CN112784862A - Fault diagnosis and identification method for refining process of atmospheric and vacuum distillation unit - Google Patents
Fault diagnosis and identification method for refining process of atmospheric and vacuum distillation unit Download PDFInfo
- Publication number
- CN112784862A CN112784862A CN201911083550.3A CN201911083550A CN112784862A CN 112784862 A CN112784862 A CN 112784862A CN 201911083550 A CN201911083550 A CN 201911083550A CN 112784862 A CN112784862 A CN 112784862A
- Authority
- CN
- China
- Prior art keywords
- data
- atmospheric
- fault
- refining process
- cluster
- 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 155
- 230000008569 process Effects 0.000 title claims abstract description 100
- 238000007670 refining Methods 0.000 title claims abstract description 85
- 238000005292 vacuum distillation Methods 0.000 title claims abstract description 53
- 238000003745 diagnosis Methods 0.000 title claims abstract description 44
- 238000005260 corrosion Methods 0.000 claims abstract description 56
- 230000007797 corrosion Effects 0.000 claims abstract description 56
- 238000000513 principal component analysis Methods 0.000 claims abstract description 36
- 238000003064 k means clustering Methods 0.000 claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 11
- 238000004458 analytical method Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 47
- 230000006870 function Effects 0.000 claims description 27
- 239000006185 dispersion Substances 0.000 claims description 15
- 238000009826 distribution Methods 0.000 claims description 14
- 239000000126 substance Substances 0.000 claims description 12
- 238000001914 filtration Methods 0.000 claims description 8
- 238000000926 separation method Methods 0.000 claims description 6
- 238000013519 translation Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000005054 agglomeration Methods 0.000 claims description 3
- 230000002776 aggregation Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 230000014509 gene expression Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 239000003921 oil Substances 0.000 description 4
- 238000010992 reflux Methods 0.000 description 4
- 238000004821 distillation Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000013145 classification model Methods 0.000 description 2
- 238000011033 desalting Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000003556 assay Methods 0.000 description 1
- 238000009835 boiling Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 239000003054 catalyst Substances 0.000 description 1
- 238000001311 chemical methods and process Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000010779 crude oil Substances 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005504 petroleum refining Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (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)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The embodiment of the invention provides a fault diagnosis and identification method for an atmospheric and vacuum distillation device refining process, which comprises the following steps: corrosion data x in refining process of atmospheric and vacuum device by adopting K-means clustering methodiPerforming diagnostic analysis to distinguish between normal set data and fault set data; and performing principal component analysis on the normal set data and the fault set data respectively to realize fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit. Compared with other fault diagnosis and identification methods, the method has the characteristics of accuracy, objectivity, high efficiency and the like; meanwhile, priori knowledge is not needed to be considered, the algorithm of the diagnosis and identification process consumes less time based on the information of the data, and the efficiency of fault diagnosis and identification in the refining process of the atmospheric and vacuum device can be improved.
Description
Technical Field
The invention relates to the field of fault diagnosis and identification, in particular to a fault diagnosis and identification method for an atmospheric and vacuum distillation device refining process.
Background
Due to the increasing quality requirements of petrochemical products, the automation degree and complexity of the atmospheric and vacuum devices in the refining process are greatly improved, and the quality of the products can be influenced by the abnormal change of the working conditions of the atmospheric and vacuum devices. If production equipment or instruments break down and cannot be timely and effectively eliminated, the product quality qualified rate is reduced, safety accidents can be caused, and even the life safety of personnel is threatened.
The fault diagnosis and identification can accurately position fault points through monitoring the abnormal state of the system, correct the system in time, ensure the stability, reliability and safety of the operation process and achieve the aims of improving the production efficiency, product quality and production safety of petroleum refining.
The refining process of the atmospheric and vacuum device has particularity, a large amount of historical data acquired by a plurality of sensors has acquisition noise, the range of data value ranges is large, the types of data are multiple, the data needs to be processed by adopting a reasonable data mining method, meanwhile, the diagnosis and the identification of faults through the data are very difficult, and the algorithm of the diagnosis and the identification process consumes much time. For example, the above problems are present in both patent (CN104238545B) and patent (CN 204689946U).
Disclosure of Invention
The invention aims to provide a fault diagnosis and identification method for an atmospheric and vacuum distillation unit refining process, which is accurate and objective compared with other fault diagnosis and identification methods, has high efficiency and low time consumption of an algorithm in the diagnosis and identification process, and can improve the efficiency of fault diagnosis and identification in the atmospheric and vacuum distillation unit refining process.
In order to achieve the above object, in a first aspect of the present invention, there is provided a method for diagnosing and identifying a fault in an atmospheric and vacuum distillation unit refining process, the method including:
by usingCorrosion data x of K-means clustering method in refining process of atmospheric and vacuum deviceiPerforming diagnostic analysis to distinguish between normal set data and fault set data;
and performing principal component analysis on the normal set data and the fault set data respectively to realize fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit.
Optionally, the method further includes:
collecting original corrosion data x' in the refining process of the atmospheric and vacuum distillation unit;
normalizing the original corrosion data x', and performing wavelet noise filtering on the normalized data x to obtain the corrosion data x in the refining process of the atmospheric and vacuum distillation uniti。
Optionally, the normalizing the original corrosion data x' includes:
normalizing the original corrosion data x' into a normal distribution standard data set with a mean value of 0 and a variance of 1 by zero mean normalization processing by adopting the following formula:
wherein x is the normalized data, x ' is the collected original corrosion data in the refining process of the atmospheric and vacuum device, mu is the mean value of the original corrosion data x ', and sigma is the standard deviation of the original corrosion data x '.
Optionally, the corrosion data x in the refining process of the atmospheric and vacuum distillation unit is obtained by performing wavelet noise filtering on the normalized data x by adopting the following formulai：
Where φ is defined as a function of energy limitation on (- ∞, + ∞), and φ constitutes a squared multiplicative signal space, which is denoted as φ ∈ L2(R), then generating a family of functions { φabAs shown below:
wherein Φ (t) is a wavelet function;
φab(t) is wavelet basis function, obtained by phi (t) expansion and translation, a is expansion factor, b is translation factor, f (t) epsilon L2(R)。
Optionally, the corrosion data x in the refining process of the atmospheric and vacuum distillation unit by adopting a K-means clustering methodiPerforming a diagnostic analysis to distinguish between normal set data and fault set data, comprising:
determining the number K of initial clustering centers;
selecting Euclidean distance function as a clustered target function to carry out corrosion data x in the refining process of the atmospheric and vacuum distillation unitiAnd carrying out data classification to obtain the normal set data and the fault set data.
