CN113011518B - Aluminum electrolysis cell condition health degree classification method based on combined weighted naive Bayes - Google Patents
Aluminum electrolysis cell condition health degree classification method based on combined weighted naive Bayes Download PDFInfo
- Publication number
- CN113011518B CN113011518B CN202110351099.XA CN202110351099A CN113011518B CN 113011518 B CN113011518 B CN 113011518B CN 202110351099 A CN202110351099 A CN 202110351099A CN 113011518 B CN113011518 B CN 113011518B
- Authority
- CN
- China
- Prior art keywords
- aluminum electrolysis
- electrolysis cell
- health
- aluminum
- sample
- 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.)
- Active
Links
- 229910052782 aluminium Inorganic materials 0.000 title claims abstract description 89
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 77
- 230000036541 health Effects 0.000 title claims abstract description 60
- 238000005868 electrolysis reaction Methods 0.000 title claims abstract description 55
- 238000004519 manufacturing process Methods 0.000 claims abstract description 28
- 238000011156 evaluation Methods 0.000 claims abstract description 25
- 238000013145 classification model Methods 0.000 claims abstract description 16
- 238000004364 calculation method Methods 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 31
- 230000008569 process Effects 0.000 claims description 23
- 230000000694 effects Effects 0.000 claims description 17
- 239000003792 electrolyte Substances 0.000 claims description 15
- 239000007788 liquid Substances 0.000 claims description 11
- AZDRQVAHHNSJOQ-UHFFFAOYSA-N alumane Chemical compound [AlH3] AZDRQVAHHNSJOQ-UHFFFAOYSA-N 0.000 claims description 10
- 238000012360 testing method Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 5
- 201000010099 disease Diseases 0.000 claims description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 abstract description 8
- 238000004458 analytical method Methods 0.000 abstract description 4
- 239000013598 vector Substances 0.000 description 8
- 230000007547 defect Effects 0.000 description 6
- 239000000463 material Substances 0.000 description 6
- 238000009826 distribution Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 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 1
- 206010037660 Pyrexia Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 201000009240 nasopharyngitis Diseases 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Electrolytic Production Of Metals (AREA)
Abstract
The invention relates to the technical field of aluminum electrolysis cell production, in particular to a method for classifying the health degree of aluminum electrolysis cell conditions based on combined weighted naive Bayes. According to the invention, common types of the aluminum electrolysis cell and the generated reasons and phenomena are analyzed, an aluminum electrolysis mechanism is used for building an aluminum electrolysis cell condition health evaluation index system, then a naive Bayesian classifier is weighted in a manner of combining a hierarchical analysis method and an entropy weighting method, a combined weighted Bayesian classifier-based aluminum electrolysis cell condition health classification model is built, the refinement, the rapidness and the automatic identification of the health of the aluminum electrolysis cell can be realized, and the subjectivity and the slowness problems in the traditional manual judgment are solved. The weights are obtained by combining the two modes, the unamplified Bayesian attribute condition independence assumption is weakened, the real difference between indexes is enhanced, the problem that the numerical calculation weight error is large when the sample size is small or the number is missing is solved, and the weights are more accurate and have practical significance.
Description
Technical Field
The invention relates to the technical field of aluminum electrolysis cell production, in particular to a method for classifying the health degree of aluminum electrolysis cell conditions based on combined weighted naive Bayes.
Background
In the aluminum electrolysis production process, the health condition of the aluminum electrolysis cell is related to whether the whole aluminum electrolysis production is stable and efficient to operate, and is similar to the characteristics that common cold, fever and the like can occur when the balance of a human body is broken, the health condition of the aluminum electrolysis cell is also used for analyzing the conditions of material balance, energy balance and the like in the electrolytic cell, the running state of the electrolytic cell can be judged, when the material balance and the energy balance in the electrolytic cell reach the stable state, the electrolytic cell has low effect and controllable effect, the more the fluctuation of the aluminum liquid is gentle, the higher the current efficiency is, and the better the health state of the electrolytic cell is; conversely, when the balance of the electrolytic tank is broken, the conditions of partial cold and partial heat, frequent anode effect, severe fluctuation of aluminum liquid and the like can occur, the energy consumption of the electrolytic tank is increased, the current efficiency is reduced, the service life of the electrolytic tank is shortened, and meanwhile, the burden of manual operation is increased, so that the analysis of the health condition of the aluminum electrolytic tank plays a vital role in the whole industrial process.
