CN114841295A - Industrial process fault detection method and device, electronic equipment and storage medium - Google Patents
Industrial process fault detection method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN114841295A CN114841295A CN202210776318.3A CN202210776318A CN114841295A CN 114841295 A CN114841295 A CN 114841295A CN 202210776318 A CN202210776318 A CN 202210776318A CN 114841295 A CN114841295 A CN 114841295A
- Authority
- CN
- China
- Prior art keywords
- vector
- monitoring
- moment
- value
- filtering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 85
- 238000001514 detection method Methods 0.000 title claims abstract description 41
- 238000012544 monitoring process Methods 0.000 claims abstract description 147
- 238000000034 method Methods 0.000 claims abstract description 93
- 238000001914 filtration Methods 0.000 claims abstract description 66
- 239000011159 matrix material Substances 0.000 claims abstract description 53
- 238000004364 calculation method Methods 0.000 claims abstract description 35
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 22
- 238000004590 computer program Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 9
- 238000004451 qualitative analysis Methods 0.000 abstract description 3
- 238000007405 data analysis Methods 0.000 abstract description 2
- 238000000513 principal component analysis Methods 0.000 description 23
- 239000007788 liquid Substances 0.000 description 13
- 229910052500 inorganic mineral Inorganic materials 0.000 description 12
- 239000011707 mineral Substances 0.000 description 12
- 238000010586 diagram Methods 0.000 description 7
- 230000002159 abnormal effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000011158 quantitative evaluation Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000002002 slurry Substances 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000005188 flotation Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Evolutionary Computation (AREA)
- Mathematical Optimization (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The embodiment of the application provides an industrial process fault detection method, an industrial process fault detection device, electronic equipment and a storage medium, which are applied to the technical field of data analysis and are used for solving the problems that the existing fault detection based on convex hull PCA only provides qualitative analysis and the process monitoring is inconvenient. The method comprises the following steps: performing convex hull PCA calculation on the training data set to obtain a first parameter set and obtain a second parameter set; determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor; determining the monitoring quantity at the k moment according to the upper limit of the monitoring quantity and the maximum value of the elements of the final reference vector; filtering the monitoring quantity at the k moment according to the filter coefficient; and if the filtering value at the moment k is greater than the monitoring threshold, determining that the production fault exists at the moment k. Therefore, the monitoring amount is calculated based on the convex hull PCA, the monitoring amount is subjected to filtering processing to obtain a filtering value, whether the production process breaks down or not is judged based on the filtering value, and convenience is brought to monitoring and operation.
Description
Technical Field
The present application relates to the field of data analysis technologies, and in particular, to a method and an apparatus for detecting a fault in an industrial process, an electronic device, and a storage medium.
Background
The production process of the process industry generally has the dangerous characteristics of high temperature, high pressure, flammability, explosiveness and the like, and serious consequences can be caused if faults are not found in time in the production process. The process fault detection method based on convex hull PCA is suitable for fault detection of non-normal characteristic data, has good data adaptability and is widely applied in the industry. The existing fault detection based on convex hull PCA can only provide qualitative analysis, but can not quantitatively evaluate the normal degree, abnormal degree and trend, which is not beneficial to the monitoring and operation of operators.
Disclosure of Invention
In order to solve the technical problem, embodiments of the present application provide a method and an apparatus for detecting a fault in an industrial process, an electronic device, and a storage medium.
In a first aspect, an embodiment of the present application provides a method for detecting a fault in an industrial process, where the method includes:
performing convex hull PCA calculation on a pre-acquired training data set to obtain a first parameter set, wherein the first parameter set at least comprises a load matrix, a convex hull hyperplane normal vector matrix and an offset vector;
acquiring a second parameter group, wherein the second parameter group at least comprises a relaxation factor, a monitoring amount upper limit and a filter coefficient;
calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k;
determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor;
acquiring the element maximum value of the final reference vector, and determining the monitoring amount at the k moment according to the element maximum value and the monitoring amount upper limit;
filtering the monitored quantity at the k moment according to the filter coefficient to obtain a filtered value at the k moment;
judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, wherein the monitoring threshold value is determined according to the upper limit of the monitoring amount;
and if the filtering value at the moment k is greater than the monitoring threshold value, determining that the production fault exists at the moment k.
