CN101017128A - Analysis method for localized corroding based on electrochemistry noise - Google Patents

Abstract

This invention provides one local erosion analysis method and device based on electrical and chemical noise, wherein, the device adopts four electrode detector to overcome three electrode for one measurement and generates two level signals and one current signal; the stored original noise data are computing each section through software; then integrating all analysis functions to sort all parameters; according judgment equation to analyze and give out the erosion status results.

Description

Local corrosion analytical approach and device based on electrochemistry noise

Technical field

The present invention relates to a kind of monitoring technology of metallic material corrosion, especially a kind of local corrosion monitoring device and analytical approach.

Background technology

Corrosion of Metallic Materials lost efficacy and had caused very enormous economic loss to countries in the world, also may bring serious society and environmental disaster simultaneously.The loss of corrosion failure more than 90% is to be caused by local corrosion, but existing on-the-spot corrosion on-line monitoring technique, as gravimetric method, linear polarization technique and resistance probe etc., all can only the uniform corrosion and the corrosiveness of the environment of metal material be monitored, and be difficult to realize the monitoring of local corrosion.Therefore the technology of setting up effectively monitoring and prevention and control local corrosion has great reality and economic implications.The electrochemistry noise monitoring technology can not only obtain the local corrosion information of metal material, and has and test system is not had any adding disturb and can really realize advantage such as continuous monitoring, has been subjected to attention more and more widely at home and abroad.

Aspect the electrochemistry noise monitoring technology, the common multiple electrode array formula probe acquisition noise data of using three electrode catheters or same metal material to make, yet this type of probe can only obtain the potential fluctuation between the working electrode of working electrode coupling back that two same metal materials make and truly potential fluctuation between the contrast electrode or coupling and the pseudo-contrast electrode that the 3rd metal material of the same race made, and can not obtain above-mentioned two kinds of potential fluctuation signals simultaneously.The pseudo-contrast electrode that metal material of the same race is made is easy to make and characteristics such as simple to operate are widely used in the industrial corrosion field monitoring because of having, but because this kind reference electrode can't be measured real corrosion potential, in the process of measuring the noise current potential, produced unpredictable potential drifting simultaneously because of constantly corroding, so design is a kind of stable in corrosion probe, the true contrast electrode that is difficult for being etched is monitored true corrosion potential, and the otherness of the noise current potential that two kinds of reference electrode measurements are obtained compares analysis simultaneously, and this overall status and local corrosion to system's corrosion takes place, the monitoring of evolution has important effect.

About the analytic method of electrochemistry noise signal, mainly comprise time domain statistical study, frequency-domain analysis and wavelet analysis in the bibliographical information at present.Wherein, temporal analysis is most widely used, but the selection of its analytical parameters and verification method still are in conceptual phase; The time-frequency switch technology deficiency of frequency-domain analysis, the error of calculation is big, and influenced obviously by human factor; And be in the desk study stage based on the noise analysis technology of wavelet transformation.The problems referred to above make the widespread use of electrochemistry noise monitoring technology in the industry spot corrosion monitoring be subjected to significant limitation.At present, the monitoring device that has only the transient state method for feature analysis that adopts time domain statistical study, frequency analysis and noise to resolve the electrochemistry noise data emerges, and as yet also will be based on the local corrosion cluster of electrochemistry noise, the report that discriminant analysis method is used for monitoring device relevant for using four electrode catheters to monitor two kinds of electric potential signals simultaneously.

Summary of the invention

The objective of the invention is to set up a kind of local corrosion monitoring device and analytical approach based on electrochemistry noise.Corrosion monitor of the present invention adopts a kind of four electrode catheters, overcome that three electrode catheters can not record true corrosion potential and the noise current potential that obtains in contain the shortcoming of the dc shift of introducing by pseudo-contrast electrode at random, can collect two kinds of electric potential signals and a kind of current signal simultaneously, and be provided with the automatic segmentation raw data, calculate the software of all statistics of each section, cluster analysis and discriminatory analysis.This device is a kind of simple to operate, local corrosion monitoring device and analytical approach based on electrochemistry noise that the analysis and distinguishing precision is high.

