CN107065834A - The method for diagnosing faults of concentrator in hydrometallurgy process - Google Patents
The method for diagnosing faults of concentrator in hydrometallurgy process Download PDFInfo
- Publication number
- CN107065834A CN107065834A CN201710379067.4A CN201710379067A CN107065834A CN 107065834 A CN107065834 A CN 107065834A CN 201710379067 A CN201710379067 A CN 201710379067A CN 107065834 A CN107065834 A CN 107065834A
- Authority
- CN
- China
- Prior art keywords
- msub
- evidence
- mrow
- rule
- confidence level
- 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
- 238000000034 method Methods 0.000 title claims abstract description 115
- 230000008569 process Effects 0.000 title claims abstract description 41
- 238000009854 hydrometallurgy Methods 0.000 title claims abstract description 39
- 238000003745 diagnosis Methods 0.000 claims abstract description 66
- 230000004927 fusion Effects 0.000 claims abstract description 25
- 238000004519 manufacturing process Methods 0.000 claims abstract description 24
- 238000010606 normalization Methods 0.000 claims description 13
- 239000000203 mixture Substances 0.000 claims description 12
- 238000005070 sampling Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 230000006870 function Effects 0.000 description 12
- 230000003044 adaptive effect Effects 0.000 description 8
- 229910052500 inorganic mineral Inorganic materials 0.000 description 8
- 239000011707 mineral Substances 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 238000011161 development Methods 0.000 description 6
- 239000007788 liquid Substances 0.000 description 6
- 238000011112 process operation Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 5
- 238000005406 washing Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 239000000047 product Substances 0.000 description 3
- 238000012552 review Methods 0.000 description 3
- 239000007787 solid Substances 0.000 description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 2
- BQCADISMDOOEFD-UHFFFAOYSA-N Silver Chemical compound [Ag] BQCADISMDOOEFD-UHFFFAOYSA-N 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000007717 exclusion Effects 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 229910052709 silver Inorganic materials 0.000 description 2
- 239000004332 silver Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 239000002932 luster Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007257 malfunction Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000009329 sexual behaviour Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000011343 solid material Substances 0.000 description 1
- 239000012265 solid product Substances 0.000 description 1
- 239000002904 solvent Substances 0.000 description 1
- 238000003756 stirring Methods 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- 208000011580 syndromic disease Diseases 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The present invention provides a kind of method for diagnosing faults of concentrator in hydrometallurgy process, including:Obtain online qualitative information and online quantitative data that hydrometallurgy concentrator is used to recognize a kind of failure;For online qualitative information, the confidence level of each event is obtained using the method for the rule-based reasoning based on confidence level, first evidence is obtained;For online quantitative data, the similarity of follow-up conclusion example is obtained using the reasoning by cases method based on data similarity, Article 2 evidence is obtained;The inhomogeneity data being made up of online quantitative data used when follow-up conclusion example is progress reasoning by cases;According to D S evidence theory fusions rule, two evidences are merged, the failure diagnosis information of concentrator in hydrometallurgy process is obtained, this method can be such that operating personnel are adjusted in time according to fault diagnosis result information, and then accident rate is effectively reduced, improve production security.
Description
Technical field
The invention belongs to a kind of fault diagnosis side of concentrator in hydrometallurgical technology, more particularly to hydrometallurgy process
Method.
Background technology
With the development of China's process of industrialization, resource problem turns into one of subject matter of restriction China development.Mineral products
Resource plays basic effect as the main source of the raw material of industry in socio-economic development.Due to mineral resources
Extensive and a large amount of consumption, cause China to face the serious problem in short supply of mineral resources, the reserves of high-grade mineral resources are just
Increasingly reduce, situation very severe.The rich reserves of China's low-grade mineral resource, extract mineral products money from poor, thin, matrix
Source becomes the inexorable trend of future development, and how the utilization low-grade mineral resource of economical and efficient is for China's economic society
Sustainable development is significant.
As the continuous reduction of the grade of ore and the requirement to environment are increasingly strict, hydrometallurgy is in low-grade mineral resource
Development and utilization in play an important role.The dense washing of hydrometallurgical flowsheets is the mistake that separation of solid and liquid is carried out using gravity
Journey, can save mass energy, and leading indicator is underflow density.Dense washing process is one of crucial work of hydrometallurgy process
Sequence.In the industrial production, generally solid material is dissolved in solvent, different component is separated, be i.e. wet split, selected
Product is the suspension of solid-liquid two-phase, for the water for obtaining aqueous less solid product He being substantially free of solid, most of feelings
Separation of solid and liquid will be carried out under condition.
At present, most of fault diagnosis of concentrator washing process by operating personnel subjectivity realize, automatization level compared with
It is low.Dense washing process complex process, production environment are severe, and the features such as with big inertia, large dead time, many influence factors, plus
Artificial subjective factor influence, be difficult to realize accurate fault diagnosis.In real process, many variables be real-time change and
Randomness is big, and changes frequent, and this make it that realization is more difficult to the fault diagnosis of dense washing process.
The content of the invention
For existing technical problem, the present invention provides a kind of fault diagnosis side of concentrator in hydrometallurgy process
Method, this method can be such that operating personnel are adjusted in time according to fault diagnosis result information, and then effectively reduce accident rate, carry
High production security.
The method for diagnosing faults of concentrator in the hydrometallurgy process of the present invention, including:
Obtain online qualitative information and online quantitative data that hydrometallurgy concentrator is used to recognize a kind of failure;
For online qualitative information, the confidence level of each event is obtained using the method for the rule-based reasoning based on confidence level,
And the confidence level of each event is modified to the evidence form of evidence theory, obtain first evidence;The event is failure
The failure determined in identification process by online qualitative information occurs or not occurred online in event, or failure cause trace back process
Cause the reason for failure occurs in qualitative information;(i.e. in Fault Identification, the event is that the failure occurs or do not sent out
It is raw;When failure cause is reviewed, the event is to cause the reason for failure occurs)
For online quantitative data, the similar of follow-up conclusion example is obtained using the reasoning by cases method based on data similarity
Spend, and the similarity of follow-up conclusion example is modified to the evidence form of evidence theory, obtain Article 2 evidence;It is described to wait to diagnose
The vector being made up of online quantitative data used when case is progress reasoning by cases;
According to D-S evidence theory fusion rule, two evidences are merged, concentrator in hydrometallurgy process is obtained
Failure diagnosis information, the failure diagnosis information includes Fault Identification and failure cause is reviewed.
Alternatively, the confidence level of each event is obtained using the method for the rule-based reasoning based on confidence level, and this is each
The confidence level of event is modified to the evidence form of evidence theory, obtains first evidence, including:
If causing the confidence level cf of n event of failure generation1, cf2…cfn;
Normalization formula when then obtaining first evidence is expressed as follows:
Wherein, the form of first evidence is { m11,m12…m1n};m11,m12…m1nFor the rule-based reasoning based on confidence level
The basic probability assignment of each element in the identification framework of method.
Alternatively, the similarity of follow-up conclusion example is obtained using the reasoning by cases method based on data similarity, including:
The similarity of follow-up conclusion example and source case is determined by following formula one;
Formula one:
Wherein, ck=(x1k,x2k,...xik...xjk), k=1,2 ... n, the source case in case library is represented, n is case
The sum of source case in storehouse;xik(i=1,2 ..., be j) i-th of variate-value in the case of kth bar source, j is variable number;
X=(x1,x2,....,xj), represent follow-up conclusion example;sim(xi,xik) it is variable xiWith variable xikSimilarity;
xi,xik∈ [α, β], α, β are respectively variable xiMinimum value in historical statistics value and
Maximum.
