley spectrum to obtain time-frequency power density distribution, and determining a maximum value P of the time-frequency power density of the wavelet approximate component of the N layerN(ii) a N is a positive integer greater than 0;
the fourth step, judge if PN≤0.9PN-1If so, N-1 is the optimal wavelet decomposition scale and is recorded as M, namely M is N-1; if PN>0.9PN-1And if the N is equal to N +1, returning to the third step.
2. A failure feature enhancement method based on the optimal wavelet decomposition scale determination method of claim 1, characterized by: the method comprises the following steps:
acquiring dynamic signal data of a speed signal and a propeller control signal of an underwater robot, and intercepting the dynamic signal by adopting a sliding time window, wherein the intercepting length L of the speed signal is1Control signal intercept length L2＝L1+1；
Secondly, determining the optimal wavelet decomposition scale M for the speed signal data based on a time-frequency power density maximum method; for the propeller control signal, the control signal change rate is obtained based on a discrete difference method, and after difference, the data length of the control signal change rate in a time window is L1The difference formula is shown as formula (1);
wherein c (n) is the propeller control signal, n is the time beat number, n ∈ [1L ]1]And Δ t is the sampling period,
the control signal change rate of the nth beat;
step three, controlling the rate of change of the speed signal u (n) in the time window
Respectively performing conventional wavelet decomposition with the decomposition layer number being the optimal wavelet decomposition scale M to obtain the M-th layer wavelet approximate component u of the velocity signalMA(n), and control signal rate of change Mth layer wavelet approximation component
Step four, based on the conventional modified Bayes algorithm, the wavelet approximate component u of the velocity signal obtained in the step three is processedMA(n), and control signal rate of change wavelet approximation component
Processing to increase the amplitude of the singular behavior in the signal, the result of the processing being duMA(n) and dcMA(n)；
Step five, comparing the result d obtained in the step fouruMA(n) and dcMA(n) performing a convolution operation to concentrate the energy in the signal to obtain a convolution calculation result uconv(n) and cconv(n)；uconv(n) and cconv(n) the calculation process is the same, and the convolution calculation is shown as formula (2):
uconv(n)＝duMA(n)*duMA(n) (2)
in the formula uconv(n) is duMA(n) the convolution operation result;
step six, the convolution calculation result u of the speed signal obtained in the step fiveconv(n), determining the positions of all minimum values, namely the positions of wave troughs, and determining u between two adjacent minimum valuesconv(n) accumulating and summing the amplitude values, taking the obtained result as the energy contained in the wave peak between two adjacent wave troughs, namely the energy of the wave peak area, and then selecting the maximum value of the energy of the wave peak area as the fault characteristic value of the propeller so as to obtain the energy fault characteristic value Eu of the wave peak area of the speed signal; according to the same process pair cconvAnd (n) obtaining the energy fault characteristic value Ec of the peak region of the control signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001135.2A CN109633269B (en) | 2019-01-02 | 2019-01-02 | Wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum value |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001135.2A CN109633269B (en) | 2019-01-02 | 2019-01-02 | Wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum value |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109633269A CN109633269A (en) | 2019-04-16 |
| CN109633269B true CN109633269B (en) | 2020-10-16 |
Family
ID=66056471
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910001135.2A Active CN109633269B (en) | 2019-01-02 | 2019-01-02 | Wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum value |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109633269B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113780355B (en) * | 2021-08-12 | 2024-02-09 | 上海理工大学 | Deep convolution neural network learning method for fault identification of deep sea submersible propeller |
| CN113655778B (en) * | 2021-08-16 | 2024-03-26 | 江苏科技大学 | Underwater propeller fault diagnosis system and method based on time-frequency energy |
| CN117992836A (en) * | 2024-04-07 | 2024-05-07 | 华能澜沧江水电股份有限公司 | Power quality signal noise reduction method and device, electronic equipment and storage medium |
Family Cites Families (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08329046A (en) * | 1995-05-31 | 1996-12-13 | Toshiba Corp | Signal analyzing device using wavelet conversion |
| US6865492B2 (en) * | 2000-01-24 | 2005-03-08 | The Cielo Institute, Inc. | Algorithmic design of peptides for binding and/or modulation of the functions of receptors and/or other proteins |
| CN1601518A (en) * | 2004-10-22 | 2005-03-30 | 北京交通大学 | Road traffic stream data integration method and device based on signal processing |
