CN102230904A - Method for detecting oil applying rate of fiber - Google Patents
Method for detecting oil applying rate of fiber Download PDFInfo
- Publication number
- CN102230904A CN102230904A CN2011101723095A CN201110172309A CN102230904A CN 102230904 A CN102230904 A CN 102230904A CN 2011101723095 A CN2011101723095 A CN 2011101723095A CN 201110172309 A CN201110172309 A CN 201110172309A CN 102230904 A CN102230904 A CN 102230904A
- Authority
- CN
- China
- Prior art keywords
- fiber
- signal
- wavelet
- low
- magnetic resonance
- 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.)
- Pending
Links
Images
Landscapes
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention discloses a method for detecting the oil applying rate of fiber. The method comprises the following steps of: 1, acquiring a low-field nuclear magnetic resonance signal of standard fiber and performing linear fitting to obtain a calibration curve; 2, acquiring a low-field nuclear magnetic resonance signal of tested fiber and performing wavelet soft-threshold filtering processing to obtain a filtered reconstruction signal; and 3, substituting the filtered reconstruction signal into the calibration curve to obtain the oil applying rate of the fiber. According to the method, the detection accuracy of the oil applying rate of the fiber is increased, and the detection time is reduced to be less than 1 minute from the original 4 minutes.
Description
Technical field
The present invention relates to the fiber rate detection method that oils, particularly relate to a kind of based on the wavelet filtering technology, be applied to the method that the low-field nuclear magnetic resonance fiber rate that oils detects.
Background technology
Fiber oil the rate detection speed and as a result accuracy be very important in fiber production process.The low-field nuclear magnetic resonance detection technique makes shorten to several minutes by a few hours detection time in the oil application of rate context of detection of fiber, and accuracy, the stability of testing result also can be greatly improved simultaneously.The external interior market in this field is mainly by German Brooker company and England Oxford company monopolizing at present.Owing to relating to reasons such as trade secret and nuclear magnetic resonance bottom data are difficult to obtain, cause being difficult to find the document of associated fiber low-field nuclear magnetic resonance signal Processing aspect at home and abroad.Because the fiber oleaginousness is very low, reasons such as the low employed magnet field intensity of magnetic resonance is lower, add and influenced by terrestrial magnetic field, neighbourhood noise, noise of equipment etc., make the signal to noise ratio (S/N ratio) of the oil content signal of being gathered very low, the actual signal that detects shows as extremely faint signal, and common signal processing method is difficult to be applicable in the low-field nuclear magnetic resonance signal Processing of fiber.In actual applications, for faint oil content input is come out, need carry out hundreds of inferior stacks to signal collected usually, thereby improve signal collected signal to noise ratio (S/N ratio) with the influence of elimination random noise.But carry out hundreds of time stacks to signal collected, though can detect faint fiber oil content signal, fiber is oiled detection time of rate significantly increases (about 4 minutes), can not satisfy fiber manufacturing enterprise and will be controlled at requirement in the 1min detection time.
Summary of the invention
The objective of the invention is to overcome above-mentioned the deficiencies in the prior art, provide a kind of based on the wavelet filtering technology, be applied to the method that the low-field nuclear magnetic resonance fiber rate that oils detects, can under the prerequisite that guarantees detection accuracy, be shortened to 1 minute by 4 minutes detection time.
Technical solution of the present invention is as follows:
A kind of fiber rate detection method that oils is characterized in that this method comprises the steps:
The first step is gathered the low-field nuclear magnetic resonance signal of standard fibers and is obtained calibration curve as linear fit;
In second step, the low-field nuclear magnetic resonance signal of collecting test fiber is also done small echo soft-threshold Filtering Processing and is obtained filtered reconstruction signal;
In the 3rd step, the described calibration curve of reconstruction signal substitution after the described filtering is obtained the fiber rate that oils.