Alternatively, if the similarity of follow-up conclusion example and n source case in case library is sim1, sim2…simn;
Normalization formula when then obtaining Article 2 evidence is expressed as follows:
Wherein, the form of Article 2 evidence is { m21,m22…m2n};m21,m22…m2nFor in the identification framework of reasoning by cases
The basic probability assignment of each element.
Alternatively, according to D-S evidence theory fusion rule, two evidences are merged, obtained in hydrometallurgy process
The failure diagnosis information of concentrator, including:
If first evidence and Article 2 evidence have identical identification framework { A1,A2,…An, and first evidence and
The confidence level difference of Article 2 evidence within a preset range, then
Merged using following formula,
Wherein, m1For first evidence, m2For Article 2 evidence, 1≤i, j≤n, Ai∩Aj=C represents that event C can be by
Event AiAnd AjIt is intersecting to obtain;It is normalization factor, is to weigh evidence m1And m2Between contradiction
The index of size.
Alternatively, if first evidence and Article 2 evidence have identical identification framework { A1,A2,…An, and first
The confidence level difference of bar evidence and Article 2 evidence is more than preset range, then
Determine the weight of first evidence, and Article 2 evidence weight;
It will averagely obtain average evidence using following formula q bar evidences weighteds, q=2,
Wherein,For the basic probability assignment of event A in average evidence, mi(A) it is the elementary probability in i-th evidence
Distribution,For the weight of i-th evidence;
And, according to formulaAverage evidence is merged;
Wherein, mfThe Basic probability assignment function after fusion is represented,The symbol that two evidences are merged is represented,For
Pass throughThe average evidence obtained.
Optionally it is determined that the weight of first evidence, and Article 2 evidence weight, including:
N before at the time of obtaining the online quantitative data data sampled are time window, are determined each in window
The average of sampled dataAnd
The total amount that follow-up conclusion example deviates normal range (NR) is expressed as:
xjJ-th of variable of example of being settled a lawsuit for follow-up,For the average value of the variable in the time window of sampling;
The then weight w of Article 2 evidencedataIt is expressed as:wdata=e-kσ;
Wherein, k is coefficient;
The weight w of first evidenceruleIt is expressed as:wrule=1-wdata。
Alternatively, methods described also includes:
The faulty failure diagnosis information of institute that concentrator may occur is obtained using traversal method, it is faulty by institute
Failure diagnosis information, determines final conclusion.
Alternatively, online qualitative information and online quantitative data that hydrometallurgy concentrator is used to recognize a kind of failure are obtained
The step of before, methods described also includes:
The offline historical variable for obtaining hydrometallurgy concentrator in preset time period;
Priori, acquisition with reference to field of hydrometallurgy expert and associative operation personnel is multiple for carrying out concentrator event
Hinder the Expert Rules of diagnosis;
According to multiple Expert Rules for being used to carry out concentrator fault diagnosis, Expert Rules storehouse is set up;
Wherein, each concentrator Fault diagnosis expert rule includes:Each regular regular former piece, consequent, rule
Intensity;Regular former piece, consequent and the rule intensity of each rule have corresponding relation;
And/or,
In the method for rule-based reasoning based on confidence level, the form of production rule is:IF E THEN H WITH cf
(H/E)
Wherein, E represents a simple premise, or multiple simple premise logical combinations logical combination premise, H be one or
Multiple conclusions, cf (H/E) is the confidence level for occurring H based on the E;
Correspondingly, the step of confidence level of each event being obtained using the method for the rule-based reasoning based on confidence level, including:
Regular former piece E is set up, and the confidence level of establishment is cf (H/E);
Then each event e of online qualitative information composition confidence level cf (H/e) is:
Cf (H/e)=cf (H/E) × max { cf (E/e), 0 };
Wherein, cf (E/e) represents event e and regular former piece matching degree, and cf (H/E) represents rule intensity.
Alternatively, online qualitative information and online quantitative data that hydrometallurgy concentrator is used to recognize a kind of failure are obtained
The step of before, methods described also includes:
Abnormal data occur for hydrometallurgy concentrator in collected offline preset time period, and no exceptions number
According to;
According to the data of collection, case library is set up.
Beneficial effect:
The method for diagnosing faults of concentrator, can not only use the knowledge and warp of expert in the hydrometallurgy process of the present invention
Test, the mass data information accumulated also in relation with production process, improve the utilization rate of information, and in real time according to data accuracy
The weight of different aforementioned sources is calculated, can be with Real time identification concentrator operating condition, diagnosis concentrator operation troubles is simultaneously out of order
Reason, rationally reliable Operating Guideline suggestion is provided for operative employee, suitably to be adjusted to process operation state in time,
And then accident rate is effectively reduced, improve production security, it is ensured that Business Economic Benefit and production efficiency.The above method can have
Effect avoids the lag issues manually evaluated, and makes appropriate adjustment to current production cycle process operation state in time.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the inventive method;
Fig. 2 is three kinds of diagnostic methods recognition effect comparison diagram higher to concentration in concentrator production process.
Embodiment
In order to preferably explain the present invention, in order to understand, below in conjunction with the accompanying drawings, by embodiment, to this hair
It is bright to be described in detail.
Device of the present invention includes hydrometallurgy concentrator intelligent Fault Diagnose Systems, host computer, PLC, scene biography
Feel pick-up part and human-computer interaction device.Wherein sensing pick-up part in scene includes the instrumentations such as concentration, pressure, flow.
In hydrometallurgy process in-site installation instrumentation, instrumentation is total by Profibus-DP by the signal of collection
Line is sent to PLC, and collection signal is sent to host computer by PLC by Ethernet timing, and the data of reception are passed to wet method by host computer
Metallurgical concentrator intelligent Fault Diagnose Systems, carry out process operating mode's switch and failure are diagnosed.
The functions of said apparatus:
1. scene senses pick-up part:Including the instrumentations such as concentration, pressure, electric current by sensor group into, be responsible for process
The collection and transmission of data;
②PLC:It is responsible for the signal A/D of collection to change, and host computer is transmitted signals to by Ethernet;PLC is controlled
Device is using the CPU 414-2 of the series of Simens 400, and with Profibus, DP mouthfuls connect distribution I/O.
Ethernet communication module is equipped with for PLC, plc data is accessed for host computer.PLC and ethernet communication mould
Block is placed in the PLC rack in central control room.
3. host computer:Local plc data is collected, hydrometallurgy concentrator intelligent Fault Diagnose Systems is sent to, carried out
Journey state recognition is simultaneously diagnosed, and provide production operation guiding opinion to failure;
4. human-computer interaction device:It is responsible for the experience of expert and observation being converted into computer language, transfers in host computer
The Mixture of expert knowledge system fault diagnosis system processing merged based on D-S, and then, pass through Mixture of expert knowledge fault diagnosis system
D-S evidence fusions function in system realizes the fusion to two conclusions.
The embodiment of the present invention is mainly for knowledge wide variety and the not high flow system of the service data degree of accuracy.In failure
Diagnose in Mixture of expert knowledge system, the more commonly used is rule-based expert knowledge system and the expertise based on case
The mixing of system.
Rule-based fault diagnosis system stresses the knowledge experience of domain expert being refined into rule, its logical expression and
It is explanatory strong, readily appreciate.The expert knowledge system of Process Based uses the subjective information that operating personnel provide, change frequency
Rate is low, but robustness is good.The uncertain problem being introduced into confidence level processing reasoning process, is adapted to the operating habit of workman, but by mistake
Report is failed to report more.
