CN102681473A - Fault detecting method for sulfur flotation process on basis of texture unit distribution - Google Patents

Fault detecting method for sulfur flotation process on basis of texture unit distribution Download PDF

Info

Publication number
CN102681473A
CN102681473A CN2012100947093A CN201210094709A CN102681473A CN 102681473 A CN102681473 A CN 102681473A CN 2012100947093 A CN2012100947093 A CN 2012100947093A CN 201210094709 A CN201210094709 A CN 201210094709A CN 102681473 A CN102681473 A CN 102681473A
Authority
CN
China
Prior art keywords
sulphur
statistic
pivot
floatation process
texture cell
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN2012100947093A
Other languages
Chinese (zh)
Inventor
桂卫华
朱红求
何明芳
阳春华
凌弈秋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Central South University
Original Assignee
Central South University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Central South University filed Critical Central South University
Priority to CN2012100947093A priority Critical patent/CN102681473A/en
Publication of CN102681473A publication Critical patent/CN102681473A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

The invention discloses a fault detecting method for sulfur flotation process on the basis of texture unit distribution. Textural features of real-time froth image during a sulfur flotation process is extracted on the basis of a texture unit, nonparametric kernel density estimation operator is designed to approach the texture unit distribution of sulfur flotation froth to obtain dynamic weight coefficient, a pivot element model based on the dynamic weight coefficient is established by a principal component analysis method, and faults during the sulfur floatation process can be effectively detected on the T2-statistic of the pivot element model. The fault detecting method for sulfur flotation process on the basis of the texture unit distribution is simple, convenient, fast, applicable to the sulfur flotation process, can detect faults of the sulfur flotation process caused by unreasonable operations, and are significant for reducing erroneous judgment rate of working conditions, improving grade of sulfur concentrate and realizing optimization of the sulfur flotation processing operation.