This metallic material local corrosion monitor comprises: by four electrode catheters that electrode I, electrode II, pseudo-contrast electrode III and true contrast electrode IV form, impedance transformer, DC compensator, binary channels chopper-zero-stabilized operational data amplifier, zero resistance galvanometer, hyperchannel low-pass filter, A/D converter, data-storage system and embedded microprocessor (ARM) or digital signal processor (DSP);

Wherein four electrode catheters are used to produce original electrochemistry noise current potential and current signal, and impedance transformer is used to reduce the impedance of the true contrast electrode IV and the noise current potential that pseudo-contrast electrode III is produced of four electrode catheters; DC compensator is used to eliminate the DC level part of the noise current potential that true contrast electrode IV gathers, and is sent to the hyperchannel low-pass filter after binary channels chopper-zero-stabilized operational data amplifier amplifies the pseudo noise current potential that comes from electrode III and the true noise electric potential signal that comes from electrode IV respectively and carries out filtering; The zero resistance galvanometer detects from the noise current signal of the electrode I of four electrode catheters and electrode II and is sent to the filtering of hyperchannel low-pass filter; A/D converter is converted to noise current potential and noise current digital signal and is stored in data-storage system; Last embedded microprocessor (ARM) or digital signal processor (DSP) carry out statistical computation, cluster analysis and discriminatory analysis to noise current potential and noise current data, finish the analysis of unknown noise data and provide the judged result of etch state.

The electrode I of described composition four electrode catheters, electrode II, pseudo-contrast electrode III are that metal material of the same race is made, true contrast electrode IV is Ag/AgCl solid reference electrode IV, and this electrode is the Ag/AgCl solid reference electrode that has wrapped up the saturated potassium chloride of one deck densification and the film that macromolecule resin mixes on Ag/AgCl solid reference electrode surface.

A kind of analytical approach of the local corrosion based on electrochemistry noise is implemented by the local corrosion monitoring device, and this method may further comprise the steps at least:

The first step, compute statistics

Calculate average, variance, standard deviation and the noise resistance of current potential and electric current respectively by the routine analyzer of establishment in advance; The pretreated method of these data is:

To real potential E ₁, E ₂..., E _m, pseudo-current potential E ' ₁, E ' ₂..., E ' _mAnd electric current I ₁, I ₂..., I _mCarry out segmentation, the number of every segment data is 2 ⁿ, wherein n is the positive integer more than or equal to 11, m represents the number of the noise data that each experiment collects.

With I ₁, I ₂..., I _mBe transformed to log|I ₁|, log|I ₂| ..., log|I _m|

According to formula $\stackrel{\‾}{X}=\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{x}_i$ , ask E ₁, E ₂..., E _m, E ' ₁, E ' ₂..., E ' _mAnd I ₁, I ₂..., I _mMean value, be designated as respectively

With

According to formula ${\mathrm{\σ}}^2=\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{(x_i-\stackrel{\‾}{X})}^2$ , ask E ₁, E ₂..., E _m, E ' ₁, E ' ₂..., E ' _mAnd I ₁, I ₂..., I _mVariance, be designated as σ E respectively ², σ E ' ²With σ I ²,

According to formula $\mathrm{\σ}=\sqrt{{\mathrm{\σ}}^2},$ Ask E ₁, E ₂..., E _m, E ' ₁, E ' ₂..., E ' _mAnd I ₁, I ₂..., I _mStandard deviation, be designated as σ respectively _E, σ _E' and σ ₁,

According to formula R _n=σ _E/ σ ₁And R ' _n=σ _E'/σ ₁, ask noise resistance, be designated as R respectively _nAnd R ' _n

Second step, cluster analysis

At first K-mean cluster analysis program is to original noise data or K congealing point of statistical parameter primary election of being calculated (poly-heart, be center of gravity), allow sample to nearest poly-heart cohesion, form preliminary classification, revise irrational classification by the minimum distance principle then, such center of gravity can change thereupon, after new cluster centre forms, the previous sample of having done check and having classified is readjusted again, up to all samples all till a certain center recently, obtain preliminary cluster result at last: this cluster process is selected square metric range of the Euclidean distance of minimum deviation sum of squares approach, if use x _IjkJ of i the element of representation class k refers to target value, and then the quadratic sum of the Euclidean distance of class k is expressed as:

${{\mathrm{<figure-callout\; id="2"\; label="D\; k"\; filenames="A20061012475600111.png,A20061012475600151.png"\; state="\{\{state\}\}">D</figure-callout>}}_k}^2=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{m_k}{\mathrm{\&Sigma;}}}{(x_{\mathrm{ijk}}-\stackrel{\&OverBar;}{x_{\&CenterDot;\mathrm{jk}}})}^2---\left(1\right)$

The 3rd step, discriminatory analysis

Classification results according to cluster analysis, the operation discriminant analysis program is with in the space that the independent variable combined projection in the higher space of dimension was lower to latitude originally, in the space of low dimension, classify again then, make that the deviation in each class is as far as possible little, and the deviation of projection is big as far as possible between inhomogeneity, at last cluster analysis result is returned and declares and obtain the discriminatory analysis equation; Just can classify according to discriminant equation, obtain the etch state of material this moment, handle, show, store and export above-mentioned all information at last the noise data group that unknown process produces.

The present invention makes it compare with existing electrochemistry noise monitoring device owing to adopted above analytical technology, have following tangible advantage and and characteristics:

1, forms by four electrode catheters owing to test probe, so can obtain the noise current potential of local corrosion system simultaneously with respect to true contrast electrode and pseudo-contrast electrode, both can obtain the corrosion potential and the Changing Pattern thereof of monitored system, can and change abundanter corrosion generation, the information of development of obtaining by the difference between pseudo-current potential and the real potential again.

2, because whole monitoring device mainly is made up of data acquisition system (DAS) and data analysis system, do not need extra auxiliary monitoring equipment,, be easy to the realization of on-the-spot corrosion monitoring engineering practical application so whole device is simple to operate.

3, owing to contain self-editing statistical computation, cluster analysis and discriminatory analysis software among ARM or the DSP, can in real time, accurately analyze noise raw data and statistical parameter, thereby obtain corrosion information more accurately, the result who comprises corrosion potential, electrochemistry noise resistance, cluster and discriminatory analysis, not only can judge the order of severity of uniform corrosion, and can judge that local corrosion takes place, the state of development, the concrete enforcement that therefore can be the corrosion protection measure provides effective foundation.

Description of drawings

Fig. 1 is based on the structural representation of the local corrosion monitoring device of electrochemistry noise, can carry out the collection and the analysis operation of noise data according to monitoring device of the present invention, I, II and III represent three working electrodes of being made by metal material of the same race respectively among the figure, the pseudo-contrast electrode made as metal material of the same race of III wherein, IV represents the Ag/AgCl solid reference electrode, be called true contrast electrode again, form four electrode corrosion probes by these four electrodes;

Fig. 2 adopts the process flow diagram of the analytical approach of monitoring device shown in Figure 1, and it can embody method of the present invention;

True noise current potential is schemed the x axle over time in after 8 hours 1024 seconds that monitor in Fig. 3 a embodiment of the present invention: time (s), y axle: real potential E (V);

The pseudo noise current potential is schemed the x axle over time in after 8 hours 1024 seconds that monitor in Fig. 3 b embodiment of the present invention: time (s), y axle: pseudo-current potential E (V);

Noise current is schemed the x axle over time in after 8 hours 1024 seconds that monitor in Fig. 3 c embodiment of the present invention: time (s), y axle: electric current I (A);

Fig. 4 a true noise potential data that embodiment obtained of the present invention is at the distribution results figure of different time points, x axle: time (h), y axle: current potential E (V);

Original noise current data are at the distribution results figure of different time points in Fig. 4 b example, x axle: time (h), y axle: electric current log|I|;

The classification results figure of data behind the K-mean cluster analysis among Fig. 5 Fig. 4, x axle: time (h), y axle: the classification results of cluster analysis;

Data set 1 that the unknown is classified in this example of Fig. 6 and 2 discriminatory analysis be figure as a result, x axle: the result of calculation of nonstandardized technique discriminant equation (y ' value), y axle: the frequency of y ' value;

The discriminatory analysis of the data set 3 that the unknown is classified in this example of Fig. 7 is figure as a result, x axle: the result of calculation of nonstandardized technique discriminant equation (y ' value), y axle: the frequency of y ' value.