The fault diagnosis system of case-based reasioning uses Process History data, and similar examine is found by data similarity
Conclusion example, obtained conclusion is directly perceived credible with reference to property.And the expert knowledge system of case-based reasioning uses real-time quantitative number
According to the reasoning results reliability is higher.But the much noise in data may cause the decision confidence of case-based reasioning to reduce.
And case similarity also includes the uncertainty as caused by data reliability in a way.
Therefore, the embodiment of the present invention proposes a kind of hybrid knowledge diagnosis method for system fault merged based on D-S, such as Fig. 1
Shown, the hydrometallurgy concentrator intelligent failure diagnosis method of the present embodiment includes:
The first step, acquisition hydrometallurgy concentrator are used for the online qualitative information for recognizing a kind of failure and online quantitative number
According to;
Second step, for online qualitative information, each event is obtained using the method for the rule-based reasoning based on confidence level
Confidence level, and the confidence level of each event is modified to the evidence form of evidence theory, obtain first evidence.
The failure that event in this step can be regarded as being determined by online qualitative information during Fault Identification occurs or event
Barrier does not occur, or causes the reason for failure occurs in failure cause trace back process.Formula (15) or formula (16) described as follows
The information of illustration.
For Fig. 1, i.e., first card is obtained in based on confidence level/confidence level silver rule-based reasoning expert system
According to.
For example, before the step is performed, following sub-steps be can perform:
A01, the in advance offline historical variable for obtaining hydrometallurgy concentrator in preset time period;
A02, the priori with reference to field of hydrometallurgy expert and associative operation personnel, acquisition are multiple dense for carrying out
The Expert Rules of machine fault diagnosis;
A03, according to multiple it is used to carry out the Expert Rules of concentrator fault diagnosis, sets up Expert Rules storehouse;
Wherein, each concentrator Fault diagnosis expert rule includes:Each regular regular former piece, consequent, rule
Intensity;Regular former piece, consequent and the rule intensity of each rule have corresponding relation.
In a particular application, in the method for the rule-based reasoning based on confidence level, the form of production rule is:IF E
THEN H WITH cf(H/E)
Wherein, E represents a simple premise, or multiple simple premise logical combinations logical combination premise, H be one or
Multiple conclusions, cf (H/E) is the confidence level for occurring H based on the E;
Correspondingly, the step of confidence level of each event being obtained using the method for the rule-based reasoning based on confidence level, including:
Regular former piece E is set up, and the confidence level of establishment is cf (H/E);
Then each event e of online qualitative information composition confidence level cf (H/e) is:
Cf (H/e)=cf (H/E) × max { cf (E/e), 0 };
Wherein, cf (E/e) represents event e and regular former piece matching degree, and cf (H/E) represents rule intensity.
3rd step, for online quantitative data, follow-up conclusion is obtained using the reasoning by cases method based on data similarity
The similarity of example, and the settle a lawsuit similarity of example of the follow-up is modified to the evidence form of evidence theory, obtain Article 2 evidence;Institute
Follow-up conclusion example is stated to carry out the vector being made up of online quantitative data used during reasoning by cases.
It for Fig. 1, can be handled in based on data similarity reasoning by cases expert system, obtain Article 2 card
According to.
In a particular application, can also hydrometallurgy be dense in collected offline preset time period in advance before performing the step
Abnormal data occur for machine, and no exceptions data;
According to the data of collection, case library is set up.
The case of various events is store in the present embodiment in case library, the label of various events is different, each event example
Form described as follows, certain failure, certain failure cause etc..
4th step, according to D-S evidence theory fusion rule, two evidences are merged, obtain hydrometallurgy process in
The failure diagnosis information of concentrator, the failure diagnosis information includes Fault Identification and failure cause is reviewed.
In actual applications, the faulty fault diagnosis letter of institute that concentrator may occur can be obtained using traversal method
Breath, by the faulty failure diagnosis information of institute, determines final conclusion.
During implementing, if causing the confidence level cf of n event of failure generation1, cf2…cfn;
Normalization formula when then obtaining first evidence is expressed as follows:
Wherein, the form of first evidence is { m11,m12…m1n};m11,m12…m1nFor the rule-based reasoning based on confidence level
The basic probability assignment of each element in the identification framework of method.
Correspondingly, if the similarity of follow-up conclusion example and n source case in case library is sim1, sim2…simn;
Normalization formula when then obtaining Article 2 evidence is expressed as follows:
Wherein, the form of Article 2 evidence is { m21,m22…m2n};m21,m22…m2nFor in the identification framework of reasoning by cases
The basic probability assignment of each element.
If first evidence and Article 2 evidence have identical identification framework { A1,A2,…An, and first evidence and
The confidence level difference of Article 2 evidence within a preset range, then
Merged using following formula,
Wherein, m1For first evidence, m2For Article 2 evidence, 1≤i, j≤n, Ai∩Aj=C represents that event C can be by
Event AiAnd AjIt is intersecting to obtain;It is normalization factor, is to weigh evidence m1And m2Between lance
The index of shield size.
In addition, if first evidence and Article 2 evidence have identical identification framework { A1,A2,…An, and first
The confidence level difference of evidence and Article 2 evidence is more than preset range, then
Determine the weight of first evidence, and Article 2 evidence weight;
Average evidence (q=2 in the present embodiment) will be averagely obtained using following formula q bar evidences weighteds,
Wherein,For the basic probability assignment of event A in average evidence, mi(A) it is the elementary probability in i-th evidence
Distribution,For the weight of i-th evidence;
And, according to formulaAverage evidence is merged;
Wherein, mfThe Basic probability assignment function after fusion is represented,The symbol that two evidences are merged is represented,For
Pass throughThe average evidence obtained.
In the present embodiment, the above method can not only use the knowledge and experience of expert, also in relation with production process accumulation
Mass data information, improve the utilization rate of information, and calculate the weight of different aforementioned sources according to data accuracy in real time, can
With Real time identification concentrator operating condition, diagnose concentrator operation troubles and provide failure cause, provide and rationally may be used for operative employee
The Operating Guideline suggestion leaned on, suitably to be adjusted to process operation state in time, and then effectively reduces accident rate, carries
High production security, it is ensured that Business Economic Benefit and production efficiency.The above method can be prevented effectively from the lag issues manually evaluated,
And appropriate adjustment is made to current production cycle process operation state in time.
To understand the scheme of the embodiment of the present invention in more detail, describe in detail as follows:
In the present embodiment, online qualitative information is that operating personnel such as expert etc. can be input to by human-computer interaction device and be based on
In the rule-based reasoning knowledge system of confidence level.
1) the rule-based reasoning knowledge based on confidence level/confidence level silver/online qualitative information of expert system acquisition is corresponding
First evidence
The main fuzzy Qualitative Knowledge provided using expert in the present embodiment, by the way that fault diagnosis Heuristics is summarized as
Rule, so reasoning it is concluded that.
Complexity of the different flows when extracting expertise and experience is also different.Some key variables of process industry
It can not measure or measure inaccurate, but the rough qualitative information of process operation plays an important roll to fault diagnosis.These are thick
Whether slightly information includes intensity of sound, material color and luster, ore pulp and bubbles etc..Under many circumstances, operating personnel are difficult to obtain in time
The prior probability that the event of obtaining occurs, the probability size that event occurs often is described with subjective quantitative.According to this operation
Custom, in the rule-based reasoning based on expertise, the uncertain problem of rule is handled using certain factor.
