CN101126744A - Ceramic carrier embrittlement detection method in ternary catalytic converter assembly - Google Patents

Abstract

The utility model provides a detection method of embrittlement cracking of ceramics carrier in a three way catalytic assembly in acoustic signal detection technical field. The method comprises steps as follow: the likelihood frames detection of embrittlement cracking signal is processed in signal-to-noise ratio threshold value method. Utilizing discrete wavelet transformation in the detected likelihood frames, the genesis and discontinuity of short time characteristic acoustic signal is detected and located in time domain. In the symmetry neighbor region of discontinuity signal peak value in set rated time limit, the signal of acoustic frames is processed with noise reduction using strict stein unbiased likelihood estimation algorithm. Then the fundamental frequency of the acoustic signal and the cepstrum envelope of discontinuity signal are calculated. The cepstrum envelope difference value of the likelihood signal is matched with the cepstrum envelope difference value template of the characteristic acoustic signal in local database according to the real-time matching method of cepstrum envelope difference value. Then the similar degree between the likelihood acoustic signal and the standard acoustic signal of embrittlement cracking can be assessed. In the condition of the threshold of the characteristic matching degree between the likelihood signal and the standard acoustic signal of embrittlement cracking is fixed as 0.85, the correct detection possibility of the invention is up to 94% and the possibility of false alarm is less than 5%.

Description

Ceramic carrier embrittlement detection method in the ternary catalytic converter assembly

Technical field

The present invention relates to the method in a kind of detection technique field, relate in particular to ceramic carrier embrittlement detection method in a kind of ternary catalytic converter assembly.

Background technology

The vehicle exhaust three unique catalytic converter is mounted in the outer waste gas purification apparatus of machine in the gas outlet, and the automobile that three-way catalytic converter is housed can be converted to non-toxic gas with 90% above harmful gas in the tail gas and enter atmosphere.The triple mode catalytic converter assembly is made of housing, buffer layer, catalyst support and catalyzer four parts, wherein the general whole porous honeycomb ceramic body that adopts salic coating of carrier.On the ternary catalytic converter assembly production line, non-plastic fracture can take place in ceramic monolith in binding, the pinpoint welding procedure.Binding, in the spot-welding technology ceramic monolith embrittlement can take place is under the multiplex (MUX) plants parameter (holding force, seizing force etc.) effect, multivariate coupling and a large amount of process of uncertain factor are at random arranged, the characteristics that also have himself simultaneously: embrittlement occurs in the housing, and being in closed state can't observe; Product is metal shell and porous medium class Ceramic Composite goods, and conventional nondestructiving detecting means lost efficacy; The embrittlement time of origin is extremely short, and signal extraction is difficulty relatively; Ground unrest complexity under the operating mode, it is big to detect the identification difficulty, be the focal issue that scientific research technician in various countries' pays close attention to for many years always, international existing detection technique scheme is that product is surveyed sample, the not only time-consuming cost height of this scheme, and checking efficiency is low, can't accurately judge for the embrittlement problem that has certain randomness.

Find through literature search prior art, for the cracked detection problem of pottery, document " with the thermal shock damage of acoustic emission monitor(ing) stupalith " (Chinese pottery, 2001 (5), 37:pp34-36) damage strength and crack propagation process when utilizing acoustic emission to detect the stupalith Quench have been introduced, though the method can detect the characteristics of Acoustic Emission in the ceramic body rupture process, but when this method is used for three unique catalytic converter production line embrittlement detection, face three big difficulties: it is short and the sound intensity is indefinite that first ceramic monolith can take place by the embrittlement time, levies unified model and the standard of ceramic embrittlement incident shortage with index attenuation sine V wave table; It two is that acoustical signal can't be located, so the embrittlement likelihood signal can't accurately be extracted; Its three be the method only but can't be indefinite to sound intensity difference, duration to detect acoustical signal, the unconspicuous likelihood signal of zero passage frequecy characteristic carries out identification; Therefore the generation of on-line real time monitoring ceramic carrier embrittlement phenomenon must find a new way.