Description

A kind of sulphur floatation process fault detection method that distributes based on texture cell
Technical field
The invention belongs to fields such as image processing techniques and probability statistics, be specially a kind of sulphur floatation process fault detection method that distributes based on texture cell.
Background technology
Floatation process is the beneficiation method of widespread use in the mineral processing, and in ore pulp, adding floating agent usually improves mineral surface wettability difference, makes the strong useful purpose mineral of hydrophobicity from useless gangue, separate, and reaches the purpose that improves head grade.The sulfide mineral surface moist is very little, and hydrophobicity is strong, need not to add any medicament in the sulphur flotation, obtains the sulphur concentrate than high-grade through regulating liquid level and the air amount of blasting.
The height of sulphur concentrate grade depends on that sulphur floatation process operating mode is good and bad, and the detection of sulphur floatation process fault directly has influence on the sulphur concentrate grade.Therefore, researching and developing a kind of effective fault detection method has great importance for improving the sulphur flotation quality of production.
At present, the sulphur floatation process adopts the method for artificial fixed point observation flotation froth to come the failure judgement operating mode, and the artificial experience operation is unreasonable, can not in time adjust fault condition, is prone to cause the sulphur concentrate grade low.Therefore; To sulphur floatation process characteristics, research adopts a kind of method that distributes based on texture cell that sulphur flotation fault is detected based on the fault detection method of machine vision; For operating personnel provide simple and direct fault detect means; For stablizing sulphur flotation operating mode, improve the sulphur concentrate grade, realize that the optimal control of sulphur floatation process has important meaning.
Summary of the invention
The objective of the invention is to solve the detection problem of sulphur floatation process fault, proposed a kind of sulphur floatation process fault detection method that distributes based on texture cell.Main contents of the present invention are following:
A kind of sulphur floatation process fault detection method that distributes based on texture cell is characterized in that, may further comprise the steps:
The first step; Image library according to sulphur floatation foam image structure under the nominal situation; Based on the texture cell texture feature extraction; The design norm of nonparametric kernel density estimates that operator approaches the actual texture cell distribution of foam and obtains dynamic weight coefficient, adopts the method for pivot analysis to set up the principal component model based on dynamic weight coefficient, and is last according to the T based on principal component model 2Statistic obtains failure determination threshold value.
Step 1: extract the foam textural characteristics based on texture cell (TU);
Figure 2012100947093100002DEST_PATH_IMAGE001
be 3
Figure 451541DEST_PATH_IMAGE002
3 neighborhoods; The gray-scale value of this neighborhood is designated as successively; Wherein is the gray-scale value of regional center pixel, the calculating of texture cell value
Figure DEST_PATH_IMAGE005
:
Figure 417452DEST_PATH_IMAGE006
Figure DEST_PATH_IMAGE007
Wherein
Figure 766394DEST_PATH_IMAGE008
;
Figure DEST_PATH_IMAGE009
;
Figure 218104DEST_PATH_IMAGE010
,
Figure DEST_PATH_IMAGE011
are the constant of confirming according to sulphur flotation operating mode;
Step 2: according to the texture cell value of calculating, the design norm of nonparametric kernel density estimates that operator approaches actual texture cell and distributes, and obtains describing the dynamic weight coefficient of textural characteristics;
According to the approximation of function principle, estimate that with the norm of nonparametric kernel density of following design operator approaches actual texture cell and distributes:
Figure 627088DEST_PATH_IMAGE012
Figure DEST_PATH_IMAGE013
is sulphur flotation froth texture cell distribution function in the formula;
Figure 164249DEST_PATH_IMAGE014
is the texture cell value; is the control input; It is the liquid level regulated value in the sulphur floatation process;
Figure 988985DEST_PATH_IMAGE016
is the weight coefficient of i kernel function;
Figure DEST_PATH_IMAGE017
is i kernel function;
Figure 663244DEST_PATH_IMAGE018
is the x axle mid point of i kernel function, and
Figure DEST_PATH_IMAGE019
is the window width of kernel function;
Step 3: adopt the method for pivot analysis to set up the principal component model based on dynamic weight coefficient, rule of thumb method is confirmed the pivot number;
If model is input as dynamic weight coefficient vector
Figure 367763DEST_PATH_IMAGE020
;
Figure DEST_PATH_IMAGE021
is
Figure 392220DEST_PATH_IMAGE022
ties up matrix;
Figure DEST_PATH_IMAGE023
is the data sample number, and
Figure 145281DEST_PATH_IMAGE024
is the input variable number., earlier modeling data is carried out normalization and handle to result's influence be convenient to mathematical processing for fear of the different dimensions of variable.Input variable after the note normalization is ; carried out pivot analysis; Adopt empirical method to confirm the pivot number,
Figure 62651DEST_PATH_IMAGE026
can be decomposed into:
Figure DEST_PATH_IMAGE027
Figure 761354DEST_PATH_IMAGE028
Figure DEST_PATH_IMAGE029
is the pivot matrix of loadings in the formula;
Figure 114844DEST_PATH_IMAGE030
gets sub matrix for pivot;
Figure DEST_PATH_IMAGE031
is residual matrix, and
Figure 520287DEST_PATH_IMAGE032
is the pivot number;