Described second step comprises the steps:
1. determine wavelet basis and wavelet basis exponent number: select only wavelet basis at signal characteristic;
2. determine to decompose number of times: find the decomposition number of times by repeatedly testing;
3. wavelet decomposition;
4. wavelet coefficient being carried out soft-threshold handles;
5. wavelet reconstruction.
Compared with prior art, beneficial effect of the present invention is as follows:
(1) little a lot, the mean value of the data fluctuations scope (maximal value subtract minimum value) of filtered data before than filtering more near actual value, mean square deviation is littler, degree of stability (mean square deviation/mean value) better.
(2) along with the increasing of stacking fold, the fluctuation range of (back) data can reduce before the fiber filtering on the same group, mean value is more near actual value, and the mean square deviation of data is littler after the filtering, degree of stability is better.
(3) mean value and the fiber real quality of 32 signals of stack have than mistake, and fluctuation range is bigger, filtered mean value and the fiber real quality is more approaching, mean square deviation is littler, degree of stability is also better, and the acquisition testing time is need 0.5 minute only.
(4) because noise effect, 64 signals of superposeing have bigger than mistake, fluctuation range with the fiber real quality before filtering.After filtering, the accuracy of signal and degree of stability have all had large increase, and mean value, the degree of stability of the original signal that obtains for 256 times with stack are all very approaching, and the acquisition testing time only needs 1 minute.
(5) in actual detected, 256 times the signal of superposeing can obtain meeting the result that the accuracy degree requires, and through after the filtering, the accuracy of signal can be further improved.
(6) before the filtering, signal need be superposeed 256 times, just can obtain the testing result that satisfies accuracy requirement in about 4 minutes consuming time, the resultant error of superpose 32 times and 64 times is all bigger; After the filtering, signal only need superpose 64 times, and about 1 minute consuming time, the close result that can obtain and superpose 256 times increased substantially detection speed.
Description of drawings
Fig. 1 is the oil process flow diagram of rate detection method of fiber of the present invention;
Fig. 2 is that fiber of the present invention oils and gathers standard fibers low-field nuclear magnetic resonance signal in the rate detection method and do the calibration curve process flow diagram;
Fig. 3 is that fiber of the present invention oils that collecting test fiber low-field nuclear magnetic resonance signal obtains filtered reconstruction signal process flow diagram in the rate detection method.
Fig. 4 is the fiber of the present invention small echo soft-threshold Filtering Processing process flow diagram in the rate detection method that oils;
Fig. 5 is the synoptic diagram that signal is carried out 1 wavelet decomposition;
Fig. 6 is the synoptic diagram that signal is carried out data storage method in 3 wavelet decomposition and the decomposable process;
Fig. 7 is the synoptic diagram that the data after the threshold process is carried out 1 wavelet reconstruction.
Embodiment
The invention will be further described below in conjunction with embodiment and accompanying drawing, but should not limit protection scope of the present invention with this.
Please consult Fig. 1 earlier, Fig. 1 is the oil process flow diagram of rate detection method of fiber of the present invention, and as described in Figure, a kind of fiber rate detection method that oils comprises step: the first step, and gather the low-field nuclear magnetic resonance signal of standard fibers and obtain calibration curve as linear fit; In second step, the low-field nuclear magnetic resonance signal of collecting test fiber is also done small echo soft-threshold Filtering Processing and is obtained filtered reconstruction signal;
In the 3rd step, the described calibration curve of reconstruction signal substitution after the described filtering is obtained the fiber rate that oils.
Fig. 2 is that fiber of the present invention oils and gathers standard fibers low-field nuclear magnetic resonance signal in the rate detection method and do the calibration curve process flow diagram, as seen from the figure, is that the signal of gathering standard fibers (fiber of the known rate that oils) stack 256 times is used for doing calibration curve.In actual detected, can use ready-made standard calibration curve, directly detect according to shown in Figure 3.