Summary of the invention

The present invention is directed to the deficiencies in the prior art, ceramic carrier embrittlement detection method in a kind of ternary catalytic converter assembly is provided, at first analyze waveform, short-time energy and the cepstrum envelope difference of voice signal sample under the different situations when making its actual use, cepstrum envelope template is carried out matching operation determining whether ceramic monolith embrittlement takes place in characteristic frequency section cepstrum envelope difference when extraction is broken and taken place and the database,, low cost realization simple, efficient with the method for time domain and the frequency domain combination detection of breaking.

The present invention is achieved by the following technical solutions, the present invention is directed to the feature of sudden change acoustical signal under the complex background noise floor, a kind of method based on time-frequency domain compound detection and identification has been proposed, effectively detect under the complicated noise background situation, the location, extract and finally pick out the embrittlement signal, thereby reach the purpose of rejecting waste product.

The present invention is achieved by the following technical solutions, the present invention includes following steps:

The first step, adopt the snr threshold method to carry out embrittlement signal likelihood frames and detect:

Second step, embrittlement acoustical signal peak point time domain is accurately located in the likelihood frames, in the detected likelihood frames of the first step, utilizes the wavelet transform technology accurately to detect the generation and the catastrophe point of short-time characteristic acoustical signal, and it is located in time domain, for subsequent step provides signal in the focus neighborhood;

The 3rd step, in the jump signal peak value symmetric neighborhood of setting the specified time limit, adopt Rigorous SURE (strict Si Tanyin does not have partial likelihood and estimates) algorithm that the audio frame signal de-noising is handled, calculate acoustical signal fundamental frequency and jump signal cepstrum envelope then;

The 4th step, according to the real-time matching process of cepstrum envelope difference diagnostic acoustic signal cepstrum envelope difference template in likelihood signal cepstrum envelope difference and the local data base is carried out matching operation, can judge the similarity degree of likelihood acoustical signal and standard embrittlement acoustical signal.

Described employing snr threshold method is carried out embrittlement signal likelihood frames and is detected, and is meant: short, faint sound intensity signal when belonging to because of the signal that breaks, by the sound detection hardware unit acoustical signal in the production run is sampled.Acoustical signal has the advantages that the sound intensity is higher than ground unrest owing to break, so can according under the different situations (as motor start suddenly, metal object bump) waveform, the short-time energy of voice signal sample, obtain time domain sound intensity snr threshold through statistical computation, set snr threshold parameter θ herein _SNR=10db, this threshold value is a little more than ground unrest.To signal to noise ratio (S/N ratio) in the audio signal frame in short-term greater than θ _SNRFrame be judged to be the characteristic signal likelihood frames, likelihood frames has the time domain sudden change feature of signal to be detected, can effectively distinguish ground unrest and sudden change acoustical signal, short signal when this step is intended to detect the sound intensity a little more than the burst of ground unrest.

Embrittlement acoustical signal peak point time domain accurately is meant the location in the described likelihood frames: in feature jump signal Singularity Detection and location, use infinitely smooth, infinite time can little Mexico straw hat small echo decomposition and reconstruction original signal under different scale, thereby accurately detect and locate the time domain catastrophe point that likelihood signal takes place under the complicated noise floor.The present invention selects for use Mexico's straw hat (Mexican hat) small echo first-harmonic to generate continuous wavelet family of functions, this process is abandoned Mallat commonly used (horse traction spy) discrete wavelet transformer scaling method, directly continuous wavelet family of functions discretize is obtained the discrete wavelet base later on, therefore having avoided every in the Mallat algorithm decomposes and need data be reduced by half once basic two extractions through scalping, can carry out the wavelet inverse transformation signal reconstruction on the wavelet scale space arbitrarily with this wavelet basis, make that sign mutation point time domain location is more accurate.

Described cepstrum envelope difference template characteristic vector, be meant: in the jump signal peak value neighborhood of setting the specified time limit, adopt harmonic wave auto-correlation algorithm and MFCC (Mel cepstrum) to calculate acoustical signal fundamental frequency and jump signal cepstrum respectively, calculate envelope envelope difference up and down then; In the n that presets a responsive frequency range interval envelope difference discrete value is asked on average, the n dimensional feature vector that the back constitutes is cepstrum envelope difference template characteristic vector.