Production rule is widely used because the logical thinking custom of people is met.In view of the uncertain of knowledge
Property, the expression-form with regular probabilistic confidence level production rule is:
if E then H with(cf(H/E)) (1)
Wherein E is the former piece of rule, represents precondition.H is the consequent of rule, represents corresponding conclusion.Cf (H/E) table
Show it is regular confidence level, be also rule intensity.
These rule intensities can not only be obtained by analysis of history data, can also subjectivity is provided by rule of thumb by expert.
The inaccurate information i.e. qualitative information that live Real Time Observation is provided is represented with e, its form is typically expressed as e:Regular former piece E into
Vertical, the confidence level of establishment is cf (H/E).Then the confidence level of conclusion is asked for using formula (2):
Cf (H/e)=cf (H/E) × max { cf (E/e), 0 } (2)
In above formula, cf (H/e) represents the confidence level of conclusion, is the matching degree cf (E/e) by field data and regular former piece
Obtained with rule intensity cf (H/E).
By obtaining the confidence level that each event occurs based on uncertain inference fault diagnosis knowledge system, it is assumed that there is n
The confidence level cf of individual event1, cf2…cfnUnified identification can be built with the D-S evidence theories introduced in following fusion process
Framework, this n confidence level is normalized using formula (3):
Wherein, m11,m12…m1nFor the basic probability assignment of each element in the identification framework of rule-based reasoning.
Event in this method in identification framework is all the probability of event and for 1 in mutual exclusion, therefore identification framework, but by
The probability for inferring the event come in two methods adds and not for 1, it is necessary to carry out the scaling of equal proportion.Just there is normalization to walk
Suddenly.
In addition, online quantitative data can be collected by host computer, and then it is dense to send the hydrometallurgy with case library to
Machine intelligent Fault Diagnose Systems.
2) case library of concentrator fault diagnosis is established, and case-based reasioning obtains online quantitative data corresponding the
Two evidences
The reasoning by cases of the present embodiment is artificial intelligence field problem solving and machine learning method, complete reasoning by cases
Theory includes the steps such as matching, multiplexing, amendment, study.A large amount of creation datas can be obtained by sensor in process industry, these
Data can reflect the running status of production.The data characteristic of normal production status and all kinds of malfunctions is different, therefore, it can
Inhomogeneity data characteristic, which is extracted, as case is used for reasoning by cases.It is general that case matching is carried out by data similarity, and it is nearest
Adjacent method is to calculate one of most common method in data similarity.Its main thought is under the definition of certain distance so that
The similarity of case to be identified and closest source case is maximum, and similarity is then smaller between distant case.
When the method for case-based reasioning is used for fault diagnosis, the source case in case library can be expressed as follows:
ck=(x1k,x2k,...xik...xjk), k=1,2 ... n (4)
Wherein, n is source case sum;xik(i=1,2 ..., be j) i-th of variate-value in the case of kth bar source, j is variable
Number.Obtaining follow-up conclusion example x=(x1,x2,....,xj) after, x and c is calculated by formula (5)kIn each case most
Neighbour's similarity:
Above formula represents that the similarity of follow-up conclusion example and source case is the weighted sum of all variable similarities.Wherein wiFor
Variable x is assigned according to procedural knowledgeiWeight, the weighted of each variable in different cases, weight is also stored in case library
In CROSS REFERENCE in.sim(xi,xik) it is variable xiWith variable xikSimilarity, its calculation formula pass through formula (6) obtain:
Wherein, xi,xik∈ [α, β], α, β are respectively variable xiMinimum value and maximum in historical statistics value.
Follow-up conclusion example and the similarity of each case in case library are calculated using formula (5), tries to achieve wait to diagnose respectively
The similarity sim of case and n source case1, sim2…simnIf they can return into same identification framework, it is necessary to pass through
This n similarity numerical value is normalized formula (7), obtains the Basic probability assignment function of evidence theory:
Wherein, m21,m22…m2nFor the basic probability assignment of each element in reasoning by cases identification framework.
3) D-S that first evidence and Article 2 evidence are carried out into adaptive weighting is merged
The D-S evidence theory of the present embodiment can merge the process failure diagnosis conclusion that a variety of inference methods are obtained, and reduce
Conclusion difference after multi-source knowledge reasoning.
Most important concept is identification framework and Basic probability assignment function in D-S evidence theory.
Identification framework θ is a set being made up of the event of multiple independent mutual exclusions.θ all subsets composition set
Power set is done, with 2θRepresent.Basic probability assignment function m is defined in 2θOn mapping function, make
Wherein Φ is empty set, and A is any component in power set.M (A) is elementary probabilities of the event A in identification framework.
The difficult point of D-S evidence theory is the concrete form for the Basic probability assignment function for how determining each evidence, general root
There are different determination modes according to the difference of practical problem.In the case based on the regular expert system of confidence level and based on similarity
Expert system, asks for the evidence form of its conclusion using formula (3) and (7) respectively.
If two evidences have identical identification framework { A1,A2,…An, the most important fusion formula of D-S evidence theory can
To be defined as:
Wherein, m1And m2It is two evidences of separate sources.1≤i, j≤n, Ai∩Aj=C represents that event C can be by thing
Part AiAnd AjIt is intersecting to obtain.
It is normalization factor, is to weigh evidence m1And m2Between contradiction size it is important
Index.When the contradiction and larger difference between evidence, the confidence level of actually evidence occurs in that difference, is contemplated that different syndrome
According to Weight.Q bar evidence weighteds are averagely obtained into average evidence first, average evidence calculates such as formula (10):
Wherein,For the basic probability assignment of event A in average evidence, mi(A) it is the elementary probability in i-th evidence
Distribution,For the weight of i-th evidence.
Then average evidence is merged again, the Basic probability assignment function after note fusion is mf, then will be flat by (9)
Equal evidence fusion is that the procedural representation of final evidence is:
Wherein, provideIt is the symbol that two evidences are merged,For the average evidence obtained by (10).
Further, it is necessary to which explanation, during application on site, the reliability of different aforementioned sources is changed over time, and is
Improve the accuracy rate of fault diagnosis, the present embodiment also propose it is a kind of judged by analyze data feature reliability of evidence from
Adapt to Weight Value Distributed Methods.This method can accomplish that when data are accurate based on reasoning by cases result, data do not pass through on time
The weight of reasoning by cases is reduced to reduce its influence in final conclusion.Regulation is when data deviation normal fluctuation range is larger
In the relatively low state of data reliability.
During application on site, it is time window to take n before the current time data sampled, and asks for each variable in window
AverageThen the total amount of case deviation normal range (NR) to be matched can be represented such as by the degrees of offset of this j variable
Under:
xjFor j-th of variable of case to be matched,For the average value of the variable in sampling window.Expressed using e index
The weight w of reasoning by cases Knowledge Sourcedata:
wdata=e-kσ (13)
In above formula, k is coefficient, different according to actual conditions value.It is then determined that the power of Process Based expert system
Weight wrule:
wrule=1-wdata (14)
The above method realizes the fusion of two kinds of conclusions, and then can rationally provide failure diagnosis information, so as to operator
Member can be adjusted according to failure diagnosis information in time, it is ensured that production safety.