In the actual aluminum electrolysis production process, the aluminum electrolysis condition is often roughly judged by a technical worker through single data or single phenomenon, but the mode has the following defects that firstly, the aluminum electrolysis mechanism is complex, a strong coupling relation exists between data, multiple factors possibly have the same effect on a certain item of data, and if a large number of data features cannot be combined for analysis, serious production accidents are easily caused because of inaccurate judgment. Second, during the production process, there are often data changes and special phenomena caused by normal operation. For example, when an operator performs operations such as lifting a bus bar and changing an anode, a series of data changes are temporarily caused, and the data are normally restored to normal levels after a period of time. This makes it easy for inexperienced workers to misjudge the data changes caused by normal production operations and the special phenomena that occur as unhealthy tank conditions, thereby affecting normal production. Thirdly, the subjectivity and the slowness exist in manual judgment, the manual judgment not only increases the labor cost, but also the judgment method is too dependent on manual experience, the problems of manpower shortage and the like exist in the current aluminum electrolysis workshop, and the human resources can be well saved by automatically identifying the health state of the electrolytic cell through a computer. In practice, the health condition of the aluminum electrolysis cell can be reflected by some data indexes, such as swing, needle vibration, anode effect coefficient, voltage deviation, electrolyte temperature and the like, and the health degree of the aluminum electrolysis cell is poor, which indicates that the aluminum electrolysis cell has poor regulation capability, material balance and energy balance are difficult to stabilize, the effect is easy to generate, the fluctuation of aluminum liquid is more severe, and technicians can take measures according to different types of health degree in the production process, so that the whole control of the health condition of the aluminum electrolysis cell has important significance for improving the production efficiency and ensuring the production safety.
Currently, little research is done on the health status of aluminum electrolysis cells, mainly by indirectly analyzing the conditions of the aluminum electrolysis cells through some production indexes of the electrolysis cells, such as anode current fluctuation characteristics, cell voltage and cell resistance signals, superheat degree and the like. The methods only study the state of a single tank, but do not deeply study the health degree by combining a plurality of tank conditions, and the comprehensive reflection of the whole tank health condition is difficult. In order to judge the energy balance and material balance of the cell conditions, the health condition of the electrolytic cell can be indirectly analyzed by analyzing the cold cell, the hot cell, the anode effect and the fluctuation of aluminum liquid. If the fluctuation of the aluminum liquid can indirectly reflect the energy balance condition to a certain extent, the effect occurrence condition can indirectly reflect the material balance condition to a certain extent, and the like.
Aiming at the problems, how to effectively analyze information according to mass production historical data and relevant experience knowledge to judge the health condition of the electrolytic tank is researched, and the method can be stably applied to actual production for a long time, and has important significance for realizing stable and efficient operation of the aluminum electrolytic tank, improving production efficiency and product quality and ensuring production safety.
Disclosure of Invention
Based on the technical problems, the invention aims at the technical problems, from the standpoint of material balance and energy balance, constructs an aluminum electrolysis cell condition health degree evaluation index system, puts forward a combined weighted naive Bayes classification method on the basis of a naive Bayes classifier to construct a classification model, and obtains the classification for the aluminum electrolysis cell condition health through training.
The invention provides a combined weighted naive Bayes-based aluminum electrolysis cell condition health degree classification method, which specifically comprises the following steps:
constructing an aluminum electrolysis cell condition health degree evaluation index system according to the types of aluminum electrolysis cells and the generated reasons and phenomena;
according to the evaluation index system, carrying out combined weighting on the naive Bayes classifier by an analytic hierarchy process and an entropy weighting process, and constructing an aluminum electrolysis cell condition health classification model based on the combined weighted Bayes classifier;
the historical evaluation index data and the corresponding cell condition health categories are acquired, and the aluminum cell condition health classification model is trained to obtain a classification model which is used for classifying the cell condition health of the aluminum cell.
Further, the evaluation index system comprises a disease groove type and an evaluation index;
the types of the disease grooves comprise a cold groove, a hot groove, a multi-effect groove and an aluminum liquid fluctuation groove;
the evaluation index includes: aluminum level, electrolyte temperature, coefficient of effect, cell voltage, swing, molecular ratio, needle vibration.