Described cepstrum envelope difference template is mated, be meant: according to cepstrum envelope Real Time Matching Algorithm envelope difference template about the diagnostic acoustic signal cepstrum envelope in envelope difference about the likelihood signal cepstrum envelope and the local data base is carried out matching operation, check whether measured jump signal is diagnostic acoustic signal to be detected; Before using, system in local data base, sets up diagnostic acoustic signal n rank cepstrum envelope difference template characteristic vector earlier by test, to be checked measure likelihood signal finish the first step and second the step obtain surveying likelihood acoustical signal n rank cepstrum envelope difference template characteristic vector after, diagnostic acoustic signal n rank cepstrum envelope difference template characteristic vector and actual measurement likelihood acoustical signal n rank cepstrum envelope difference template characteristic vector are obtained cepstrum envelope difference value tag matching degree according to the cosine law comparison, provide the affirmation that likelihood signal is a diagnostic acoustic signal according to the matching degree setting threshold, still be the final decision result of false alarm.

Short signal is usually expressed as the class pulse signal with main peak value during embrittlement on time domain, and all kinds of jump signal temporal signatures that various enchancement factors cause are not obvious, and whether can't carry out likelihood signal by the time-domain analysis means is the judgement of characteristic signal; Simultaneously, Chang Gui time-frequency combination analytical approach (as wavelet analysis) also because the time short signal frequency domain components complexity, different frequency range energy have undulatory property and lost efficacy; The embrittlement acoustical signal is the sudden change peak signal, its spectral range is almost wide with noise background, but what be different from background noise is, comparatively concentrates at Low Medium Frequency section embrittlement energy, cepstrum envelope difference amplitude is bigger, can be used as and distinguishes after characteristic spectra and standard embrittlement acoustical signal contrast.Cepstrum envelope difference can concentrated expression different frequency range signal energy distribute, and short signal is at the frequency domain character of particular low frequency section when being fit to detect.Thereby, under the prerequisite that detects acoustical signal appearance sudden change, can judge further whether jump signal is desired characteristic signal.N rank cepstrum envelope difference template characteristic vector C1n in the embrittlement occurrence characteristics acoustical signal characteristic spectra is charged to (experimental result shows n desirable 0～19 totally 20 rank coefficients) in the local data base, and actual measurement likelihood acoustical signal n rank cepstrum envelope difference value tag vector is designated as C2 _n, according to the matching degree of cosine sciagraphy (1) formula calculating with cepstrum envelope difference template

$\mathrm{\&rho;}=\frac{<{C1}_n,{C2}_n>}{\left|{C1}_n\right|\left|{C2}_n\right|}=\frac{{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A20071004646100111.png"\; state="\{\{state\}\}">C</figure-callout>}1}_1{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A20071004646100111.png"\; state="\{\{state\}\}">C</figure-callout>}2}_1+{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A20071004646100111.png"\; state="\{\{state\}\}">C</figure-callout>}1}_2{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A20071004646100111.png"\; state="\{\{state\}\}">C</figure-callout>}2}_2+L{C1}_n{C2}_{\mathrm{<figure-callout\; id="1"\; label="n\; C"\; filenames="A20071004646100111.png"\; state="\{\{state\}\}">n</figure-callout>}}}{\sqrt{{C_{}}^{}}}---\left(1\right)$

Determine matching degree threshold value ρ according to the matching degree of likelihood signal cepstrum envelope difference and masterplate _Th(being set at 0.866 in the actual production) is if actual computation matching degree ρ 〉=ρ _Th, then thinking has characteristic signal to produce, and is judged to be three unique catalytic converter ceramic monolith generation embrittlement, otherwise is the false-alarm signal.

Compared with prior art, the employing burst sound signal cepstrum envelope difference value tag matching method that proposes among the present invention, can under strong noise background, detect and pick out the embrittlement acoustical signal, can be sensitiveer detect acoustical signal sudden change time of origin point, have in fiducial probability (matching degree) mode to judge whether draw institute's detection signal serves as the function of expectation characteristic signal simultaneously, can finish real-time online and detect the embrittlement problem that takes place in the three unique catalytic converter ceramic monolith assembly manufacture process.Evidence is under 0.85 condition setting likelihood signal and standard embrittlement acoustical signal characteristic matching degree threshold value, correct detection probability reaches more than 94%, false alarm probability the invention enables three unique catalytic converter assembly Assembling Production efficient and product quality reliability to significantly improve, produce and detect cost and declines to a great extent less than 5%.