Embodiment

The present invention will be further described below in conjunction with accompanying drawing and embodiment.Based on the structural representation of the local corrosion apparatus for evaluating of electrochemistry noise as shown in Figure 1, four electrode catheters 1, impedance transformer 2, DC compensator 3, binary channels chopper-zero-stabilized operational data amplifier 4, zero resistance galvanometer 5, hyperchannel low-pass filter 6 and A/D converter 7 constitute the electrochemistry noise monitoring system, the original noise data that the storage of collected that is used for data-carrier store 8 arrives, embedded microprocessor (ARM) or digital signal processor (DSP) the 9th, the analytic system of noise data, and keyboard, LCD constitutes interactive system, can adjust in case of necessity, correction has related parameter.

Describe analytical approach of the present invention below in conjunction with accompanying drawing 2, Fig. 2 is the process flow diagram that utilizes the data analysing method of monitoring device shown in Figure 1.Four electrode systems 1 to be tested are put into 0.50mol/L HCO ₃ ^-+ 0.10mol/LCl ^-Mixed solution after, while opening entry noise data, wherein noise current comes from two work electrode I being linked to each other by zero resistance galvanometer 5 and the current fluctuation signal between II, and two kinds of noise current potentials come from I after the coupling and II respectively with truly between contrast electrode IV and the potential fluctuation signal between coupling electrode and III.The voltage signal of simulation converts digital voltage signal to through impedance transformer 2, binary channels chopper-zero-stabilized operational data amplifier 4, hyperchannel low-pass filter 6 and A/D converter 7, and wherein the real potential signal is flat through eliminating its direct current base by DC compensator 3 behind the resistance transformer 2; Analog current signal is transformed into the digital current signal through zero resistance galvanometer 5, hyperchannel low-pass filter 6 and A/D converter 7.Then, with 8 storages of three kinds of signal data memory inputs, Fig. 3 a, 3b and 3c are respectively that the real potential after 8 hours in this example, pseudo-current potential and electric current are schemed over time simultaneously.Three kinds of noise signals all show typical spot corrosion metastable state noise fluctuations feature among the figure, the correspondence at current potential and current noise peak is obvious, real potential is similar with the whole fluctuation characteristic of pseudo-current potential, but both have the difference of the order of magnitude on concrete numerical value, from the real potential collection of illustrative plates as can be known the corrosion potential of system remain on-0.220～-fluctuate between the 0.260V, show that corrosion system is in passive state generally.At last, respectively three kinds of signal datas are carried out the automatic segmentation processing and every segment data is carried out statistical computation, with the initial object of result as further data analysis, this example select in 12 hours true noise current potential E and based on the pre-service amount log|I| of noise current I research object as cluster analysis and discriminatory analysis, selected variable rule is over time seen respectively shown in accompanying drawing 4a and Fig. 4 b.

According to the distribution characteristics of data among the corrosion characteristics of example and Fig. 4, determine initializaing variable: K=2, maximum iteration time=30, convergence parameter=0.02.Program is selected the initial position at 2 class centers automatically by the distribution characteristics of primary data, operation K-mean cluster method.

Almost do not change through two class centers after 9 iteration, therefore program is according to convergence parameter value (0.02), be the maximal value that changes of class centre distance less than 2% of the initial center coordinate figure of minimum, this criterion finishes cluster process, simultaneously, program generates the classification and the The results of analysis of variance of analyzed sample, exports the etch state of cluster result and each group data representative at last.(see shown in Figure 5) as can be known by classification results: preceding 7 groups are divided fully to class 1 in 12 groups of (24576) data, and expression electrode this moment is in the metastable state corrosion stage; The 3 groups of data in back belong in the class 2 fully, illustrates that electrode was in and stablize the corrosion stage this moment, promptly macroscopical putting the generation stage; Wherein 8,9 liang of group data points appearances that interweave between two classes illustrate that the process of spot corrosion at this moment is in by the transition period of metastable state to the stable state transformation.Above-mentioned conclusion is consistent with the etch state of detecting head surface, and as seen, the change procedure of cluster analysis result and spot corrosion state has significant correlation, can think that cluster analysis result has objectively responded actual extent of corrosion.At last with the The results of analysis of variance of program output, i.e. the test statistics F value of each parameter is with F _α(α=0.005, the gained of tabling look-up) relatively obtains: F＞＞F _0.005, and the value-at-risk Sig. value of each parameter is equal to 0.00, thus proved that index E and log|I| have contribution to classification, and have significant difference, and this time cluster result has higher reliability, and this program has the meaning of practical application.