Fig. 4 is the fiber of the present invention small echo soft-threshold Filtering Processing process flow diagram in the rate detection method that oils, as seen from the figure, in second step, the low-field nuclear magnetic resonance signal of collecting test fiber is also done small echo soft-threshold Filtering Processing and is obtained filtered reconstruction signal and can be divided into 1. and to determine wavelet basis and wavelet basis exponent number; 2. determine to decompose number of times; 3. wavelet decomposition; 4. wavelet coefficient being carried out soft-threshold handles; 5. wavelet reconstruction five goes on foot totally.
1, determines wavelet basis and wavelet basis exponent number
Wavelet transformation is exactly the bank of filters of being made up of a low-pass filter and logical (high pass) wave filter of a series of band in essence.The character of low-pass filter and logical (high pass) wave filter of band is to be determined by the exponent number of selected wavelet basis and wavelet basis in the bank of filters, the wavelet basis kind is more, the character of having nothing in common with each other is being carried out in the filtering signal, select only wavelet basis at signal characteristic; The wavelet basis of the same race that exponent number is different, the ability of characterization signal local features is different, and the ability of the high more characterization signal of exponent number part is strong more, and it is big that calculated amount also can the phase strain.Through actual detected checking, adopt the sym8 wavelet basis that the low-field nuclear magnetic resonance signal of fiber is carried out wavelet decomposition and can obtain very desirable effect.
2, determine decomposition scale (number of times)
Free induction decay (Free Induction Decay, FID) the wavelet coefficient modulus maximum of signal increases along with the increase of wavelet transform dimension, and the modulus maximum of white noise reduces along with the increase of wavelet transform dimension.When noise is very strong in the FID signal, decompose number of times and want big, promptly the number of times of wavelet decomposition is more, but calculated amount is also can the phase strain big.Otherwise the number of times of wavelet decomposition will lack, and calculated amount also can correspondingly reduce.Therefore, certain signals and associated noises is often needed to find best decomposition number of times by repeatedly testing.
3, wavelet decomposition
The calculated amount of signal being carried out wavelet transformation is very big, and in order to improve computing velocity, the fast algorithm that the present invention adopts Mallat to propose is as the scale coefficient A that knows a certain decomposition scale
jAfter [k], can obtain the scale coefficient A of higher yardstick by further decomposing
J+1[k] and wavelet coefficient D
J+1[k]:
In actual computation, often directly use the sample sequence f (kT of signal f (t)
s) as the scale coefficient A of lowest scale
0[k].In the following formula, A
j[m] is the scale coefficient value at m place constantly of j decomposition scale.The process of signal being carried out a wavelet decomposition with computing machine as shown in Figure 5.
Fig. 5 is the process of a binary channels filtering (wavelet decomposition), h and g are the filter coefficient of wavelet transformation, can regard FIR(Finite Impulse Response as, finite impulse response) unit impulse response of digital filter, the filter coefficient that different wavelet basiss is corresponding different.The filter coefficient of the kind of wavelet basis and correspondence thereof is very many, can select only filter coefficient at different signals.H has low-pass characteristic, and g has high pass characteristic, and their output is the low-frequency approximation (scale coefficient) and the high frequency details (wavelet coefficient) of corresponding discrete signal respectively.Scale coefficient and wavelet coefficient total length are the twice of raw data, and information is redundant, need carry out two extractions, make their total length and the equal in length of original signal.In like manner, can be from A
1[k] further obtains the low-frequency approximation and the high frequency details of the wavelet decomposition second layer by this process.So repeat, can obtain the small echo expansion coefficient of a series of different scales by original signal f (t), Figure 6 shows that three wavelet decomposition and decompose the storage mode of back data, the active computer programming realizes in the process of wavelet decomposition, available storage of array data that equate with the raw data array size, add the data number that 1 storage of array is decomposed each yardstick of back with another size for the wavelet decomposition yardstick, the space complexity of algorithm is O (n).