Step 4: according to T 2Statistic characterizes a kind of characteristics of estimating of principal component model interior change, calculates the T based on principal component model 2Statistic;
In order to detect the fluctuation of sulphur floatation process, adopt T based on principal component model 2The fault detection method of statistic.For
Figure DEST_PATH_IMAGE033
The dynamic weight coefficient of individual sample
Figure 584187DEST_PATH_IMAGE034
, T 2Statistic is defined as:
Figure DEST_PATH_IMAGE035
Figure 379973DEST_PATH_IMAGE036
Wherein
Figure 615783DEST_PATH_IMAGE023
is the data sample number;
Figure DEST_PATH_IMAGE037
be in
Figure 751098DEST_PATH_IMAGE038
matrix
Figure DEST_PATH_IMAGE039
OK;
Figure 843688DEST_PATH_IMAGE038
is made up of the score vector of
Figure 2137DEST_PATH_IMAGE032
the individual pivot that constitutes pca model,
Figure 510478DEST_PATH_IMAGE040
by the diagonal angle average of being formed with preceding
Figure 703562DEST_PATH_IMAGE032
pairing eigenwert of individual pivot.
Step 5: distribution obtains T based on F 2The control limit of statistic is limit as sulphur floatation process failure determination threshold value with this control;
T 2The calculating of the control limit of statistic:
Figure DEST_PATH_IMAGE041
Wherein
Figure 967053DEST_PATH_IMAGE023
is the data sample number;
Figure 612798DEST_PATH_IMAGE032
is the pivot number that is kept;
Figure 930690DEST_PATH_IMAGE042
is insolation level;
Figure DEST_PATH_IMAGE043
is corresponding to insolation level is
Figure 102915DEST_PATH_IMAGE042
; Degree of freedom is
Figure 412673DEST_PATH_IMAGE032
,
Figure 608031DEST_PATH_IMAGE044
critical value that distributes of F under the condition.Make T 2The control limit of statistic
Figure DEST_PATH_IMAGE045
As failure determination threshold value.
Second step, failure determination threshold value that the principal component model of setting up according to normal froth images storehouse obtains and real-time froth images T 2Sulphur floatation process state is judged in the comparison of statistics value.
Step 1: the T that uses preceding 4 step calculating real-time froth images in the first step 2Statistic
Step 2: according to current T 2Statistics value and threshold value
Figure 625852DEST_PATH_IMAGE045
Relatively detection sulphur floatation process fault.
If
Figure DEST_PATH_IMAGE047
can judge that then less fluctuation just takes place the sulphur floatation process.if ; Then alarm sulphur floatation process breaks down; Can in time instruct the sulphur floating operation through this statistic, take effective liquid level to regulate strategy.
According to the texture characteristic distributions of sulphur flotation froth, when approximating curve, selected 25 kernel functions for use, kernel function is a prototype with the Gaussian kernel function, makes up the kernel function that meets sulphur floatation process system:
Figure DEST_PATH_IMAGE049
What the present invention proposed estimates that with norm of nonparametric kernel density operator approaches actual texture cell and distributes; Adopt the method for pivot analysis to set up principal component model based on texture distribution dynamic weight coefficient; Can select suitable pivot number to represent the fluctuation information of sulphur flotation operating mode according to the actual requirements, and based on T 2Statistic detects fault in the pivot subspace, when the principal component model of sulphur floatation process texture distribution weight coefficient and foundation is not inconsistent, can judge that then existing fault takes place in this process.
Utilize the sulphur floatation process fault detection method that distributes based on texture of the present invention,, can make the sulphur concentrate grade improve 10% ~ 30% with respect to the artificial experience operation.
The present invention has the characteristics simply and easily of calculating; Be suitable for the detection of sulphur floatation process fault condition; Have stronger practicality, make the site operation personnel can quick and precisely judge sulphur floatation process fault condition, to reducing sulphur flotation operating mode False Rate; Optimize the sulphur floating operation, improve the sulphur concentrate grade and have great importance.
Description of drawings
Fig. 1 approaches the norm of nonparametric kernel density method of estimation figure that texture cell distributes;
Normal sulphur flotation froth of Fig. 2 and fault sulphur flotation froth figure;
Fig. 3 (a) and Fig. 3 (b) are respectively normal sulphur flotation froth and fault sulphur floatation foam image texture cell distribution plan;
The 3D mesh figure that texture cell distributed before and after Fig. 4 sulphur floatation process fault took place;
Fig. 5 is based on T 2The sulphur floatation process fault detect result of statistic;
Fig. 6 uses comparing result based on the fault detection method that texture cell distributes.
Embodiment
Technical scheme for a better understanding of the present invention further describes embodiment of the present invention below in conjunction with the accompanying drawing of instructions:
The first step; Image library according to sulphur floatation foam image structure under the nominal situation; Adopt the texture cell texture feature extraction; Design norm of nonparametric kernel density and estimate that operator approaches the actual texture cell distribution of foam and obtains dynamic weight coefficient, adopt the method for pivot analysis to set up principal component model, last T based on principal component model based on dynamic weight coefficient 2Statistic obtains failure determination threshold value.
Step 1: adopt the texture cell value to describe the foam textural characteristics;
One 3
Figure 629951DEST_PATH_IMAGE002
3 neighborhoods that texture cell (TU) is made up of each pixel and its eight fields pixel; The gray-scale value of this neighborhood is designated as