In one embodiment, the determining the monitoring amount at the time k according to the maximum value of the element and the upper limit of the monitoring amount includes:
calculating the k moment monitoring quantity according to the following formula;
wherein t (k) represents the monitored quantity at the time k, a represents the upper limit of the monitored quantity, and d (k) represents the maximum value of the elements of the reference vector.
In an embodiment, the filtering the monitored quantity at the time k according to the filter coefficient to obtain a filtered value at the time k includes:
calculating the k moment filtering value according to the following formula;
wherein,
represents the filtered value at the time instant k,
representing the filtered value at time k-1,
represents the monitoring amount at the k time, and α represents the filter coefficient.
In one embodiment, the determining a final reference vector according to the pivot vector, the convex hull hyperplane normal vector matrix, the offset vector, and the relaxation factor includes:
determining the final reference vector according to the following formula;
wherein w (k) represents the final reference vector, t (k) represents the pivot vector, V represents the convex hull hyperplane normal vector matrix, b represents the offset vector, epsilon represents the relaxation factor, epsilon >0, I represents a vector with each element being 1, epsilon I represents a relaxation vector, wherein the number of elements of the relaxation vector and the offset vector is the same.
In an embodiment, the calculating a principal component vector according to the load matrix and a pre-obtained sample vector to be analyzed at time k includes:
calculating the pivot vector according to the following formula;
wherein,
representing the vector of the pivot element(s),
representing the sample vector to be analyzed, P representing the load matrix.
In one embodiment, the obtaining of the training data set comprises:
obtaining multi-dimensional historical data of normal working conditions;
carrying out normalization calculation on the multi-dimensional historical data to obtain the training data set;
the obtaining of the sample vector to be analyzed comprises:
collecting process data at the k moment, and performing normalization calculation on the process data at the k moment to obtain the sample vector to be analyzed, wherein the dimension number of the process data at the k moment is the same as the dimension number of the multi-dimensional historical data.
In an embodiment, the method further comprises:
and if the filtering value at the moment k is less than or equal to the monitoring threshold, determining that the production at the moment k is normal.
In a second aspect, an embodiment of the present application provides an industrial process fault detection apparatus, including:
the first calculation module is used for performing convex hull PCA calculation on a pre-acquired training data set to obtain a first parameter set, wherein the first parameter set at least comprises a load matrix, a convex hull hyperplane normal vector matrix and an offset vector;
a first obtaining module, configured to obtain a second parameter set, where the second parameter set at least includes a relaxation factor, a monitoring amount upper limit, and a filter coefficient;
the second calculation module is used for calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k;
a determining module, configured to determine a final reference vector according to the pivot vector, the convex hull hyperplane normal vector matrix, the offset vector, and the relaxation factor;
the second obtaining module is used for obtaining the element maximum value of the final reference vector and determining the monitoring quantity at the k moment according to the element maximum value and the monitoring quantity upper limit;
the filtering processing module is used for carrying out filtering processing on the k moment monitoring quantity according to the filtering coefficient to obtain a k moment filtering value;
the judgment module is used for judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, and the monitoring threshold value is determined according to the upper limit of the monitoring amount;
and the fault capturing module is used for determining that a production fault exists at the moment k if the filtered value at the moment k is greater than the monitoring threshold value.
In a third aspect, an embodiment of the present application provides an electronic device, which includes a memory and a processor, where the memory is used to store a computer program, and the computer program executes the industrial process fault detection method provided in the first aspect when the processor runs.
In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program, and the computer program, when running on a processor, executes the method for detecting industrial process faults provided in the first aspect.
According to the industrial process fault detection method, the industrial process fault detection device, the electronic equipment and the storage medium, convex hull PCA calculation is carried out on a pre-acquired training data set to obtain a first parameter set, and a second parameter set is obtained; calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k; determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor; acquiring the element maximum value of the final reference vector, and determining the monitoring amount at the k moment according to the element maximum value and the monitoring amount upper limit; filtering the monitored quantity at the k moment according to the filter coefficient to obtain a filtered value at the k moment; judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, wherein the monitoring threshold value is determined according to the upper limit of the monitoring amount; and if the filtering value at the moment k is greater than the monitoring threshold value, determining that the production fault exists at the moment k. Therefore, the monitoring quantity of the principal component convex hull detection is constructed on the basis of the convex hull PCA calculation, the monitoring quantity can be limited within a certain range, the monitoring quantity is subjected to filtering processing to obtain a filtering value, whether a fault occurs in the production process is judged based on the filtering value, the quantitative evaluation of the current production condition by an operator is facilitated, and convenience is brought to monitoring and operation.