| CN100590417C (en) * | 2008-11-20 | 2010-02-17 | 北京航空航天大学 | Plant blade physiologic index non-destructive test method |
| CN101936767B (en) * | 2010-09-27 | 2013-01-09 | 上海交通大学 | Method for extracting engineering machine running characteristic signals |
| CN102539150B (en) * | 2012-01-17 | 2014-07-16 | 电子科技大学 | Self-adaptive failure diagnosis method of rotary mechanical component based on continuous wavelet transformation |
| CN103217213B (en) * | 2013-02-21 | 2014-12-10 | 北京工业大学 | Modal parameter identification method based on response signal time-frequency joint distribution characteristics |
| US8958750B1 (en) * | 2013-09-12 | 2015-02-17 | King Fahd University Of Petroleum And Minerals | Peak detection method using blind source separation |
| CN105640505B (en) * | 2016-01-13 | 2018-06-19 | 昆明理工大学 | A kind of pulse signal random noise denoising method based on AR model spectra estimations |
| CN106019076B (en) * | 2016-05-16 | 2019-03-26 | 西安理工大学 | A kind of high voltage direct current transmission line fault location method |
| CN106404386A (en) * | 2016-08-29 | 2017-02-15 | 北京工业大学 | Method for acquiring, extracting and diagnosing early stage fault characteristic signals of gear case |
| CN107860958B (en) * | 2017-09-27 | 2020-05-26 | 江苏科技大学 | Noise reduction method for real-time monitoring current value of underwater robot propeller and real-time monitoring system thereof |
| CN108470018A (en) * | 2018-02-22 | 2018-08-31 | 中国铁道科学研究院 | Smoothing method and device based on the intrinsic mode functions that empirical mode decomposition decomposes |
| CN109033965B (en) * | 2018-06-22 | 2021-10-01 | 江苏科技大学 | Underwater robot propeller fault time-frequency feature enhancement method |
| CN108880604B (en) * | 2018-06-28 | 2021-05-14 | 河南机电职业学院 | Multi-component frequency modulation interference suppression method and device in spread spectrum communication system |
-
2019
- 2019-01-02 CN CN201910001135.2A patent/CN109633269B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109633269A (en) | 2019-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109633269B (en) | Wavelet decomposition scale determination and fault feature enhancement method based on time-frequency power spectral density maximum value | |
| CN106646406B (en) | Based on the outer trajectory velocity radar power spectrum detection method for improving wavelet threshold denoising | |
| CN110151175A (en) | Surface electromyogram signal noise-eliminating method based on CEEMD and improvement wavelet threshold | |
| CN103995950A (en) | Wavelet coefficient partial discharge signal noise elimination method based on related space domain correction threshold values | |
| CN106706320B (en) | A kind of Bearing Initial Fault Diagnosis method based on feedforward control accidental resonance | |
| CN105785330A (en) | Cognitive minor lobe interference suppression method | |
| CN110716532A (en) | Underwater robot propeller weak fault identification method based on wavelet packet energy and FFT | |
| CN108008187B (en) | Power grid harmonic detection method based on variational modal decomposition | |
| CN113702968A (en) | Life detection method and device | |
| CN110542926B (en) | Autonomous detection and suppression method for seismic data spike noise cluster | |
| CN115951328A (en) | Wind speed estimation method and device of wind lidar based on probability density constraint | |
| CN105785346A (en) | Unknown target line spectrum detection method based on phase variance weighting and system thereof | |
| Malik et al. | Automatic threshold optimization in nonlinear energy operator based spike detection | |
| CN103915102B (en) | Method for noise abatement of LFM underwater sound multi-path signals | |
| CN113655778A (en) | Underwater propeller fault diagnosis system and method based on time-frequency energy | |
| Gryllias et al. | A peak energy criterion (pe) for the selection of resonance bands in complex shifted morlet wavelet (csmw) based demodulation of defective rolling element bearings vibration response | |
| CN109765425B (en) | Pulse signal frequency analysis method based on time difference between peak values and energy weighting | |
| CN105652166B (en) | A kind of Weighted Threshold wavelet de-noising method for partial discharge on-line monitoring | |
| CN107908592B (en) | Background noise frequency division analysis method for hypersonic wind tunnel | |
| CN112817046B (en) | OBS data bubble suppression method and processing terminal | |
| RU2719626C1 (en) | Method for determining noise-generating object maneuver | |
| CN103487794B (en) | A kind of underwater reverberation suppression method based on wavelet package transforms | |
| Islam et al. | Frequency recognition of steady-state visually evoked potentials using binary subband canonical correlation analysis with reduced dimension of reference signals | |
| Qingxin et al. | Research on chaotic character of ship-radiated noise based on phase space reconstruction | |
| CN113376698B (en) | Low-missing-detection seismic data peak noise detection and suppression method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |