CN109033586A - Determination method and determining system based on the alloy grain size of mapping monotonicity - Google Patents
Determination method and determining system based on the alloy grain size of mapping monotonicity Download PDFInfo
- Publication number
- CN109033586A CN109033586A CN201810768923.XA CN201810768923A CN109033586A CN 109033586 A CN109033586 A CN 109033586A CN 201810768923 A CN201810768923 A CN 201810768923A CN 109033586 A CN109033586 A CN 109033586A
- Authority
- CN
- China
- Prior art keywords
- parameter
- ultrasound detection
- crystallite dimension
- selection
- coefficient
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
The present invention discloses the determination method and determining system of a kind of alloy grain size based on mapping monotonicity.The present invention uses the effective ultrasound detection parameter of relativity measurement criterion Stepwise Screening, then optimization aim is up to monotonicity, crystallite dimension soft-sensing model is established according to the effective ultrasound detection parameter sets finally screened, so that with crystallite dimension ordered arrangement, effective ultrasound detection parameter can not only keep the form of monotonic increase or monotone decreasing, while can be realized the validity tested outside sample set.Therefore, the crystallite dimension soft-sensing model established using determining method provided by the invention and determining system, measurement accuracy is high, and monotonicity is good, and error is small, and the result figure of model is intuitive complete and has stable evaluation effect.
Description
Technical field
The present invention relates to ultrasonic detecting technology and its characterization alloy grain size fields, more particularly to one kind based on mapping
The determination method of the alloy grain size of monotonicity and determining system.
Background technique
Titanium alloy has elastoresistance, corrosion resistance, fatigue resistance and good heat resistance, is widely used in Advanced Aircraft, flies
In the defence equipments such as ship, high thrust ratio aero-engine, ship, aerial engine fan, compressor disk and blade etc. are served as
Zero component of key position is known as " space metal ".Titanium alloy primary alpha phase crystallite dimension to yield strength, fatigue behaviour and
Corrosion resistance suffers from certain influence, and each macro property all can form corresponding feature with the different variations of crystallite dimension
Response.Because the particularity and importance of titanium alloy application are all irreplaceable, effective symbolized so design is a set of
The method of titanium alloy primary alpha phase crystallite dimension is vital.
Existing crystallite dimension detection method, which can be divided into, to be damaged and non-destructive testing two categories.Damaging detection mainly has metallographic
The methods of detection, electron backscatter diffraction (EBSD) detection.Nondestructive evaluation has ultrasound detection], the methods of EDDY CURRENT.Damage method
Although detection accuracy is high, detection process is cumbersome, detection efficiency is low and irreversible destruction can be caused to test specimen.By comparison
Compared with nondestructive determination also ensures higher detection efficiency in the case where not destroying examined workpiece, therefore constructs a kind of characterization
The Nondestructive Evaluation method of material grains size is the critical issue of current research.Ultrasonic non-destructive rating method have penetration capacity it is big,
The advantages that flaw detection sensitivity height and inspection easy to automate, present convergence are lossless in high temperature alloy, titanium alloy crystallite dimension
It is the most commonly used in characterization.
There is different degrees of influence for the linear ultrasonic detection parameters velocity of sound, attenuation coefficient etc. in crystallite dimension.With
The continuous improvement of particularity and sensitive requirements that people respond microstructure, more and more researchers are by mesh
Light is gathered in the characterization of relation of ultrasound non-linear parameter Yu microcosmic crystallite dimension.However, only with single ultrasound detection parameter and crystalline substance
Particle size establish linear relationship, because entrained by the characteristic parameters such as the velocity of sound, attenuation coefficient and nonlinear factor about crystal grain ruler
Very little information is variant, lacks accuracy so as to cause evaluation model.For existing multi-parameter evaluation of programme, due to quilt
The increase of the microcosmic crystallite dimension complex information of sample material, with the crystallite dimension ultrasonic evaluation method of error minimum target building without
The non-fully dull linearity curve of the crystallite dimension of method Efficient Characterization complex alloys material, formation makes evaluation precision lower, even
Evaluation effect can be lost.
Therefore, the accurate a set of effective non-destructive testing side for determining complex alloys material grains size how is established
Method, the technical issues of becoming those skilled in the art's urgent need to resolve.
Summary of the invention
The object of the present invention is to provide a kind of determination method of alloy grain size based on mapping monotonicity and determine system
System, the crystallite dimension soft-sensing model established using the determining method and determining system, measurement accuracy is high, and monotonicity is good, accidentally
Difference is small, and the result figure of model is intuitive complete and has stable evaluation effect.
To achieve the above object, the present invention provides following schemes:
A kind of determination method of the alloy grain size based on mapping monotonicity, the determining method include:
Obtain ultrasound the fixed point scanning signal, average thickness values, crystallite dimension value of each experiment sample;
The each super of each experiment sample is determined according to the average thickness values and the ultrasound fixed point scanning signal
Sound detection parameter value;
Obtain section step-length, lowest threshold and current selection moment;
The selection moment corresponding selection section is determined according to the section step-length and the selection moment;
According to the lowest threshold, the selection section, each ultrasound detection parameter value and the crystallite dimension value,
Effective ultrasound detection parameter at each selection moment is determined using relativity measurement criterion;
Final effectively ultrasound detection parameter set is determined according to effective ultrasound detection parameter at each selection moment;
It is up to optimization aim with monotonicity, the soft survey of crystallite dimension is established according to the final effectively ultrasound detection parameter set
Measure model;
The crystallite dimension of tested alloy is determined using the crystallite dimension soft-sensing model.
Optionally, described that the selection moment corresponding selection area is determined according to the section step-length and the selection moment
Between, it specifically includes:
According to formula:Determine the selection moment corresponding selection section, wherein t is indicated
The moment is selected,Indicate section step-length, θtIndicate the corresponding selection section selection moment t.
Optionally, described according to the lowest threshold, the selection section, each ultrasound detection parameter value and described
Crystallite dimension value is determined effective ultrasound detection parameter at each selection moment using relativity measurement criterion, specifically included:
Judge whether the maximum value in the selection section is more than or equal to lowest threshold, obtains the first judging result;
When first judging result indicates the maximum value in the selection section more than or equal to lowest threshold, use
Pearson correlation coefficients analysis method calculates separately each of the crystallite dimension value of each experiment sample and each experiment sample
The related coefficient of a ultrasound detection parameter value obtains each size-parameter related coefficient;
Select the corresponding ultrasound detection ginseng of ultrasound detection parameter value of the size-parameter related coefficient in the selection section
Number is used as primary election ultrasound detection parameter;
Each primary election ultrasound detection parameter of each experiment sample is calculated separately using Pearson correlation coefficients analysis method
Average correlation coefficient;
Average correlation coefficient is selected to be less than the primary election ultrasound detection parameter and average phase of the minimum value in the selection section
It closes property coefficient and is greater than the maximum primary election ultrasound detection parameter of average correlation coefficient for selecting section minimum value as effectively
Ultrasound detection parameter;
The selection moment is updated, and " selection section is determined according to the section step-length and the selection moment " described in return.
Optionally, described that optimization aim is up to monotonicity, it is established according to the final effectively ultrasound detection parameter set
Crystallite dimension soft-sensing model, specifically includes:
According to each effective ultrasound detection parametric configuration multidimensional actual parameter vector;
Dimensionality reduction mapping function is constructed, and is dropped the multidimensional actual parameter vector at one-dimensional using the dimensionality reduction mapping function
Actual parameter;
The one-dimensional actual parameter is normalized, normalization one-dimensional actual parameter is obtained;
The first fitting function is constructed, the dependent variable of first fitting function is crystallite dimension, first fitting function
Independent variable be normalization one-dimensional actual parameter;
To first fitting function carry out inverse transformation, obtain the second fitting function, second fitting function because become
Amount is normalization one-dimensional actual parameter, and the independent variable of second fitting function is crystallite dimension;
Difference with the corresponding dependent variable of adjacent independent variable of second fitting function is all positive number or is all negative
Maximum number is target, constitution optimization function;
The majorized function is solved using adaptive differential evolution algorithm, obtain make second fitting function it is adjacent from
The difference of the corresponding dependent variable of variable be all positive number or be all negative the maximum optimal dimensionality reduction coefficient of number and optimal fitting system
Number, wherein the optimal dimensionality reduction coefficient is the optimal coefficient of the dimensionality reduction mapping function, and the optimal fitting coefficient is described the
The optimal coefficient of one fitting function;
The optimal fitting coefficient is substituted into first fitting function, obtains crystallite dimension soft-sensing model.
A kind of determination system of the alloy grain size based on mapping monotonicity, the determining system include:
Sample parameter obtains module, for obtaining ultrasound the fixed point scanning signal, average thickness values, crystalline substance of each experiment sample
Particle size value;
Ultrasound detection parameter value determining module, for true according to the average thickness values and the ultrasound fixed point scanning signal
Each ultrasound detection parameter value of fixed each experiment sample;
Selection parameter obtains module, for obtaining section step-length, lowest threshold and current selection moment;
Section determining module is selected, for determining the selection moment pair according to the section step-length and the selection moment
The selection section answered;
Effective ultrasound detection parameter determination module, for according to the lowest threshold, the selection section, described each surpass
Sound detection parameter value and the crystallite dimension value determine effective ultrasound detection at each selection moment using relativity measurement criterion
Parameter;
Final effectively ultrasound detection parameter set determining module, for effective ultrasound detection parameter according to each selection moment
Determine final effectively ultrasound detection parameter set;
Crystallite dimension soft-sensing model establishes module, for being up to optimization aim with monotonicity, is finally had according to described
Effect ultrasound detection parameter set establishes crystallite dimension soft-sensing model;
Crystallite dimension measurement module, for determining the crystal grain ruler of tested alloy using the crystallite dimension soft-sensing model
It is very little.
Optionally, selection section determining module is according to formula:When determining the selection
Carving corresponding selection section, wherein t indicates the selection moment,Indicate section step-length, θtIndicate the corresponding selection area selection moment t
Between.
Optionally, effective ultrasound detection parameter determination module specifically includes:
First judging unit is obtained for judging whether the maximum value in the selection section is more than or equal to lowest threshold
Obtain the first judging result;
Size-parameter related coefficient determination unit, for indicating the selection section most when first judging result
When big value is more than or equal to lowest threshold, the crystalline substance of each experiment sample is calculated separately using Pearson correlation coefficients analysis method
The related coefficient of each ultrasound detection parameter value of particle size value and each experiment sample, obtains each size-parameter phase
Relationship number;
Primary election ultrasound detection choice of parameters unit, for selecting size-parameter related coefficient in the selection section
The corresponding ultrasound detection parameter of ultrasound detection parameter value is as primary election ultrasound detection parameter;
Average correlation coefficient calculation unit, for calculating separately each experiment using Pearson correlation coefficients analysis method
The average correlation coefficient of each primary election ultrasound detection parameter of sample;
Effective ultrasound detection parameter selection unit, for selecting average correlation coefficient to be less than the minimum in the selection section
The primary election ultrasound detection parameter and average correlation coefficient of value are greater than the average correlation coefficient of the selection section minimum value most
Big primary election ultrasound detection parameter is as effective ultrasound detection parameter;
Moment updating unit is selected, for updating the selection moment.
Optionally, the crystallite dimension soft-sensing model is established module and is specifically included:
Multidimensional actual parameter vector structural unit, for according to each effective ultrasound detection parametric configuration multidimensional actual parameter
Vector;
Dimensionality reduction unit is effectively joined the multidimensional for constructing dimensionality reduction mapping function, and using the dimensionality reduction mapping function
Number vector is dropped into one-dimensional actual parameter;
Normalized unit, for the one-dimensional actual parameter to be normalized, obtaining normalization one-dimensional has
Imitate parameter;
Fitting function structural unit, for constructing the first fitting function, the dependent variable of first fitting function is crystal grain
Size, the independent variable of first fitting function are normalization one-dimensional actual parameter;
Inverse transformation block, for first fitting function carry out inverse transformation, obtain the second fitting function, described second
The dependent variable of fitting function is normalization one-dimensional actual parameter, and the independent variable of second fitting function is crystallite dimension;
Majorized function structural unit, for the difference of the corresponding dependent variable of adjacent independent variable of second fitting function
Be all positive number or be all negative maximum number be target, constitution optimization function;
Adaptive differential evolution algorithm solves unit, for solving the optimization letter using adaptive differential evolution algorithm
Number, obtaining makes the difference of the corresponding dependent variable of adjacent independent variable of second fitting function be all positive number or be all a of negative
The maximum optimal dimensionality reduction coefficient of number and optimal fitting coefficient, wherein the optimal dimensionality reduction coefficient is the dimensionality reduction mapping function
Optimal coefficient, the optimal fitting coefficient are the optimal coefficient of first fitting function;
Soft-sensing model determination unit obtains brilliant for the optimal fitting coefficient to be substituted into first fitting function
Particle size soft-sensing model.
The specific embodiment provided according to the present invention, the invention discloses following technical effects:
The determination method and determining system of a kind of alloy grain size based on mapping monotonicity provided by the invention, use
The effective ultrasound detection parameter of relativity measurement criterion Stepwise Screening, is then up to optimization aim with monotonicity, according to finishing screen
Effective ultrasound detection parameter set of choosing establishes crystallite dimension soft-sensing model, so that with crystallite dimension ordered arrangement the case where
Under, effective ultrasound detection parameter can not only keep the form of monotonic increase or monotone decreasing, while can be realized outside sample set
The validity of test.Therefore, the crystallite dimension soft-sensing model established using determining method provided by the invention and determining system,
Measurement accuracy is high, and monotonicity is good, and error is small, and the result figure of model is intuitive complete and has stable evaluation effect.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without creative efforts, can also obtain according to these attached drawings
Obtain other attached drawings.
Fig. 1 is a kind of determination method for alloy grain size based on mapping monotonicity that the embodiment of the present invention 1 provides
Flow chart;
Fig. 2 is a kind of determination system for alloy grain size based on mapping monotonicity that the embodiment of the present invention 2 provides
Structural block diagram;
Fig. 3 is that the different forging temperatures that provide of the embodiment of the present invention 3, the TC4 titanium alloy under different forging deformation amount are typical
Microstructure morphology;
Fig. 4 is the crystallite dimension average value validity schematic diagram that the embodiment of the present invention 3 provides;
Fig. 5 is the schematic diagram of calculating monotonicity and non-monotonic situation that the embodiment of the present invention 3 provides;
Fig. 6 is the model and fit correlation curve for 5 kinds of evaluation crystallite dimension average value that the embodiment of the present invention 3 provides;
Fig. 7 is the model and fit correlation curve for 5 kinds of evaluation crystallite dimension standard deviations that the embodiment of the present invention 3 provides.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The object of the present invention is to provide a kind of determination method of alloy grain size based on mapping monotonicity and determine system
System, the crystallite dimension soft-sensing model established using the determining method and determining system, measurement accuracy is high, and monotonicity is good, accidentally
Difference is small, and the result figure of model is intuitive complete and has stable evaluation effect.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
Embodiment 1:
Fig. 1 is a kind of determination method for alloy grain size based on mapping monotonicity that the embodiment of the present invention 1 provides
Flow chart.As shown in Figure 1, a kind of determination method of the alloy grain size based on mapping monotonicity, the determining method include:
Step 101: obtaining ultrasound the fixed point scanning signal, average thickness values, crystallite dimension value of each experiment sample;
Step 102: determining each experiment sample according to the average thickness values and the ultrasound fixed point scanning signal
Each ultrasound detection parameter value;
Step 103: obtaining section step-length, lowest threshold and current selection moment;
Step 104: according to the section step-length and the selection moment, using formula:Really
Determining the selection moment corresponding selection section, wherein t indicates the selection moment,Indicate section step-length, θtIndicate selection moment t
Corresponding selection section;
Step 105: according to the lowest threshold, the selection section, each ultrasound detection parameter value and the crystalline substance
Particle size value determines effective ultrasound detection parameter at each selection moment using relativity measurement criterion;
Step 106: final effectively ultrasound detection parameter set is determined according to effective ultrasound detection parameter at each selection moment;
Step 107: optimization aim being up to monotonicity, crystal grain is established according to the final effectively ultrasound detection parameter set
Size soft-sensing model;
Step 108: the crystallite dimension of tested alloy is determined using the crystallite dimension soft-sensing model.
Specifically, the step 105: according to the lowest threshold, the selection section, each ultrasound detection parameter
Value and the crystallite dimension value determine effective ultrasound detection parameter at each selection moment using relativity measurement criterion, specifically
Include:
Judge whether the maximum value in the selection section is more than or equal to lowest threshold, obtains the first judging result;
When first judging result indicates the maximum value in the selection section more than or equal to lowest threshold, use
Pearson correlation coefficients analysis method calculates separately each of the crystallite dimension value of each experiment sample and each experiment sample
The related coefficient of a ultrasound detection parameter value obtains each size-parameter related coefficient;
Select the corresponding ultrasound detection ginseng of ultrasound detection parameter value of the size-parameter related coefficient in the selection section
Number is used as primary election ultrasound detection parameter;
Each primary election ultrasound detection parameter of each experiment sample is calculated separately using Pearson correlation coefficients analysis method
Average correlation coefficient;
Average correlation coefficient is selected to be less than the primary election ultrasound detection parameter and average phase of the minimum value in the selection section
It closes property coefficient and is greater than the maximum primary election ultrasound detection parameter of average correlation coefficient for selecting section minimum value as effectively
Ultrasound detection parameter;
The selection moment is updated, and returns to the step 104: selection is determined according to the section step-length and the selection moment
Section;
When first judging result indicates the maximum value in the selection section less than lowest threshold, step 106 is executed:
Final effectively ultrasound detection parameter set is determined according to effective ultrasound detection parameter at each selection moment.
Specifically, the step 107: being up to optimization aim with monotonicity, according to the final effectively ultrasound detection ginseng
Manifold establishes crystallite dimension soft-sensing model, specifically includes:
According to each effective ultrasound detection parametric configuration multidimensional actual parameter vector;
Dimensionality reduction mapping function is constructed, and is dropped the multidimensional actual parameter vector at one-dimensional using the dimensionality reduction mapping function
Actual parameter;
The one-dimensional actual parameter is normalized, normalization one-dimensional actual parameter is obtained;
The first fitting function is constructed, the dependent variable of first fitting function is crystallite dimension, first fitting function
Independent variable be normalization one-dimensional actual parameter;
To first fitting function carry out inverse transformation, obtain the second fitting function, second fitting function because become
Amount is normalization one-dimensional actual parameter, and the independent variable of second fitting function is crystallite dimension;
Difference with the corresponding dependent variable of adjacent independent variable of second fitting function is all positive number or is all negative
Maximum number is target, constitution optimization function;
The majorized function is solved using adaptive differential evolution algorithm, obtain make second fitting function it is adjacent from
The difference of the corresponding dependent variable of variable be all positive number or be all negative the maximum optimal dimensionality reduction coefficient of number and optimal fitting system
Number, wherein the optimal dimensionality reduction coefficient is the optimal coefficient of the dimensionality reduction mapping function, and the optimal fitting coefficient is described the
The optimal coefficient of one fitting function;
The optimal fitting coefficient is substituted into first fitting function, obtains crystallite dimension soft-sensing model.
Embodiment 2:
Fig. 2 is a kind of determination system for alloy grain size based on mapping monotonicity that the embodiment of the present invention 2 provides
Structural block diagram.As shown in Fig. 2, a kind of determination system of alloy grain size, the determining system include:
Sample parameter obtains module 201, for obtaining ultrasound the fixed point scanning signal, average thickness of each experiment sample
Value, crystallite dimension value;
Ultrasound detection parameter value determining module 202, for according to the average thickness values and the ultrasound fixed point scanning letter
Number determine each ultrasound detection parameter value of each experiment sample;
Selection parameter obtains module 203, for obtaining section step-length, lowest threshold and current selection moment;
Section determining module 204 is selected, for according to formula:Determine the selection moment
Corresponding selection section, wherein t indicates the selection moment,Indicate section step-length, θtIndicate the corresponding selection area selection moment t
Between;
Effective ultrasound detection parameter determination module 205, for according to the lowest threshold, the selection section, it is described respectively
A ultrasound detection parameter value and the crystallite dimension value determine effective ultrasound at each selection moment using relativity measurement criterion
Detection parameters;
Final effectively ultrasound detection parameter set determining module 206, for effective ultrasound detection according to each selection moment
Parameter determines final effectively ultrasound detection parameter set;
Crystallite dimension soft-sensing model establishes module 207, for being up to optimization aim with monotonicity, according to described final
Effective ultrasound detection parameter set establishes crystallite dimension soft-sensing model;
Crystallite dimension measurement module 208, for determining the crystal grain of tested alloy using the crystallite dimension soft-sensing model
Size.
Specifically, effective ultrasound detection parameter determination module 205 specifically includes:
First judging unit is obtained for judging whether the maximum value in the selection section is more than or equal to lowest threshold
Obtain the first judging result;
Size-parameter related coefficient determination unit, for indicating the selection section most when first judging result
When big value is more than or equal to lowest threshold, the crystalline substance of each experiment sample is calculated separately using Pearson correlation coefficients analysis method
The related coefficient of each ultrasound detection parameter value of particle size value and each experiment sample, obtains each size-parameter phase
Relationship number;
Primary election ultrasound detection choice of parameters unit, for selecting size-parameter related coefficient in the selection section
The corresponding ultrasound detection parameter of ultrasound detection parameter value is as primary election ultrasound detection parameter;
Average correlation coefficient calculation unit, for calculating separately each experiment using Pearson correlation coefficients analysis method
The average correlation coefficient of each primary election ultrasound detection parameter of sample;
Effective ultrasound detection parameter selection unit, for selecting average correlation coefficient to be less than the minimum in the selection section
The primary election ultrasound detection parameter and average correlation coefficient of value are greater than the average correlation coefficient of the selection section minimum value most
Big primary election ultrasound detection parameter is as effective ultrasound detection parameter;
Moment updating unit is selected, for updating the selection moment.
Specifically, the crystallite dimension soft-sensing model is established module 207 and is specifically included:
Multidimensional actual parameter vector structural unit, for according to each effective ultrasound detection parametric configuration multidimensional actual parameter
Vector;
Dimensionality reduction unit is effectively joined the multidimensional for constructing dimensionality reduction mapping function, and using the dimensionality reduction mapping function
Number vector is dropped into one-dimensional actual parameter;
Normalized unit, for the one-dimensional actual parameter to be normalized, obtaining normalization one-dimensional has
Imitate parameter;
Fitting function structural unit, for constructing the first fitting function, the dependent variable of first fitting function is crystal grain
Size, the independent variable of first fitting function are normalization one-dimensional actual parameter;
Inverse transformation block, for first fitting function carry out inverse transformation, obtain the second fitting function, described second
The dependent variable of fitting function is normalization one-dimensional actual parameter, and the independent variable of second fitting function is crystallite dimension;
Majorized function structural unit, for the difference of the corresponding dependent variable of adjacent independent variable of second fitting function
Be all positive number or be all negative maximum number be target, constitution optimization function;
Adaptive differential evolution algorithm solves unit, for solving the optimization letter using adaptive differential evolution algorithm
Number, obtaining makes the difference of the corresponding dependent variable of adjacent independent variable of second fitting function be all positive number or be all a of negative
The maximum optimal dimensionality reduction coefficient of number and optimal fitting coefficient, wherein the optimal dimensionality reduction coefficient is the dimensionality reduction mapping function
Optimal coefficient, the optimal fitting coefficient are the optimal coefficient of first fitting function;
Soft-sensing model determination unit obtains brilliant for the optimal fitting coefficient to be substituted into first fitting function
Particle size soft-sensing model.
Embodiment 3:
The present embodiment is strategy with the influence of monotonic performance, and preferably multiple ultrasound detection parameter drops are at one-dimensional parameter and normalizing
Change processing, with primary alpha phase crystallite dimension through first order be fitted and formulate monotonicity (drop at one-dimensional ultrasound detection parameter sample point according to
Secondary sequence difference is all positive or negative) it is up to optimization aim, combining adaptive differential evolution algorithm (self-adaptive
Differential evolution, SADE) algorithm determines that optimization problem carrys out optimization aim, solution obtain corresponding mapping function with
The undetermined coefficient of fitting function, to establish a kind of TC4 primary alpha phase crystallite dimension ultrasonic evaluation method for mapping monotonicity.
Step 1: ultrasound fixed point scanning signal, average thickness values, crystallite dimension that experiment obtains each experiment sample are carried out
Value, and determine that each ultrasound of each experiment sample is examined according to the average thickness values and the ultrasound fixed point scanning signal
Survey parameter value.
It treats and inspects sample through different forging temperatures (920-990 DEG C) and different forging deformation amounts (23-42%) to prepare.Into
The experiment of row ultrasound detection, is respectively adopted pulse reflection method (5077PR impulse generator, set of frequency 10MHz) and conllinear harmonic wave
Method (RAM-5000-SNAP non-linear ultrasonic test macro, receives frequency 5MHz at tranmitting frequency 2.5MHz) carries out detection sample,
The original A of linear ultrasonic is extracted to sweep signal (ultrasound fixed point scanning signal) and utilize EVA data processing software storage information, and according to
Following formula calculates the velocity of sound and attenuation coefficient:
In formula, CLThe expression velocity of sound, transit time of the Δ t between surface echo and a Bottom echo peak value,For sample
Average thickness, PSEFor surface echo peak value, PF1For a Bottom echo peak value, α indicates attenuation coefficient.
It is as follows with the calculation formula of the nonlinear factor β of non-linear ultrasonic detection mode extraction:
In formula, x is propagation distance, and k is wave number, A0For fundamental voltage amplitude, A2For secondary harmonic amplitude.Usually be approximately considered x and
K is definite value, using relative nonlinear factor beta ' replace β, i.e.,
The ultrasound detection parameter of extraction, which uses, to be averaged with the statistical of standard deviation and carries out to each sample corresponding
Calculating, the characteristic parameter of extraction is respectively as follows: the velocity of sound, including velocity of sound average valueWith velocity of sound standard deviationAttenuation coefficient, including
Attenuation coefficient average valueWith attenuation coefficient standard deviationBottom echo frequency peak PF1With secondary bottom wave frequency rate peak value
PF2, primary bottom frequency offset amountWith secondary bottom frequency offset amountRelative nonlinear coefficient, including relative nonlinear
Coefficient average valueIt is poor with relative nonlinear factor standardThe corresponding ultrasound detection parameter of the present embodiment is shown in Table 1.
The ultrasound detection parameter list that table 1 extracts
Metallographic preparation is carried out to sample, it then follows to 4 sample interception, polishing, polishing, corrosion steps.Wherein metallographic sand
Paper is KmTBCr15Mo silicon carbide paper, the corrosive liquid of preparation be HF:HNO3:H2O=3:8:89 ratio carry out surface corrosion and
Optical microphotograph microscopic observation sample tissue.TC4 titanium alloy typical case under gained difference forging temperature, different forging deformation amounts is microcosmic
Tissue topography is as shown in Figure 3, wherein part (a) of Fig. 3 indicates 920 DEG C of forging temperature, the TC4 titanium under forging deformation amount 23%
Part (b) of alloy representative microstructure pattern, Fig. 3 indicates 920 DEG C of forging temperature, and the TC4 titanium under forging deformation amount 38% closes
Golden representative microstructure pattern, part (c) of Fig. 3 indicate 940 DEG C of forging temperature, the TC4 titanium alloy under forging deformation amount 26%
Representative microstructure pattern, part (d) of Fig. 3 indicate 940 DEG C of forging temperature, the TC4 titanium alloy allusion quotation under forging deformation amount 40%
Part (e) of type microstructure morphology, Fig. 3 indicates 990 DEG C of forging temperature, and the TC4 titanium alloy under forging deformation amount 26% is typical
Microstructure morphology, 990 DEG C of the part (f) forging temperature of Fig. 3, the TC4 titanium alloy under forging deformation amount 42% are microcosmic group typical
Knit pattern.It is carried out using metallographic microscope of the ImageJ software to TC4 titanium alloy sample nascentαPhase white area proportion grading simultaneously measures
Its area S, in conjunction with calculation formulaStatistics obtains equivalent primary alpha phase crystallite dimension (average valueStandard deviation
), the present embodiment indicates TC4 titanium alloy primary alpha phase crystallite dimension with crystallite dimension.The ginseng of the related process obtained by metallographical measurement
Number is shown in Table 2.
2 TC4 of table is through the primary alpha phase crystallite dimension under different forging temperatures, forging deformation amount
Step 2: section step-length, lowest threshold and current selection moment are obtained, according to the section step-length and the choosing
It selects and determines the selection moment corresponding selection section constantly, and according to the lowest threshold, the selection section, described each
Ultrasound detection parameter value and the crystallite dimension value determine that effective ultrasound at each selection moment is examined using relativity measurement criterion
Survey parameter.
The ultrasound detection parameter of extraction is expressed as Y={ Y in the form of variable1,Y2,…,Yk, primary alpha phase crystallite dimension
Average and standard deviation be expressed asWherein,Indicate the average value of primary alpha phase crystallite dimension,Indicate primary α
The standard deviation of phase crystallite dimension.Model is directly established with crystallite dimension because of multi-dimensional ultrasound detection parameters can seem information redundancy and not
Certainty influences, and need to reject invalid, interference parameter from multi-Dimensional parameters with a kind of strategy.The present embodiment is with Pearson
(Pearson) correlation analysis method carries out relativity measurement, and the formula of Pearson (Pearson) correlation analysis method is as follows
In formula, ρXYIllustrate related coefficient.X illustrates crystallite dimension, and Y illustrates ultrasound detection parameter,Indicate crystal grain
Average value of the size in sample,Indicate average value of the ultrasound detection parameter in sample.
Crystallite dimension X and ultrasound detection parameter sets Y correlation analysis form are denoted as
Indicate the relative coefficient of crystallite dimension X and i-th of ultrasound detection parameter.
Ultrasound detection parameter interdependency analytical form is as follows:
In formula, matrix ρYYIndicate the correlation between ultrasound detection parameter,Indicate i-th of ultrasound detection parameter Yi
With j-th of ultrasound detection parameter YjCorrelation.
In formula,Indicate i-th of ultrasound detection parameter YiThat sums and be calculated with remaining dependence on parameter absolute value is flat
Related property coefficient, k indicate the total number of all ultrasound detection parameters.
When the present embodiment is according to correlation analysis ultrasound detection parameter, selection section is setIt formulates
2 relativity measurement criterion carry out assisting sifting and go out effective ultrasound detection parameter, and criterion is as follows
Criterion 1: the correlation ρ of ultrasound detection parameter Y and crystallite dimension XXYPositioned at the selection section θ of t momenttIt is interior.Formula
It is as follows:
In formula, t indicates the current set moment, and k is all ultrasound detection parameter total numbers,For current time
Interval range, according to ultrasound detection parameter sets selected by criterion 1 be Yt={ Yt1,Yt2,…,Ytp, YtFor ultrasound detection
The correlation of parameter and crystallite dimension belongs to the characteristic parameter in selection section.
Criterion 2: average correlation coefficient between Selecting All ParametersIt is lower thanUltrasound detection parameter, calculation formula is such as
Under:
In formula,It is less than the primary election ultrasound detection parameter of the minimum value in the selection section for average correlation coefficient, i.e.,
It is criterion 2 from YtThe effective ultrasound detection parameter of the part 1 of middle selection, p are set YtContained in number of parameters.
And for being higher thanUltrasound detection parameter in preferably the strongest ultrasound detection parameter of correlation and all have
The formula for imitating parameter sets is as follows:
In formula, QtIndicate YtIn be all larger than by average correlation coefficient calculated valueParameter sets.
In the present embodiment, useIt indicates from parameter sets QtThat strongest ultrasound parameter of the average correlation of middle selection is super
Sound detection parameter is as the effective ultrasound detection parameter of part 2.
In formula,As from Y in rule 2tIn the effective ultrasound detection parameter of part 2 selected.Most for the present embodiment
The effective ultrasound detection parameter sets determined eventually.
The detailed process for choosing effective characteristic parameters from multi-dimensional ultrasound detection parameters according to association rules 1,2 is as follows:
Step1: initiation parameter gives ultrasound detection parameter Y, initial time t=1, section step-lengthMinimum threshold
Value ρυ。
Step2: according to formula:Determine the selection moment corresponding selection section.
Step3: determine current intervalWhether ρ is greater thanυ;
If so, continuing to execute operation Step3.
Step4: in t moment, section isRelevant parameter Y is selected according to correlation criterion 1 (formula 7)t。
Step5:, can be from set Y according to correlation criterion 2 (formula 8~10)tIn select essential feature parameter
Step6: updating section isStep2 is returned to, necessary characteristic parameter is selected in continuation from Y, and is classified as
Set
The phase of table 3 and table 4 between ultrasound detection parameter and crystallite dimension (average value, standard deviation) and ultrasound detection parameter
Relationship number.Using selection method provided in this embodiment, k 10,Value 0.2, ρυValue is 0.4.With crystallite dimension average value
When for object, it can be selected according to Step4And it can be obtained according to Step5For call parameter, foundation Stpe3~Step6
It can obtainAF2、It also corresponds to require, then chooseIt is (average with crystallite dimension as ultrasound
Value) modeling input parameter.
When similarly, using crystallite dimension standard deviation as object, it can selectAs ultrasound and crystalline substance
Particle size (standard deviation) modeling input parameter.
The correlation of 3 ultrasonic feature parameter of table and primary alpha phase crystallite dimension (average value, standard deviation)
Correlation between 4 ultrasonic feature parameter inside of table
Step 3: being up to optimization aim with monotonicity, is established according to final effectively ultrasound detection parameter set based on dullness
The multi-parameter ultrasound Evaluation model of property is as crystallite dimension soft-sensing model.
Existing research is focused on optimal to determine with the minimum target of error between ultrasound detection parameter and match value
Evaluation model, but the list increased with various interference informations such as material grains size irregular distributions to cannot be guaranteed model
Tonality.Fig. 4 is the crystallite dimension average value validity schematic diagram that the embodiment of the present invention 3 provides.Fig. 4 shows minimum with error
The evaluation model (a kind of special failure case) of target, specifically with (velocity of sound standard deviation, attenuation coefficient average value, secondary bottom wave
Frequency peak, relative nonlinear factor standard are poor) with crystallite dimension average value establish the evaluation mould with the minimum target of error
Type, the certain very little of error in figure between gained ultrasound detection parameter optimization value and match value, but there is following drawback:
1) it is gone from global angle, model is formed by fit line and is on a horizontal line, easily causes provide phase in this way
When ultrasound detection parameter (sample value distinguishes unconspicuous parameter) answered, the crystallite dimension deviation reflected is very big, loses substantially
Evaluation effect.
2) it is gone from local angle, monotonicity is also not exclusively presented in optimizing lines, and such part evaluation will also result in
To the results abnormity of crystallite dimension.
From the above analysis, when determining ultrasound detection parameter evaluation model, list will form uncertain using error as target
Property and part, the overall situation etc. can not comment phenomenon, so having lacked feasibility.It need to be commented from another angle (monotonicity) target to construct
Valence model, so that the ultrasound detection parameter of mapping not only keeps monotonic increase with crystallite dimension ordered arrangement
Or the form of monotone decreasing, while can be realized the validity tested outside sample set.
TC4 titanium alloy primary alpha phase crystallite dimension is more sensitive to the response of ultrasound parameter, constructs ultrasonic nonodestruction evaluation mode
It is to form ultrasonic feature parameter can directly characterize the mapping relations of crystallite dimension.Have for by the evaluation model of target of error
The uncertain problem of effect property, monotonicity is up to optimization aim by the present invention, and carrys out solving optimization problem in conjunction with optimization algorithm
Measurement model is obtained, process includes: to introduce dimensionality reduction mapping function, is formed individually by the mapping function conversion containing undetermined coefficient
One-dimensional actual parameter;Then gained one-dimensional actual parameter is normalized and is introduced fitting function, formulate a ultrasound parameter and crystalline substance
The particle size sample point maximum optimization aim of monotonicity that difference is positive or is negative simultaneously one by one, and then will be by dimensionality reduction mapping function
Determine that the process of single ultrasound parameter is converted into optimization problem with the undetermined coefficient of fitting function.It is strategy with optimization aim and combines
SADE algorithm solving optimization problem, last Solve problems simultaneously find optimal mapping function and fitting function undetermined coefficient, determine
Multi-parameter ultrasound Evaluation (Multi-parameter Ultrasonic Evaluation Based on based on monotonicity
Monotonicity, MUEBM) model, to obtain crystallite dimension soft-sensing model.
In order to which multiple ultrasonic feature parameters are converted to one-dimensional parametric form, it is convenient for one-dimensional ultrasonic feature and crystallite dimension
(average value, standard deviation) establishes the corresponding evaluation model of internal specimen.It, will using quadratic polynomial as mapping function original form
The Y ' chosen by correlation is as input parameter and combination SADE algorithm optimization inputs parameter and searches out and constitutes mapping function
Best undetermined coefficient λij, obtain mapping function f i.e. and can determine one-dimensional ultrasonic feature parameter Z.
The dimensionality reduction mapping function form of the present embodiment construction is as follows:
In formula, (λi1,λi2,λi3) indicate dimensionality reduction mapping function coefficient, wherein i=1,2 ..., m;
Contain 4 ultrasonic feature parameter vectors for what is chosen according to correlation;Z indicates one-dimensional corresponding with sample contained by crystallite dimension
Effective ultrasound parameter.
Information content as contained by each ultrasonic feature parameter of extraction is different and not in unified dimension, introduces normalization
Method restricts the ultrasound parameter of dimension difference within the limits prescribed convenient for analysis and subsequent modeling.The present embodiment will pass through
Dimensionality reduction mapping function f drop has carried out the control of characteristic dimension at the one-dimensional actual parameter of one-dimensional and has done normalized, obtains
Normalize one-dimensional actual parameter.Scope control is normalized at (N, M), the calculation formula of normalized is as follows:
In formula, M, N are normalized maxima and minima, respectively 0.99,0.01, Z be to normalized single ultrasound
Parameter, normalization one-dimensional actual parameter areMin (Z) indicates that the minimum component of one-dimensional ultrasonic feature vector Z, max (Z) indicate
The largest component of one-dimensional ultrasonic feature vector Z.
The relational model between one-dimensional ultrasonic feature parameter Z and crystallite dimension X constructed in formula (11) in order to obtain, draws
Enter least square method fitting function to draft the linear relationship of Z and X.Under the effect of this fitting function, one-dimensional ultrasound parameter Z and X
Forming primary polynomial linear expression-form usually in the relationship of ultrasonic nonodestruction evaluation is reflected by crystallite dimension
The various acoustic characteristics of ultrasound.Then independent variable made with X in fit correlation and Z makees the corresponding form of dependent variable, input is corresponding
Crystallite dimension be just able to reflect out corresponding ultrasound parameter.
Public affairs are seen for being fitted average grain size and normalizing the first fitting function of one-dimensional actual parameter in the present embodiment
Formula (13), dependent variable are crystallite dimension, and independent variable is normalization one-dimensional actual parameter;
In formula, X*It indicates according to the crystallite dimension obtained after normalization one-dimensional actual parameter fitting;λaAnd λbIntend for undetermined
Collaboration number.Formula (13) is inversely transformed into the second fitting function, can indicate the ultrasound detection parameter changed with crystallite dimension X,
Formula is as follows:
Z*=F ' (X)=ξ1X+ξ2(14)
In formula, Z*It indicates to be fitted the normalization one-dimensional actual parameter of acquisition using the crystallite dimension of experiment sample as input,
ξ1=1/ λa、ξ2=-λb/λaRespectively indicate the undetermined coefficient of fitting function.
Below by establishing with acquire and table 1 that primary Calculation obtains shown in ultrasonic feature parameter input, with crystalline substance in table 2
The average and standard deviation of particle size exports and can reflect microcosmic grain size and the discrete implementations that are evenly distributed
Ultrasound Evaluation model as final crystallite dimension soft-sensing model.In order to determine optimization aim maximum and optimal evaluation mould
Type needs accurately to obtain each undetermined coefficient of dimensionality reduction mapping function, fitting function, in order to reach higher measurement accuracy.
The present embodiment withWith X-shaped at relation curve in successively difference between samples be all positive number or be all negative maximum
Number is optimization aim, i.e. monotonicity target.When this optimization aim is intended to maximum value, gained relation curve will form list
Adjust incremental or monotone decreasing trend even can reach total monotonicity, so that being formed in the case where X is constant regular
'sArrangement.The undetermined coefficient that corresponding 2 groups of functions are determined by above-mentioned optimization method, that is, can determine crystallite dimension hard measurement
Model, the error between match value and calculated value is small, and the measurement accuracy of model is improved.
Fig. 5 is the schematic diagram of calculating monotonicity and non-monotonic situation that the embodiment of the present invention 3 provides.Curve is the in Fig. 5
The relation curve of two fitting functions, as seen from Figure 5, abscissa is the complete increasing trend that rule is presented in figure, and ordinate is not
Decline trend is presented completely.It is that 1 dull section is presented that ordinate difference between point 2 and point 1, which is negative and counts, is clicked through to 10
Going and successively calculating down to obtain 9 monotone decreasing sections is to meet total monotonicity.And 5 Hes of dotted line frame region performance point in figure
Ascendant trend is presented in point 8, has violated whole monotonic trend to cause and only have 7 monotone decreasing sections in Fig. 7, and
Violating will cause fitting lines news commentary valence result appearance exception and deviates situation in region.Construct the calculation formula of monotonicity target
It is as follows:
Wherein,l′numIndicate the number that difference is positive, n is monotonicity sample
The total number of interior normalization one-dimensional actual parameter, kiIndicate that normalization one-dimensional of second fitting function in monotonicity sample is effective
Parameter Z*Successively item number difference;lmax-numThe maximum value for being positive or being negative for the difference number of successively item number.
When the dull section number of searching is identical, i.e., successively the difference number of item number is positive or is negative maximum value
lmax-numWhen identical, consider that the formula of minimal error is as follows:
In formula,Indicate when the monotonicity number searched out is identical, preferentially select mean absolute error the smallest for
Optimal monotonicity target number.
The monotonicity obtained according to monotonicity strategy shows that dull number is bigger, and corresponding error is smaller, gained crystal grain ruler
Very little soft-sensing model is better;Dull number is smaller, and corresponding error is bigger, and crystallite dimension soft-sensing model is poorer.For optimization
Target problem reflects to guarantee to be intended to error of the good monotonicity to restrict crystallite dimension soft-sensing model by dimensionality reduction
The dimensionality reduction of function and the first fitting function, optimizing and conversion in fit procedure are penetrated, 2 groups of optimization systems are found in conjunction with SADE algorithm
Number.The formula of gained optimization aim is as follows:
In formula, λ=(λi1,λi2,λi3), ξ=(ξ1,ξ2), λ, ξ show respectively dimensionality reduction mapping function coefficient and the second fitting
Function coefficients, lmax-numIndicate normalization one-dimensional actual parameter Z of second fitting function in monotonicity sample*Successively item number
The maximum value that difference number is positive or is negative.
The present embodiment is directed to ultrasonic feature parameter and crystallite dimension, establishes the TC4 primary alpha phase crystallite dimension based on monotonicity
When ultrasonic soft-sensing model, building considers that multi-parameter ultrasound is responded using monotonicity strategy as the processing mode of optimization aim and simultaneously
Primary alpha phase crystallite dimension ultrasound soft-sensing model implementing procedure it is as follows:
(1) excessive for the redundancy of multiple ultrasonic feature parameter Y, preliminary screening is limited with correlation analysis and section
Required ultrasonic feature parameter Y ', while determining crystallite dimensionWith the strong correlation of ultrasound parameter.
(2) Y ' of selection is subjected to dimensionality reduction by the quadratic polynomial dimensionality reduction mapping function f of (formula 11) construction, obtained new
The effective ultrasound parameter Z of one-dimensional.
(3) using normalized shown in formula (12), so that Z is transformed under same scaleConvenient for model
It establishes.
(4) willWithEvaluation model is established, the first fitting function of introducing is that single order shown in formula (13) is fitted letter
Number F, is fitted processing using formula (13) and formula (14), the effective ultrasound parameter Z of the one-dimensional being newly fitted*。
(5) the effective ultrasound parameter Z of the monotonicity target according to shown in formula (15) and (16), i.e. one-dimensional*Contained sample point
The dull number of increasing or decreasing is maximum, so that it is determined that calculating the mode of monotonicity.
(6) optimization aim shown in constructive formula (17), with SADE algorithm to monotonicity target lmax-numIt optimizes, seeks
Ideal dimensionality reduction mapping function coefficient lambda and fitting function coefficient ξ are looked for, so that it is determined that corresponding dimensionality reduction mapping function f and fitting letter
Number F.
Step 4: experimental verification and analysis
(1) model experiment and interpretation of result in sample set
The ultrasound detection parameter (containing 10 samples) and crystallite dimension extracted using table 1,2 is experiment basis, with selectionFor input, the undetermined coefficient of function f can be obtained according to formula (11), and
Fitting function coefficient is obtained by solving optimization problem to establish crystallite dimension soft-sensing model.
When establishing the evaluation model based on monotonicity, all characteristic parameter sample values are all as crystallite dimension has made new row
Sequence and ultrasound parameter Y ' is inputted while choosing 10 samples and being basis, while also calculated with the minimum target of error
Multi-parameter ultrasound Evaluation (Multi-parameter Ultrasonic Evaluation, MUE) model.Establish ultrasound and crystalline substance
The relation curve of particle size (average value, standard deviation) is expressed as f=ξ1X+ξ2, obtain corresponding dimensionality reduction mapping function coefficient and be shown in Table 5
With table 7.Using sample data in table 1 as foundation, respectively by the velocity of sound average value of extraction, attenuation coefficient average value and nonlinear factor
Average value is corresponding with crystallite dimension and draws relation curve, is directly built each ultrasound parameter with crystallite dimension with least square method
Vertical model of fit, obtained fitting evaluation model form are denoted as Γ (τ)=ξ1X+ξ2, wherein (τ=1,2,3) illustrates 3 kinds of tradition
Measure the classification of evaluation model.The undetermined coefficient value of gained model, dull number, error amount and the related coefficient such as table of fitting
6 and table 8 shown in.Fig. 6 is the model and fit correlation curve for 5 kinds of evaluation crystallite dimension average value that the embodiment of the present invention 3 provides,
Wherein, part (a) of Fig. 6 indicates MUEBM evaluation relation curve, and part (b) of Fig. 6 indicates MUE evaluation relation curve, Fig. 6's
(c) part indicates the relation curve of the velocity of sound and crystallite dimension average value, and part (d) of Fig. 6 indicates attenuation coefficient and crystallite dimension
Part (e) of average value relation curve, Fig. 6 indicates nonlinear factor and crystallite dimension average value relation curve.Fig. 7 is the present invention
The model and fit correlation curve for 5 kinds of evaluation crystallite dimension standard deviations that embodiment 3 provides, wherein part (a) of Fig. 7 indicates
MUEBM evaluation relation curve, part (b) of Fig. 7 indicate MUE evaluation relation curve, and part (c) of Fig. 7 indicates the velocity of sound and crystal grain
Part (d) of the relation curve of dimensional standard difference, Fig. 7 indicates attenuation coefficient and crystallite dimension standard deviation relation curve, Fig. 7's
(e) part indicates nonlinear factor and crystallite dimension standard deviation relation curve.
Table 5 is using crystallite dimension average value as the undetermined coefficient of MUEBM, MUE model mapping function of target
Table 6 is using crystallite dimension average value as the evaluation model parameter of target
Table 7 is using crystallite dimension standard deviation as the undetermined coefficient of MUEBM, MUE model mapping function of target
Table 8 is using crystallite dimension standard deviation as the evaluation model parameter of target
Several Parameters in Mathematical Model established for ultrasound parameter and crystallite dimension average value are shown in Table 6.With the primary of foundation
For fitting of a polynomial model,The mean absolute error value of model is 0.0729,Model and other 3 kinds of models point
It Wei 0.0969,0.2068,0.1999,0.1983, it is seen thatModel is for average exhausted between calculated value and match value
To difference minimum, whereinModel is also an advantage over other 3 kinds of models.The sample point contained by the calculated value successively monotonic increase or is passed
From the point of view of the maximum number statistical conditions subtracted, sample point used in model is 10,Model andModel meets most
Big dullness number is 7 and 5, and other 3 kinds of models are respectively 5,5,3, it is clear thatModel has the advantage of monotonicity,
Meanwhile the related coefficient of model provided by the invention is 0.9307, all shows advantage than other 4 kinds of models.AlthoughWithIt can
To be fitted with 2 order polynomials, obtained mean absolute error value less than 1 rank model of fit, but from (c) of Fig. 7 partially and
(e) it partially can be seen that 2 rank fit solutions will cause evaluation failure, it is contemplated that 1 rank fit correlation model is more appropriate.
It can be seen from figure 7 thatCompactness it is very strong and the monotonic performance that meets is preferable, and other 4 kinds of models
Discrimination it is very big, and the state of dull sex expression is poor, in additionThe registration of model is also relatively strong, but shown list
Tune trend is inferior toAngularly evaluation model is judged from monotonicity, correlation, discrimination, mean absolute error,Model is an advantage over other 4 kinds of models, and the error shown is minimum, and correlation is most strong and monotonic performance is best, and comprehensive point
Analysis show that crystallite dimension soft-sensing model provided by the invention is to have stronger characterization ability in sample set.
Several Parameters in Mathematical Model established for ultrasound parameter and crystallite dimension standard deviation are shown in Table 8.According to related in table
Parameter analysis it is found that for establishing a fitting of a polynomial model,The mean absolute error value of model is
0.0342, and other 3 kinds of models are respectively 0.0820,0.2325,0.1912, hence it is evident that find outModel for calculated value with
The error amount counted between match value is minimum, and wherein the error amount of attenuation coefficient model is maximum and characterization ability is on the weak side.From calculating
Sample point contained by value is successively from the point of view of monotonic increase or the number statistical conditions successively decreased,The monotonicity of model meets completely single
Tonality and number are 9, in addition Γs(1)、Γs(2) and Γs(3) dull number is respectively 6,5,4, it is seen that the dullness of this model
Performance be it is best, make model level off to good situation substantially, related coefficient is -0.9868 phase for being also an advantage over other models
Relationship number and substantially withModel is at the same level.WithModel compares,The mean absolute error value of model is
0.0247 smaller and slightly aobvious advantage, butThe dull section that model is presented is only 6, shows as non-fully monotonicity and is inferior toThe total monotonicity of model.Similarly, in Fig. 7 in (d), (e) and table 8 under 2 rank fit solutions, though it is flat to obtain model
Equal absolute error value becomes smaller, but will cause the phenomenon that can not commenting and evaluation result is caused to fail, and then only considers 1 rank model of fit
Situation.
As can be seen from Figure 7,The fit line and optimizing line of model are substantially at coincidence status, and dull pass is presented
Subtract trend.And the lines discrimination of the actual value and match value of attenuation coefficient and nonlinear system exponential model is very big.For other 3 kinds
For model, only velocity of sound model two lines item is slightly proximal to.In additionThe registration of model is stronger, but is displayed as non-fully
Monotone decreasing trend.Angularly evaluation model is judged from monotonicity, correlation, discrimination, mean absolute error, this hair
Bright offerModel is that have stronger characterization ability, the result figure of comprehensive each performance show better than other 3
Kind model.
(2) the model verifying and interpretation of result outside sample set
The model established to the average and standard deviation of corresponding ultrasonic feature parameter and crystallite dimension carries out outside sample set
Verifying and interpretation of result.WithWithFor 2 test samples of crystallite dimension average value, respectively 11.67 μm and 13.64 μ
M, withWithFor 2 test samples of crystallite dimension standard deviation, respectively 3.64 μm and 5.79 μm.By T1、T2Test sample
Ultrasonic feature parameter inputs, and is calculated respectively with MUEBM method, MUE method, sound velocity method, attenuation coefficient method and nonlinear factor method
The associated ultrasonic parameter of test sample obtains corresponding crystallite dimension (average value, standard deviation), by acquired results and uses metallographic method
Measured crystallite dimension (average value, standard deviation) is made comparisons analysis.It is to judge with the accuracy of evaluation result and relative error magnitudes
The foundation of model validation, each model are shown in Table 9 and table 10 to the evaluation result of the average and standard deviation of crystallite dimension respectively.
Table 9 is using crystallite dimension average value as 5 kinds of model evaluation Comparative results of target
Table 10 is using crystallite dimension standard deviation as 5 kinds of model evaluation Comparative results of target
ForSample, nonlinear factor method model evaluation result error is maximum, reaches 4.14 μm, MUEBM model evaluation
As a result minimum is only 0.26 μm, and the evaluation result precision of crystallite dimension soft-sensing model (MUEBM model) provided by the invention is
Most preferably.ForSample, the evaluation result of MUEBM model are 0.06 μm, be an advantage over general 3 kinds of models, and MUE model is commented
Valence result is also to be inferior to MUEBM model.From relative error angle analysis, for 2 test samples, the phase of MUEBM model
It is respectively -2.23%, 0.44% to error amount, is all shown as simultaneously compared to other 4 kinds of models best.It can be seen that crystal grain ruler
The MUEBM model that very little average value the is established advantage small with evaluation result precision height and relative error.
Similarly, forThe evaluation result of sample, attenuation coefficient method is best, and MUEBM model evaluation is the result is that be better than other 3
Kind of model, and the performance of the evaluation result of MUE model is worst reaches 3.65 μm.ForSample, attenuation coefficient model evaluation result
Deviation is maximum, and up to 4.64 μm, sound velocity method model evaluation result error is minimum, and only 0.52 μm, and the evaluation result of MUEBM method is excellent
Maintain an equal level in other methods and substantially with sound velocity method.From the angle comprehensive analysis of relative error, wherein attenuation coefficient method model (Sample
This), sound velocity method model (Sample) relative error is smaller, but the more unilateral test knot for not ensuring that 2 samples of evaluation result
The aobvious advantage of fruit.And MUEBM model is the most reliable and the most stable to the evaluation result of 2 test samples and better than traditional 3 kinds of model,
The result that wherein the verification result precision and relative error of 2 samples is presented in MUE model is inferior to this model.It can be seen that crystalline substance
The MUEBM model overall assessment result precision height and relative error magnitudes that particle size standard deviation is established are smaller, show that MUEBM model exists
The validity of the outer validation test of sample set.
(3) MUEBM model validation is analyzed
Comprehensive analysis is it is found that accurate for MUEBM model evaluation result constructed by crystallite dimension (average value, standard deviation)
Property is much better than MUE model, using error as the Optimized model of target due to structure is complicated for crystallite dimension microcosmic for TC4 titanium alloy
It only can guarantee the reliability of model in sample and cannot be guaranteed the validity of evaluation result.Then the excellent of monotonicity has been drafted again
Change target, so that optimizing line and fit line can guarantee dull orderly trend, to ensure that the validity of measurement result, with list
3 kinds of models based on one parameter have equally shown advantage compared to result.This is because the microcosmic crystallite dimension of material and defect are mutual
Phase separation difference will cause the variation of ultrasound parameter, and gained ultrasonic feature parameter includes that the information of primary alpha phase crystallite dimension is different.
MUEBM evaluation method has comprehensively considered crystallite dimension (average value) to velocity of sound standard deviation, attenuation coefficient average value, secondary bottom wave frequency
The response of the poor 4 ultrasonic feature parameters of rate peak value, relative nonlinear factor standard also considers crystallite dimension (standard deviation) simultaneously
It is special to velocity of sound standard deviation, attenuation coefficient average value, secondary bottom frequency offset amount, poor 4 ultrasounds of relative nonlinear factor standard
Influencing and being that important policy goals carry out Optimized model with monotonicity for sign parameter, by dimensionality reduction, fitting and optimization processing, is rejected
The interference of redundancy ultrasound parameter.For constructing measurement model, although with the minimum optimizing evaluation model found of error
Error very little, but will form the even invalid model of nonmonotonic model;Opposite, it is found using monotonicity by target
Error is small in optimizing evaluation model M UEBM model, monotonicity is good and evaluation result outside sample set is good, and model presents
The result precision of regular monotonic trend and evaluation is small and robustness is good.
It is strong and have good that MUEBM measurement model has merged effective multi-parameter ultrasonic feature and crystallite dimension correlation
Anti-interference ability.In 3 kinds of methods that single ultrasound parameter constructs, join although choosing the ultrasound strong with crystallite dimension correlation
Number, but single ultrasound parameter contain crystallite dimension detection acoustic information it is less, the acoustics all contained can not be covered
Information content, the model of fit anti-interference ability directly established is weak and characterization result is extremely unstable, is applied to and detects and comment in practice
Valence is undesirable.Opposite, MUEBM evaluation method characterizes ability with regard to ideal, and comprehensive analysis simultaneously symbolizes crystallite dimension
Size and dispersion degree distributing homogeneity.
The beneficial effect that the present embodiment can be realized is:
1) for TC4 titanium alloy primary alpha phase crystallite dimension, structure is complicated that information increases, and MUE model and MUEBM model have
There is error similitude.The former evaluation result is undesirable and cannot be guaranteed good evaluation effect.Using monotonicity as the evaluation of target
It is notably complete monotonicity and evaluation result precision is high, relative error is small that the monotonicity that model is presented is most ideal.Illustrate pair
Consider that error has lacked validity for modeling target in such Titanium Alloys for Aviation material for having complex information crystallite dimension, and
It introduces monotonicity target and guarantees that evaluation effect obtained from model error minimum is best.
2) when building is using monotonicity as the Optimized model of target, the response of multi-parameter need to be considered simultaneously.That is: preferably multiple
Ultrasonic feature parameter is handled through means such as mapping, normalization, fittings, is formulated monotonicity and is optimization aim and combines SADE algorithm excellent
Change this target, determine optimal mapping function and fitting function coefficient, ideal MUEBM evaluation model can be obtained.
3) through experimental analysis, MUEBM measurement model has highlighted more superior performance.Compared with MUE method, MUEBM considers
With monotonicity is important influence factor and has shown good evaluation result;With velocity of sound model, attenuation coefficient model and
Nonlinear system exponential model is compared, and new method combines the global information to crystallite dimension response, can not only characterize crystallite dimension
True value, and the dispersion degree that crystallite dimension deviates true value, i.e. uniformity distribution situation can be analyzed.With preferred special
Sign parameter constructs the evaluation model that precision is high, monotonicity is good and error is small, and the result figure of gained model is intuitive complete and has
Stable evaluation effect.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For system disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
Used herein a specific example illustrates the principle and implementation of the invention, and above embodiments are said
It is bright to be merely used to help understand method and its core concept of the invention;At the same time, for those skilled in the art, foundation
Thought of the invention, there will be changes in the specific implementation manner and application range.In conclusion the content of the present specification is not
It is interpreted as limitation of the present invention.
Claims (8)
1. a kind of determination method of the alloy grain size based on mapping monotonicity, which is characterized in that the determining method includes:
Obtain ultrasound the fixed point scanning signal, average thickness values, crystallite dimension value of each experiment sample;
Each ultrasound inspection of each experiment sample is determined according to the average thickness values and the ultrasound fixed point scanning signal
Survey parameter value;
Obtain section step-length, lowest threshold and current selection moment;
The selection moment corresponding selection section is determined according to the section step-length and the selection moment;
According to the lowest threshold, the selection section, each ultrasound detection parameter value and the crystallite dimension value, use
Relativity measurement criterion determines effective ultrasound detection parameter at each selection moment;
Final effectively ultrasound detection parameter set is determined according to effective ultrasound detection parameter at each selection moment;
It is up to optimization aim with monotonicity, crystallite dimension hard measurement mould is established according to the final effectively ultrasound detection parameter set
Type;
The crystallite dimension of tested alloy is determined using the crystallite dimension soft-sensing model.
2. determining method according to claim 1, which is characterized in that it is described according to the section step-length and the selection when
It carves and determines the selection moment corresponding selection section, specifically include:
According to formula:Determine the selection moment corresponding selection section, wherein t indicates selection
Moment,Indicate section step-length, θtIndicate the corresponding selection section selection moment t.
3. determining method according to claim 2, which is characterized in that described according to the lowest threshold, the selection area
Between, each ultrasound detection parameter value and the crystallite dimension value, each selection moment is determined using relativity measurement criterion
Effective ultrasound detection parameter, specifically include:
Judge whether the maximum value in the selection section is more than or equal to lowest threshold, obtains the first judging result;
When first judging result indicates the maximum value in the selection section more than or equal to lowest threshold, using Pierre
Gloomy correlation analysis method calculates separately the crystallite dimension value of each experiment sample and each of each experiment sample surpasses
The related coefficient of sound detection parameter value obtains each size-parameter related coefficient;
Ultrasound detection parameter value corresponding ultrasound detection parameter of the size-parameter related coefficient in the selection section is selected to make
For primary election ultrasound detection parameter;
The flat of each primary election ultrasound detection parameter of each experiment sample is calculated separately using Pearson correlation coefficients analysis method
Related property coefficient;
Average correlation coefficient is selected to be less than the primary election ultrasound detection parameter and average correlation of the minimum value in the selection section
Coefficient is greater than the maximum primary election ultrasound detection parameter of average correlation coefficient for selecting section minimum value as effective ultrasound
Detection parameters;
The selection moment is updated, and " selection section is determined according to the section step-length and the selection moment " described in return.
4. determining method according to claim 1, which is characterized in that it is described that optimization aim is up to monotonicity, according to
The final effectively ultrasound detection parameter set establishes crystallite dimension soft-sensing model, specifically includes:
According to each effective ultrasound detection parametric configuration multidimensional actual parameter vector;
Dimensionality reduction mapping function is constructed, and using the dimensionality reduction mapping function that multidimensional actual parameter vector drop is effective at one-dimensional
Parameter;
The one-dimensional actual parameter is normalized, normalization one-dimensional actual parameter is obtained;
Construct the first fitting function, the dependent variable of first fitting function is crystallite dimension, first fitting function from
Variable is normalization one-dimensional actual parameter;
Inverse transformation is carried out to first fitting function, obtains the second fitting function, the dependent variable of second fitting function is
One-dimensional actual parameter is normalized, the independent variable of second fitting function is crystallite dimension;
Difference with the corresponding dependent variable of adjacent independent variable of second fitting function be all positive number or be all negative maximum
Number is target, constitution optimization function;
The majorized function is solved using adaptive differential evolution algorithm, obtains the adjacent independent variable for making second fitting function
The difference of corresponding dependent variable be all positive number or be all negative the maximum optimal dimensionality reduction coefficient of number and optimal fitting coefficient,
In, the optimal dimensionality reduction coefficient is the optimal coefficient of the dimensionality reduction mapping function, and the optimal fitting coefficient is described first quasi-
Close the optimal coefficient of function;
The optimal fitting coefficient is substituted into first fitting function, obtains crystallite dimension soft-sensing model.
5. a kind of determination system of the alloy grain size based on mapping monotonicity, which is characterized in that the determining system includes:
Sample parameter obtains module, for obtaining ultrasound the fixed point scanning signal, average thickness values, crystal grain ruler of each experiment sample
Very little value;
Ultrasound detection parameter value determining module, it is every for being determined according to the average thickness values and the ultrasound fixed point scanning signal
Each ultrasound detection parameter value of a experiment sample;
Selection parameter obtains module, for obtaining section step-length, lowest threshold and current selection moment;
Section determining module is selected, for determining that the selection moment is corresponding according to the section step-length and the selection moment
Select section;
Effective ultrasound detection parameter determination module, for being examined according to the lowest threshold, the selection section, each ultrasound
Parameter value and the crystallite dimension value are surveyed, determines that effective ultrasound detection at each selection moment is joined using relativity measurement criterion
Number;
Final effectively ultrasound detection parameter set determining module, for being determined according to effective ultrasound detection parameter at each selection moment
Final effectively ultrasound detection parameter set;
Crystallite dimension soft-sensing model establishes module, for being up to optimization aim with monotonicity, according to described final effectively super
Sound detection parameter set establishes crystallite dimension soft-sensing model;
Crystallite dimension measurement module, for determining the crystallite dimension of tested alloy using the crystallite dimension soft-sensing model.
6. determining system according to claim 5, which is characterized in that selection section determining module is according to formula:Determining the selection moment corresponding selection section, wherein t indicates the selection moment,Indicate area
Between step-length, θtIndicate the corresponding selection section selection moment t.
7. determining system according to claim 6, which is characterized in that effective ultrasound detection parameter determination module is specific
Include:
First judging unit obtains for judging whether the maximum value in the selection section is more than or equal to lowest threshold
One judging result;
Size-parameter related coefficient determination unit, for indicating the maximum value in the selection section when first judging result
When more than or equal to lowest threshold, the crystal grain ruler of each experiment sample is calculated separately using Pearson correlation coefficients analysis method
The related coefficient of very little value and each ultrasound detection parameter value of each experiment sample, obtains each size-parameter phase relation
Number;
Primary election ultrasound detection choice of parameters unit, for selecting ultrasound of the size-parameter related coefficient in the selection section
The corresponding ultrasound detection parameter of parameter values for detection is as primary election ultrasound detection parameter;
Average correlation coefficient calculation unit, for calculating separately each experiment sample using Pearson correlation coefficients analysis method
Each primary election ultrasound detection parameter average correlation coefficient;
Effective ultrasound detection parameter selection unit, for selecting average correlation coefficient to be less than the minimum value for selecting section
The average correlation coefficient that primary election ultrasound detection parameter and average correlation coefficient are greater than the selection section minimum value is maximum
Primary election ultrasound detection parameter is as effective ultrasound detection parameter;
Moment updating unit is selected, for updating the selection moment.