Figure 386555DEST_PATH_IMAGE003
successively; Wherein
Figure 100433DEST_PATH_IMAGE004
is the gray-scale value of regional center pixel, the calculating of
Figure 264524DEST_PATH_IMAGE001
:
? （1）
Wherein
Figure 421202DEST_PATH_IMAGE009
;
Figure 750552DEST_PATH_IMAGE010
; is the constant of confirming according to sulphur flotation operating mode; Because each unit of texture cell have 3 kinds of possibility values; So plant different values for each texture cell has
Figure DEST_PATH_IMAGE051
; We represent these value combinations with number
Figure 664336DEST_PATH_IMAGE005
, are designated as the texture cell value:
Figure 47913DEST_PATH_IMAGE006
（2）
Step 2: according to the texture cell value of calculating, the design norm of nonparametric kernel density estimates that operator approaches actual texture cell and distributes, and obtains describing the dynamic weight coefficient of textural characteristics;
According to the approximation of function principle, estimate that with the norm of nonparametric kernel density of following design operator approaches actual texture cell and distributes:
Figure 598980DEST_PATH_IMAGE012
（3）
Figure 726205DEST_PATH_IMAGE013
is sulphur flotation froth texture cell distribution function in the formula;
Figure 311907DEST_PATH_IMAGE014
is the texture cell value;
Figure 544348DEST_PATH_IMAGE015
is the control input; It is the liquid level regulated value in the sulphur floatation process;
Figure 848291DEST_PATH_IMAGE016
is the weight coefficient of i kernel function;
Figure 451310DEST_PATH_IMAGE017
is i kernel function;
Figure 829202DEST_PATH_IMAGE018
is the x axle mid point of i kernel function, and
Figure 492264DEST_PATH_IMAGE019
is the window width of kernel function;
Texture cell according to foam distributes, and when approximating curve, has selected 25 kernel functions for use, and kernel function is a prototype with the Gaussian kernel function, makes up the kernel function that meets sulphur floatation process system:
Figure 17924DEST_PATH_IMAGE049
（4）
Consider actual sulphur flotation operating mode; Select suitable
Figure 922612DEST_PATH_IMAGE019
value; Each kernel function all immobilizes; Can obtain the corresponding weight coefficient of each kernel function
Figure 490996DEST_PATH_IMAGE052
; Being used for characterizing the output texture cell distributes; The kernel method that designs approaches the situation (wherein, 1,2 be respectively actual texture distribution and estimate from the kernel method that designs) as shown in Figure 1 that actual texture distributes.
The error of bringing when the kernel function of considering design is approached actual texture cell distribution curve; if
Figure DEST_PATH_IMAGE053
; ;
Figure DEST_PATH_IMAGE055
Introduce error function :
Figure DEST_PATH_IMAGE057
（5）
Description to sulphur flotation froth texture cell distributes just changes into one group of dynamic weight coefficient.
Step 3: adopt the method for pivot analysis to set up the principal component model based on dynamic weight coefficient, rule of thumb method is confirmed the pivot number;
Adopt the method for pivot analysis to set up principal component model; If model is input as dynamic weight coefficient vector
Figure 253788DEST_PATH_IMAGE020
;
Figure 993074DEST_PATH_IMAGE021
is ties up matrix;
Figure 241838DEST_PATH_IMAGE023
is the data sample number, and
Figure 511146DEST_PATH_IMAGE024
is the input variable number., earlier modeling data is carried out normalization and handle to result's influence be convenient to mathematical processing for fear of the different dimensions of variable.If the mean vector of is ; The standard deviation vector is
Figure DEST_PATH_IMAGE059
; Then the input vector after the normalization is
Figure 978662DEST_PATH_IMAGE060
;
Figure DEST_PATH_IMAGE061
; The input variable of note after the normalization be
Figure 368056DEST_PATH_IMAGE025
, remembers mean vector and covariance matrix are and
Figure DEST_PATH_IMAGE063
For the data after the normalization;
Figure 138937DEST_PATH_IMAGE064
; , .
Figure 900406DEST_PATH_IMAGE026
carried out pivot analysis;
Figure 924862DEST_PATH_IMAGE024
individual eigenwert
Figure 490973DEST_PATH_IMAGE068
of at first trying to achieve
Figure DEST_PATH_IMAGE067
and corresponding unit orthogonal characteristic vector
Figure DEST_PATH_IMAGE069
thereof;
Figure 651696DEST_PATH_IMAGE066
; Wherein
Figure 340166DEST_PATH_IMAGE070
; Note
Figure DEST_PATH_IMAGE071
,
Figure 592199DEST_PATH_IMAGE072
.Definition has according to characteristic root:
Figure DEST_PATH_IMAGE073
（6）
Figure 367443DEST_PATH_IMAGE024
the individual pivot of
Figure 148951DEST_PATH_IMAGE026
is defined as
Figure 226815DEST_PATH_IMAGE074
;
Figure 897967DEST_PATH_IMAGE066
; And
Figure DEST_PATH_IMAGE075
arranged the variance of (
Figure 930514DEST_PATH_IMAGE076
for
Figure DEST_PATH_IMAGE077
),
Figure 65829DEST_PATH_IMAGE078
.
Owing to have high correlation between each input variable, the individual pivot of preceding
Figure 526461DEST_PATH_IMAGE032
(
Figure 684910DEST_PATH_IMAGE028
) with
Figure DEST_PATH_IMAGE079
just can be represented the most of fluctuation information (as more than 90%) in
Figure 521148DEST_PATH_IMAGE079
.The method of confirming
Figure 714232DEST_PATH_IMAGE032
can be used empirical method, promptly gets make
Figure 295572DEST_PATH_IMAGE080
.
Then
Figure 873184DEST_PATH_IMAGE026
can be decomposed into:
Figure 292850DEST_PATH_IMAGE028
?（7）
Figure 160312DEST_PATH_IMAGE029
is the pivot matrix of loadings in the formula;
Figure 276035DEST_PATH_IMAGE030
gets sub matrix for pivot;
Figure 178132DEST_PATH_IMAGE031
is residual matrix, and
Figure 726968DEST_PATH_IMAGE032
is the pivot number.
Step 4: according to T 2Statistic characterizes a kind of characteristics of estimating of principal component model interior change, calculates the T based on principal component model 2Statistic;