Optionally, determining the number K of the initial clustering centers by using a Calinski-Harabasz index;
wherein, the Calinski-Harabasz index describes the compactness through an intra-class dispersion matrix, and describes the separation through an inter-class dispersion matrix, and the Calinski-Harabasz index is defined as shown in the following formula:
wherein n is the number of clusters, K is the current class, trB (K) is the trace of the inter-class dispersion matrix, and trW (K) is the trace of the intra-class dispersion matrix;
the larger the Calinski-Harabasz index is, the tighter the representative class is, the more the classes are dispersed, and the optimal clustering result is obtained, so that the maximum K of the Calinski-Harabasz index is the number K of the initial clustering centers.
Optionally, determining the number K of the initial clustering centers by using a contour coefficient Silhouuette;
wherein the atmospheric and vacuum distillation unit smeltsCorrosion data x in chemical processiThe contour coefficient of (1) is defined as follows:
wherein S is a contour coefficient Silhouuette; a is corrosion data x in the refining process of the atmospheric and vacuum distillation unitiAverage distance from other samples in the same cluster, i.e. degree of agglomeration; b is corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe average distance, i.e., degree of separation, from all samples in the nearest cluster;
the definition of the nearest cluster is shown as follows:
wherein p is a certain cluster CkThe sample of (1);
the average contour coefficient Silhouette is the average value of the contour coefficients Silhouette of all samples, the value range of the average contour coefficient Silhouette is [ -1,1], and the closer the distance of the samples in the cluster is, the farther the distance of the samples between the clusters is; the larger the average contour coefficient Silhouuette is, the better the clustering effect is, and the maximum K of the average contour coefficient Silhouuette is made to be the number K of the initial clustering centers.
Optionally, determining the number K of the initial clustering centers by using a Davies-Bouldin index;
the calculation step for obtaining the Davies-Bouldin index comprises the following steps:
s1) calculating a ratio of the intra-cluster distance sum to the inter-cluster distance sum as shown in the following formula:
Sirepresenting the degree of scatter, X, of the metric data points in the ith classjRepresenting the jth data point in the ith class; a. theiRepresents the center of the ith class; t isiIs shown asThe number of data points in the i class; the degree of dispersion has two attributes: the mean value of the distances from each point to the center and the standard deviation of the distances from each point to the center; q is 1 and represents the mean value of the distances from each point to the center, and q is 2 and represents the standard deviation of the distances from each point to the center;
s2) calculating a distance value M indicating a distance between the ith and jth classesijAs shown in the following formula:
akivalue of the Kth attribute, M, representing the center point of the ith classijIs the distance from the center of class i to class j;
s3) calculating a value R for measuring the similarity between the ith class and the jth classijAs shown in the following formula:
s4) taking the maximum RijThe maximum similarity value in the similarity between the ith class and other classes, and the Davies-Bouldin index is the mean value of the maximum similarity of each class;
wherein, the smaller the Davies-Bouldin index is, the better the classification is; and K when the Davies-Bouldin index is the minimum is the number K of the initial clustering centers.
Optionally, the euclidean distance function is selected as a target function of clustering to corrosion data x in the refining process of the atmospheric and vacuum distillation unitiPerforming data classification to obtain the normal set data and the fault set data, including:
establishing K initial clustering centers by adopting a K-means clustering method, distributing each point in the data set to the nearest centroid corresponding to the required cluster number, updating the centroid of each cluster based on the point distributed to the cluster through each iteration until the iteration is finished when the square error and the minimum requirement are met, finishing clustering and realizing data classification; wherein the euclidean distance function represents the actual distance between two points;
given a specified number of clusters K in the sample data, each cluster C is made to be a cluster C by a K-means algorithmkEach mean value mu ofkThe sum of squared errors SSE between them is minimal, given by:
wherein SSE represents the sum of squared errors, CkDenotes a specified cluster, μkThe mean value of the distance within each cluster is indicated.
Optionally, the performing principal component analysis on the normal set data and the fault set data respectively to implement fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit includes:
establishing a principal component analysis principal component model;
calculating principal component analysis model statistic T of the normal set data2And the square prediction error SPE and the control limit thereof, and calculating principal component analysis model statistic T of the fault set data2And the square prediction error SPE is used for diagnosing whether a fault occurs in the refining process of the atmospheric and vacuum device;
calculating the statistics T for each process variable in the fault set data2And the contribution rate of the square prediction error SPE, and identifying the variable with the largest contribution rate as the variable causing the fault.
Optionally, the establishing a principal component analysis principal component model includes:
determining a principal component space by covariance decomposition, wherein a covariance matrix S is given by:
where the original matrix X is an m-dimensional dataset containing n samples, Λ is an eigenvalue matrix of the covariance matrix S, and the diagonal elements satisfy λ1≥λ2≥……≥λm；
Vm×mIs the feature matrix of S, P is the first A column of V, contains the information of all the pivot elements,
is the remaining m-A column of V, containing non-pivot information;
decomposing the original matrix X to obtain a principal element subspace matrix T and a residual error subspace matrix E, as shown in the following formula:
Tn×A＝Xn×m·Pm×A
wherein the content of the first and second substances,
is a scoring matrix, Pm×AAs a load matrix, Pm×AConsisting of the first a eigenvectors of S.