According to cluster analysis result, behind keyboard selection Fisher diagnostic method and Wilks ' Lambda fitting process, the operation discriminant analysis program is derived standard form discriminant function (shown in the square journey 2), be used to judge the parameter that need give priority to, specify a kind of decision rule simultaneously, promptly set up non-standard discriminant function (shown in the square journey 3), be used for determining that the affiliated classification of unknown sample makes the false determination ratio minimum, export the etch state and Chi-square Test (chi-square check) result of unknown sample representative at last.

y＝1.13E+1.92log|I| (2)

y＝57.46E+7.34log|I|+68.55 (3)

According to the coefficient magnitude before two parameters in the mahjong (2) of program output as can be known, the coefficient value of variable log|I| is greater than the coefficient of variable E, so y is subjected to the influence of log|I| more obvious, promptly needs to give special heed to the variation characteristic of electric current in monitoring spot corrosion process.Chi-square Test result by program output: χ ²＞＞χ ² _0.005, the Sig.=0.000 discriminant equation of discriminatory analysis extraction as can be known has statistical significance.

The nonstandardized technique discriminant function that utilization is tried to achieve is classified to 3 groups of (6144) data that the electrochemistry noise acquisition system newly collects in different spot corrosion systems, and the result is respectively shown in Fig. 6 and 7.Wherein the 1st group of all data are all declared to class 1; The 2nd data are declared to class 2; And have 88.57% to belong in the class 1 in the 3rd group of data, remaining 11.43% belongs in the class 2.Electrode was in the spot corrosion metastable state phase when above-mentioned differentiation presentation of results produced the 1st group of data, and the trend that does not almost develop to macroscopic view point; The 2nd group of data declaration electrode surface seriously corroded this moment produced macroscopic erosion point; The corrosion process of the 3rd group of data characterization electrode surface this moment just changes the stable state corrosion stage over to by the metastable state stage, and macroscopical point will appear in its surface at short notice.According to the etch state of experiment condition and electrode surface as can be known, the the 1st and 2 group of data come from the corrosion process after typical metastable state and the appearance of macroscopic view point respectively, though and the 3rd group of data come from the metastable state process, but macroscopical point appears in electrode surface in this later 2h, therefore can think that this group data characterization is by the transition period of metastable state to the stable state transformation, so discriminatory analysis result and experimental fact match, as seen the discrimination formula of having set up is feasible to same corrosion system in practical application in the differentiation of the noise data of different phase generation.

Claims (3)

1. the monitoring device of metallic material local corrosion, it is characterized in that, this monitoring device comprises: by four electrode catheters (1) that electrode (I), electrode (II), pseudo-contrast electrode (III) and true contrast electrode (IV) are formed, impedance transformer (2), DC compensator (3), binary channels chopper-zero-stabilized operational data amplifier (4), zero resistance galvanometer (5), hyperchannel low-pass filter (6), A/D converter (7), data-storage system (8) and embedded microprocessor or digital signal processor (9);

Wherein four electrode catheters (1) are used to produce original electrochemistry noise current potential and current signal, and impedance transformer (2) is used to reduce the impedance of the true contrast electrode (IV) and the noise current potential that pseudo-contrast electrode (III) is produced of four electrode catheters (1); DC compensator (3) is used to eliminate the DC level part of the noise current potential that true contrast electrode (IV) gathers, and is sent to hyperchannel low-pass filter (6) after binary channels chopper-zero-stabilized operational data amplifier (4) amplifies the pseudo noise current potential that comes from electrode (III) and the true noise electric potential signal that comes from electrode (IV) respectively and carries out filtering; Zero resistance galvanometer (5) detects from the noise current signal of the electrode (I) of four electrode catheters (1) and electrode (II) and is sent to hyperchannel low-pass filter (6) filtering; A/D converter (7) is converted to noise current potential and noise current digital signal and is stored in data-storage system (8); Last embedded microprocessor or digital signal processor (9) carry out statistical computation, cluster analysis and discriminatory analysis to noise current potential and noise current data, finish the analysis of unknown noise data and provide the judged result of etch state.