T 2Statistic is the standard quadratic sum of pivot score vector, indicates each sampled point on variation tendency and amplitude, to depart from the degree of model, is a kind of the estimating that characterizes the principal component model interior change.In order to detect the fluctuation of sulphur floatation process, design is based on T 2The fault detection method of statistic.For
Figure 284988DEST_PATH_IMAGE033
The dynamic weight coefficient of individual sample
Figure 1141DEST_PATH_IMAGE034
, T 2Statistic is defined as:
Figure 147957DEST_PATH_IMAGE035
Figure 861835DEST_PATH_IMAGE036
（8）
Wherein is the data sample number;
Figure 161415DEST_PATH_IMAGE037
be in
Figure 38104DEST_PATH_IMAGE038
matrix
Figure 860567DEST_PATH_IMAGE039
OK;
Figure 252234DEST_PATH_IMAGE038
is made up of the score vector of
Figure 451134DEST_PATH_IMAGE032
the individual pivot that constitutes principal component model, by the diagonal angle average of being formed with preceding
Figure 238011DEST_PATH_IMAGE032
pairing eigenwert of individual pivot.T 2Statistic reflects the situation that multivariate changes through the fluctuation of the inner principal component vector mould of principal component model.
Step 5: distribution obtains T based on F 2The control limit of statistic is limit as sulphur floatation process failure determination threshold value with this control;
T 2The control limit of statistic can be utilized F to distribute and calculate:
Figure 54657DEST_PATH_IMAGE041
（9）
Wherein
Figure 57248DEST_PATH_IMAGE023
is the data sample number;
Figure 722845DEST_PATH_IMAGE032
is the pivot number that is kept;
Figure 949427DEST_PATH_IMAGE042
is insolation level;
Figure 253369DEST_PATH_IMAGE043
is corresponding to insolation level is
Figure 856389DEST_PATH_IMAGE042
; Degree of freedom is ,
Figure 959660DEST_PATH_IMAGE044
critical value that distributes of F under the condition.Make T 2The control limit of statistic
Figure 423003DEST_PATH_IMAGE045
As failure determination threshold value.
Second step, failure determination threshold value that the principal component model of setting up according to normal froth images storehouse obtains and real-time froth images T 2Sulphur floatation process state is judged in the comparison of statistics value.
Step 1: the T of calculating real-time froth images 2Statistic
Figure 829713DEST_PATH_IMAGE046
, concrete steps like step 1 in the first step to shown in the step 4.
Step 2: if
Figure 186745DEST_PATH_IMAGE047
can judge that then less fluctuation just takes place the sulphur floatation process.if
Figure 20709DEST_PATH_IMAGE048
; Then alarm sulphur floatation process breaks down; Can in time instruct the sulphur floating operation through this statistic, take effective liquid level to regulate strategy.
Embodiment 1: certain plumbous zinc factory Zinc hydrometallurgy process sulphur floatation process is that example is explained superiority of the present invention.The first step is gathered the principal component model that the sulphur floatation foam image is set up based on the texture cell distribution under the nominal situation and is calculated sulphur floatation process failure determination threshold value; In second step, calculate T according to real-time froth images 2The statistics value compares with failure determination threshold value, judges sulphur floatation process state.Sulphur floatation foam image when normal sulphur floatation foam image that cleaner cell is gathered and fault take place is as shown in Figure 2.It is higher that the fault here refers to liquid level, causes turning over the ore pulp phenomenon.Normal sulphur floatation foam image that the kernel method that employing designs certainly approaches and fault sulphur floatation foam image texture distribution plan are shown in Fig. 3 (a) and Fig. 3 (b).Shown in Figure 4 is in a certain period continuous time, and along with the adjusting of liquid level, the 3D mesh figure (wherein, 3 are the place that breaks down) that the output texture cell distributed before and after fault took place is based on T 2The fault detect result of statistic (wherein, 4 are the place that breaks down, and a is the nominal situation sample, and b is the fail-safe control limit, and c is the fault condition sample) as shown in Figure 5, the testing result analysis is shown in subordinate list 1.Can know that from subordinate list 1 employing can detect sulphur floatation process fault based on the fault detection method that texture cell distributes, provide alarm, guiding operation personnel in time adjust liquid level, eliminate fault.
A sulphur concentrate grade value is chemically examined by the sky in sulphur flotation production scene, the on-the-spot sulphur concentrate grade contrast of these method application front and back (wherein, 5,6 being respectively the preceding and use back of use) as shown in Figure 6.From figure, can find out the fault detection method that use distributes based on texture cell; Operating personnel can in time find fault and adjust; Make outlet sulphur concentrate grade fluctuation range narrow down to 14.33% from 28.64%; And the sulphur concentrate grade is stabilized in more than 60%, and monthly concentrate grade value brings up to 72.98% from 59.78%.
Its result shows that method proposed by the invention has taken into full account the singularity that actual sulphur flotation froth texture distributes, and adopts the kernel function that designs certainly to approach the output texture cell and distributes the T based on principal component model of foundation 2The fault detection method of statistic can accurately be judged sulphur flotation fault fast, has stablized sulphur flotation production run, has improved the sulphur concentrate grade.
Subordinate list 1:
Project Sample number The correct number that detects False Rate
Normal picture 150 144 4%
The fault graph picture 106 95 10.38%
Add up to 256 239 6.64%