Further, the analytic hierarchy process specifically includes:
constructing a plurality of judgment matrixes according to the evaluation index system, comparing the judgment matrixes in pairs, and assigning values to the judgment matrixes according to the importance degree of indexes in the judgment matrixes;
consistency of the judgment matrix is performed according to the formula cr=ci/RIChecking, wherein CR is the random consistency ratio of the judgment matrix; CI is the consistency index of the judgment matrix, CI= (lambda) max -n)/(n-1)，λ max Judging the maximum characteristic root of the matrix; RI is the average random consistency index of the judgment matrix;
according to the formula rw=λ max Calculating weight of W, wherein W is a characteristic value lambda max Is described.
Further, the entropy weighting method specifically includes:
the production data is formulatedColumn normalization processing, x ij For the value of the jth attribute in the ith sample, min (x j ) For the minimum value of the j-th attribute in all samples, max (x j ) The maximum value of the j-th attribute in all samples;
according to the formulaCalculating information entropy of data, n represents the number of samples, and p ij Representing the probability of occurrence of the jth attribute value in the ith sample, < >>
The weight coefficient calculation of each attribute is according to the formulaAnd (5) calculating.
Further, the weight calculation formula of the combined weighting is as follows:
w＝αw 1 +βw 2
wherein w is the final combining weight, w 1 Representing the weight calculated by analytic hierarchy process, alpha represents subjective preference coefficient, w 2 Representing the weight obtained by the entropy weighting method, and β representing the objective preference coefficient, and having α=1 to β.
Further, the aluminum electrolysis cell condition health classification model based on the combined weighted Bayes classifier is as follows:
wherein θ represents the last classification result of the sample to be classified, c represents the number of classes, y k For the kth category, n k For the number of samples belonging to class k, N is the number of samples of sample space X, w ij For the combined weight obtained by combining the analytic hierarchy process and the entropy weight process, x ij Valuing the jth attribute of the ith test sample, u ij (k) For training the sample mean value (sigma) of the jth attribute in the kth sample ij (k) ) 2 Sample variance for the j-th attribute in the K-th sample class is trained.
The beneficial effects are that:
(1) According to the method, the types of the diseased cells and the generated reasons and phenomena are analyzed, a layered aluminum cell condition health evaluation index system is built by combining the mechanism knowledge of the aluminum cell, and relevant characteristics are extracted according to a layered mechanism model.
(2) The invention weakens the independent assumption of the naive Bayes attribute condition, enhances the real difference between indexes by absorbing expert experience, solves the problem of larger error caused by numerical calculation weight when the sample size is smaller or the sample is missing, avoids the defects of the two methods by combining the weights obtained by the two methods, and ensures that the weight is more accurate and has practical significance.
(3) According to the invention, the knowledge of the aluminum electrolysis mechanism is combined with the experience of an expert, and the intelligent learning model of the cell health degree is constructed based on a large amount of actual production data, so that the refinement, the rapidness and the automatic identification of the cell health degree can be realized, the subjectivity and the slowness problems in the traditional manual judgment are solved, the labor cost is saved, the calculation complexity is low, and the memory requirement is less; in actual production, technicians can take measures according to the judged health types, and the method has important significance for realizing stable and efficient operation of the aluminum electrolysis cell, improving production efficiency and product quality and ensuring production safety.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for classifying the health of aluminum electrolysis cells based on combined weighted naive Bayes provided by an embodiment of the invention;
FIG. 2 is a graph of the relationship between the health evaluation index systems of the aluminum electrolysis cell provided by the embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
As shown in fig. 1, in the embodiment of the present invention, a flow chart of a method for classifying the health of aluminum electrolysis cell conditions based on a combined weighted naive bayes is provided, which specifically includes the following steps:
step S101, constructing an aluminum electrolysis cell condition health evaluation index system according to the types of the aluminum electrolysis cell and the generated reasons and phenomena.
In the embodiment of the invention, the types of the aluminum electrolysis cell include: the method comprises the steps of cold tank, hot tank, multi-effect tank, aluminum liquid fluctuation tank and the like, analyzing the reasons of the unhealthy tank conditions and phenomena occurring during the production, building an aluminum electrolysis tank condition health degree evaluation index system through knowledge of an aluminum electrolysis mechanism, wherein the indexes selected in the embodiment of the invention are aluminum level, electrolyte temperature, effect coefficient, tank voltage, swing, molecular ratio and needle vibration, and the built aluminum electrolysis tank condition health degree evaluation index system is shown in figure 2.