2. according to the monitoring device of claim 1, it is characterized in that, the electrode (I) of described composition four electrode catheters (1), electrode (II), pseudo-contrast electrode (III) metal material of the same race are made, true contrast electrode (IV) is Ag/AgCl solid reference electrode (IV), and this electrode is the Ag/AgCl solid reference electrode that has wrapped up the saturated potassium chloride of one deck densification and the film that macromolecule resin mixes on Ag/AgCl solid reference electrode surface.

3. the analytical approach based on the local corrosion of electrochemistry noise is implemented by the local corrosion monitoring device, it is characterized in that this method may further comprise the steps at least:

The first step, compute statistics

Calculate average, variance, standard deviation and the noise resistance of current potential and electric current respectively by the routine analyzer of establishment in advance; The pretreated method of these data is:

To real potential E ₁, E ₂..., E _m, pseudo-current potential E ' ₁, E ' ₂..., E ' _mAnd electric current I ₁, I ₂..., I _mCarry out segmentation, the number of every segment data is 2 ⁿ, wherein n is the positive integer more than or equal to 11, m represents the number of the noise data that each experiment collects;

With I ₁, I ₂..., I _mBe transformed to log|I ₁|, log|I ₂| ..., log|I _m|

According to formula $\stackrel{-}{X}=\frac{1}{m}\underset{l=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_i,$ Ask E ₁, E ₂..., E _m, E ' ₁, E ' ₂..., E ' _mAnd I ₁, I ₂..., I _mMean value, be designated as respectively

With

According to formula ${\mathrm{\&sigma;}}^2=\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{(x_i-\stackrel{-}{X})}^2,$ Ask E ₁, E ₂..., E _m, E ' ₁, E ' ₂..., E ' _mAnd I ₁, I ₂..., I _mVariance, be designated as σ respectively _E ², σ _{E '} ²And σ _I ²,

According to formula $\mathrm{\&sigma;}=\sqrt{{\mathrm{\&sigma;}}^2},$ Ask E ₁, E ₂..., E _m, E ' ₁, E ' ₂..., E ' _mAnd I ₁, I ₂..., I _mStandard deviation, be designated as σ respectively _E, σ _{E '}And σ ₁,

According to formula R _n=σ _E/ σ _IAnd R ' _n=σ _{E '}/ σ _I, ask noise resistance, be designated as R respectively _nAnd R ' _n

Second step, cluster analysis

At first K-mean cluster analysis program is to original noise data or K congealing point of statistical parameter primary election of being calculated, the promptly poly-heart or center of gravity, allow sample to nearest poly-heart cohesion, form preliminary classification, revise irrational classification by the minimum distance principle then, such center of gravity can change thereupon, after new cluster centre forms, previous done check and the sample of classification is readjusted again, all till a certain center recently, obtained preliminary cluster result at last up to all samples; This cluster process is selected square metric range of the Euclidean distance of minimum deviation sum of squares approach, if use x _IjkJ of i the element of representation class k refers to target value, and then the quadratic sum of the Euclidean distance of class k is expressed as: $D_k^2=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{m_k}{\mathrm{\&Sigma;}}}{(x_{\mathrm{ijk}}-\stackrel{-}{x_{\&CenterDot;\mathrm{jk}}})}^2$

The 3rd step, discriminatory analysis

Classification results according to cluster analysis, the operation discriminant analysis program is with in the space that the independent variable combined projection in the higher space of dimension was lower to latitude originally, in the space of low dimension, classify again then, make that the deviation in each class is as far as possible little, and the deviation of projection is big as far as possible between inhomogeneity, at last cluster analysis result is returned and declares and obtain the discriminatory analysis equation; Just can classify according to discriminant equation, obtain the etch state of material this moment, handle, show, store and export above-mentioned all information at last the noise data group that unknown process produces.

Images