8. determining system according to claim 5, which is characterized in that the crystallite dimension soft-sensing model establishes module tool
Body includes:
Multidimensional actual parameter vector structural unit, for according to each effective ultrasound detection parametric configuration multidimensional actual parameter to
Amount;
Dimensionality reduction unit, for constructing dimensionality reduction mapping function, and using the dimensionality reduction mapping function by the multidimensional actual parameter to
Amount drop is at one-dimensional actual parameter;
Normalized unit obtains normalization one-dimensional and effectively joins for the one-dimensional actual parameter to be normalized
Number;
Fitting function structural unit, for constructing the first fitting function, the dependent variable of first fitting function is crystallite dimension,
The independent variable of first fitting function is normalization one-dimensional actual parameter;
Inverse transformation block obtains the second fitting function, second fitting for carrying out inverse transformation to first fitting function
The dependent variable of function is normalization one-dimensional actual parameter, and the independent variable of second fitting function is crystallite dimension;
Majorized function structural unit, the difference for the corresponding dependent variable of adjacent independent variable with second fitting function are all
Positive number or the maximum number for being all negative are target, constitution optimization function;
Adaptive differential evolution algorithm solves unit, for solving the majorized function using adaptive differential evolution algorithm, obtains
Must make the corresponding dependent variable of adjacent independent variable of second fitting function difference be all positive number or be all the number of negative most
Big optimal dimensionality reduction coefficient and optimal fitting coefficient, wherein the optimal dimensionality reduction coefficient is the optimal of the dimensionality reduction mapping function
Coefficient, the optimal fitting coefficient are the optimal coefficient of first fitting function;
Soft-sensing model determination unit obtains crystal grain ruler for the optimal fitting coefficient to be substituted into first fitting function
Very little soft-sensing model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810768923.XA CN109033586B (en) | 2018-07-13 | 2018-07-13 | Method and system for determining alloy grain size based on mapping monotonicity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810768923.XA CN109033586B (en) | 2018-07-13 | 2018-07-13 | Method and system for determining alloy grain size based on mapping monotonicity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109033586A true CN109033586A (en) | 2018-12-18 |
CN109033586B CN109033586B (en) | 2022-08-12 |
Family
ID=64641056
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810768923.XA Active CN109033586B (en) | 2018-07-13 | 2018-07-13 | Method and system for determining alloy grain size based on mapping monotonicity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109033586B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109726493A (en) * | 2019-01-04 | 2019-05-07 | 厦门大学 | A kind of gas component concentrations online soft sensor method in shell residual oil gasifier burner hearth |
CN111044614A (en) * | 2019-12-16 | 2020-04-21 | 南昌航空大学 | High-temperature alloy grain size circle-like mapping ultrasonic evaluation method |
CN113017650A (en) * | 2021-03-12 | 2021-06-25 | 南昌航空大学 | Electroencephalogram feature extraction method and system based on power spectral density image |
CN113139356A (en) * | 2021-04-27 | 2021-07-20 | 中国矿业大学 | Structural parameter optimization method for cylindrical switched reluctance electric linear motor |
CN113362909A (en) * | 2021-06-02 | 2021-09-07 | 燕山大学 | Method for evaluating grain structure uniformity in alloy steel forging |
CN113804591A (en) * | 2021-09-03 | 2021-12-17 | 南昌航空大学 | High-dimensional ultrasonic evaluation method for grain size of nickel-based alloy |
CN114047211A (en) * | 2021-11-10 | 2022-02-15 | 北京理工大学 | Method for detecting austenite grain diameter of elastic steel material based on EBSD |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE19839512A1 (en) * | 1998-08-29 | 2000-11-30 | Christoph Berthold | Process to obtain statements about the geometry (grain shape) of the particles to be measured with the help of laser diffraction devices, which are commonly used for grain size measurement |
CN101220425A (en) * | 2008-01-24 | 2008-07-16 | 东北大学 | High-strength nano-level crystal nickel material and method of manufacturing the same |
CN104297110A (en) * | 2014-09-19 | 2015-01-21 | 中南大学 | Crystal grain size ultrasonic non-destructive evaluation method without thickness measurement |
CN104749251A (en) * | 2015-04-09 | 2015-07-01 | 中南大学 | Grain size ultrasonic evaluation method without influence of underwater sound distance |
-
2018
- 2018-07-13 CN CN201810768923.XA patent/CN109033586B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE19839512A1 (en) * | 1998-08-29 | 2000-11-30 | Christoph Berthold | Process to obtain statements about the geometry (grain shape) of the particles to be measured with the help of laser diffraction devices, which are commonly used for grain size measurement |
CN101220425A (en) * | 2008-01-24 | 2008-07-16 | 东北大学 | High-strength nano-level crystal nickel material and method of manufacturing the same |
CN104297110A (en) * | 2014-09-19 | 2015-01-21 | 中南大学 | Crystal grain size ultrasonic non-destructive evaluation method without thickness measurement |
CN104749251A (en) * | 2015-04-09 | 2015-07-01 | 中南大学 | Grain size ultrasonic evaluation method without influence of underwater sound distance |
Non-Patent Citations (3)
Title |
---|
WEI WU等: "Ultrasonic evaluation of microstructure of titanium alloy TC4 based on optimization", 《15TH ASIA PACIFIC CONFERENCE FOR NON-DESTRUCTIVE TESTING (APCNDT2017)》 * |
XI CHEN等: "Study of the relationship between ultrasonic properties and microstructure of nickel-based superalloy GH706", 《ULTRASONICS》 * |
于漫漫: "TC4钛合金组织超声信号特征研究", 《中国优秀硕士学位论文全文数据库 工程科技I辑》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109726493A (en) * | 2019-01-04 | 2019-05-07 | 厦门大学 | A kind of gas component concentrations online soft sensor method in shell residual oil gasifier burner hearth |
CN109726493B (en) * | 2019-01-04 | 2021-03-23 | 厦门大学 | Online soft measurement method for gas component concentration in hearth of residual oil gasification furnace |
CN111044614A (en) * | 2019-12-16 | 2020-04-21 | 南昌航空大学 | High-temperature alloy grain size circle-like mapping ultrasonic evaluation method |
CN111044614B (en) * | 2019-12-16 | 2022-03-29 | 南昌航空大学 | High-temperature alloy grain size circle-like mapping ultrasonic evaluation method |
CN113017650A (en) * | 2021-03-12 | 2021-06-25 | 南昌航空大学 | Electroencephalogram feature extraction method and system based on power spectral density image |
CN113139356A (en) * | 2021-04-27 | 2021-07-20 | 中国矿业大学 | Structural parameter optimization method for cylindrical switched reluctance electric linear motor |
CN113362909A (en) * | 2021-06-02 | 2021-09-07 | 燕山大学 | Method for evaluating grain structure uniformity in alloy steel forging |
CN113362909B (en) * | 2021-06-02 | 2022-07-01 | 燕山大学 | Method for evaluating grain structure uniformity in alloy steel forging |
CN113804591A (en) * | 2021-09-03 | 2021-12-17 | 南昌航空大学 | High-dimensional ultrasonic evaluation method for grain size of nickel-based alloy |
CN113804591B (en) * | 2021-09-03 | 2023-05-12 | 南昌航空大学 | High-dimensional ultrasonic evaluation method for nickel-based alloy grain size |
CN114047211A (en) * | 2021-11-10 | 2022-02-15 | 北京理工大学 | Method for detecting austenite grain diameter of elastic steel material based on EBSD |
Also Published As
Publication number | Publication date |
---|---|
CN109033586B (en) | 2022-08-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109033586A (en) | Determination method and determining system based on the alloy grain size of mapping monotonicity | |
CN105930976B (en) | Node voltage sag severity comprehensive evaluation method based on weighted ideal point method | |
CN108490081B (en) | Alloy grain size determination method and system based on multiple parameters | |
Sause | Identification of failure mechanisms in hybrid materials utilizing pattern recognition techniques applied to acoustic emission signals | |
Li et al. | Laser cladding state recognition and crack defect diagnosis by acoustic emission signal and neural network | |
CN111488713A (en) | Method, system and storage medium for predicting early carbonization of concrete | |
CN110722285A (en) | Laser hot wire welding seam forming quality prediction method, system and medium | |
Sause et al. | Quantification of the uncertainty of pattern recognition approaches applied to acoustic emission signals | |
CN106650250B (en) | Infrared Multi-Target threat sequercing method based on equivalent evaluation points | |
Karimian et al. | Acoustic emission signal clustering in CFRP laminates using a new feature set based on waveform analysis and information entropy analysis | |
CN114692302B (en) | Fatigue crack detection method and system based on Gaussian mixture model | |
Chen et al. | A multi-parameter ultrasonic evaluation of mean grain size using optimization | |
Li et al. | Structural condition monitoring and identification of laser cladding metallic panels based on an acoustic emission signal feature optimization algorithm | |
Wang et al. | Tracing fracture damage evolution and identifying damage patterns in cast steel using advanced acoustic emission analysis | |
CN103984856A (en) | System and method for predicting longitudinal crack width of wall body based on neural network | |
CN113947280A (en) | Combined evaluation method based on feedback adjustment weight | |
CN109255394B (en) | Forced oscillation identification method based on mode similarity | |
CN116579657A (en) | Method for evaluating quality of metal lattice sandwich structure product in additive manufacturing | |
Chiou et al. | Techniques and software tools for estimating ultrasonic signal-to-noise ratios | |
Sun et al. | Physics-based deep learning framework for terahertz thickness measurement of thermal barrier coatings | |
Kłysz et al. | Good practice for fatigue crack growth curves description | |
CN114580884A (en) | Aerospace target threat assessment method based on multi-characteristic hierarchical structure and self-adaptive approximation | |
CN115171816A (en) | Comprehensive evaluation method for application verification of key materials of aviation equipment | |
Zhao et al. | Automated quantification of small defects in ultrasonic phased array imaging using AWGA-gcForest algorithm | |
CN112085316A (en) | Steel rail quality integral evaluation 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 |