Optionally, the principal component analysis model statistic T of the normal set data is calculated2And the square prediction error SPE and the control limit thereof, and calculating principal component analysis model statistic T of the fault set data2And the square prediction error SPE is used for diagnosing whether a fault occurs in the refining process of the atmospheric and vacuum device, and the method comprises the following steps:
calculating a statistic T of the normal set data2The statistic T is compared with the control limit of the square prediction error SPE2The control limit of the SPE and the square prediction error is used as a threshold value for distinguishing a normal working condition statistic value from an abnormal working condition statistic value;
calculating a statistic T of the fault set data2And a square prediction error SPE, if the statistic T of the fault set data2A value greater than the threshold, and/or a squared prediction error, SPE, value of the fault set data greater than the threshold, thenDiagnosing that the refining process of the atmospheric and vacuum device has faults, otherwise, diagnosing that the refining process of the atmospheric and vacuum device has no faults;
wherein the statistic T2Is used for measuring corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe change in principal component space is shown by the following equation:
wherein Λ ═ diag { λ ═ λ1,λ2,...,λA}，
T being confidence a2The control limit is shown as the following formula:
wherein, FA,n-A,αF distribution value with A and n-A freedom degrees and alpha confidence coefficient;
wherein the square prediction error SPE is used for measuring corrosion data x in the refining process of the atmospheric and vacuum deviceiThe projection variation in residual space is mathematically expressed as:
wherein the content of the first and second substances,
the control limit of the squared prediction error SPE with the confidence coefficient alpha is expressed as follows:
wherein the content of the first and second substances,
the covariance matrix eigenvalue is X; cαIs a threshold value of the standard normal distribution at the confidence level alpha.
Optionally, calculating the statistics T for each process variable in the fault set data2And the contribution rate of the square prediction error SPE, wherein the variable with the largest contribution rate is identified as the variable causing the fault, and the variable comprises the following components:
the statistic T2And the value of the variable contribution of the squared prediction error SPE is defined as follows:
wherein the content of the first and second substances,
the variable with the largest contribution ratio Cont value is the variable causing the refining process fault of the atmospheric and vacuum device.
According to the fault diagnosis and identification method for the refining process of the atmospheric and vacuum device, firstly, zero mean (z-score) standardization is carried out on training set data, variables with different dimensions are converted into dimensionless expressions, and normal distribution data with the mean value of 1 and the variance of 0 are met. And then, determining the number of initial clustering centers during K-means clustering, and diagnosing normal working conditions and fault working conditions through clustering. And determining the control limits of the normal working condition statistics T2 and SPE and the contribution rate of each variable to the fault by a Principal Component Analysis (PCA), wherein the larger the contribution rate is, the more likely the cause of the fault is, so as to realize fault identification. Compared with other fault diagnosis and identification methods, the method has the characteristics of accuracy, objectivity, high efficiency and the like; meanwhile, priori knowledge is not needed to be considered, the algorithm of the diagnosis and identification process consumes less time based on the information of the data, and the efficiency of fault diagnosis and identification in the refining process of the atmospheric and vacuum device can be improved.
Additional features and advantages of embodiments of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the embodiments of the invention without limiting the embodiments of the invention. In the drawings:
FIG. 1 is a flow chart of the fault diagnosis and identification of the refining process of the atmospheric and vacuum distillation unit provided by one embodiment of the invention;
FIG. 2 is a flow chart of an atmospheric and vacuum distillation unit refining process provided by an alternative embodiment of the invention;
fig. 3 is a graph of the contribution rate of a variable in fault identification provided by an alternative embodiment of the invention.
Technical term interpretation:
the K-means clustering method comprises the following steps: the K-means clustering algorithm (K-means clustering algorithm) is a clustering analysis algorithm for iterative solution.
Principal component analysis method (PCA): principal Component Analysis (PCA), a statistical method, converts a set of variables that may have correlation into a set of linearly uncorrelated variables by orthogonal transformation, and the converted set of variables is called Principal components. Principal component analysis is also referred to as principal component analysis.
F1 Score: the F1 score is an index used for measuring the accuracy of the two classification models in statistics and takes the accuracy and the recall rate of the classification models into consideration.
SPE: the prediction error is squared.
z-score: zero mean value; also called standard score, is a process of dividing the difference between a number and a mean by a standard deviation, and can truly reflect a relative standard distance between a score and the mean.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present invention, are given by way of illustration and explanation only, not limitation.
In the embodiments of the present invention, unless otherwise specified, the use of directional terms such as "upper, lower, top, and bottom" is generally used with respect to the orientation shown in the drawings or the positional relationship of the components with respect to each other in the vertical, or gravitational direction.
The first embodiment is as follows:
fig. 1 is a flow chart for diagnosing and identifying faults in an atmospheric and vacuum distillation unit refining process according to an embodiment of the invention. As shown in fig. 1, in a first aspect of the present invention, there is provided a method for diagnosing and identifying a fault in an atmospheric and vacuum distillation unit refining process, the method comprising:
corrosion data x in refining process of atmospheric and vacuum device by adopting K-means clustering methodiPerforming diagnostic analysis to distinguish between normal set data and fault set data; corrosion data x in the refining process of the atmospheric and vacuum distillation unitiIs the preprocessed data, and the specific preprocessing process is described in the following section.
The steps of the K-means clustering method comprise: randomly selecting K objects as initial clustering centers, then calculating the distance between each object and each seed clustering center, and allocating each object to the closest clustering center; the cluster centers and the objects assigned to them represent a cluster. The cluster center of a cluster is recalculated for each sample assigned based on the objects existing in the cluster. This process will be repeated until some termination condition is met; the termination condition may be that no (or minimum number) objects are reassigned to different clusters, no (or minimum number) cluster centers are changed again, and the sum of squared errors is locally minimal. The implementation steps in the technical solution of the present invention are described in the following section.
And performing principal component analysis on the normal set data and the fault set data respectively to realize fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit. When multivariate variables are studied using statistical analysis, too many variables increase computational complexity. It is generally desirable to obtain a large amount of information with a small number of variables. In many cases, there is a certain correlation between variables, and when there is a certain correlation between two variables, it can be interpreted that there is a certain overlap of information reflecting the subject. The principal component analysis is to eliminate redundant repeated variables (closely related variables) for all the variables originally proposed, and to establish new variables as few as possible, so that the new variables are unrelated in pairs, and the new variables keep original information as much as possible in terms of reflecting the information of the subject.
Optionally, the method further includes:
collecting original corrosion data x' in the refining process of the atmospheric and vacuum distillation unit;
normalizing the original corrosion data x', and performing wavelet noise filtering on the normalized data x to obtain the corrosion data x in the refining process of the atmospheric and vacuum distillation uniti。
Optionally, the normalizing the original corrosion data x' includes:
the raw corrosion data x' was normalized to a normal distribution standard data set with a mean of 0 and a variance of 1 by a zero mean (z-score) normalization process using the following formula:
wherein x is the normalized data, x ' is the collected original corrosion data in the refining process of the atmospheric and vacuum device, mu is the mean value of the original corrosion data x ', and sigma is the standard deviation of the original corrosion data x '. The normal distribution standard data set represents the distance of the original data from the mean, and the standard of the distance measure is standard deviation. If the statistics are sufficiently large, the zero mean (z-score) distribution is satisfied, with 99% of the data distributed between-1 and 1, 95% between-2 and 2, and 68% between-3 and 3. When the data distribution is too messy, the maximum value and the minimum value cannot be judged, or excessive singular points exist in the data, the data can be subjected to normalized processing by adopting a zero mean value (z-score).