Step S102, carrying out combined weighting on the naive Bayes classifier according to the evaluation index system and by a hierarchical analysis method and an entropy weighting method, and constructing an aluminum electrolysis cell condition health classification model based on the combined weighted Bayes classifier.
Aiming at the defect that the weight is calculated by only utilizing the association degree between the data indexes and the information quantity provided by each index, in the embodiment of the invention, the weight is obtained by combining a subjective method and an objective method, namely the weight is obtained by fusing data knowledge and mechanism knowledge, the intention of a decision maker is reflected by the weight determined by a subjective weighting method, the importance degree of each index is judged according to the theory and experience of a senior expert, and corresponding assignment is obtained, so that the method has a certain subjectivity; the objective weighting method mainly depends on complete mathematical theory and method, obtains weight by learning data sample distribution, association degree and the like from objective data, ignores subjective information of a decision maker, and ignores the actual condition without considering the difference of indexes. The weights obtained in the two modes are combined, so that the defects of the two methods are avoided, and the calculated weights are more accurate and have better interpretability. In the method, a subjective weighting method adopts an analytic hierarchy process, corresponding weights are obtained by adopting a pairwise comparison method through a plurality of experts with deeper seniority, and consistency is calculated by utilizing a judgment matrix to obtain the weights. The objective weighting method adopts an entropy weighting method, and determines objective weights according to the size of index variability.
In the embodiment of the invention, the analytic hierarchy process specifically comprises the following steps:
step S201, a plurality of judgment matrixes are constructed according to the evaluation index system, the judgment matrixes are compared in pairs, and the judgment matrixes are assigned according to the importance degree of indexes in the judgment matrixes.
In the embodiment of the invention, a judging matrix R is respectively constructed by a cold tank, a hot tank, a multi-effect tank and an aluminum liquid fluctuation tank 1 Constructing a judgment matrix R by using the aluminum level, the electrolyte temperature, the cell voltage and the molecular ratio 2 Constructing a judgment matrix R by using an effect coefficient and a slot voltage 3 Constructing a judgment matrix R by using aluminum level, electrolyte level, cell voltage, swing and needle vibration 4 The expert with deeper qualification is obtained by adopting a pairwise comparison method, and a scale R is given to the judgment matrix according to the importance degree of the index ij (on a 1-9 scale) R ij Representing the importance of index i relative to j, wherein matrix R is determined 1 、R 2 、R 3 、R 4 As shown in tables 1, 2 and 3, respectively.
TABLE 1 matrix R 1
| R 1 | Cold/hot tank | Multi-effect trough | Aluminum liquid fluctuation tank |
| Cold/hot tank | 1 | 1/2 | 1/2 |
| Multi-effect trough | 2 | 1 | 1 |
| Aluminum liquid fluctuation tank | 2 | 1 | 1 |
TABLE 2 matrix R 2
| R 2 | Aluminum level | Electrolyte level | Electrolyte temperature | Cell voltage | Molecular ratio |
| Aluminum level | 1 | 1 | 1/3 | 1/2 | 1/3 |
| Electrolyte level | 1 | 1 | 1/3 | 1/2 | 1/3 |
| Electrolyte temperature | 3 | 3 | 1 | 2 | 2 |
| Cell voltage | 2 | 2 | 1/2 | 1 | 1 |
| Molecular ratio | 3 | 3 | 1/2 | 1 | 1 |
TABLE 3 matrix R 3
| R 3 | Coefficient of effect | Cell voltage |
| Coefficient of effect | 1 | 5 |
| Cell voltage | 1/5 | 1 |
TABLE 4 matrix R 4
| R 4 | Aluminum level | Electrolyte level | Cell voltage | Swinging movement | Needle vibration |
| Aluminum level | 1 | 1 | 1/3 | 1/4 | 1/4 |
| Electrolyte level | 1 | 1 | 1/3 | 1/4 | 1/4 |
| Cell voltage | 3 | 3 | 1 | 1 | 1 |
| Swinging movement | 4 | 4 | 1 | 1 | 1 |
| Needle vibration | 4 | 4 | 1 | 1 | 1 |
Step S202, consistency test is carried out on the judgment matrix according to a preset method.
In the embodiment of the invention, the judgment matrix R is judged by adopting a formula (1) 1 、R 2 、R 3 、R 4 And (5) performing consistency test.
CR＝CI/RI (1)
Wherein CR is the random consistency ratio of the judgment matrix; CI is the consistency index of the judgment matrix, CI= (lambda) max -n)/(n-1)，λ max Judging the maximum characteristic root of the matrix; RI is the average random consistency index of the judgment matrix, and the value is shown in Table 5.
Table 5 average random uniformity index of judgment matrix
| Matrix order | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| RI | 0 | 0 | 0.58 | 0.9 | 1.12 | 1.24 | 1.32 | 1.41 | 1.45 | 1.49 |
R is calculated 1 、R 2 、R 3 、R 4 The uniformity ratio CR was less than 0.1, and R was considered to be 1 、R 2 、R 3 、R 4 The degree of inconsistency of (c) is within the allowable range, and satisfactory consistency is achieved, and consistency test is passed.
Step S203, according to the formula rw=λ max Calculating weight of W, wherein W is a characteristic value lambda max Is described.
In the embodiment of the invention, the weight obtained by the analytic hierarchy process is calculated by adopting the formula (2).
RW＝λ max W (2)
Wherein lambda is max Is the maximum characteristic root of R, W is the characteristic value lambda max Can be used as a weight vector.
In the embodiment of the present invention, the obtained weight vectors are shown in table 6.
TABLE 6 analytic hierarchy process right vector
| Index (I) | W 1 |
| Cell voltage | 0.21132 |
| Coefficient of effect | 0.33332 |
| Needle vibration | 0.11684 |
| Swinging movement | 0.11684 |
| Aluminum level | 0.05066 |
| Electrolyte level | 0.05066 |
| Temperature (temperature) | 0.07252 |
| Molecular ratio | 0.04794 |
The historical evaluation index data and the corresponding cell condition health categories are acquired, and the aluminum cell condition health classification model is trained to obtain a classification model which is used for classifying the cell condition health of the aluminum cell.
The analytic hierarchy process adopted by the invention carries out assignment of the judgment matrix by acquiring the importance judgment of the experienter on each index, and obtains the weight value of each index after the judgment matrix passes the consistency test to obtain the weight vector.
In the embodiment of the invention, the entropy weight method specifically comprises the following steps:
step S301, row and column normalization is performed on the data to obtain standardized data.
Performing row and column normalization processing on production data according to a formula (3):
wherein x is ij For the value of the jth attribute in the ith sample, min (x j ) For the minimum value of the j-th attribute in all samples, max (x j ) Is the maximum value of the j-th attribute in all samples.
Step S302, calculating the information entropy of the data according to the formula (4).
Wherein n represents the number of samples, p ij Representing the probability of occurrence of the j-th attribute value in the i-th sample,
in step S303, the weight coefficient calculation of each attribute is calculated according to formula (5).
Wherein k represents the number of attributes, and weight vectors obtained by the entropy weight method in the embodiment of the invention are shown in table 7.
Table 7 entropy weight normal weight vector
| Index (I) | W 2 |
| Cell voltage | 0.09881 |
| Coefficient of effect | 0.54967 |
| Needle vibration | 0.09695 |
| Swinging movement | 0.07010 |
| Aluminum level | 0.02913 |
| Electrolyte level | 0.12171 |
| Temperature (temperature) | 0.01351 |
| Molecular ratio | 0.02012 |
The entropy weight method adopted by the invention is an objective weight method, mainly depends on the complete backup mathematical theory and method, starts from objective data, learns and obtains weights by utilizing data sample distribution, association degree and the like, and can obtain corresponding weight vectors from the objective data.
In the embodiment of the invention, the weight determination method based on the combination weighting is calculated by adopting a formula (6).
w＝αw 1 +βw 2 (6)
Where w is the final combining weight, w 1 Representing the weight calculated by analytic hierarchy process, alpha represents the subjective preference coefficient, w2 represents the weight obtained by entropy weighting process, beta represents the objective preference coefficient, and has alpha=1-beta, taking alpha=0.5, beta=0.5 and w 1 And w 2 Substituting the combination weights w to obtain the calculation results are shown in table 8.
Table 8 combined weight vector
| Index (I) | W |
| Cell voltage | 0.15506 |
| Coefficient of effect | 0.44150 |
| Needle vibration | 0.10690 |
| Swinging movement | 0.09347 |
| Aluminum level | 0.03990 |
| Electrolyte level | 0.08618 |
| Temperature (temperature) | 0.04301 |
| Molecular ratio | 0.03403 |
In the prior art, for the bayesian classification method, the performance of the classifier is improved mainly through attribute weighting, for example, weighting parameters are learned from training data by using gain ratios, correlation relations, rough sets and the like. Most of researches in recent years are to obtain an optimal weight through an optimization algorithm, and although the classification accuracy of naive Bayes is continuously improved, some defects exist. First, the physical meaning of each data index in the production process is weakened through the weight calculation of the data, and the difference of the indexes is ignored. The method has wide applicability, but weak pertinence, and the physical interpretability of naive Bayes is greatly reduced by data calculation; second, the weights calculated from the data are not accurate enough for data sets with a smaller sample size or samples with more missing data. Third, for weight optimization with excessive redundancy attributes, if the attributes are not selected, the efficiency and complexity of the optimization algorithm will be increased. The conventional method also calculates the redundant relation among the attributes through the correlation relation among the data, so that the relation among the attributes and the class attributes is difficult to interpret, and the efficiency is low.
In the embodiment of the invention, a naive Bayes classification method with combined weighting is provided on the basis of a naive Bayes classifier, and aiming at the problem that the independent assumption of the naive Bayes model is difficult to be established in reality, the classification performance of the naive Bayes is improved by utilizing attribute weighting.
The naive Bayes classifier classifies samples to be classified into categories with maximum posterior probability according to posterior probability, and the model is as follows:
wherein θ represents the last classification result of the sample to be classified, c represents the number of classes, y k For the kth category, P (y k ) The prior probability of the kth category can be calculated from the proportion of samples belonging to the category, namely P (w k )＝n k N, nk is the number of samples belonging to class k, N is the number of samples of sample space X, P (X) i |y k ) For category label y k Time division sample X i Conditional probability of P (X) i ) Representing a sample as X i For all classes, P (X i ) The value of (2) is fixed, so the model can be simplified as:
to weaken the constraint of attribute condition independence assumption, P (X i |y k ) Can be decomposed into:
wherein m represents the number of attributes, w ij For the combining weights obtained by combining the analytic hierarchy process with the entropy weight process in table 8, a naive bayes classification model based on the combining weights can be obtained:
for a continuity attribute feature, the class conditional probability may be represented by a gaussian distribution, namely:
wherein x is ij Valuing the jth attribute of the ith test sample, u ij (k) For training the sample mean value (sigma) of the jth attribute in the kth sample ij (k) ) 2 Sample variance for the j-th attribute in the K-th sample class is trained.
In the embodiment of the invention, the class conditional probability of each attribute of the test sample can be obtained through the formula (10), and the probability of the sample belonging to each class can be obtained through the formulas (10) and (11), wherein the class with the highest probability is the classification result of the final health degree.
According to the invention, the true difference between indexes is enhanced by weakening the independent assumption of the naive Bayesian attribute condition and sucking the experience of an expert, meanwhile, the problem of large error caused by numerical calculation weight when the sample quantity is small or the sample is missing is solved, and the respective defects of the two methods are avoided by combining the weights obtained by the two methods, so that the weight is more accurate and has practical significance.
Step S103, historical evaluation index data and corresponding cell condition health categories are obtained, and the aluminum cell condition health classification model is trained to obtain a classification model for classifying the cell condition health of the aluminum cell.
According to the invention, the knowledge of the aluminum electrolysis mechanism is combined with the expert experience, and the intelligent learning model of the cell health degree is constructed based on a large amount of actual production data, so that the refinement, the rapidness and the automatic identification of the cell health degree can be realized, the subjectivity and the slowness problems existing in the traditional manual judgment are solved, the labor cost is saved, the calculation complexity is low, and the memory requirement is less.
The foregoing examples illustrate only a few embodiments of the invention and are described in detail herein without thereby limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any adaptations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It should be understood that, although the steps in the flowcharts of the embodiments of the present invention are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in various embodiments may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor do the order in which the sub-steps or stages are performed necessarily performed in sequence, but may be performed alternately or alternately with at least a portion of the sub-steps or stages of other steps or other steps.
