CN107038481A - A kind of case-based reasoning system building method towards metallurgical mine field - Google Patents
A kind of case-based reasoning system building method towards metallurgical mine field Download PDFInfo
- Publication number
- CN107038481A CN107038481A CN201710196559.XA CN201710196559A CN107038481A CN 107038481 A CN107038481 A CN 107038481A CN 201710196559 A CN201710196559 A CN 201710196559A CN 107038481 A CN107038481 A CN 107038481A
- Authority
- CN
- China
- Prior art keywords
- case
- library
- new problem
- solution
- new
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000011156 evaluation Methods 0.000 claims abstract description 40
- 238000012544 monitoring process Methods 0.000 claims abstract description 24
- 238000012423 maintenance Methods 0.000 claims description 19
- 238000000605 extraction Methods 0.000 claims description 5
- 238000007667 floating Methods 0.000 description 23
- 238000005188 flotation Methods 0.000 description 18
- 238000004519 manufacturing process Methods 0.000 description 12
- 239000012141 concentrate Substances 0.000 description 11
- 238000005516 engineering process Methods 0.000 description 8
- 239000000284 extract Substances 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 238000011524 similarity measure Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000001149 cognitive effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- CWYNVVGOOAEACU-UHFFFAOYSA-N Fe2+ Chemical compound [Fe+2] CWYNVVGOOAEACU-UHFFFAOYSA-N 0.000 description 1
- 229920002472 Starch Polymers 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- CNOILWYCLUVYET-UHFFFAOYSA-N ethyl n-[4-[benzyl(2-phenylethyl)amino]-2-(3-methylphenyl)-1h-imidazo[4,5-c]pyridin-6-yl]carbamate Chemical compound N=1C(NC(=O)OCC)=CC=2NC(C=3C=C(C)C=CC=3)=NC=2C=1N(CC=1C=CC=CC=1)CCC1=CC=CC=C1 CNOILWYCLUVYET-UHFFFAOYSA-N 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 235000019698 starch Nutrition 0.000 description 1
- 239000008107 starch Substances 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Educational Administration (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of case-based reasoning system building method towards metallurgical mine field, it is possible to increase the validity and accuracy of case reasoning.Methods described includes:Generate case library;Monitoring Data is obtained, if the excursion of certain characteristic value exceeds default range threshold in the Monitoring Data, the Monitoring Data is converted into new problem;If the case that the case library is not matched completely with the new problem, similarity is chosen from the case library more than the alternative case that the case of the first predetermined threshold value is reused as case;According to obtained alternative case, the solution of the new problem is calculated;If the solution for calculating the obtained new problem meets evaluation criterion, the solution of the new problem and the new problem is then constituted into a new case, judge whether the new case and the highest similarity of case in the case library are more than the second predetermined threshold value, if more than the second predetermined threshold value, by new case storage into the case library.The present invention is applied to metallurgical mine field.
Description
Technical field
The present invention relates to automatic control technology field, a kind of case-based reasoning system towards metallurgical mine field is particularly related to
Building method.
Background technology
Bargh's production is numerous due to its Consideration, and there is complicated coupled relation, its dynamic dispatching between factor
Problem is particularly difficult.Mining production complex process, production equipment are more, logistics is crisscross.Not only to consider ore reach when
Between deviation, equipment fault, and need consider ore property how, floating agent adjustment etc. dynamic disturbances event.So, move
State scheduling is the key link of bargh's wisdom.
Solving dynamic scheduling problem needs rich experience knowledge, and setting up experts database, rule base or case library not only can be with
Mitigate workload, while can avoid making floating concentrate grade fluctuation because technical staff's know-how is low or lacks experience, very
To there are quality problems.Case-based reasoning technology provides a kind of in the occasion for lacking system model but having wide experience
Under the method that is solved to problem.Meanwhile, CBR increases with the case of preservation, and case library is increasingly perfect, pushes away
The validity and accuracy of reason can be lifted increasingly.
During expert system (Expert System, ES) is artificial intelligence (Artifcial Intelligence, AI) field
One of most successful research field.The knowledge model that expert system is generally based on display expression carrys out Solve problems, whether
Superficial knowledge or deep knowledge, all must extract and realize the model of the domain knowledge of a display expression.One expert system of exploitation
The technology generally used of uniting is RBR, and the foundation of rule-based system passes through leads to from retrievable extracting data
With the rule of property, and applied in the situation of current problem, be only the simple analog to mankind's abstract thinking.Gathered around at those
There is the field enriched one's knowledge, the system of Process Based (Rule-Based Reasoning, RBR) has become one relatively
Ripe technology.However, people encounter these following problems in the development process of RBR systems:
There is knowledge acquisition bottleneck (knowledge elicitation bottleneck) in RBR systems;
The realization of RBR systems be difficult process, it is necessary to special skill and generally to consume substantial amounts of manpower and
The time of several years;The generally operation of RBR systems is slower;
RBR systems seem fragile in large-scale information processing;
RBR systems it is difficult in maintenance.
In order to overcome the weakness of RBR systems, enlightened by human cognitive process, if people have experience in process problem
And gear shaper without theoretical can equally solve problem, then brainstrust puts forward a kind of inference method of new Solve problems --- based on case
Reasoning (Case-Based Reasoning, CBR).
CBR abbreviation reasoning by cases, also referred to as Case Based Reasoning, cite a precedent reasoning and case-based reasoning, as artificial intelligence by top layer
Machine imitates to a kind of developing form of mind over machine of deep layer and has obtained cognitive science and artificial intelligence study person increasingly
Many attention.CBR employs extremely different method, with reference of the specific case data as current problem solution.Expert
Substantial amounts of thinking in images has been used in decision-making.The solution of problem, is not that simple dependence rule can solve the problem that, in addition it is also necessary to specially
The experience of family.Compared with traditional method based on model, CBR need not show the domain model of expression, bad suitable for solving
Structure problem.CBR solves new problem by adjusting the successful answer of conventional relevant issues.
Itself professional knowledge is all placed under the framework of CBR and comprehensive by the research of current all kinds of specific fields
The knowledge such as mathematics, statistics, form and solve coping mechanism of problems in the field.In spite of cognitive science and artificial intelligence
Deng the support of some correlation theories, but so far, CBR technologies, still without a set of tight system is formed, are walked to each in its course of work
It is rapid adopt how, the unified mature technology of neither one and theoretical direction.Can only be tried in numerous theoretical researches
Selection is tested, the best approach that this area studies a question is directed to reach.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of case-based reasoning system construction side towards metallurgical mine field
Method, with solve present in prior art without provide towards metallurgical mine field reasoning by cases method the problem of.
In order to solve the above technical problems, the embodiment of the present invention provides a kind of case-based reasoning system towards metallurgical mine field
Building method, including:
Case library is generated, wherein, each case in case library is the structural description to a problem in application field;
Each case includes:The Expressive Features of problem and the solution of problem;
Monitoring Data is obtained, if the excursion of certain characteristic value exceeds default range threshold in the Monitoring Data,
The Monitoring Data is converted into by new problem according to the structure of case;
Calculate the similarity between each case in the new problem and the case library, if the case library not with institute
The case that new problem is matched completely is stated, then similarity is chosen from the case library is used as case more than the case of the first predetermined threshold value
The alternative case that example is reused;
The alternative case reused according to obtained case, calculates the solution of the new problem;
The solution of the new problem obtained according to default evaluation criterion to calculating is evaluated;
If the solution for calculating the obtained new problem meets evaluation criterion, by the new problem and the solution of the new problem
A new case is constituted, judges whether the highest similarity of the new case and case in the case library is more than the second default threshold
Value, if more than the second predetermined threshold value, by new case storage into the case library.
Further, the generation case library includes:
According to the Expressive Features of problem and the solution of problem, extract corresponding original from the data warehouse pre-set
Data;
The initial data to extraction is clustered, using the cluster centre of each class as respective classes representative case
Example, extracts the case that represents and enters case library.
Further, the generation case library includes:
According to the Expressive Features of problem and the solution of problem, adjustment case is formulated;
According to evaluation result of the expert to the adjustment case, it is that excellent adjustment case enters case to choose evaluation result
Storehouse.
Further, the similarity calculated in the new problem and the case library between each case includes:
Similarity in the new problem and the case library between each case, described first are calculated by the first formula
Formula is expressed as:
Wherein, sim (D, Dk) represent new problem D and kth bar case D in case librarykBetween similarity;fiExpression is newly asked
Inscribe D i-th of Expressive Features;M represents the number of Expressive Features;fI, kRepresent kth bar case D in case librarykI-th description
Feature;sim(fi, fI, k) represent new problem D i-th of Expressive Features fiWith kth bar case D in case librarykI-th of description it is special
Levy fI, kBetween similarity;ωiRepresent the weight of i-th of Expressive Features;K represents the number of case in case library.
Further, if fi, fI, kFor continuous variable, then sim (fi, fI, k) be expressed as:
sim(fi, fI, k)=1- | fi-fI, k|/max(fi, fI, k)。
Further, if fi, fI, kFor discrete variable, then sim (fi, fI, k) be expressed as:
Further, the alternative case that the case that the basis is obtained is reused, calculating the solution of the new problem includes:
The alternative case reused according to obtained case, obtains the solution J=(J of the new problem1, J2..., Jl), wherein,
JlIt is expressed as:
Wherein, wkThe weights of k-th of alternative case are represented, r represents the number of alternative case, Jl,kRepresent k-th of alternative case
L-th of solution of example, JlRepresent l-th of solution of new problem.
Further, wk=Simk;
Wherein, SimkRepresent the similarity between k-th of alternative case and the new problem.
Further, methods described also includes:
If the solution for calculating the obtained new problem is unsatisfactory for evaluation criterion, the new problem is modified;
The solution of the revised new problem is evaluated according to default evaluation criterion.
Further, methods described also includes:
Collect single case, the daily record of whole case library, the maintenance daily record of whole case-based reasoning system;
Wherein, the daily record of single case includes:Record the number of times of each case successful application, the number of times that failure is applied, comment
Value, last adjustment result;
The daily record of whole case library includes:The size of case library is monitored, the Case No. called daily is recorded and calls case
The time in storehouse;
The maintenance daily record of whole case-based reasoning system includes:The maintenance time of whole case-based reasoning system, maintenance mode, touch
Whether hair reason, expert operate.
The above-mentioned technical proposal of the present invention has the beneficial effect that:
In such scheme, by generating case library, wherein, each case in case library is that one in application field is asked
The structural description of topic;Each case includes:The Expressive Features of problem and the solution of problem;Monitoring Data is obtained, if the monitoring
The excursion of the characteristic value of certain in data exceeds default range threshold, then is converted the Monitoring Data according to the structure of case
For new problem;Calculate the similarity between each case in the new problem and case library, if the case library not with it is described
The case that new problem is matched completely, then choose similarity from the case library and be used as case more than the case of the first predetermined threshold value
The alternative case reused;The alternative case reused according to obtained case, calculates the solution of the new problem;Evaluated according to default
The solution for the new problem that standard is obtained to calculating is evaluated;If the solution for calculating the obtained new problem, which is met, evaluates mark
Standard, then constitute a new case by the solution of the new problem and the new problem, judges in the new case and the case library
Whether the highest similarity of case is more than the second predetermined threshold value, if more than the second predetermined threshold value, the new case storage is arrived
In the case library, increasingly perfect case library, it is possible to increase the validity and accuracy of case reasoning, so as to be bargh
Dynamic dispatching provides support.
Brief description of the drawings
Fig. 1 shows for the flow of the case-based reasoning system building method provided in an embodiment of the present invention towards metallurgical mine field
It is intended to;
Fig. 2 is the detailed stream of the case-based reasoning system building method provided in an embodiment of the present invention towards metallurgical mine field
Journey schematic diagram;
Fig. 3 is the case hierarchical structure schematic diagram of floating operation.
Embodiment
To make the technical problem to be solved in the present invention, technical scheme and advantage clearer, below in conjunction with accompanying drawing and tool
Body embodiment is described in detail.
The present invention is for existing without there is provided one kind the problem of offer towards the reasoning by cases method in metallurgical mine field
Towards the case-based reasoning system building method in metallurgical mine field.
Referring to shown in Fig. 1, the case-based reasoning system building method provided in an embodiment of the present invention towards metallurgical mine field,
Including:
S11, generates case library, wherein, each case in case library is the structuring to a problem in application field
Description;Each case includes:The Expressive Features of problem and the solution of problem;
S12, obtains Monitoring Data, if the excursion of certain characteristic value exceeds default scope threshold in the Monitoring Data
Value, then be converted into new problem according to the structure of case by the Monitoring Data;
S13, calculates the similarity between each case in the new problem and the case library, if the case library does not have
The case matched completely with the new problem, the then case that similarity is chosen from the case library more than the first predetermined threshold value is made
The alternative case reused for case;
S14, the alternative case reused according to obtained case calculates the solution of the new problem;
S15, the solution of the new problem obtained according to default evaluation criterion to calculating is evaluated;
S16, if the solution for calculating the obtained new problem meets evaluation criterion, by the new problem and the new problem
Solution constitute a new case, judge whether the highest similarity of the new case and case in the case library pre- more than second
If threshold value, if more than the second predetermined threshold value, by new case storage into the case library.
The case-based reasoning system building method towards metallurgical mine field described in the embodiment of the present invention, by generating case
Storehouse, wherein, each case in case library is the structural description to a problem in application field;Each case includes:Ask
The Expressive Features of topic and the solution of problem;Monitoring Data is obtained, if the excursion of certain characteristic value is beyond pre- in the Monitoring Data
If range threshold, then the Monitoring Data is converted into by new problem according to the structure of case;Calculate the new problem and case
Similarity in storehouse between each case, if the case that the case library is not matched completely with the new problem, from described
Similarity is chosen in case library more than the alternative case that the case of the first predetermined threshold value is reused as case;According to obtained case
The alternative case reused, calculates the solution of the new problem;The new problem obtained according to default evaluation criterion to calculating
Solution is evaluated;If the solution for calculating the obtained new problem meets evaluation criterion, by the new problem and the new problem
Solution constitute a new case, judge whether the highest similarity of the new case and case in the case library pre- more than second
If threshold value, if more than the second predetermined threshold value, by new case storage into the case library, increasingly perfect case library,
The validity and accuracy of case reasoning can be improved, so as to provide support for bargh's dynamic dispatching.
In the present embodiment, each case in case library is the structural description to a problem in application field, one
Case-based reasoning is largely to rely on the structure and content of collected case, and scheduling case representation technology is generally wrapped
Include:The Expressive Features of problem and the solution of problem.
In the foregoing embodiment towards the case-based reasoning system building method in metallurgical mine field, further
Ground, the generation case library includes:
According to the Expressive Features of problem and the solution of problem, extract corresponding original from the data warehouse pre-set
Data;
The initial data to extraction is clustered, using the cluster centre of each class as respective classes representative case
Example, extracts the case that represents and enters case library.
In the present embodiment, according to the description of case, according to the Expressive Features of problem and the solution of problem, set from advance
The data warehouse put extracts corresponding initial data, and clustering is carried out to initial data, using the class center of cluster result as
The representative case of respective classes, extracts the case that represents and enters case library.If initial data is directly placed into case library, make
For the case in case library, then because the difference between initial data is smaller, and the larger retrieval for being unfavorable for case of data volume, institute
The representative case of different is obtained by cluster with the present embodiment, it is possible to increase the speed and otherness of Case Retrieval.
In the foregoing embodiment towards the case-based reasoning system building method in metallurgical mine field, further
Ground, the generation case library includes:
According to the Expressive Features of problem and the solution of problem, adjustment case is formulated;
According to evaluation result of the expert to the adjustment case, it is that excellent adjustment case enters case to choose evaluation result
Storehouse.
In the present embodiment, because early stage case bar number is less, it is possible to according to the Expressive Features of problem and the solution of problem
Certainly scheme, the adjustment case of some special operation conditions is formulated by expert, the case library to enrich initial stage.By different adjustment cases
It is presented to expert to be evaluated, chooses wherein scoring and enter case library for excellent adjustment case.
In the present embodiment, the case in the case library be the class center case (representing case) that is obtained by cluster and
The top grade adjustment case of expert opinion.Need to store the importance (weight) of Expressive Features in the case in case library simultaneously, with
Continue searching step after an action of the bowels to use, the importance (weight) of Expressive Features in the case can be specified by expert, and can be adjusted at any time.
In the foregoing embodiment towards the case-based reasoning system building method in metallurgical mine field, further
Similarity in ground, the calculating new problem and the case library between each case includes:
Similarity in the new problem and the case library between each case, described first are calculated by the first formula
Formula is expressed as:
Wherein, sim (D, Dk) represent new problem D and kth bar case D in case librarykBetween similarity;fiExpression is newly asked
Inscribe D i-th of Expressive Features;M represents the number of Expressive Features;fI, kRepresent kth bar case D in case librarykI-th description
Feature;sim(fi, fI, k) represent new problem D i-th of Expressive Features fiWith kth bar case D in case librarykI-th of description it is special
Levy fI, kBetween similarity;ωiRepresent the weight of i-th of Expressive Features;K represents the number of case in case library.
In the present embodiment, after generation case library, the case-based reasoning system obtains Monitoring Data, if in the Monitoring Data
The excursion of certain characteristic value exceeds default range threshold, then is converted into the Monitoring Data according to the structure of case and newly asks
Topic, then retrieved by case library.
In the present embodiment, the whether successful criterion of case-based reasoning system retrieves phase depending on the case-based reasoning system
Like case to provide the ability of new problem solution.The substantive characteristics of case is necessary in the substantive characteristics and case library of new problem
It is the basis that case-based reasoning system is solved with similarity relationships.Similarity problem have impact on each of case-based reasoning system reasoning
Aspect, so the definition of similarity is particularly important.
In the present embodiment, according to the Expressive Features of new problem, it is possible to use the K that neighbouring method allows in new problem and case library is individual
Case is compared one by one, calculates K to the similarity between case.
In the present embodiment, new problem, D can be represented with DkKth bar case in case library is represented, new problem D description is special
Levy as F={ fi(i=1 ..., m), kth bar case D in case librarykExpressive Features be F_k={ fI, k(k=1,2 ..., K),
K is the quantity of case in case library.Preferably, D and D can be calculated by the first formulakBetween similarity sim (D, Dk), institute
The first formula is stated to be expressed as:
Wherein, sim (D, DkRepresent new problem D and kth bar case D in case librarykBetween similarity;fiRepresent new problem
D i-th of Expressive Features;M represents the number of Expressive Features;fI, kRepresent kth bar case D in case librarykI-th of description it is special
Levy;sim(fi, fI, k) represent kth bar case D in new problem D i-th of Expressive Features fi and case librarykI-th of Expressive Features
fI, kBetween similarity;ωiRepresent the weight of i-th of Expressive Features;K represents the number of case in case library.
In the present embodiment, it is preferable that if fi, fI, kFor continuous variable, then sim (fi, fI, k) be expressed as:
sim(fi, fI, k)=1- | fi-fI, k|/max(fi, fI, k)。
In the present embodiment, it is preferable that if fi, fI, kFor discrete variable, then sim (fi, fI, k) be expressed as:
In the present embodiment, according to the first formula, calculate similar between new problem D and each case in case library one by one
After degree, be ranked up display according to the descending case in case library of similarity, if the case library not with it is described newly
The case (Similarity value is 99%) that problem is matched completely, then choose similarity more than the first predetermined threshold value from the case library
The alternative case that is reused as case of case, for example, first predetermined threshold value can be 80%, as shown in Figure 2.
In the present embodiment, it is assumed that the case { F that r similarity is more than 80% is retrieved from case library11,F2,…,Fr,
{F11,F2,…,FrAlternately case, wherein, Fk(k=1 ..., it is r) Sim with the similarity of new problemk, FkCorresponding solution
For Jk=(J1,k, J2,k..., Jl,k), then the solution of new problem is J=(J1, J2..., Jl), wherein,
Wherein, wkThe weights of k-th of alternative case are represented, r represents the number of alternative case, Jl,kRepresent k-th of alternative case
L-th of solution of example, JlRepresent l-th of solution of new problem, it is preferable that wk=Simk, (k=1,2 ..., r).
In the present embodiment, as shown in Fig. 2 because the alternative case retrieved is generally not exclusively consistent with practical problem, needing
The solution for the new problem to be obtained to calculating is evaluated, and to judge whether to need to be modified the new problem, is corrected
Afterwards so that the solution of new problem does not deviate by correct solution too much, so, just it can apply among production.
In the present embodiment, amendment refers to be adjusted the solution of new problem so that the solution of the new problem after adjustment is close to just
Really solve.
In the present embodiment, as shown in Fig. 2 the solution for the new problem that can be obtained according to default evaluation criterion to calculating
(scheme evaluation) is evaluated, if the solution for calculating the obtained new problem meets evaluation criterion, by the new problem and institute
The solution for stating new problem constitutes a new case, is directly entered case storage link;If calculating the solution of the obtained new problem not
Evaluation criterion is met, then the new problem is modified, is modified during amendment using the rule in default rule system
(early stage can be provided with manual amendment, later stage by regulation engine).
In the present embodiment, after amendment, the new problem is reapplied in production, such as obtains satisfactory result (that is, amendment
The solution of the new problem afterwards meets evaluation criterion), Case-based adaptation terminates;As do not obtained satisfactory solution, then entered by man-machine interface
Terminate after row amendment.
In the present embodiment, in case storage ring section, determine whether new case stores using the method based on similarity, specifically
, it can be judged by following rule:
If max (sim (M, Mi))<α then store this case
If max (sim (M, Mi))>α then do not store this case
Wherein, max (sim (M, Mi)) new case and the highest similarity of case in case library are represented, M is new case, Mi
(i=1,2 ... K) are i-th of case in case library, and K is the number of case in case library, and α is the second predetermined threshold value.
Case library in the present embodiment is the study work(in the main repository in case-based reasoning system, case-based reasoning system
Can be constantly toward increasing new case in case library.When case library constantly increases, have the advantage that and be easy to find out phase
With case or similar cases, number of times and the time of amendment are reduced.In general knowledge base is bigger, and knowledge is abundanter, so, case
The problem of inference system can solve more, embodies its level of intelligence.But along with the continuous expansion of case library, it can cause
The retrieval time of similar cases greatly increases, and retrieval time is increasingly longer, so as to have impact on the ability of case-based reasoning system, thus
The ability and efficiency of case-based reasoning system are triggered.The maintenance of case-based reasoning system case library needs to realize that daily record stores work(
, single case, whole case library, the maintenance information of whole case-based reasoning system can be collected, case is solved by log analysis
Example recall precision problem.
In the present embodiment, the daily record of single case includes:Record number of times, time of failure application of each case successful application
Number, evaluation of estimate, last adjustment result, as shown in table 1;The daily record of whole case library includes:Monitor size, the record of case library
The Case No. that calls daily and the time for calling case library;The maintenance daily record of whole case-based reasoning system includes:Whole case is pushed away
Whether the maintenance time of reason system, maintenance mode, triggering reason, expert operate.
The daily record of the single case of table 1
Case No | Case number of success | The case frequency of failure | Evaluation of estimate | The floating concentrate grade of last time adjustment | The floating tailings grade of last time adjustment |
CaesNo. | CaseSucceedNum | CaseFailNum | Valuation | last_fujing | last_fuwei |
1 | 4 | 1 | 0.76 | 68.6 | 15.7 |
2 | 3 | 0 | 0.64 | 67.9 | 16.8 |
3 | 7 | 1 | 0.92 | 68.3 | 17.3 |
In the present embodiment, institute's evaluation values refer to the evaluation to case adjustment result;The computational methods of evaluation of estimate are:
When adjustment failure, then evaluation of estimate Valuation is 0.
Essence floating tail is floated after adjustment in reasonable interval, then by following method Calculation Estimation value Valuation:
Valuation=0.3 × Vlast+0.7×Vthis
Wherein, VlastRepresent the evaluation of estimate of last adjustment, VthisThe evaluation of estimate of this adjustment is represented, Valuation is represented
General comment is worth, γjkAnd γwkFor current floating concentrate grade and the actual value of floating tailings grade, γ 'jkIdeal value for floating concentrate grade (can
Configuration), γj_maxAnd γj_minTo float concentrate grade bound (configurable), γw_maxThe upper limit allowed to float tailings grade (can match somebody with somebody
Put), adjust the floating weight (configurable) of concentrate grade and floating tailings grade in evaluation with α.
In the present embodiment, maintenance time is divided into:Regularly handle, conditional processing, the processing under particular requirement.It is all
Log content can be obtained by back-stage management interface queries.Log system includes case library daily record, case-based reasoning system
Safeguard daily record, case daily record etc..
In the present embodiment, the redundancy of case library can also be found and handled by different similarity threshold values, so as to
Selectively to delete the unnecessary similar cases for meeting threshold values, it is ensured that case library possesses higher coverage;Reduce simultaneously
The amount of calculation of similarity in case library maintenance process.Its flow is:(1) system is found for a long time using the extremely low case of rate automatically
Example, is deleted, deletion record is stored in daily record.(2) the inconsistent case (periodic maintenance) of system discovery two, two cases
The Adjusted Option (adjustment result) that example is drawn is entirely different, then two cases is prompted into scheduling together with their Adjusted Option
Personnel, by its amendment.(3) case valuation value substantially reduces (excursion of setting evaluation of estimate) after system discovery adjustment, by phase
Close case and be prompted to dispatcher's amendment, improve the value of case.(4) when there is adjusting unsuccessful case, by new problem
Dispatcher is pointed out, is supplemented in time if the representative case for lacking problems in case library.
In the present embodiment, for a further understanding of the present embodiment, the typical process produced with metallurgical mine --- flotation work
Exemplified by sequence quality management and control problem, the case-based reasoning system building method towards metallurgical mine field is described in detail:
It is assumed that flotation circuit have altogether including:Three systems (1C, 2C, 3C), each system is divided into two loops of A, B, therefore
The data source of flotation case-based reasoning system has six loops, it is necessary to make a distinction.Each loop is required for independent progress case
Reasoning, can share a case library, it is assumed that the case library is flotation production wisdom operation case library, but each loop
Data are separated, and ore dressing plant floating operation can be described with structure shown in Fig. 3.The case towards metallurgical mine field
The specific steps of example inference system building method can include:
The first step, determine case in flotation production wisdom operation case library structure (Expressive Features of problem and problem
Solution).The data extracted are needed to have in flotation link:
1) Expressive Features of problem include but is not limited to:Floating concentrate grade, floating tailings grade, mixed magnetic concentrate grade, granularity, raw ore product
Position, ferrous, optional sex index, concentration;
2) solution of problem includes but is not limited to:The smart flow of the thick flows of LKY, LKY, starch flow, CaO flows.
Current data with existing composition flotation production wisdom peration data warehouse (by taking the system A loops of flotation one as an example), forms
Case structure.
Second step, extraction represents case.The structure of case in wisdom operation case library is produced according to flotation, from flotation production
Initial data is extracted in wisdom peration data warehouse to carry out after k-means clusters (for example, cluster number k value is 20), is obtained
Number of cases of appearing in court is more than 50 a total of 9 class of classification, takes the cluster centre of each class as case is represented, forms initial case, just
The structure of beginning case is as shown in table 2:
The structure of the initial case of table 2
3rd step, builds flotation production wisdom operation case library and sets Expressive Features weight.In hadoop databases
Flotation production wisdom operation case library is set up in (Hadoop Database, HBase), and the representative case of extraction is put in storage, then
The weights formulated according to expert set the weighted value of Expressive Features.Temporarily the weighted value of the Expressive Features of the data of nothing is 0 at present.Power
The scope of weight is 0 to 1, and summation is 1.
In the present embodiment, the hadoop databases are a kind of distributed memory systems.
4th step, the similarity to new problem in flotation is retrieved:The prison at current time is pushed to case-based reasoning system
Data are surveyed, if the excursion of the characteristic value of certain in Monitoring Data exceeds default range threshold, case-based reasoning system is according to case
The Monitoring Data is converted into the description of new problem by the structure of example.New problem data are as shown in table 3 in flotation:
New problem data in the flotation of table 3
In the present embodiment, case in new problem and case library is subjected to Similarity Measure one by one, with No. 1 case in case library
Exemplified by, No. 1 case data is as shown in table 4:
No. 1 case data in the case library of table 4
Single Expressive Features Similarity Measure:
sim(f1,f1,1)=1- | 26.36-25.22 |/max (26.36,25.22)=0.9568
sim(f2,f1,2)=1- | 68.33-68.69 |/max (68.33,68.69)=0.9977
sim(f3,f1,3)=1- | 6.505-5.95 |/max (6.505,5.95)=0.9147
sim(f4,f1,4)=1- | 47.40-46.49 |/max (47.40,46.49)=0.9808
sim(f6,f1,6)=1- | 16.72-16.08 |/max (16.72,16.08)=0.9617
sim(f7,f1,7)=1- | 47.39-47.82 |/max (47.39,47.82)=0.9910
The Similarity Measure of new problem and No. 1 case:
The present embodiment, the Expressive Features of problem are 7, that is to say, that m is 7, but is due to Expressive Features f4,f1,4Between
There is no similarity, so calculating sim (D, Dk), m values are also rational for 6.
Thus the similarity for calculating new problem and No. 1 case is 0.9727.New problem and each case have been calculated one by one
Similarity after, descending sequencing display is carried out to case according to similarity, this retrieval without matching (similarity completely
More than case 99%), then choose similarity and be more than the alternative case that 80% case is reused as case.
5th step, the reuse of flotation case.The alternative case that similarity is more than 80% is chosen after Case Retrieval, according to
Above-mentioned formula calculates the solution of new problem, and the solution of this new problem is pushed into execute-in-place workman and referred to.
6th step, the evaluation and amendment of flotation case.The solution of new problem after site operation personnel reuses according to case
Carry out the adjustment of production operation amount, it is assumed that after the stipulated time, the Grade change for floating the floating tail of essence is 67.9 and 15.3.It is special by scheduling
Family sets the bound (evaluation criterion) of floating concentrate grade and floating tailings grade in background interface, it is assumed that floating concentrate grade lower limit γj_min
=67.5, float concentrate grade upper limit γj_max=68.9, float the upper limit γ that tailings grade allowsw_max=25.
It is possible thereby to judge:γj_min<67.9<γj_max and 15.3≤γw_max
It is that, in reasonable interval, new problem and its solution can be constituted into new case to illustrate the floating tailings grade of floating essence.
It is assumed that detecting floating smart or floating tail not in reasonable interval after adjustment, then this new problem is prompted to dispatcher
The amendment solved, then site operation personnel is pushed to again, hereafter judge the same above-mentioned steps of evaluation procedure.
7th step, stores flotation new case.New problem is adjusted with after evaluation procedure, obtains new case.By calculating,
New case is 0.89 with the maximum similarity that case is had in case library, it is assumed that the second predetermined threshold value is 0.85, then meets new case
The condition of example storage, this case is added in case library and numbered.
8th step, writes daily record.The daily record of single case needs to record the number of times of each case successful application, unsuccessfully applied
Number of times, evaluation of estimate, last adjustment result (needing to store result how long);The daily record of whole case library is mainly prison
Case No. and allocating time that size, record depending on case library are called daily;The maintenance daily record of whole CBR system, including safeguard
Whether time, maintenance mode, triggering reason, expert operate.