Optionally, the corrosion data x in the refining process of the atmospheric and vacuum distillation unit is obtained by performing wavelet noise filtering on the normalized data x by adopting the following formulai：
Where φ is defined as a function of energy limitation on (- ∞, + ∞), and φ constitutes a squared multiplicative signal space, which is denoted as φ ∈ L2(R), then generating a family of functions { φabAs shown below:
wherein Φ (t) is a wavelet function;
φab(t) is wavelet basis function, obtained by phi (t) expansion and translation, a is expansion factor, b is translation factor, f (t) epsilon L2(R) in the presence of a catalyst. Typically, the data collected by the sensor will contain measurement noise, and a filter may be selected to filter out the noise. The wavelet transform is a time-frequency analysis of signals, has the characteristic of multi-resolution, and can conveniently extract original signals from signals mixed with strong noise. The noise generated during data acquisition can be effectively filtered by performing wavelet noise filtering on the original data set.
Optionally, the corrosion data x in the refining process of the atmospheric and vacuum distillation unit by adopting a K-means clustering methodiPerforming a diagnostic analysis to distinguish between normal set data and fault set data, comprising:
determining the number K of initial clustering centers;
selecting Euclidean distance function as a clustered target function to the atmospheric and vacuum pressure deviceCorrosion data x in refiningiAnd carrying out data classification to obtain the normal set data and the fault set data.
And obtaining the number K of the clustering centers by adopting any index of Calinski-Harabasz, Silhouette and Davies-Bouldin. As follows:
optionally, determining the number K of the initial clustering centers by using a Calinski-Harabasz index;
wherein, the Calinski-Harabasz index describes the compactness through an intra-class dispersion matrix, and describes the separation through an inter-class dispersion matrix, and the Calinski-Harabasz index is defined as shown in the following formula:
wherein n is the number of clusters, K is the current class, trB (K) is the trace of the inter-class dispersion matrix, and trW (K) is the trace of the intra-class dispersion matrix;
the larger the Calinski-Harabasz index is, the tighter the representative class is, the more the classes are dispersed, and the optimal clustering result is obtained, so that the maximum K of the Calinski-Harabasz index is the number K of the initial clustering centers.
Optionally, determining the number K of the initial clustering centers by using a contour coefficient Silhouuette;
wherein, the corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe contour coefficient of (1) is defined as follows:
wherein S is a contour coefficient Silhouuette; a is corrosion data x in the refining process of the atmospheric and vacuum distillation unitiAverage distance from other samples in the same cluster, i.e. degree of agglomeration; b is corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe average distance, i.e., degree of separation, from all samples in the nearest cluster;
the definition of the nearest cluster is shown as follows:
wherein p is a certain cluster CkThe sample of (1);
the average contour coefficient Silhouette is the average value of the contour coefficients Silhouette of all samples, the value range of the average contour coefficient Silhouette is [ -1,1], and the closer the distance of the samples in the cluster is, the farther the distance of the samples between the clusters is; the larger the average contour coefficient Silhouuette is, the better the clustering effect is, and the maximum K of the average contour coefficient Silhouuette is made to be the number K of the initial clustering centers.
Optionally, determining the number K of the initial clustering centers by using a Davies-Bouldin index;
the calculation step for obtaining the Davies-Bouldin index comprises the following steps:
s1) calculating a ratio of the intra-cluster distance sum to the inter-cluster distance sum as shown in the following formula:
Sirepresenting the degree of scatter, X, of the metric data points in the ith classjRepresenting the jth data point in the ith class; a. theiRepresents the center of the ith class; t isiRepresenting the number of data points in the ith class; the degree of dispersion has two attributes: the mean value of the distances from each point to the center and the standard deviation of the distances from each point to the center; q is 1 and represents the mean value of the distances from each point to the center, and q is 2 and represents the standard deviation of the distances from each point to the center;
s2) calculating a distance value M indicating a distance between the ith and jth classesijAs shown in the following formula:
akivalue of the Kth attribute, M, representing the center point of the ith classijIs the distance from the center of class i to class j;
s3) calculating a value R for measuring the similarity between the ith class and the jth classijAs shown in the following formula:
s4) taking the maximum RijThe maximum similarity value in the similarity between the ith class and other classes, and the Davies-Bouldin index is the mean value of the maximum similarity of each class;
wherein, the smaller the Davies-Bouldin index is, the better the classification is; and K when the Davies-Bouldin index is the minimum is the number K of the initial clustering centers.
Optionally, the euclidean distance function is selected as a target function of clustering to corrosion data x in the refining process of the atmospheric and vacuum distillation unitiPerforming data classification to obtain the normal set data and the fault set data, including:
establishing K initial clustering centers by adopting a K-means clustering method, distributing each point in the data set to the nearest centroid corresponding to the required cluster number, updating the centroid of each cluster based on the point distributed to the cluster through each iteration until the iteration is finished when the square error and the minimum requirement are met, finishing clustering and realizing data classification; wherein the euclidean distance function represents the actual distance between two points;
given a specified number of clusters K in the sample data, each cluster C is made to be a cluster C by a K-means algorithmkEach mean value mu ofkThe sum of squared errors SSE between them is minimal, given by:
wherein SSE represents the sum of squared errors, CkDenotes a specified cluster, μkThe mean value of the distance within each cluster is indicated.
F1-score can be used for evaluating the diagnosis capability of the K-means clustering method for the faults of the refining process of the atmospheric and vacuum devices. F1-score assesses the accuracy of the model by combining Precision (Precision) and Recall (Recall), the expression of which is shown in (11):
the quantities in the above equation are determined from the confusion matrix, which is shown in table 1.