In addition, the Mixture of expert knowledge system method for diagnosing faults that following explanation is merged based on D-S
Fault diagnosis is generally divided into two steps, first tracing trouble, then reviews failure cause.Both use in Fig. 1
Method, but have difference again in concrete operations., it is necessary to set the identification frame of D-S theory forms for every kind of failure during fault diagnosis
Frame.These fault diagnosis frameworks are all comprising two elements, as shown in formula (15):
Wherein s is known failure mode number, is passed through formula (15), it can be determined that whether all known faults of process are sent out
It is raw.
During practical application, in rule-based knowledge system and knowledge system based on case, respectively reasoning obtain therefore
Hinder after the confidence level occurred and failure similarity, the nonevent possibility of failure is tried to achieve and similar by complementary events new probability formula
Degree.When the information that failure is caused by underlying causes and reason is reviewed is abundant, identification framework can be further set up former to failure
Because being reviewed, such as shown in formula (16):
{ failure cause 1, failure cause 2 ... failure cause t } (16)
Wherein, t is number the reason for causing certain failure, and different failures, t value is simultaneously different.
In actual applications, the step of Mixture of expert knowledge system fault diagnosis merged based on D-S, is explainable as follows:
(11) rule base and case library can be set up offline.
In the present embodiment, different expert systems handle different types of knowledge and data.If data have missing, lead to
Cross historical data completion.
Specifically, concentrator Fault diagnosis expert rule base is established offline:The foundation of rule base according to procedural knowledge and
Expertise, is summarized experience when expert carries out operating mode's switch and fault diagnosis to process, is given expression in the form of if-then
Come, many rules are collected for Fault diagnosis expert rule base.These rules have uncertainty degree, and rule are expressed with confidence level
Credibility then.
In addition, setting up rule base offline:Available fault diagnosis variable is selected, concentrator operation mechanism is analysed in depth, point
Analysis main exception and failure.The variable of concentrator running status can be fully demonstrated by being found out in multivariable of comforming, based on this,
Carry out the extraction of abnormal and diagnosis rule;Process data under different production status differs from one another, and data are classified with this,
Set up case library.
(12) rule-based reasoning and reasoning by cases are carried out respectively, obtain each self-corresponding evidence.
(13) assign evidence different weights according to the order of accuarcy of data.
(14) final conclusion is obtained by weight D-S fusions, assert that the elementary probability of distribution exceedes the correspondence of given threshold
Failure occurs.
(15) whether traversal judgement institute is faulty occurs, and obtains finally running conclusion.
In the present embodiment, the step of the step of failure cause is reviewed is with fault diagnosis is essentially identical, is proceeding to (14)
Step, is accordingly inferred to after failure cause, failure cause reviews end.
During application on site, expert will be known the observation of evidence by human-computer interaction device input system and entrance Mixture of expert
The rule-based system for knowing fault diagnosis system makes inferences, and obtains final conclusion.
Online acquisition to data directly input the case-based reasoning system carry out case of Mixture of expert knowledge fault diagnosis system
Example reasoning, it is also corresponding it is concluded that.
Fusion to two conclusions is realized by the D-S evidence fusions function in Mixture of expert knowledge fault diagnosis system.
During fusion, weight is distributed to two expert systems according to the order of accuarcy of data.
Institute's extracting method is verified using certain dense flow of refinery of enterprise hydrometallurgy as research object below.The process influences
The major failure of production has:Dashpot emits groove, underflow Flow Fault, underflow density failure etc. before pressure rake, filter press.
The measurable variable of concentrator production status, which can be reflected, 10, is respectively:Harrow bottom pressure 1, rake bottom pressure 2, in
Dashpot liquid level rate of change, Pulp pump electric current, Pulp pump before dashpot liquid level, filter press before heart stirring motor electric current, filter press
Frequency, excess flow, overflow turbidity, underflow flow.Store in case library is used for fault diagnosis by what this 10 variables were constituted
The case reviewed with failure cause.
In addition, operating personnel also have the method for a set of judgement critical process operating index.According to domain expert and operator
The experience and knowledge of member, sums up dense process failure diagnosis Expert Rules.Here, only enumerating the failure that part can be used for emulation
Diagnostic rule, as shown in table 1:
The concentrator Failure Diagnostic Code of table 1
In order to verify the validity of put forward method for diagnosing faults, from the factory, dense flow gathers the lunar system of in August, 2015-12
Row data are used for emulation experiment, and sample data is handled after filtering.Choose 100 groups and only exist the higher fault sample of concentration
Row fault diagnosis.Fault diagnosis result is as shown in Figure 2.Each number of regulation in figure in ordinate corresponding table 1,0 representative sample
Belong to nominal situation, 3 represent the higher failure of concentration.Abscissa is sampling sequence number.
Ideally, 100 samplings are all the higher fault samples of concentration, are the 3rd classes.From figure 2, it is seen that due to behaviour
The information change frequency provided as personnel is smaller, so the conclusion change frequency of Process Based is small and has mass data to be examined
Break to be normal.And the conclusion obtained by reasoning by cases is that, based on real time data, data movement is more frequent, but also occur
Situation about reporting by mistake more.The conclusion of two expert systems is merged by adaptive weighting D-S, only a small amount of samples is diagnosed as normally
Point, conclusion accuracy rate can reach 94%, and use rule-based reasoning and reasoning by cases merely, the accuracy of its fault diagnosis all compared with
Low, fusion diagnosis accuracy rate is significantly improved, and can meet field requirement.
So that dashpot emits groove before certain moment filter press as an example, this method answering in failure cause trace back process is explained in detail
With.Corresponding uncertain rule is as shown in table 2:
Dashpot emits groove reason throw back rule before the filter press of table 2
Sequence number | Regular former piece | Consequent | Rule intensity |
5 | Groove is emitted, less, flow is little for pulp density | Other reasonses cause to emit groove | 0.8 |
6 | Groove is emitted, pulp density is bigger than normal, and flow is little | Concentration height causes to emit groove | 0.8 |
7 | Groove is emitted, less, flow is bigger than normal for pulp density | Flow causes to emit groove greatly | 0.8 |
8 | Groove is emitted, pulp density is bigger than normal, and flow is bigger than normal | Concentration is high and flow causes to emit groove greatly | 0.8 |
(1) reasoning based on confidence level rule
Liquid level, which transfinites, first is to determine sexual behavior part.Confidence level per rule is all set to 0.8, the initial evidence that expert provides
Confidence level as shown in Table 3:
The event confidence level that the expert of table 3 provides
Event | The confidence level that expert provides |
Concentration is bigger than normal | 0.78 |
Concentration is little | -0.78 |
Flow is bigger than normal | 0.81 |
Flow is little | -0.81 |
Confidence level and the normalization for four kinds of possible causes for causing to emit groove are calculated, the card based on confidence level rule-based reasoning is obtained
According to:
m1={ only concentration is high, and only flow is big, and concentration is high and flow is big, other reasonses }
={ 0,0,1,0 }
(2) reasoning by cases based on data similarity
Determine and Process Based identical failure cause identification framework first.Ask for waiting to diagnose by arest neighbors method
Similarity and normalization of the case with each failure cause case, ask for its last evidence:
m2={ 0.04,0.23,0.73,0 }
(3) weight fusion conclusion
Analyze data, asks for the weight of different inference method evidences respectively, and this moment process operation is normal, all to survey change
Amount is all in historical data normal range (NR), and by formula (13) and formula (14), parameter k=1 respectively obtains reasoning by cases knot
It is 0.9 by weight, rule-based reasoning evidence weight is 0.1.Calculating average evidence by formula (10) is:
It is right by formula (9)Merged.Basic probability assignment function after fusion is:
M={ only concentration is high, and only flow is big, and concentration is high and flow is big, other reasonses }
={ 0,0.07,0.93,0 }
The rule-based reasoning conclusion for being based purely on confidence level thinks that concentration is high and flow is the unique original for causing to emit groove failure greatly
Cause.And the reasoning by cases based on data similarity then thinks that other reasonses are contributed to failure in various degree, this reality of more fitting
Border situation, can provide more accurate diagnosis for follow-up self-healing control.Conclusion is merged more by adaptive weighting D-S
Accurately.