4, wavelet coefficient being carried out soft-threshold handles
4.1 wavelet threshold filter method
The present invention adopts small echo soft-threshold filter method.Its basic thought is the wavelet coefficient of removing by a small margin, and the wavelet coefficient bigger to amplitude shrinks.
4.2 determining of soft-threshold
Estimation to threshold value when carrying out the filtering of small echo soft-threshold is very difficult, the generic threshold value that the present invention adopts Donoho to propose
, wherein
Be the variance of signal noise, N is the length of original signal.Can only obtain containing the signal of noise in the actual detected, can not obtain pure noise or purified signal, the variance of noise
Be unknown, can estimate it according to formula 3.
Use W
f(j, n) expression yardstick j goes up the wavelet transform of the signals and associated noises f of n place, position,
Expression W
f(j has eliminated the value behind the actual signal drastic change point in n), then in the formula (3)
N is the length of signal, and K is W
f(j, the n) number of middle actual signal drastic change point, g
n 0It is the Hi-pass filter coefficient of 0 yardstick.The variance of the noise of the process wavelet transformation on yardstick 1
Go up Noise Variance Estimation by the provable yardstick m of the decomposition texture of wavelet decomposition
In the following formula,
The norm of expression signal f (x), * represents convolution, h
n xThe low-pass filter coefficients of expression x yardstick.
5, wavelet reconstruction
Wavelet reconstruction can be described as inverse wavelet transform again, is the inverse process of wavelet decomposition.Mallat wavelet reconstruction algorithm can be expressed as with mathematical formulae:
Be illustrated in figure 7 as the process of carrying out a wavelet reconstruction.In like manner, can be with A
J-1With the D after threshold process
J-1Obtain A according to this step
J-2So repeat, can obtain the signal after the reconstruct.
The results showed, adopt the fiber of the present invention rate detection method that oils that the fiber low-field nuclear magnetic resonance signal of less collection stacking fold is carried out small echo soft-threshold Filtering Processing, under the prerequisite that does not influence accuracy of detection, improve the signal to noise ratio (S/N ratio) of signal, and increase substantially detection efficiency.The oil detection accuracy rate of rate of fiber is improved, and reduced to less than 1 minute by original 4 minutes detection time.
Claims (2)
1. fiber rate detection method that oils is characterized in that this method comprises the steps:
The first step is gathered the low-field nuclear magnetic resonance signal of standard fibers and is obtained calibration curve as linear fit;
In second step, the low-field nuclear magnetic resonance signal of collecting test fiber is also done small echo soft-threshold Filtering Processing and is obtained filtered reconstruction signal;
In the 3rd step, the described calibration curve of reconstruction signal substitution after the described filtering is obtained the fiber rate that oils.