TABLE 1 confusion matrix
Where TP represents correctly classified normal data, FP represents incorrectly classified normal data, TN represents correctly classified fault data, and FN represents incorrectly classified fault data. The mathematical expressions corresponding to the accuracy and the recall degree are shown in formulas (12) and (13):
optionally, the performing principal component analysis on the normal set data and the fault set data respectively to implement fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit includes:
establishing a principal component analysis principal component model;
calculating principal component analysis model statistic T of the normal set data2And the square prediction error SPE and the control limit thereof, and calculating principal component analysis model statistic T of the fault set data2And the square prediction error SPE is used for diagnosing whether a fault occurs in the refining process of the atmospheric and vacuum device;
calculating the statistics for each process variable pair in the fault set dataT2And the contribution rate of the square prediction error SPE, and identifying the variable with the largest contribution rate as the variable causing the fault.
Optionally, the establishing a principal component analysis principal component model includes:
determining a principal component space by covariance decomposition, wherein a covariance matrix S is given by:
where the original matrix X is an m-dimensional dataset containing n samples, Λ is an eigenvalue matrix of the covariance matrix S, and the diagonal elements satisfy λ1≥λ2≥……≥λm；
Vm×mIs the feature matrix of S, P is the first A column of V, contains the information of all the pivot elements,
is the remaining m-A column of V, containing non-pivot information;
decomposing the original matrix X to obtain a principal element subspace matrix T and a residual error subspace matrix E, as shown in the following formula:
Tn×A＝Xn×m·Pm×A
wherein the content of the first and second substances,
is a scoring matrix, Pm×AAs a load matrix, Pm×AConsisting of the first a eigenvectors of S.
Optionally, the principal component analysis model statistic T of the normal set data is calculated2Is and flatSquare prediction error SPE and control limit thereof, and principal component analysis model statistic T for calculating fault set data2And the square prediction error SPE is used for diagnosing whether a fault occurs in the refining process of the atmospheric and vacuum device, and the method comprises the following steps:
calculating a statistic T of the normal set data2The statistic T is compared with the control limit of the square prediction error SPE2The control limit of the SPE and the square prediction error is used as a threshold value for distinguishing a normal working condition statistic value from an abnormal working condition statistic value;
calculating a statistic T of the fault set data2And a square prediction error SPE, if the statistic T of the fault set data2If the value is larger than the threshold value and/or the SPE value is larger than the threshold value, diagnosing that the refining process of the atmospheric and vacuum device has a fault, otherwise, diagnosing that the refining process of the atmospheric and vacuum device has no fault;
wherein the statistic T2Is used for measuring corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe change in principal component space is shown by the following equation:
wherein Λ ═ diag { λ ═ λ1,λ2,...,λA}，
T being confidence a2The control limit is shown as the following formula:
wherein, FA,n-A,αF distribution value with A and n-A freedom degrees and alpha confidence coefficient;
wherein the square prediction error SPE is used for measuring corrosion data x in the refining process of the atmospheric and vacuum deviceiProjection variation in residual space, mathematical tablesThe following formula is shown:
wherein the content of the first and second substances,
the control limit of the squared prediction error SPE with the confidence coefficient alpha is expressed as follows:
wherein the content of the first and second substances,
the covariance matrix eigenvalue is X; cαIs a threshold value of the standard normal distribution at the confidence level alpha.
Optionally, calculating the statistics T for each process variable in the fault set data2And the contribution rate of the square prediction error SPE, wherein the variable with the largest contribution rate is identified as the variable causing the fault, and the variable comprises the following components:
the statistic T2And the value of the variable contribution of the squared prediction error SPE is defined as follows:
wherein the content of the first and second substances,
the variable with the largest value of the contribution rate Cont isBecomes a variable of the refining process fault of the atmospheric and vacuum distillation unit.
Example two:
fig. 2 shows a flow chart of the refining process of the atmospheric and vacuum distillation unit, and the embodiment according to fig. 2 is as follows:
the atmospheric and vacuum distillation device generally comprises a primary distillation tower, an atmospheric tower and a vacuum tower.
The crude oil enters an electric desalting tank for desalting and dewatering after the heat exchange in the tank area, and enters a distillation tower after the heat exchange again. Primary top oil is separated from the top of the distillation tower, and bottom oil is heated by a normal pressure furnace and then sent to a normal pressure tower. The atmospheric tower is generally provided with three lateral lines, namely a normal line, a normal line and a normal line from top to bottom. Each side draw is sent out of the device after being stripped. Atmospheric towers are typically provided with one top reflux and two mid-stream refluxes, each of which takes away residual heat to reduce energy consumption. The tower bottom oil of the atmospheric tower is heated by a vacuum furnace and then sent into a vacuum tower. The vacuum is drawn at the top of the vacuum column to reduce the pressure at the top of the column, thereby reducing the boiling point of the component oil. The vacuum tower is generally provided with three lateral lines, namely a first line, a second line and a third line from top to bottom. The vacuum tower is generally provided with a top circulation reflux and two middle reflux streams, and the vacuum residue is sent out of the device from the bottom of the vacuum tower for secondary processing.
The collected original corrosion data in the refining process of the atmospheric and vacuum distillation unit mainly comprises production real-time process data, raw material and product sampling data and water quality analysis and assay data, and diagnosis and identification of different working conditions are carried out by taking a corrosion control parameter 'Fe ion concentration in cut water at the top of an atmospheric tower' of the atmospheric and vacuum distillation unit as an example. Different working conditions can occur in the reaction process of the atmospheric and vacuum device, the data volume of normal (or optimized) working conditions is large, the data fluctuation range is small, and the operation is stable; the data amount of the fault (or degradation) working condition is small, and the data fluctuation is large. By the method of combining K-means clustering and principal component analysis, different working conditions can be diagnosed, and main factors influencing the working conditions are identified.
And carrying out normalization and wavelet noise filtering on the original corrosion data in the refining process of the atmospheric and vacuum device.
Training normal working condition data and determining statistic T of variable2And the contribution rate of the SPE to generating a certain fault, wherein variables with larger contribution rates are the most likely causes of the fault, and a list of ten factors ranking the contribution rates is shown in Table 2. The factors are shown in fig. 3 as a graph of the contribution rate of the variables in fault identification.