The present embodiment realizes to be clustered by initial data and forms initial case, initial case storage is stored as into case library
(or knowledge base), event triggering (if the excursion of certain characteristic value exceeds default range threshold in the Monitoring Data)
Afterwards, the similarity retrieval of new problem and sequence, can also push, show highest similarity case and at back-stage management interface
The Core Feature of the case-based reasoning systems such as weight is set, a solution is provided for bargh's dynamic scheduling problem.
Described above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, on the premise of principle of the present invention is not departed from, some improvements and modifications can also be made, these improvements and modifications
It should be regarded as protection scope of the present invention.
Claims (10)
1. a kind of case-based reasoning system building method towards metallurgical mine field, it is characterised in that including:
Case library is generated, wherein, each case in case library is the structural description to a problem in application field;Each
Case includes:The Expressive Features of problem and the solution of problem;
Monitoring Data is obtained, if the excursion of certain characteristic value exceeds default range threshold, basis in the Monitoring Data
The Monitoring Data is converted into new problem by the structure of case;
Calculate the similarity between each case in the new problem and the case library, if the case library not with it is described new
The case that problem is matched completely, then choose similarity from the case library and be used as case weight more than the case of the first predetermined threshold value
Alternative case;
The alternative case reused according to obtained case, calculates the solution of the new problem;
The solution of the new problem obtained according to default evaluation criterion to calculating is evaluated;
If the solution for calculating the obtained new problem meets evaluation criterion, the solution of the new problem and the new problem is constituted
One new case, judges whether the new case and the highest similarity of case in the case library are more than the second predetermined threshold value,
If more than the second predetermined threshold value, by new case storage into the case library.
2. the case-based reasoning system building method according to claim 1 towards metallurgical mine field, it is characterised in that institute
Stating generation case library includes:
According to the Expressive Features of problem and the solution of problem, corresponding original number is extracted from the data warehouse pre-set
According to;
The initial data to extraction is clustered, and using the cluster centre of each class as the representative case of respective classes, is carried
The case that represents is taken to enter case library.
3. the case-based reasoning system building method according to claim 1 or 2 towards metallurgical mine field, its feature exists
In the generation case library includes:
According to the Expressive Features of problem and the solution of problem, adjustment case is formulated;
According to evaluation result of the expert to the adjustment case, it is that excellent adjustment case enters case library to choose evaluation result.
4. the case-based reasoning system building method according to claim 1 towards metallurgical mine field, it is characterised in that institute
Stating the similarity calculated in the new problem and the case library between each case includes:
Similarity in the new problem and the case library between each case, first formula are calculated by the first formula
It is expressed as:
Wherein, sim (D, Dk) represent new problem D and kth bar case D in case librarykBetween similarity;fiRepresent new problem D's
I-th of Expressive Features;M represents the number of Expressive Features;fI, kRepresent kth bar case D in case librarykI-th of Expressive Features;
sim(fi, fI, k) represent new problem D i-th of Expressive Features fiWith kth bar case D in case librarykI-th of Expressive Features fI, k
Between similarity;ωiRepresent the weight of i-th of Expressive Features;K represents the number of case in case library.
5. the case-based reasoning system building method according to claim 4 towards metallurgical mine field, it is characterised in that if
fi, fI, kFor continuous variable, then sim (fi, fI, k) be expressed as:
sim(fi, fI, k)=1- | fi-fI, k|/max(fi, fI, k)。
6. the case-based reasoning system building method according to claim 4 towards metallurgical mine field, it is characterised in that if
fi, fI, kFor discrete variable, then sim (fi, fI, k) be expressed as:
7. the case-based reasoning system building method according to claim 1 towards metallurgical mine field, it is characterised in that institute
The alternative case reused according to obtained case is stated, calculating the solution of the new problem includes:
The alternative case reused according to obtained case, obtains the solution J=(J of the new problem1, J2..., Jl), wherein, JlRepresent
For:
Wherein, wkThe weights of k-th of alternative case are represented, r represents the number of alternative case, Jl,kRepresent k-th alternative case
L-th of solution, JlRepresent l-th of solution of new problem.
8. the case-based reasoning system building method according to claim 7 towards metallurgical mine field, it is characterised in that wk
=Simk;
Wherein, SimkRepresent the similarity between k-th of alternative case and the new problem.
9. the case-based reasoning system building method according to claim 1 towards metallurgical mine field, it is characterised in that institute
Stating method also includes:
If the solution for calculating the obtained new problem is unsatisfactory for evaluation criterion, the new problem is modified;
The solution of the revised new problem is evaluated according to default evaluation criterion.