In order to verify that adaptive weighting D-S fusion methods are more preferable than the diagnosis effect that fixed weight is merged.To 200 groups
Judge that the sample of the higher failure of concentration carries out failure cause and reviews and failure cause is reviewed into conclusion and log contrast.System
The meter failure cause rate of false alarm reviewed compares the quality of two kinds of fusion methods.To show the generalization ability of adaptive weighting method
Preferably, four kinds of situations are divided into and carry out l-G simulation test.As shown in Table 4, in each case, adaptive weighting D-S fusion after knot
All decreased by rate of false alarm.
Use and conclusion fusion of the raising of hybrid system fault diagnosis accuracy mainly by multi-source information are realized
, compared to the expert knowledge system of single reasoning, hybrid system has carried out dual identification to failure.Although adding calculating
Amount.But not cause obvious Diagnostic Time delayed.
The adaptive weighting D-S of table 4 merges contrast with fixed weight D-S
Fixed weight is reported by mistake | Adaptive weighting is reported by mistake | |
Rule Information is lacked | 6.5% | 4.0% |
Data message is lacked | 10% | 6.5% |
Noise 10% | 5.25% | 4.0% |
Data are more accurate | 5% | 3.5% |
It is average | 6.69% | 4.50% |
The technical principle of the present invention is described above in association with specific embodiment, these descriptions are intended merely to explain the present invention's
Principle, it is impossible to be construed to limiting the scope of the invention in any way.Based on explaining herein, those skilled in the art
Would not require any inventive effort can associate other embodiments of the present invention, and these modes fall within this hair
Within bright protection domain.
Claims (10)
1. the method for diagnosing faults of concentrator in a kind of hydrometallurgy process, it is characterised in that including:
Obtain online qualitative information and online quantitative data that hydrometallurgy concentrator is used to recognize a kind of failure;
For online qualitative information, the confidence level of each event is obtained using the method for the rule-based reasoning based on confidence level, and will
The confidence level of each event is modified to the evidence form of evidence theory, obtains first evidence;The event is Fault Identification
During the failure that is determined by online qualitative information occur or do not occur to cause failure in event, or failure cause trace back process
The reason for generation;
For online quantitative data, the similarity of follow-up conclusion example is obtained using the reasoning by cases method based on data similarity,
And the similarity of follow-up conclusion example is modified to the evidence form of evidence theory, obtain Article 2 evidence;The follow-up conclusion
The vector being made up of online quantitative data used when example is progress reasoning by cases;
According to D-S evidence theory fusion rule, two evidences are merged, the failure of concentrator in hydrometallurgy process is obtained
Diagnostic message, the failure diagnosis information includes Fault Identification and failure cause is reviewed.
2. according to the method described in claim 1, it is characterised in that obtain every using the method for the rule-based reasoning based on confidence level
The confidence level of one event, and the confidence level of each event is modified to the evidence form of evidence theory, first evidence is obtained,
Including:
If causing the confidence level cf of n event of failure generation1, cf2…cfn;
Normalization formula when obtaining first evidence is expressed as follows:
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<msub>
<mi>cf</mi>
<mn>1</mn>
</msub>
</mrow>
<msub>
<mi>m</mi>
<mn>11</mn>
</msub>
</mfrac>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>cf</mi>
<mn>2</mn>
</msub>
</mrow>
<msub>
<mi>m</mi>
<mn>12</mn>
</msub>
</mfrac>
<mo>=</mo>
<mo>...</mo>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>cf</mi>
<mi>n</mi>
</msub>
</mrow>
<msub>
<mi>m</mi>
<mrow>
<mn>1</mn>
<mi>n</mi>
</mrow>
</msub>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>m</mi>
<mn>11</mn>
</msub>
<mo>+</mo>
<msub>
<mi>m</mi>
<mn>12</mn>
</msub>
<mo>+</mo>
<mn>...</mn>
<mo>+</mo>
<msub>
<mi>m</mi>
<mrow>
<mn>1</mn>
<mi>n</mi>
</mrow>
</msub>
<mo>=</mo>
<mn>1</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>;</mo>
</mrow>
Wherein, the form of first evidence is { m11,m12…m1n};m11,m12…m1nFor the rule-based reasoning method based on confidence level
Identification framework in each element basic probability assignment.
3. method according to claim 2, it is characterised in that obtained using the reasoning by cases method based on data similarity
The similarity of follow-up conclusion example, including:
The similarity of follow-up conclusion example and source case is determined by following formula one;
Formula one:
Wherein, ck=(x1k,x2k,...xik...xjk), k=1,2 ... n represents the source case in case library, during n is case library
The sum of source case;xik(i=1,2 ..., be j) i-th of variate-value in the case of kth bar source, j is variable number;
X=(x1,x2,...,xj), represent follow-up conclusion example;sim(xi,xik) it is variable xiWith variable xikSimilarity;
α, β are respectively variable xiMinimum value and maximum in historical statistics value.
4. method according to claim 3, it is characterised in that
If follow-up is settled a lawsuit, the similarity of example and n source case in case library is sim1, sim2…simn;Then
Normalization formula when obtaining Article 2 evidence is expressed as follows:
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<msub>
<mi>sim</mi>
<mn>1</mn>
</msub>
</mrow>
<msub>
<mi>m</mi>
<mn>21</mn>
</msub>
</mfrac>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>sim</mi>
<mn>2</mn>
</msub>
</mrow>
<msub>
<mi>m</mi>
<mn>22</mn>
</msub>
</mfrac>
<mo>=</mo>
<mo>...</mo>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>sim</mi>
<mi>n</mi>
</msub>
</mrow>
<msub>
<mi>m</mi>
<mrow>
<mn>2</mn>
<mi>n</mi>
</mrow>
</msub>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>m</mi>
<mn>21</mn>
</msub>
<mo>+</mo>
<msub>
<mi>m</mi>
<mn>22</mn>
</msub>
<mo>+</mo>
<mn>...</mn>
<mo>+</mo>
<msub>
<mi>m</mi>
<mrow>
<mn>2</mn>
<mi>n</mi>
</mrow>
</msub>
<mo>=</mo>
<mn>1</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>;</mo>
</mrow>
Wherein, the form of Article 2 evidence is { m21,m22…m2n};m21,m22…m2nFor each in the identification framework of reasoning by cases
The basic probability assignment of element.
5. method according to claim 4, it is characterised in that according to D-S evidence theory fusion rule, two evidences are entered
Row fusion, obtains the failure diagnosis information of concentrator in hydrometallurgy process, including:
If first evidence and Article 2 evidence have identical identification framework { A1,A2,…An, and first evidence and second
The confidence level difference of bar evidence within a preset range, then
Merged using following formula,
<mrow>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>C</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>K</mi>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>A</mi>
<mi>i</mi>
</msub>
<mo>&cap;</mo>
<msub>
<mi>A</mi>
<mi>j</mi>
</msub>
<mo>=</mo>
<mi>C</mi>
</mrow>
</munder>
<msub>
<mi>m</mi>
<mn>1</mn>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>A</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<msub>
<mi>m</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>A</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein, m1For first evidence, m2For Article 2 evidence, 1≤i, j≤n, Ai∩Aj=C represents that event C can be by event Ai
And AjIt is intersecting to obtain;It is normalization factor, is to weigh evidence m1And m2Between contradiction size
Index.