2. the fiber according to claim 1 rate detection method that oils is characterized in that, described second step comprises the steps:
1. determine wavelet basis and wavelet basis exponent number: select only wavelet basis at signal characteristic;
2. determine to decompose number of times: find the decomposition number of times by repeatedly testing;
3. wavelet decomposition;
4. wavelet coefficient being carried out soft-threshold handles;
5. wavelet reconstruction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101723095A CN102230904A (en) | 2011-06-24 | 2011-06-24 | Method for detecting oil applying rate of fiber |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101723095A CN102230904A (en) | 2011-06-24 | 2011-06-24 | Method for detecting oil applying rate of fiber |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102230904A true CN102230904A (en) | 2011-11-02 |
Family
ID=44843492
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011101723095A Pending CN102230904A (en) | 2011-06-24 | 2011-06-24 | Method for detecting oil applying rate of fiber |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102230904A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103674993A (en) * | 2012-09-19 | 2014-03-26 | 五邑大学 | Method for measuring surface oil of natural protein fibers |
| CN106591971A (en) * | 2016-12-27 | 2017-04-26 | 南通醋酸纤维有限公司 | Device and method for testing oiling performance of cellulose acetate tow |
| CN109030533A (en) * | 2018-08-06 | 2018-12-18 | 苏州纽迈分析仪器股份有限公司 | Method that is a kind of while measuring chemical fibre regain and oil content |
-
2011
- 2011-06-24 CN CN2011101723095A patent/CN102230904A/en active Pending
Non-Patent Citations (3)
| Title |
|---|
| 宣一岷等: "用核磁共振法快速测定DTY含油率", 《化纤与纺织技术》 * |
| 张一鸣等: "低场脉冲核磁共振分析测量仪及其应用", 《现代科学仪器》 * |
| 郑传行等: "实验低场脉冲核磁共振仪数据接收与处理", 《现代科学仪器》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103674993A (en) * | 2012-09-19 | 2014-03-26 | 五邑大学 | Method for measuring surface oil of natural protein fibers |
| CN106591971A (en) * | 2016-12-27 | 2017-04-26 | 南通醋酸纤维有限公司 | Device and method for testing oiling performance of cellulose acetate tow |
| CN109030533A (en) * | 2018-08-06 | 2018-12-18 | 苏州纽迈分析仪器股份有限公司 | Method that is a kind of while measuring chemical fibre regain and oil content |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108345033A (en) | A kind of microseism signal time-frequency domain first arrival detection method | |
| CN107144879A (en) | A kind of seismic wave noise-reduction method combined based on adaptive-filtering with wavelet transformation | |
| CN102053276A (en) | Two-dimensional filtering method for a plurality of gathers of digital seismic signal | |
| CN109633761B (en) | Magnetic resonance signal power frequency noise reduction method based on wavelet transformation modulus maximum value method | |
| CN109142863B (en) | Power system frequency measurement method and system | |
| CN117009870B (en) | Pump cavitation state identification method for frequency domain improved SDP diagram | |
| CN113723171A (en) | Electroencephalogram signal denoising method based on residual error generation countermeasure network | |
| CN111650654A (en) | Ground magnetic resonance signal peak noise elimination method combining EMD and WT algorithms | |
| CN102230904A (en) | Method for detecting oil applying rate of fiber | |
| CN110987431B (en) | Bearing state monitoring and fault diagnosis method based on TQWT (TQWT-assisted SPC) | |
| CN110943954B (en) | Greedy algorithm based mixing sequence optimization method for modulation broadband converter | |
| CN105841867A (en) | Measuring method for tooth groove torque of permanent magnet motor | |
| CN105004363B (en) | Multi thresholds small wave sensor performance online test device and method are based under strong jamming | |
| CN110160778A (en) | Gearbox fault state identification method based on sequential hypothesis testing | |
| Zhang et al. | Sparsity-guided multi-scale empirical wavelet transform and its application in fault diagnosis of rolling bearings | |
| CN111623968A (en) | Fault feature extraction method based on adaptive morphological filtering | |
| CN114993671A (en) | Vibration fault diagnosis method and system based on Q factor wavelet transform | |
| CN110287853A (en) | A kind of Transient Signal Denoising based on wavelet decomposition | |
| CN103344988B (en) | Based on the vibroseis signal phase detection method that K-L decomposes | |
| CN113655534B (en) | Nuclear magnetic resonance FID signal noise suppression method based on multi-linear singular value tensor decomposition | |
| CN114398922A (en) | CNN-BilSTM-based fault diagnosis method for looseness of high-voltage shunt reactor winding | |
| CN113686577A (en) | Bearing fault diagnosis method based on rapid nonlinear sparse spectrum | |
| CN103376108B (en) | Star sensor noise reduction method | |
| CN112464855A (en) | While-drilling mud positive pulse signal processing method and device based on EEMD | |
| He et al. | Sparsity-assisted signal representation for rotating machinery fault diagnosis using the tunable Q-factor wavelet transform with overlapping group shrinkage |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20111102 |