Table 2 contribution rate ranked top ten factor table
The method is suitable for fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit. First, zero mean (z-score) normalization is performed on training set data, variables of different dimensions are converted into dimensionless expressions, and normal distribution data with a mean of 1 and a variance of 0 is satisfied. And then, determining the number of initial clustering centers during K-means clustering, and diagnosing normal working conditions and fault working conditions through clustering. And determining the control limits of the normal working condition statistics T2 and SPE and the contribution rate of each variable to the fault by a Principal Component Analysis (PCA), wherein the larger the contribution rate is, the more likely the cause of the fault is, so as to realize fault identification. Compared with other fault diagnosis and identification methods, the method has the characteristics of accuracy, objectivity, high efficiency and the like; meanwhile, priori knowledge is not needed to be considered, the algorithm of the diagnosis and identification process consumes less time based on the information of the data, and the efficiency of fault diagnosis and identification in the refining process of the atmospheric and vacuum device can be improved.
While the embodiments of the present invention have been described in detail with reference to the accompanying drawings, the embodiments of the present invention are not limited to the details of the above embodiments, and various simple modifications can be made to the technical solution of the embodiments of the present invention within the technical idea of the embodiments of the present invention, and the simple modifications are within the scope of the embodiments of the present invention.
It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. In order to avoid unnecessary repetition, the embodiments of the present invention will not be described separately for the various possible combinations.
Those skilled in the art will appreciate that all or part of the steps in the method for implementing the above embodiments may be implemented by a program, which is stored in a storage medium and includes several instructions to enable a single chip, a chip, or a processor (processor) to execute all or part of the steps in the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In addition, any combination of the various embodiments of the present invention is also possible, and the same should be considered as disclosed in the embodiments of the present invention as long as it does not depart from the spirit of the embodiments of the present invention.
Claims (13)
1. A fault diagnosis and identification method for an atmospheric and vacuum distillation unit refining process is characterized by comprising the following steps:
corrosion data x in refining process of atmospheric and vacuum device by adopting K-means clustering methodiPerforming diagnostic analysis to distinguish between normal set data and fault set data;
and performing principal component analysis on the normal set data and the fault set data respectively to realize fault diagnosis and identification in the refining process of the atmospheric and vacuum distillation unit.
2. The fault diagnosis and identification method according to claim 1, characterized in that the method further comprises:
collecting original corrosion data x' in the refining process of the atmospheric and vacuum distillation unit;
normalizing the original corrosion data x', and performing wavelet noise filtering on the normalized data x to obtain the corrosion data x in the refining process of the atmospheric and vacuum distillation uniti。
3. The method for diagnosing and identifying faults according to claim 2, wherein the normalizing the raw corrosion data x' comprises:
normalizing the original corrosion data x' into a normal distribution standard data set with a mean value of 0 and a variance of 1 by zero mean normalization processing by adopting the following formula:
wherein x is the normalized data, x ' is the collected original corrosion data in the refining process of the atmospheric and vacuum device, mu is the mean value of the original corrosion data x ', and sigma is the standard deviation of the original corrosion data x '.
4. The method for diagnosing and identifying faults according to claim 3, wherein the corrosion data x in the refining process of the atmospheric and vacuum distillation device is obtained by performing wavelet noise filtering on the normalized data x by adopting the following formulai：
Where φ is defined as a function of energy limitation on (- ∞, + ∞), and φ constitutes a squared multiplicative signal space, which is denoted as φ ∈ L2(R), then generating a family of functions { φabAs shown below:
wherein Φ (t) is a wavelet function;
φab(t) is wavelet basis function, obtained by phi (t) expansion and translation, a is expansion factor, b is translation factor, f (t) epsilon L2(R)。
5. The method of claim 1The fault diagnosis and identification method is characterized in that the corrosion data x in the refining process of the atmospheric and vacuum device by adopting a K-means clustering methodiPerforming a diagnostic analysis to distinguish between normal set data and fault set data, comprising:
determining the number K of initial clustering centers;
selecting Euclidean distance function as a clustered target function to carry out corrosion data x in the refining process of the atmospheric and vacuum distillation unitiAnd carrying out data classification to obtain the normal set data and the fault set data.
6. The fault diagnosis and identification method according to claim 5, wherein the number K of initial cluster centers is determined using Calinski-Harabasz criteria;
wherein, the Calinski-Harabasz index describes the compactness through an intra-class dispersion matrix, and describes the separation through an inter-class dispersion matrix, and the Calinski-Harabasz index is defined as shown in the following formula:
wherein n is the number of clusters, K is the current class, trB (K) is the trace of the inter-class dispersion matrix, and trW (K) is the trace of the intra-class dispersion matrix;
the larger the Calinski-Harabasz index is, the tighter the representative class is, the more the classes are dispersed, and the optimal clustering result is obtained, so that the maximum K of the Calinski-Harabasz index is the number K of the initial clustering centers.
7. The fault diagnosis and identification method according to claim 5, characterized in that the number K of the initial cluster centers is determined by using a contour coefficient Silhouuette;
wherein, the corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe contour coefficient of (1) is defined as follows:
wherein S is a contour coefficient Silhouuette; a is corrosion data x in the refining process of the atmospheric and vacuum distillation unitiAverage distance from other samples in the same cluster, i.e. degree of agglomeration; b is corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe average distance, i.e., degree of separation, from all samples in the nearest cluster;
the definition of the nearest cluster is shown as follows:
wherein p is a certain cluster CkThe sample of (1);
the average contour coefficient Silhouette is the average value of the contour coefficients Silhouette of all samples, the value range of the average contour coefficient Silhouette is [ -1,1], and the closer the distance of the samples in the cluster is, the farther the distance of the samples between the clusters is; the larger the average contour coefficient Silhouuette is, the better the clustering effect is, and the maximum K of the average contour coefficient Silhouuette is made to be the number K of the initial clustering centers.
8. The fault diagnosis and identification method according to claim 5, wherein the initial cluster center number K is determined using a Davies-Bouldin index;
the calculation step for obtaining the Davies-Bouldin index comprises the following steps:
s1) calculating a ratio of the intra-cluster distance sum to the inter-cluster distance sum as shown in the following formula:
Sirepresenting the degree of scatter, X, of the metric data points in the ith classjRepresenting the jth data point in the ith class; a. theiRepresents the center of the ith class; t isiRepresenting the number of data points in the ith class; the degree of dispersion has two attributes:the mean value of the distances from each point to the center and the standard deviation of the distances from each point to the center; q is 1 and represents the mean value of the distances from each point to the center, and q is 2 and represents the standard deviation of the distances from each point to the center;
s2) calculating a distance value M indicating a distance between the ith and jth classesijAs shown in the following formula:
akivalue of the Kth attribute, M, representing the center point of the ith classijIs the distance from the center of class i to class j;
s3) calculating a value R for measuring the similarity between the ith class and the jth classijAs shown in the following formula:
s4) taking the maximum RijThe maximum similarity value in the similarity between the ith class and other classes, and the Davies-Bouldin index is the mean value of the maximum similarity of each class;
wherein, the smaller the Davies-Bouldin index is, the better the classification is; and K when the Davies-Bouldin index is the minimum is the number K of the initial clustering centers.