6. method according to claim 5, it is characterised in that
If first evidence and Article 2 evidence have identical identification framework { A1,A2,…An, and first evidence and second
The confidence level difference of bar evidence is more than preset range, then
Determine the weight of first evidence, and Article 2 evidence weight;
It will averagely obtain average evidence using following formula q bar evidences weighteds, q=2,
<mrow>
<mover>
<mi>m</mi>
<mo>&OverBar;</mo>
</mover>
<mrow>
<mo>(</mo>
<mi>A</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>q</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mi>i</mi>
<mi>q</mi>
</munderover>
<msubsup>
<mi>w</mi>
<mi>i</mi>
<mi>e</mi>
</msubsup>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>A</mi>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein,For the basic probability assignment of event A in average evidence, mi(A) it is the elementary probability point in i-th evidence
Match somebody with somebody,For the weight of i-th evidence;
And, according to formulaAverage evidence is merged;
Wherein, mfThe Basic probability assignment function after fusion is represented,The symbol that two evidences are merged is represented,To pass throughThe average evidence obtained.
7. method according to claim 6, it is characterised in that determine the weight of first evidence, and Article 2 evidence
Weight, including:
N before at the time of obtaining the online quantitative data data sampled are time window, determine each to sample in window
The average of dataAnd
The total amount that follow-up conclusion example deviates normal range (NR) is expressed as:
<mrow>
<mi>&sigma;</mi>
<mo>=</mo>
<mo>|</mo>
<msub>
<mi>x</mi>
<mn>1</mn>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>x</mi>
<mn>1</mn>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>|</mo>
<mo>+</mo>
<mo>|</mo>
<msub>
<mi>x</mi>
<mn>2</mn>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>x</mi>
<mn>2</mn>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>|</mo>
<mo>+</mo>
<mo>...</mo>
<mo>+</mo>
<mo>|</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>|</mo>
<mo>;</mo>
</mrow>
xjJ-th of variable of example of being settled a lawsuit for follow-up,For the average value of the variable in the time window of sampling;
The then weight w of Article 2 evidencedataIt is expressed as:wdata=e-kσ;
Wherein, k is coefficient;
The weight w of first evidenceruleIt is expressed as:wrule=1-wdata。
8. method according to claim 7, it is characterised in that methods described also includes:
The faulty failure diagnosis information of institute that concentrator may occur is obtained using traversal method, passes through the faulty failure of institute
Diagnostic message, determines final conclusion.
9. according to the method described in claim 1, it is characterised in that obtaining hydrometallurgy concentrator is used to recognize a kind of failure
Before the step of online qualitative information and online quantitative data, methods described also includes:
The offline historical variable for obtaining hydrometallurgy concentrator in preset time period;
Priori with reference to field of hydrometallurgy expert and associative operation personnel, obtain and multiple be used to carrying out concentrator failure examining
Disconnected Expert Rules;
According to multiple Expert Rules for being used to carry out concentrator fault diagnosis, Expert Rules storehouse is set up;
Wherein, each concentrator Fault diagnosis expert rule includes:Each regular regular former piece, consequent, rule intensity;
Regular former piece, consequent and the rule intensity of each rule have corresponding relation;
And/or,
In the method for rule-based reasoning based on confidence level, the form of production rule is:IF E THEN H WITH cf(H/E)
Wherein, E represents a simple premise, or multiple simple premise logical combinations logical combination premise, H is one or more
Conclusion, cf (H/E) is the confidence level for occurring H based on the E;
Correspondingly, the step of confidence level of each event being obtained using the method for the rule-based reasoning based on confidence level, including:
Regular former piece E is set up, and the confidence level of establishment is cf (H/E);
Then each event e of online qualitative information composition confidence level cf (H/e) is:
Cf (H/e)=cf (H/E) × max { cf (E/e), 0 };
Wherein, cf (E/e) represents event e and regular former piece matching degree, and cf (H/E) represents rule intensity.
10. according to the method described in claim 1, it is characterised in that obtaining hydrometallurgy concentrator is used to recognize a kind of failure
Online qualitative information and the step of online quantitative data before, methods described also includes:
Abnormal data occur for hydrometallurgy concentrator in collected offline preset time period, and no exceptions data;
According to the data of collection, case library is set up.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710379067.4A CN107065834B (en) | 2017-05-25 | 2017-05-25 | The method for diagnosing faults of concentrator in hydrometallurgy process |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710379067.4A CN107065834B (en) | 2017-05-25 | 2017-05-25 | The method for diagnosing faults of concentrator in hydrometallurgy process |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107065834A true CN107065834A (en) | 2017-08-18 |
CN107065834B CN107065834B (en) | 2019-01-22 |
Family
ID=59610545
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710379067.4A Expired - Fee Related CN107065834B (en) | 2017-05-25 | 2017-05-25 | The method for diagnosing faults of concentrator in hydrometallurgy process |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107065834B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108257365A (en) * | 2018-01-29 | 2018-07-06 | 杭州电子科技大学 | A kind of industrial alarm designs method based on global nonspecific evidence dynamic fusion |
CN108734197A (en) * | 2018-04-17 | 2018-11-02 | 东北大学 | A kind of Fault monitoring and diagnosis method of the dense washing process of hydrometallurgy |
CN110046651A (en) * | 2019-03-15 | 2019-07-23 | 西安交通大学 | A kind of pipeline conditions recognition methods based on monitoring data multi-attribute feature fusion |
CN110307990A (en) * | 2019-06-05 | 2019-10-08 | 中国中车股份有限公司 | A kind of intelligent diagnosing method and system of the diesel engine lubricating oil system failure |
CN111611279A (en) * | 2020-04-24 | 2020-09-01 | 中国电子科技集团公司第二十九研究所 | Microwave assembly fault diagnosis system and method based on test index similarity |
CN111667197A (en) * | 2020-06-23 | 2020-09-15 | 环胜电子(深圳)有限公司 | Production quality diagnosis method and system for surface mount technology manufacturing process |
CN112326283A (en) * | 2020-10-16 | 2021-02-05 | 山东黄金矿业(莱州)有限公司三山岛金矿 | Thickener underflow blocking fault diagnosis method based on wavelet analysis |
CN113933707A (en) * | 2021-10-12 | 2022-01-14 | 哈尔滨电机厂有限责任公司 | Self-adjusting method for judging reliability of fault reason of generator set |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5638460A (en) * | 1989-12-29 | 1997-06-10 | Canon Kabushiki Kaisha | Method of evaluating object based upon image processing and inspection apparatus using said method |
CN102930281A (en) * | 2011-08-11 | 2013-02-13 | 金宏斌 | Dempster-Shafer (DS) theory and Dezert-Smarandache (DSm) theory-based interactive self-adaptive target identification method |