Claims (2)

1. sulphur floatation process fault detection method that distributes based on texture cell is characterized in that comprising following steps:
The first step; Image library according to sulphur floatation foam image structure under the nominal situation; Based on the texture cell texture feature extraction; The design norm of nonparametric kernel density estimates that operator approaches the actual texture cell distribution of foam and obtains dynamic weight coefficient, adopts the method for pivot analysis to set up the principal component model based on dynamic weight coefficient, and is last according to the T based on principal component model 2Statistic obtains failure determination threshold value, specifically comprises:
Step 1: extract the foam textural characteristics based on texture cell;
be 3
Figure 667922DEST_PATH_IMAGE002
3 neighborhoods; The gray-scale value of this neighborhood is designated as
Figure 717787DEST_PATH_IMAGE003
successively; Wherein
Figure 4412DEST_PATH_IMAGE004
is the gray-scale value of regional center pixel, and texture cell value is calculated as:
Figure 474893DEST_PATH_IMAGE006
Figure 887420DEST_PATH_IMAGE007
Wherein ;
Figure 221635DEST_PATH_IMAGE009
; ,
Figure 435765DEST_PATH_IMAGE011
are the constant of confirming according to sulphur flotation operating mode;
Step 2: according to the texture cell value of calculating, the design norm of nonparametric kernel density estimates that operator approaches actual texture cell and distributes, and obtains describing the dynamic weight coefficient of textural characteristics;
According to the approximation of function principle, estimate that with the norm of nonparametric kernel density of following design operator approaches actual texture cell and distributes:
Figure 64192DEST_PATH_IMAGE012
Figure 365861DEST_PATH_IMAGE013
is sulphur flotation froth texture cell distribution function in the formula;
Figure 94477DEST_PATH_IMAGE014
is the texture cell value;
Figure 606230DEST_PATH_IMAGE015
is the control input; It is the liquid level regulated value in the sulphur floatation process;
Figure 100665DEST_PATH_IMAGE016
is the weight coefficient of i kernel function;
Figure 991260DEST_PATH_IMAGE017
is i kernel function;
Figure 850632DEST_PATH_IMAGE018
is the x axle mid point of i kernel function, and
Figure 521785DEST_PATH_IMAGE019
is the window width of kernel function;
Step 3: adopt the method for pivot analysis to set up the principal component model based on dynamic weight coefficient, rule of thumb method is confirmed the pivot number;
If model is input as dynamic weight coefficient vector ;
Figure 892909DEST_PATH_IMAGE021
is ties up matrix;
Figure 81631DEST_PATH_IMAGE023
is the data sample number;
Figure 855552DEST_PATH_IMAGE024
is the input variable number; For fear of the different dimensions of variable to result's influence be convenient to mathematical processing; Earlier modeling data being carried out normalization handles; Input variable after the note normalization is
Figure 720740DEST_PATH_IMAGE025
;
Figure 921914DEST_PATH_IMAGE026
carried out pivot analysis; Adopt empirical method to confirm the pivot number,
Figure 573518DEST_PATH_IMAGE026
can be decomposed into:
Figure 823234DEST_PATH_IMAGE027
Figure 870825DEST_PATH_IMAGE028
is the pivot matrix of loadings in the formula;
Figure 375941DEST_PATH_IMAGE030
gets sub matrix for pivot;
Figure 491665DEST_PATH_IMAGE031
is residual matrix, and is the pivot number;
Step 4: according to T 2Statistic characterizes a kind of characteristics of estimating of principal component model interior change, calculates the T based on principal component model 2Statistic;
In order to detect the fluctuation of sulphur floatation process, adopt T based on principal component model 2The fault detection method of statistic is for
Figure 874421DEST_PATH_IMAGE033
The dynamic weight coefficient of individual sample , T 2Statistic is defined as:
Figure 148594DEST_PATH_IMAGE035
Wherein is the data sample number;
Figure 664392DEST_PATH_IMAGE037
be in
Figure 121918DEST_PATH_IMAGE038
matrix
Figure 733028DEST_PATH_IMAGE039
OK;
Figure 883386DEST_PATH_IMAGE038
is made up of the score vector of
Figure 150420DEST_PATH_IMAGE032
the individual pivot that constitutes principal component model,
Figure 411637DEST_PATH_IMAGE040
by the diagonal angle average of being formed with preceding
Figure 877253DEST_PATH_IMAGE032
pairing eigenwert of individual pivot;
Step 5: distribution obtains T based on F 2The control limit of statistic is limit as sulphur floatation process failure determination threshold value with this control;
T 2The calculating of the control limit of statistic:
Figure 198513DEST_PATH_IMAGE041
Wherein Be the data sample number, Be the pivot number that is kept,
Figure 355452DEST_PATH_IMAGE042
Be insolation level,
Figure 909930DEST_PATH_IMAGE043
Be to do corresponding to insolation level
Figure 948293DEST_PATH_IMAGE042
, degree of freedom does
Figure 551312DEST_PATH_IMAGE032
,
Figure 381734DEST_PATH_IMAGE044
The critical value that F under the condition distributes; Make T 2The control limit of statistic As failure determination threshold value;
Second step, the T of failure determination threshold value that the principal component model of setting up according to normal froth images storehouse obtains and real-time froth images 2Sulphur floatation process state is judged in the comparison of statistics value;
Step 1: the T that uses preceding 4 step calculating real-time froth images in the first step 2Statistic
Step 2: according to current T 2Statistics value and threshold value
Figure 836221DEST_PATH_IMAGE045
Relatively detection sulphur floatation process fault;
If
Figure 396515DEST_PATH_IMAGE047
can judge that then less fluctuation just takes place the sulphur floatation process; if
Figure 964900DEST_PATH_IMAGE048
, then alarm sulphur floatation process breaks down.
2. the sulphur floatation process fault detection method that distributes based on texture cell according to claim 1; It is characterized in that: in the first step step 2; Texture characteristic distributions according to the sulphur flotation froth; When approximating curve, selected 25 kernel functions for use, kernel function is a prototype with the Gaussian kernel function, makes up the kernel function that meets sulphur floatation process system:
Figure 311611DEST_PATH_IMAGE049
。
CN2012100947093A 2012-04-01 2012-04-01 Fault detecting method for sulfur flotation process on basis of texture unit distribution Pending CN102681473A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2012100947093A CN102681473A (en) 2012-04-01 2012-04-01 Fault detecting method for sulfur flotation process on basis of texture unit distribution