10. the case-based reasoning system building method according to claim 1 towards metallurgical mine field, it is characterised in that
Methods described also includes:
Collect single case, the daily record of whole case library, the maintenance daily record of whole case-based reasoning system;
Wherein, the daily record of single case includes:Record the number of times of each case successful application, the number of times of failure application, evaluation of estimate,
Last time adjustment result;
The daily record of whole case library includes:The size of case library is monitored, the Case No. called daily is recorded and calls case library
Time;
The maintenance daily record of whole case-based reasoning system includes:The maintenance time of whole case-based reasoning system, maintenance mode, triggering are former
Whether cause, expert operate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710196559.XA CN107038481A (en) | 2017-03-29 | 2017-03-29 | A kind of case-based reasoning system building method towards metallurgical mine field |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710196559.XA CN107038481A (en) | 2017-03-29 | 2017-03-29 | A kind of case-based reasoning system building method towards metallurgical mine field |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107038481A true CN107038481A (en) | 2017-08-11 |
Family
ID=59534565
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710196559.XA Pending CN107038481A (en) | 2017-03-29 | 2017-03-29 | A kind of case-based reasoning system building method towards metallurgical mine field |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107038481A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108491444A (en) * | 2018-02-12 | 2018-09-04 | 龙马智芯(珠海横琴)科技有限公司 | The generation method and device of solution |
CN112765133A (en) * | 2020-12-25 | 2021-05-07 | 广东电网有限责任公司电力科学研究院 | Maintenance system and maintenance method for PAS case base |
CN113642733A (en) * | 2021-10-19 | 2021-11-12 | 矿冶科技集团有限公司 | Case reasoning and matching method for gene mineral separation process |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101660969A (en) * | 2009-09-25 | 2010-03-03 | 北京工业大学 | Intelligent fault diagnosis method for gear box |
CN102243628A (en) * | 2010-05-12 | 2011-11-16 | 电子科技大学 | Mineralizing case reasoning model and method |
CN104318313A (en) * | 2014-09-19 | 2015-01-28 | 东北大学 | Intelligent decision system and method for ore dressing total flow operation index on the basis of case-based reasoning |
CN104503920A (en) * | 2015-01-23 | 2015-04-08 | 桂林电子科技大学 | Software system fault detection method based on dynamic description logic and case-based reasoning |
CN106292292A (en) * | 2016-10-17 | 2017-01-04 | 鞍钢集团矿业有限公司 | The floatation of iron ore dosing Optimal Setting method and system of case-based reasioning |
-
2017
- 2017-03-29 CN CN201710196559.XA patent/CN107038481A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101660969A (en) * | 2009-09-25 | 2010-03-03 | 北京工业大学 | Intelligent fault diagnosis method for gear box |
CN102243628A (en) * | 2010-05-12 | 2011-11-16 | 电子科技大学 | Mineralizing case reasoning model and method |
CN104318313A (en) * | 2014-09-19 | 2015-01-28 | 东北大学 | Intelligent decision system and method for ore dressing total flow operation index on the basis of case-based reasoning |
CN104503920A (en) * | 2015-01-23 | 2015-04-08 | 桂林电子科技大学 | Software system fault detection method based on dynamic description logic and case-based reasoning |
CN106292292A (en) * | 2016-10-17 | 2017-01-04 | 鞍钢集团矿业有限公司 | The floatation of iron ore dosing Optimal Setting method and system of case-based reasioning |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108491444A (en) * | 2018-02-12 | 2018-09-04 | 龙马智芯(珠海横琴)科技有限公司 | The generation method and device of solution |
CN112765133A (en) * | 2020-12-25 | 2021-05-07 | 广东电网有限责任公司电力科学研究院 | Maintenance system and maintenance method for PAS case base |
CN113642733A (en) * | 2021-10-19 | 2021-11-12 | 矿冶科技集团有限公司 | Case reasoning and matching method for gene mineral separation process |
CN113642733B (en) * | 2021-10-19 | 2022-02-15 | 矿冶科技集团有限公司 | Case reasoning and matching method for gene mineral separation process |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107038481A (en) | A kind of case-based reasoning system building method towards metallurgical mine field | |
CN103400214B (en) | Based on the voltage dip prediction analysis method of multi-dimension multi-level association rules | |
CN109766921A (en) | A kind of vibration data Fault Classification based on depth domain-adaptive | |
CN110263172A (en) | A kind of evented autonomous classification method of power system monitor warning information | |
CN103544379B (en) | Landed Typhoon similarity analysis method based on GIS | |
CN106056252A (en) | Shale gas field well group division three-level optimization method | |
CN104933505A (en) | Decision and evaluation method for intelligent power distribution network group based on fuzzy assessment | |
CN111636891B (en) | Real-time shield attitude prediction system and construction method of prediction model | |
CN109902111A (en) | A kind of earth and rockfill dam security information method for digging | |
CN112734218B (en) | River health evaluation method and device based on river basin | |
CN107657369A (en) | One kind carries out flood control emergency response based on intelligent event category and knows method for distinguishing | |
CN102679935A (en) | System and method for calculating icing thickness of power transmission line | |
CN106227165B (en) | A kind of intelligence manufacture management system | |
Eseye et al. | Short-term wind power forecasting using a double-stage hierarchical hybrid GA-ANFIS approach | |
CN108448721A (en) | The method of electric system main station monitoring signal and standard signal Auto-matching | |
CN107909251A (en) | A kind of cigarette machine overall efficiency analysis of Influential Factors method and apparatus | |
CN107463620A (en) | A kind of elevator accident early-warning and predicting system based on data mining | |
CN113268370A (en) | Root cause alarm analysis method, system, equipment and storage medium | |
CN108205727B (en) | Digital workshop process decision method based on decision tree and expert system | |
CN104484728B (en) | A kind of power grid security comprehensive index system framework method | |
CN106899014A (en) | A kind of modeling and forecasting method of the electric railway load decomposed based on waveform | |
CN104318313B (en) | Intelligent decision system and method for ore dressing total flow operation index on the basis of case-based reasoning | |
CN105303470A (en) | Electric power project planning and construction method based on big data | |
CN110069773A (en) | A kind of security risk intelligent identification Method, device and storage medium | |
CN105184318A (en) | Mixing attribute data flow cluster method based on reinforcement cluster edge detection of grid |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170811 |
|
RJ01 | Rejection of invention patent application after publication |