Those skilled in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by a computer program for instructing relevant hardware, where the program may be stored in a non-volatile computer readable storage medium, and where the program, when executed, may include processes in the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
Claims (5)
1. The method for classifying the health degree of the aluminum electrolysis cell based on the combined weighted naive Bayes is characterized by comprising the following steps of:
constructing an aluminum electrolysis cell condition health degree evaluation index system according to the types of aluminum electrolysis cells and the generated reasons and phenomena;
the evaluation index system comprises a disease groove type and an evaluation index;
the types of the disease grooves comprise a cold groove, a hot groove, a multi-effect groove and an aluminum liquid fluctuation groove;
the evaluation index includes: aluminum level, electrolyte temperature, effect coefficient, cell voltage, swing, molecular ratio, needle vibration;
according to the evaluation index system, carrying out combined weighting on the naive Bayes classifier by an analytic hierarchy process and an entropy weighting process, and constructing an aluminum electrolysis cell condition health classification model based on the combined weighted Bayes classifier;
the historical evaluation index data and the corresponding cell condition health categories are acquired, and the aluminum cell condition health classification model is trained to obtain a classification model which is used for classifying the cell condition health of the aluminum cell.
2. The method for classifying the health of the aluminum electrolysis cell conditions based on the combined weighted naive bayes according to claim 1, wherein the analytic hierarchy process specifically comprises:
constructing a plurality of judgment matrixes according to the evaluation index system, comparing the judgment matrixes in pairs, and assigning values to the judgment matrixes according to the importance degree of indexes in the judgment matrixes;
consistency test is carried out on the judgment matrix according to a formula CR=CI/RI, wherein CR is the random consistency ratio of the judgment matrix; CI is the consistency index of the judgment matrix, CI= (lambda) max -n)/(n-1)，λ max Judging the maximum characteristic root of the matrix; RI is the average random consistency index of the judgment matrix;
according to the formula rw=λ max Calculating weight of W, wherein W is a characteristic value lambda max Is described.
3. The method for classifying the health of the aluminum electrolysis cell based on the combined weighted naive Bayes according to claim 1, wherein the entropy weighting method specifically comprises the following steps:
the production data is formulatedColumn normalization processing, x ij For the value of the jth attribute in the ith sample, min (x j ) For the minimum value of the j-th attribute in all samples, max (x j ) The maximum value of the j-th attribute in all samples;
according to the formulaCalculating information entropy of data, n represents the number of samples, and p ij Representing the probability of occurrence of the jth attribute value in the ith sample, < >>
The weight coefficient calculation of each attribute is according to the formulaAnd (5) calculating.
4. The method for classifying the health of the aluminum electrolysis cell based on the combined weighted naive Bayes according to claim 1, wherein the weight calculation formula of the combined weighting is as follows:
w＝αw 1 +βw 2
wherein w is the final combining weight, w 1 Representing the weight calculated by analytic hierarchy process, alpha represents subjective preference coefficient, w 2 Representing the weight obtained by the entropy weighting method, and β representing the objective preference coefficient, and having α=1 to β.
5. The method for classifying the health degree of the aluminum electrolysis cell conditions based on the combined weighted naive Bayes according to claim 1, wherein the aluminum electrolysis cell condition health classification model based on the combined weighted Bayes classifier is as follows:
wherein θ represents the last classification result of the sample to be classified, c represents the number of classes, y k For the kth category, n k For the number of samples belonging to class k, N is the number of samples of sample space X, w ij For the combined weight obtained by combining the analytic hierarchy process and the entropy weight process, x ij Valuing the jth attribute of the ith test sample, u ij (k) For training the sample mean value (sigma) of the jth attribute in the kth sample ij (k) ) 2 Sample variance for the j-th attribute in the K-th sample class is trained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110351099.XA CN113011518B (en) | 2021-03-31 | 2021-03-31 | Aluminum electrolysis cell condition health degree classification method based on combined weighted naive Bayes |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110351099.XA CN113011518B (en) | 2021-03-31 | 2021-03-31 | Aluminum electrolysis cell condition health degree classification method based on combined weighted naive Bayes |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113011518A CN113011518A (en) | 2021-06-22 |
| CN113011518B true CN113011518B (en) | 2024-03-29 |
Family
ID=76387608
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110351099.XA Active CN113011518B (en) | 2021-03-31 | 2021-03-31 | Aluminum electrolysis cell condition health degree classification method based on combined weighted naive Bayes |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113011518B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113486980B (en) * | 2021-07-29 | 2023-02-24 | 中南大学 | Aluminum electrolysis cell condition identification method, system and equipment based on self-expression dictionary pair |
| CN114369849B (en) * | 2022-01-04 | 2024-01-30 | 阳光氢能科技有限公司 | Method and device for monitoring health degree of electrolytic cell and electrolytic cell monitoring system |
| CN116862308A (en) * | 2023-07-12 | 2023-10-10 | 成都地铁运营有限公司 | Rail health state evaluation method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705373A (en) * | 2019-09-10 | 2020-01-17 | 中南大学 | Tank voltage signal classification method based on wavelet packet analysis and residual error network |
-
2021
- 2021-03-31 CN CN202110351099.XA patent/CN113011518B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705373A (en) * | 2019-09-10 | 2020-01-17 | 中南大学 | Tank voltage signal classification method based on wavelet packet analysis and residual error network |
Non-Patent Citations (9)
| Title |
|---|
| A mechanism knowledge-driven method for identifying the pseudo dissolution hysteresis coefficient in the industrial aluminium electrolysis process;Zeng Zhaohui等;《Control Engineering Practice》;1-14 * |
| ANP算法的有限元并行计算方法在铝电解槽电场仿真中的应用研究;孙晨玮;王雅琳;陈晓方;桂卫华;;计算机与应用化学(第10期);97-108 * |
| Evaluation strategy and mass balance for mass balance for making decision about the amount of aluminum fluoride addition based on superheat degree;Weichao Yue等;《Journal of Industrial and Management Optimization》;601-622 * |
| 一种铝电解过热度软测量预测模型;杨吉森等;《2017中国自动化大会(CAC2017)暨国际智能制造创新大会》;161-167 * |
| 基于博弈论和模糊综合评判的锌冶炼企业清洁生产评价;李欢;明俊桦;石晓凤;;黄金科学技术(第05期);78-82 * |
| 基于高斯混合模型的铝电解槽况聚类研究;曹丹阳;孔淑麒;高磊;;轻金属(第02期);30-34 * |
| 大型压缩机组设备群的故障预测与健康管理;徐光华等;《控制工程》;140-146+151 * |
| 湿法炼锌沉铁过程氧化速率优化控制;谢世文等;《自动化学报》;2036-2046 * |
| 通用航空机队设备可靠性动态识别模型;陈勇刚;熊升华;贺强;贺元骅;;航空学报(第03期);172-183 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113011518A (en) | 2021-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113011518B (en) | Aluminum electrolysis cell condition health degree classification method based on combined weighted naive Bayes | |
| CN111539515B (en) | Complex equipment maintenance decision method based on fault prediction | |
| CN111105104A (en) | Short-term power load prediction method based on similar day and RBF neural network | |
| CN109948920B (en) | Electric power market settlement data risk processing method based on evidence theory | |
| CN111178725A (en) | Protective equipment state early warning method based on analytic hierarchy process | |
| CN114187120A (en) | Vehicle insurance claim settlement fraud risk identification method and device | |
| CN115081749A (en) | Bayesian optimization LSTM-based shield tunneling load advanced prediction method and system | |
| CN116128544A (en) | Active auditing method and system for electric power marketing abnormal business data | |
| CN111723085A (en) | Electric energy consumption data cleaning method and system for air conditioner of public building | |
| Xu et al. | Copula-based high dimensional cross-market dependence modeling | |
| CN113344692B (en) | Method for establishing network loan credit risk assessment model with multi-information-source fusion | |
| CN113052395A (en) | Method for predicting financial data by neural network fusing network characteristics | |
| Yet et al. | Estimating criteria weight distributions in multiple criteria decision making: a Bayesian approach | |
| CN113468811A (en) | Power grid reserve capacity probabilistic dynamic evaluation method, system, terminal and readable storage medium containing new energy unit | |
| Cappelli et al. | Change point analysis of imprecise time series | |
| CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
| Wang et al. | How magic a bullet is machine learning for credit analysis? An exploration with FinTech lending data | |
| CN115115414A (en) | Second-hand car valuation method based on machine learning | |
| CN115795328A (en) | Method and system for simultaneously generating new energy output conventional scene and extreme scene | |
| CN115222454A (en) | Vehicle sales prediction method, system and storage medium based on Stacking ensemble learning algorithm | |
| CN114091908A (en) | Power distribution network comprehensive evaluation method, device and equipment considering multi-mode energy storage station | |
| CN110852536A (en) | Storage tank maintenance decision determining method and device | |
| CN114663219B (en) | Main body credit assessment method and system based on energy interconnection power market | |
| Li et al. | Early warning model of credit risk for family farms and ranches in Inner Mongolia based on Probit regression-Kmeans clustering | |
| CN110673578B (en) | Fault degradation degree determination method and device, computer equipment and storage medium |
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 |