9. The fault diagnosis and identification method according to any one of claims 6 to 8, wherein the Euclidean distance function is selected as a clustered objective function to carry out refining on corrosion data x in the atmospheric and vacuum distillation deviceiPerforming data classification to obtain the normal set data and the fault set data, including:
establishing K initial clustering centers by adopting a K-means clustering method, distributing each point in the data set to the nearest centroid corresponding to the required cluster number, updating the centroid of each cluster based on the point distributed to the cluster through each iteration until the iteration is finished when the square error and the minimum requirement are met, finishing clustering and realizing data classification; wherein the euclidean distance function represents the actual distance between two points;
given a specified number of clusters K in the sample data, each cluster C is made to be a cluster C by a K-means algorithmkEach mean value mu ofkThe sum of squared errors SSE between them is minimal, given by:
wherein SSE represents the sum of squared errors, CkDenotes a specified cluster, μkThe mean value of the distance within each cluster is indicated.
10. The fault diagnosis and identification method according to claim 1, wherein the performing principal component analysis on the normal set data and the fault set data respectively to realize fault diagnosis and identification in the refining process of the atmospheric and vacuum device comprises:
establishing a principal component analysis principal component model;
calculating principal component analysis model statistic T of the normal set data2And the square prediction error SPE and the control limit thereof, and calculating principal component analysis model statistic T of the fault set data2And the square prediction error SPE is used for diagnosing whether a fault occurs in the refining process of the atmospheric and vacuum device;
calculating the statistics T for each process variable in the fault set data2And the contribution rate of the square prediction error SPE, and identifying the variable with the largest contribution rate as the variable causing the fault.
11. The fault diagnosis and identification method according to claim 10, wherein the establishing a principal component analysis principal component model comprises:
determining a principal component space by covariance decomposition, wherein a covariance matrix S is given by:
where the original matrix X is an m-dimensional dataset containing n samples, Λ is an eigenvalue matrix of the covariance matrix S, and the diagonal elements satisfy λ1≥λ2≥……≥λm；
Vm×mIs the feature matrix of S, P is the first A column of V, contains the information of all the pivot elements,
is the remaining m-A column of V, containing non-pivot information;
decomposing the original matrix X to obtain a principal element subspace matrix T and a residual error subspace matrix E, as shown in the following formula:
Tn×A＝Xn×m·Pm×A
wherein the content of the first and second substances,
is a scoring matrix, Pm×AAs a load matrix, Pm×AConsisting of the first a eigenvectors of S.
12. The fault diagnosis and identification method according to claim 10, wherein said calculating a principal component analysis model statistic T of said normal set data2And the square prediction error SPE and the control limit thereof, and calculating principal component analysis model statistic T of the fault set data2And the square prediction error SPE is used for diagnosing whether a fault occurs in the refining process of the atmospheric and vacuum device, and the method comprises the following steps:
calculating a statistic T of the normal set data2The statistic T is compared with the control limit of the square prediction error SPE2The control limit of the SPE and the square prediction error is used as a threshold value for distinguishing a normal working condition statistic value from an abnormal working condition statistic value;
calculating a statistic T of the fault set data2And a square prediction error SPE, if the statistic T of the fault set data2If the value is larger than the threshold value and/or the SPE value is larger than the threshold value, diagnosing that the refining process of the atmospheric and vacuum device has a fault, otherwise, diagnosing that the refining process of the atmospheric and vacuum device has no fault;
wherein the statistic T2Is used for measuring corrosion data x in the refining process of the atmospheric and vacuum distillation unitiThe change in principal component space is shown by the following equation:
wherein Λ ═ diag { λ ═ λ1,λ2,...,λA}，
T being confidence a2The control limit is shown as the following formula:
wherein, FA,n-A,αF distribution value with A and n-A freedom degrees and alpha confidence coefficient;
wherein the square prediction error SPE is used for measuring corrosion data x in the refining process of the atmospheric and vacuum deviceiThe projection variation in residual space is mathematically expressed as:
wherein the content of the first and second substances,
the control limit of the squared prediction error SPE with the confidence coefficient alpha is expressed as follows:
wherein the content of the first and second substances,
the covariance matrix eigenvalue is X; cαIs a threshold value of the standard normal distribution at the confidence level alpha.