Description of drawings

Fig. 1 is a detection method schematic flow sheet of the present invention;

Fig. 2 be contain ground unrest, ironware impact noise and ceramic monolith break sound the original signal synoptic diagram;

Fig. 3 is that mexican hat wavelet 6 yardsticks of signal decompose and catastrophe point location gray scale synoptic diagram;

Fig. 4 is a focus neighborhood signal intercepting synoptic diagram;

Fig. 5 is the noise reduction process synoptic diagram;

Fig. 6 is signal cepstral analysis synoptic diagram in 100 milliseconds of focus neighborhoods of sudden change peak value

Wherein: Fig. 6 a is the signal cepstrum, and Fig. 6 b calculates for the cepstrum envelope; Fig. 6 c is a cepstrum envelope difference up and down;

Embodiment

Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

Described sound detection device comprises CHZ-16 type condenser type electret acoustic sensor, transmitter and signal regulating panel, the DAQCard-1200 data collecting card of NI company, based on the industrial control computer monitoring platform of LabVIEW-RT.Present embodiment is the production line of the stainless steel outer packed housing of a cylindric vehicle exhaust ternary catalyzing unit pottery inner core, and purpose causes inner core cracked when being to detect in packing Stainless Steel Shell process.

Whether the present embodiment technology path is to have the acoustical signal of breaking whether to take place to judge to breaking by detecting in the packaging process.As shown in Figure 1, concrete implementation step is as follows:

1. because break signal short, faint sound intensity signal when belonging to, typical temporal signatures is: by break first big amplitude sound pulse producing and forming because of some high and low frequency signal combination of crack advance and fragment extruding generation following closely of hard brittle material.To acoustical signal sampling in the production run, sample frequency is 8000 hertz of monophonys by hardware device.The audio frame width is 200 milliseconds, and interframe is overlapping to be 60 milliseconds.Every frame is that running mean is done by unit with 5 milliseconds, is higher than the characteristics of ground unrest according to the acoustical signal sound intensity that breaks, and sets snr threshold parameter θ _SNR=10db, this threshold value a little more than ground unrest to signal to noise ratio (S/N ratio) in the audio signal frame in short-term greater than θ _SNRFrame be judged to be the characteristic signal likelihood frames, likelihood frames has the time domain sudden change feature of signal to be detected, accompanying drawing 2 is the detected likelihood frames of carrying noise of noise threshold method, can find out intuitively among the figure has the sound intensity amplitude of two places apparently higher than ground unrest, therefore has reason to believe the ceramic carrier embrittlement phenomenon takes place in this signal acquisition process probably; But there are tangible difference its peak value of likelihood signal, duration in the likelihood frames, and whether the two is real ceramic embrittlement signal must further be calculated and discern.

2. the accurate positional mutation point of wavelet transform method: infinitely smooth, infinite time of Mexico's straw hat small echo can be little, so it is not to independent noise spot sensitivity; According to its unique time domain character, it can manifest the exaggeration that jump signal carries out the caricature formula, makes that the sudden change unique point that comprises information is outstanding especially, and sudden change peak value singular point is had good positioning analysis and precision analysis.Utilize wavelet analysis positional mutation point to detect effect, as shown in Figure 3, provided the positioning analysis result of two sudden change acoustical signals (peak value singular point) employing Mexican hat wavelet function, transverse axis is the time, and the longitudinal axis is that yardstick represents that mexican hat wavelet 6 yardsticks of signal decompose.The white stripes that arrow is pointed out among Fig. 3 is promptly decomposed the maximum wavelet coefficient in back, can accurately determine cracked time and frequency range thus.Online in real time is handled needs to set a threshold value in program, then wavelet coefficient that can value is bigger and corresponding time thereof and yardstick are differentiated out, and wavelet coefficient gray-scale map herein is the explanation synoptic diagram.