CN103135014A (en) * | 2012-12-14 | 2013-06-05 | 西安电子科技大学 | Transformer fault diagnosis method based on case-based reasoning |
CN103246762A (en) * | 2013-04-10 | 2013-08-14 | 哈尔滨工程大学 | Method of comprehensive evaluation for simulation credibility of electric propulsion system |
CN103810526A (en) * | 2014-01-28 | 2014-05-21 | 北京仿真中心 | Knowledge fusion method based on D-S evidence theory |
CN104991549A (en) * | 2015-05-27 | 2015-10-21 | 中国铁路总公司 | Track circuit red-light strip default diagnosis method based on FTA and multilevel fuzzy-neural sub-networks |
CN105046067A (en) * | 2015-07-03 | 2015-11-11 | 西北工业大学 | Multi-sensor information fusion method based on evidence similarity |
CN105373700A (en) * | 2015-10-30 | 2016-03-02 | 哈尔滨工程大学 | Method for mechanical fault diagnosis based on information entropies and evidence theory |
-
2017
- 2017-05-25 CN CN201710379067.4A patent/CN107065834B/en not_active Expired - Fee Related
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5638460A (en) * | 1989-12-29 | 1997-06-10 | Canon Kabushiki Kaisha | Method of evaluating object based upon image processing and inspection apparatus using said method |
CN102930281A (en) * | 2011-08-11 | 2013-02-13 | 金宏斌 | Dempster-Shafer (DS) theory and Dezert-Smarandache (DSm) theory-based interactive self-adaptive target identification method |
CN103135014A (en) * | 2012-12-14 | 2013-06-05 | 西安电子科技大学 | Transformer fault diagnosis method based on case-based reasoning |
CN103246762A (en) * | 2013-04-10 | 2013-08-14 | 哈尔滨工程大学 | Method of comprehensive evaluation for simulation credibility of electric propulsion system |
CN103810526A (en) * | 2014-01-28 | 2014-05-21 | 北京仿真中心 | Knowledge fusion method based on D-S evidence theory |
CN104991549A (en) * | 2015-05-27 | 2015-10-21 | 中国铁路总公司 | Track circuit red-light strip default diagnosis method based on FTA and multilevel fuzzy-neural sub-networks |
CN105046067A (en) * | 2015-07-03 | 2015-11-11 | 西北工业大学 | Multi-sensor information fusion method based on evidence similarity |
CN105373700A (en) * | 2015-10-30 | 2016-03-02 | 哈尔滨工程大学 | Method for mechanical fault diagnosis based on information entropies and evidence theory |
Non-Patent Citations (4)
Title |
---|
张博: "湿法冶金浓密洗涤过程监测与故障诊断", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
张春晓,严爱军,王普: "一种改进的案例推理分类方法研究", 《自动化学报》 * |
汤永利,李伟杰,于金霞,闫玺玺: "基于改进D-S 证据理论的网络安全态势", 《南京理工大学学报》 * |
聂新雨: "基于信息融合的轧制过程故障诊断方法研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108257365B (en) * | 2018-01-29 | 2020-04-24 | 杭州电子科技大学 | Industrial alarm design method based on global uncertainty evidence dynamic fusion |
CN108257365A (en) * | 2018-01-29 | 2018-07-06 | 杭州电子科技大学 | A kind of industrial alarm designs method based on global nonspecific evidence dynamic fusion |
CN108734197A (en) * | 2018-04-17 | 2018-11-02 | 东北大学 | A kind of Fault monitoring and diagnosis method of the dense washing process of hydrometallurgy |
CN110046651A (en) * | 2019-03-15 | 2019-07-23 | 西安交通大学 | A kind of pipeline conditions recognition methods based on monitoring data multi-attribute feature fusion |
CN110046651B (en) * | 2019-03-15 | 2021-01-19 | 西安交通大学 | Pipeline state identification method based on monitoring data multi-attribute feature fusion |
CN110307990A (en) * | 2019-06-05 | 2019-10-08 | 中国中车股份有限公司 | A kind of intelligent diagnosing method and system of the diesel engine lubricating oil system failure |
CN111611279B (en) * | 2020-04-24 | 2023-09-12 | 中国电子科技集团公司第二十九研究所 | Microwave component fault diagnosis system and method based on test index similarity |
CN111611279A (en) * | 2020-04-24 | 2020-09-01 | 中国电子科技集团公司第二十九研究所 | Microwave assembly fault diagnosis system and method based on test index similarity |
CN111667197A (en) * | 2020-06-23 | 2020-09-15 | 环胜电子(深圳)有限公司 | Production quality diagnosis method and system for surface mount technology manufacturing process |
CN111667197B (en) * | 2020-06-23 | 2023-05-05 | 环荣电子(惠州)有限公司 | Production quality diagnosis method and system for surface adhesion technology manufacturing process |
CN112326283A (en) * | 2020-10-16 | 2021-02-05 | 山东黄金矿业(莱州)有限公司三山岛金矿 | Thickener underflow blocking fault diagnosis method based on wavelet analysis |
CN113933707A (en) * | 2021-10-12 | 2022-01-14 | 哈尔滨电机厂有限责任公司 | Self-adjusting method for judging reliability of fault reason of generator set |
CN113933707B (en) * | 2021-10-12 | 2023-09-19 | 哈尔滨电机厂有限责任公司 | Self-adjusting method for judging reliability of fault cause of generator set |
Also Published As
Publication number | Publication date |
---|---|
CN107065834B (en) | 2019-01-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107065834A (en) | The method for diagnosing faults of concentrator in hydrometallurgy process | |
CN107436597B (en) | A kind of chemical process fault detection method based on sparse filtering and logistic regression | |
CN110320892A (en) | The sewage disposal device fault diagnosis system and method returned based on Lasso | |
CN101957889B (en) | Selective wear-based equipment optimal maintenance time prediction method | |
CN105608842B (en) | A kind of damaged online monitoring alarm device of nuclear reactor fuel | |
CN103793853A (en) | Overhead power transmission line running state assessment method based on bidirectional Bayesian network | |
CN104677997B (en) | A kind of transformer oil chromatographic on-line monitoring differentiation method for early warning | |
CN104615122B (en) | A kind of industry control signal detection system and detection method | |
CN107291830A (en) | A kind of creation method of equipment knowledge base | |
CN108921230A (en) | Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network | |
CN101021723A (en) | Melt index detection fault diagnozing system and method in propylene polymerization production | |
CN110163075A (en) | A kind of multi-information fusion method for diagnosing faults based on Weight Training | |
CN107169658B (en) | Reliability-based fault diagnosis method for hydrometallurgical thickener | |
DE102021201296A1 (en) | Prediction model for predicting product quality parameter values | |
CN112632845B (en) | Data-based mini-reactor online fault diagnosis method, medium and equipment | |
CN114519923A (en) | Intelligent diagnosis and early warning method and system for power plant | |
CN106682835A (en) | Data-driven complex electromechanical system service quality state evaluation method | |
CN101738998A (en) | System and method for monitoring industrial process based on local discriminatory analysis | |
CN106933097A (en) | A kind of Fault Diagnosis for Chemical Process method based on multi-level optimization PCC SDG | |
CN111160393B (en) | Modularized modeling method of carrier rocket health evaluation model based on data driving | |
CN114002597A (en) | Motor fault diagnosis method and system based on GRU network stator current analysis | |
CN117032165A (en) | Industrial equipment fault diagnosis method | |
Tripathy et al. | Explaining Anomalies in Industrial Multivariate Time-series Data with the help of eXplainable AI | |
CN111210085B (en) | Coal mine gas concentration early warning method based on multi-view ensemble learning | |
CN110532699B (en) | Fuzzy DCD-based fault diagnosis method for hydrometallurgy dense washing process |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190122 |
|
CF01 | Termination of patent right due to non-payment of annual fee |