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2012100947093A CN102681473A (en) 2012-04-01 2012-04-01 Fault detecting method for sulfur flotation process on basis of texture unit distribution

Publications (1)

Publication Number Publication Date
CN102681473A true CN102681473A (en) 2012-09-19

Family

ID=46813550

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2012100947093A Pending CN102681473A (en) 2012-04-01 2012-04-01 Fault detecting method for sulfur flotation process on basis of texture unit distribution

Country Status (1)

Country Link
CN (1) CN102681473A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104808652A (en) * 2015-04-20 2015-07-29 中国石油化工股份有限公司 Chemical engineering device operating state scoring method
CN107704680A (en) * 2017-09-29 2018-02-16 北京航空航天大学 A kind of computational methods of Consumer Product Safety injury probability of happening
CN108181891A (en) * 2017-12-13 2018-06-19 东北大学 Industrial big data method for diagnosing faults based on intelligent core pivot element analysis
CN109740727A (en) * 2018-12-05 2019-05-10 中国软件与技术服务股份有限公司 A kind of hydraulic turbine shaft state monitoring method neural network based and system

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5396260A (en) * 1992-12-22 1995-03-07 The Center For Innovative Technology Video instrumentation for the analysis of mineral content in ores and coal
CN1399579A (en) * 1999-11-24 2003-02-26 奥托库姆普联合股份公司 Monitoring and control of froth flotation plant
CN101334366A (en) * 2008-07-18 2008-12-31 中南大学 Flotation recovery rate prediction method based on image characteristic analysis
CN101446827A (en) * 2008-11-06 2009-06-03 西安交通大学 Process fault analysis device of process industry system and method therefor
CN101566848A (en) * 2009-05-27 2009-10-28 中南大学 Fault diagnostic method of floatation process based on statistic distribution of foam size
CN101738435A (en) * 2009-11-30 2010-06-16 浙江大学 Dynamic fault diagnostic method of polymer aggregation in gas-solid fluidized bed reactor
CN102033523A (en) * 2009-09-25 2011-04-27 上海宝钢工业检测公司 Strip steel quality forecasting, furnace condition early-warning and fault diagnosis method based on partial least square