3. the acoustical signal time domain intercepts and noise reduction process: in order to reduce the influence of noise to cracked feature, after determining the time of sudden change acoustical signal peak value generation, again to (pressing the broken feature of hard brittle material in 100 milliseconds of scopes near the catastrophe point, get the preceding 10ms of catastrophe point, get 90ms afterwards) sample sequence carry out denoising Processing, denoising Processing adopts Rigorous SURE algorithm.Fig. 4 is figure (Fig. 5) after 100 milliseconds of section interceptings of focus neighborhood signal (Fig. 4) and the noise reduction process, because Fig. 4 signal is subjected to the pollution of ambient noise signal, the small embrittlement acoustical signal of reflection crack advance is covered by noise, so the embrittlement acoustical signal cepstrum feature of non-filtered noise reduction process and the difference of unimodal value likelihood signal are not obvious; By adopting the later signal of Rigorous SURE algorithm noise reduction among Fig. 5 as can be seen, white noise signal can obviously be observed the small size decay concussion in the acoustical signal peak value field substantially by filtering.

4. calculate likelihood signal cepstrum, cepstrum envelope and cepstrum envelope envelope difference up and down, its result as shown in Figure 6, Fig. 6 a is the signal cepstrum, what cepstrum characterized is the power spectrum of log power spectrum, therefore see to have from the signal energy viewpoint peak signal is suppressed the feeble signal amplification, help launching analysis characteristic signal; Fig. 6 b is a cepstrum envelope result of calculation, and the cepstrum envelope has two envelopes, and envelope is represented the peak value and the valley of cepstrum respectively up and down; Fig. 6 c is a up and down envelope difference of cepstrum, and the envelope difference has been carried the disconnected energy information of characteristic frequency, so the feature that the envelope difference can be used as the embrittlement signal is done also identification of coupling with likelihood signal.

5. cepstrum envelope difference template matches: get up and down the envelope difference [0,30], [160,190], [290,320] average in three intervals is as embrittlement acoustical signal eigenvector (0.05 0.06 0.06), and the difference masterplate eigenvector (0.05 0.03 0.06) of three characteristic intervals calculates matching degree 0.9582 by the similarity formula.

6. determine matching degree threshold value ρ according to the matching degree of likelihood signal cepstrum envelope difference and masterplate _Th(being set at 0.866 in the actual production) is if actual computation matching degree ρ=0.9852 〉=ρ _Th, then thinking has characteristic signal to produce, and is judged to be three unique catalytic converter ceramic monolith generation embrittlement.

Experimental results show that setting likelihood signal and standard embrittlement acoustical signal characteristic matching degree threshold value be under 0.866 condition, the actual computation matching degree is assert three unique catalytic converter ceramic monolith generation embrittlement greater than characteristic matching degree threshold value; The scene is opened housing and is detected ceramic monolith, and it is cracked to observe ceramic monolith, and practice examining result detects judged result with the present invention and conforms to, thereby has verified validity of the present invention.

Claims (6)

1. ceramic carrier embrittlement detection method in the ternary catalytic converter assembly is characterized in that, may further comprise the steps:

The first step, adopt the snr threshold method to carry out embrittlement signal likelihood frames and detect:

Second step, embrittlement acoustical signal peak point time domain location in the detected likelihood frames of the first step, utilizes the generation and the catastrophe point of wavelet transform technology for detection short-time characteristic acoustical signal in the likelihood frames, and it is located in time domain, for subsequent step provides signal in the focus neighborhood;

The 3rd step, in the jump signal peak value symmetric neighborhood of setting the specified time limit, adopt strict Si Tanyin not have the partial likelihood algorithm for estimating audio frame signal de-noising is handled, calculate acoustical signal fundamental frequency and jump signal cepstrum envelope then;

The 4th step, according to the real-time matching process of cepstrum envelope difference diagnostic acoustic signal cepstrum envelope difference template in likelihood signal cepstrum envelope difference and the local data base is mated, judge the similarity degree of likelihood acoustical signal and standard embrittlement acoustical signal.