13. The fault diagnosis and identification method according to claim 10, characterized in that each process variable in the fault set data is calculated for the statistic T2And the contribution rate of the square prediction error SPE, wherein the variable with the largest contribution rate is identified as the variable causing the fault, and the variable comprises the following components:
the statistic T2And the value of the variable contribution of the squared prediction error SPE is defined as follows:
wherein the content of the first and second substances,
contribution toThe variable with the largest value of the rate Cont is a variable causing a fault in the refining process of the atmospheric and vacuum equipment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911083550.3A CN112784862A (en) | 2019-11-07 | 2019-11-07 | Fault diagnosis and identification method for refining process of atmospheric and vacuum distillation unit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911083550.3A CN112784862A (en) | 2019-11-07 | 2019-11-07 | Fault diagnosis and identification method for refining process of atmospheric and vacuum distillation unit |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112784862A true CN112784862A (en) | 2021-05-11 |
Family
ID=75748005
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911083550.3A Pending CN112784862A (en) | 2019-11-07 | 2019-11-07 | Fault diagnosis and identification method for refining process of atmospheric and vacuum distillation unit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112784862A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569950A (en) * | 2021-07-28 | 2021-10-29 | 大唐环境产业集团股份有限公司 | Power station equipment fault monitoring model generation method, system and device |
| CN113640607A (en) * | 2021-08-18 | 2021-11-12 | 江苏科技大学 | Early fault diagnosis method for inverter circuit and motor of high-speed train |
| CN114355850A (en) * | 2021-12-28 | 2022-04-15 | 汉谷云智(武汉)科技有限公司 | Atmospheric and vacuum pressure device fault diagnosis method based on queue competition algorithm |
| CN115146191A (en) * | 2022-07-21 | 2022-10-04 | 北京天防安全科技有限公司 | Method and device for identifying video monitoring assets based on AI (Artificial Intelligence) and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120160707A1 (en) * | 2010-12-28 | 2012-06-28 | Grzegorz Jan Kusinski | Processes and systems for characterizing and blending refinery feedstocks |
| CN105893700A (en) * | 2016-04-26 | 2016-08-24 | 陆新建 | Chemical production on-line fault detection and diagnosis technique based on physical-large data hybrid model |
| CN106762452A (en) * | 2016-12-08 | 2017-05-31 | 东北大学 | Fan master control system fault diagnosis and on-line monitoring method based on data-driven |
| CN106778259A (en) * | 2016-12-28 | 2017-05-31 | 北京明朝万达科技股份有限公司 | A kind of abnormal behaviour based on big data machine learning finds method and system |
| CN107608335A (en) * | 2017-09-14 | 2018-01-19 | 山东科技大学 | A kind of UAV Flight Control System fault detect and the data-driven method of fault reconstruction |
-
2019
- 2019-11-07 CN CN201911083550.3A patent/CN112784862A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120160707A1 (en) * | 2010-12-28 | 2012-06-28 | Grzegorz Jan Kusinski | Processes and systems for characterizing and blending refinery feedstocks |
| CN105893700A (en) * | 2016-04-26 | 2016-08-24 | 陆新建 | Chemical production on-line fault detection and diagnosis technique based on physical-large data hybrid model |
| CN106762452A (en) * | 2016-12-08 | 2017-05-31 | 东北大学 | Fan master control system fault diagnosis and on-line monitoring method based on data-driven |
| CN106778259A (en) * | 2016-12-28 | 2017-05-31 | 北京明朝万达科技股份有限公司 | A kind of abnormal behaviour based on big data machine learning finds method and system |
| CN107608335A (en) * | 2017-09-14 | 2018-01-19 | 山东科技大学 | A kind of UAV Flight Control System fault detect and the data-driven method of fault reconstruction |
Non-Patent Citations (2)
| Title |
|---|
| HAMIDEH ROSTAMI ET AL.: "Equipment condition diagnosis and fault fingerprint extraction in semiconductor manufacturing", 《2016 15TH IEEE INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS》, pages 534 - 539 * |
| 杨秀璋等: "《Python 网络数据爬取及分析从入门到精通 分析篇》", 30 June 2018, 北京航空航天大学出版社, pages: 88 - 89 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569950A (en) * | 2021-07-28 | 2021-10-29 | 大唐环境产业集团股份有限公司 | Power station equipment fault monitoring model generation method, system and device |
| CN113569950B (en) * | 2021-07-28 | 2024-05-28 | 大唐环境产业集团股份有限公司 | Power station equipment fault monitoring model generation method, system and device |
| CN113640607A (en) * | 2021-08-18 | 2021-11-12 | 江苏科技大学 | Early fault diagnosis method for inverter circuit and motor of high-speed train |
| CN113640607B (en) * | 2021-08-18 | 2023-02-28 | 江苏科技大学 | Early fault diagnosis method for inverter circuit and motor of high-speed train |
| CN114355850A (en) * | 2021-12-28 | 2022-04-15 | 汉谷云智(武汉)科技有限公司 | Atmospheric and vacuum pressure device fault diagnosis method based on queue competition algorithm |
| CN114355850B (en) * | 2021-12-28 | 2023-06-20 | 汉谷云智(武汉)科技有限公司 | Atmospheric and vacuum device fault diagnosis method based on queuing competition algorithm |
| CN115146191A (en) * | 2022-07-21 | 2022-10-04 | 北京天防安全科技有限公司 | Method and device for identifying video monitoring assets based on AI (Artificial Intelligence) and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112784862A (en) | Fault diagnosis and identification method for refining process of atmospheric and vacuum distillation unit | |
| CN109829402B (en) | GS-SVM-based bearing damage degree diagnosis method under different working conditions | |
| CN108762228B (en) | Distributed PCA-based multi-working-condition fault monitoring method | |
| Bevilacqua et al. | Classification and class-modelling | |
| Monroy et al. | A semi-supervised approach to fault diagnosis for chemical processes | |
| CN108388234B (en) | Fault monitoring method based on relevance division multi-variable block PCA model | |
| CN111208793B (en) | State monitoring method of non-stationary industrial process based on slow characteristic analysis | |
| Li et al. | Rotating machine fault diagnosis using dimension reduction with linear local tangent space alignment | |
| CN111898690B (en) | Power transformer fault classification method and system | |
| CN112904810B (en) | Process industry nonlinear process monitoring method based on effective feature selection | |
| CN114637263B (en) | Abnormal working condition real-time monitoring method, device, equipment and storage medium | |
| CN110427019B (en) | Industrial process fault classification method and control device based on multivariate discriminant analysis | |
| Gallos et al. | ISOMAP and machine learning algorithms for the construction of embedded functional connectivity networks of anatomically separated brain regions from resting state fMRI data of patients with Schizophrenia | |
| CN113935535A (en) | Principal component analysis method for medium-and-long-term prediction model | |
| CN111062848A (en) | Intelligent monitoring method for monitoring abnormal state of fire-fighting engineering | |
| Wang et al. | A hybrid approach for identification of concurrent control chart patterns | |
| CN111912638B (en) | Rectifying tower fault diagnosis method for online fault source identification | |
| CN115983534A (en) | Method and system for evaluating state of sewage treatment process | |
| CN115952432B (en) | Unsupervised clustering method based on diabetes data | |
| CN114118292B (en) | Fault classification method based on linear discriminant neighborhood preserving embedding | |
| CN116776181A (en) | Terminal side load identification method, medium and system based on improved fuzzy clustering | |
| CN116661410A (en) | Large-scale industrial process fault detection and diagnosis method based on weighted directed graph | |
| CN115798685A (en) | Depression diet management method based on food image segmentation | |
| CN116522993A (en) | Chemical process fault detection method based on countermeasure self-coding network | |
| Sena et al. | Multivariate statistical analysis and chemometrics |
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 |