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5396260A (en) * 1992-12-22 1995-03-07 The Center For Innovative Technology Video instrumentation for the analysis of mineral content in ores and coal
CN1399579A (en) * 1999-11-24 2003-02-26 奥托库姆普联合股份公司 Monitoring and control of froth flotation plant
CN101334366A (en) * 2008-07-18 2008-12-31 中南大学 Flotation recovery rate prediction method based on image characteristic analysis
CN101446827A (en) * 2008-11-06 2009-06-03 西安交通大学 Process fault analysis device of process industry system and method therefor
CN101566848A (en) * 2009-05-27 2009-10-28 中南大学 Fault diagnostic method of floatation process based on statistic distribution of foam size
CN102033523A (en) * 2009-09-25 2011-04-27 上海宝钢工业检测公司 Strip steel quality forecasting, furnace condition early-warning and fault diagnosis method based on partial least square
CN101738435A (en) * 2009-11-30 2010-06-16 浙江大学 Dynamic fault diagnostic method of polymer aggregation in gas-solid fluidized bed reactor

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
《中国优秀硕士学位论文全文数据库•信息科技辑》 20110315 程翠兰 基于颜色与纹理特征的矿物浮选精选泡沫分类 I138-1033 第2011年卷, 第3期 *
程翠兰: "基于颜色与纹理特征的矿物浮选精选泡沫分类", 《中国优秀硕士学位论文全文数据库•信息科技辑》, vol. 2011, no. 3, 15 March 2011 (2011-03-15), pages 138 - 1033 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104808652A (en) * 2015-04-20 2015-07-29 中国石油化工股份有限公司 Chemical engineering device operating state scoring method
CN107704680A (en) * 2017-09-29 2018-02-16 北京航空航天大学 A kind of computational methods of Consumer Product Safety injury probability of happening
CN108181891A (en) * 2017-12-13 2018-06-19 东北大学 Industrial big data method for diagnosing faults based on intelligent core pivot element analysis
CN108181891B (en) * 2017-12-13 2020-05-05 东北大学 Industrial big data fault diagnosis method based on intelligent core principal component analysis
CN109740727A (en) * 2018-12-05 2019-05-10 中国软件与技术服务股份有限公司 A kind of hydraulic turbine shaft state monitoring method neural network based and system

Similar Documents

Publication Publication Date Title
AU2018212323B2 (en) Mineral recovery control
CN102681473A (en) Fault detecting method for sulfur flotation process on basis of texture unit distribution
Alves et al. Universal fluctuations in radial growth models belonging to the KPZ universality class
CN107301657B (en) A kind of video target tracking method considering target movable information
CN104483962A (en) Intermittent process online fault detection method based on multi-direction orthogonal locality preserving projections
CN104828517A (en) Belt deviation detecting method based on visual sense
CN103092078A (en) Whole-process optimizing method of multi-stage batch production process
CN101853243A (en) Unknown adaptive Kalman filter method for system model
CN107401140B (en) The determination method of tributary lower reaches section channel forming disahcge
CN102680050A (en) Sulfur flotation liquid level measuring method based on foam image characteristic and air volume
CN108647722B (en) Zinc ore grade soft measurement method based on process size characteristics
CN106683405B (en) Road traffic state evaluation method based on road traffic density
CN105787959A (en) Method for multi-agent network object tracking based on improved adaptive particle filtering
CN106372320B (en) Method for performing large vortex simulation on highway tunnel turbulence by adopting sub-filtering scale model
CN101566848A (en) Fault diagnostic method of floatation process based on statistic distribution of foam size
CN104268600A (en) Mineral flotation froth image texture analysis and working condition identification method based on Minkowski distance
CN109934456A (en) A kind of method and system for acquisition operational system progress intelligent trouble detection
Stano et al. Estimation of the soil-dependent time-varying parameters of the hopper sedimentation model: The FPF versus the BPF
CN104001739A (en) Method and device for obtaining cold rolling temper rolling pressure
CN105372651A (en) Adaptive maneuvering target tracking method on the basis of optimal AR (Autoregressive) model
CN108226887A (en) A kind of waterborne target rescue method for estimating state in the case of observed quantity transient loss
CN107808209B (en) Wind power plant abnormal data identification method based on weighted kNN distance
CN102339381B (en) Method for tracking particle filter video motion target based on particle position adjustment
CN103700095A (en) Log end surface image partitioning algorithm for improving active contour model based on circle constraint
CN114352946B (en) Drainage engineering pipeline blocks up early warning system

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C12 Rejection of a patent application after its publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20120919