2. ceramic carrier embrittlement detection method in the ternary catalytic converter assembly according to claim 1, it is characterized in that, described employing snr threshold method is carried out embrittlement signal likelihood frames and is detected, be meant: short, the faint sound intensity signal when signal that breaks belongs to, by the sound detection hardware unit acoustical signal in the production run is sampled, according to waveform, the short-time energy of voice signal sample, obtain time domain sound intensity snr threshold through statistical computation, set snr threshold parameter θ herein _SNR=10db, this threshold value is higher than ground unrest, to signal to noise ratio (S/N ratio) in the audio signal frame in short-term greater than θ _SNRFrame be judged to be the characteristic signal likelihood frames.

3. ceramic carrier embrittlement detection method in the ternary catalytic converter assembly according to claim 1, it is characterized in that, embrittlement acoustical signal peak point time domain location in the described likelihood frames, be meant: in feature jump signal Singularity Detection and location, use infinitely smooth, infinite time can little Mexico straw hat small echo decomposition and reconstruction original signal under different scale, Mexico's straw hat small echo first-harmonic generates continuous wavelet family of functions, directly continuous wavelet family of functions discretize is obtained the discrete wavelet base later on, carrying out the wavelet inverse transformation signal reconstruction on the wavelet scale space arbitrarily with this wavelet basis, make that sign mutation point time domain location is more accurate, thereby detect and locate the time domain catastrophe point that likelihood signal takes place under the complicated noise floor.

4. ceramic carrier embrittlement detection method in the ternary catalytic converter assembly according to claim 1, it is characterized in that, described calculating acoustical signal fundamental frequency and jump signal cepstrum envelope, be meant: in the jump signal peak value neighborhood of setting the specified time limit, adopt harmonic wave auto-correlation algorithm and Mel cepstrum to calculate acoustical signal fundamental frequency and jump signal cepstrum respectively, then calculate envelope envelope difference up and down, in the n that presets a responsive frequency range interval envelope difference discrete value is asked on average, the n dimensional feature vector that the back constitutes is a cepstrum envelope difference template characteristic vector.

5. ceramic carrier embrittlement detection method in the ternary catalytic converter assembly according to claim 1, it is characterized in that, described cepstrum envelope difference template is mated, be meant: in local data base, set up diagnostic acoustic signal n rank cepstrum envelope difference template characteristic vector earlier before system uses by test, to be checked measure likelihood signal finish the first step and second the step obtain surveying likelihood acoustical signal n rank cepstrum envelope difference template characteristic vector after, diagnostic acoustic signal n rank cepstrum envelope difference template characteristic vector and actual measurement likelihood acoustical signal n rank cepstrum envelope difference template characteristic vector are obtained cepstrum envelope difference value tag matching degree according to the cosine law comparison, provide the affirmation that likelihood signal is a diagnostic acoustic signal according to the matching degree setting threshold, still be the final decision result of false alarm.

6. ceramic carrier embrittlement detection method in the ternary catalytic converter assembly according to claim 1 or 5, it is characterized in that, describedly provide the affirmation that likelihood signal is a diagnostic acoustic signal according to the matching degree setting threshold, still be the final decision result of false alarm, be specially: n rank cepstrum envelope difference template characteristic vector Cln in the embrittlement occurrence characteristics acoustical signal characteristic spectra is charged in the local data base, n gets 0～19 totally 20 rank coefficients, and actual measurement likelihood acoustical signal n rank cepstrum envelope difference value tag vector is designated as C2 _n, calculate matching degree with cepstrum envelope difference template according to the cosine sciagraphy:

$\mathrm{\&rho;}=\frac{\&lang;{C1}_n,{C2}_n\&rang;}{\left|{C1}_n\right|\left|{C2}_n\right|}=\frac{{C1}_1{C2}_1+{C1}_2{C2}_2+L{C1}_n{C2}_n}{\sqrt{{{C1}_1}^2+{{C1}_2}^2+L{{C1}_n}^2}\&CenterDot;\sqrt{{{C2}_1}^2+{{C2}_2}^2+L{{C2}_n}^2}}$

Determine matching degree threshold value ρ according to the matching degree of likelihood signal cepstrum envelope difference and masterplate _Th, ρ _ThBe set at 0.866, if actual computation matching degree ρ 〉=ρ _Th, then thinking has characteristic signal to produce, and is judged to be three unique catalytic converter ceramic monolith generation embrittlement, otherwise is the false-alarm signal.

Images