Drawings
In order to more clearly explain the technical solutions of the present application, the drawings needed to be used in the embodiments are briefly introduced below, and it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope of protection of the present application. Like components are numbered similarly in the various figures.
FIG. 1 illustrates a flow diagram of a method for industrial process fault detection provided by an embodiment of the present application;
FIG. 2 shows a process schematic of a mineral processing grinding classification process provided by an embodiment of the present application;
FIG. 3 illustrates another schematic flow diagram of a method for industrial process fault detection provided by an embodiment of the present application;
FIG. 4 is a timing diagram illustrating the filtering capacity provided by the embodiment of the present application;
fig. 5 shows a schematic structural diagram of an industrial process fault detection device provided by an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments.
The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.
Hereinafter, the terms "including", "having", and their derivatives, which may be used in various embodiments of the present application, are intended to indicate only specific features, numbers, steps, operations, elements, components, or combinations of the foregoing, and should not be construed as first excluding the existence of, or adding to, one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing.
Furthermore, the terms "first," "second," "third," and the like are used solely to distinguish one from another and are not to be construed as indicating or implying relative importance.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the present application belong. The terms (such as those defined in commonly used dictionaries) should be interpreted as having a meaning that is consistent with their contextual meaning in the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein in various embodiments.
The production process of the process industry generally has the dangerous characteristics of high temperature, high pressure, flammability, explosiveness and the like, if faults occur in the production process and cannot be found and processed in time, the production efficiency is reduced, the product quality is unqualified and other adverse effects are likely to be caused, and even the production safety problem is caused, so the real-time fault monitoring technology of the production process is strongly required by the industry. The fault monitoring is a technology for analyzing process data on line so as to determine whether process conditions are changed from a normal state to a fault state, and the technology has important significance for ensuring safe and stable operation of an industrial process.
Fault detection based on PCA (principal Component Analysis) is widely applied in the industry, and a kernel PCA method and the like are developed for nonlinear characteristics of industrial processes. However, the existing scheme has the following 3 problems in the application process: (1) monitoring statistic T 2 The threshold value of (2) depends on the normal characteristic of the pivot score vector, and many process data in the practical application process cannot meet the normal characteristic; (2) the selection and parameter adjustment of the kernel function of the kernel PCA method are complex, and an ideal monitoring effect is not easy to achieve; (3) monitoring statistic T when fault occurs 2 The value of the Square Prediction Error (SPE) is sometimes abnormal and far exceeds the threshold, which is not favorable for monitoring the human-machine interface of the statistic in the Discrete Control System (DCS)The time sequence display of (Human Machine Interface, HMI) pictures brings inconvenience to the monitoring and operation of operators. The prior art also provides a convex hull PCA method, and the detection method aiming at the principal component convex hull does not construct monitoring quantity, only can analyze whether a fault occurs or not, only provides qualitative analysis, but cannot quantitatively evaluate the normal degree, the abnormal degree and the trend, and is not beneficial to monitoring and operation of an operator.
Example 1
The embodiment of the disclosure provides an industrial process fault detection method.
Specifically, referring to fig. 1, the industrial process fault detection method includes:
step S101, convex hull PCA calculation is carried out on a pre-acquired training data set to obtain a first parameter set.
The industrial process fault detection method provided by the embodiment can be applied to the production process of the process industry, and real-time data of the production process is monitored. The process industry can be a variety of process flows, for example, the industrial process can be a mineral processing grinding classification process. Specifically, fig. 2 is a schematic diagram of a mineral processing grinding classification process. As shown in figure 2, the equipment involved in the mineral processing grinding classification process comprises a grinding pump pool 201, a slurry pump 202, a ball mill 203, raw ore 204 and a hydrocyclone unit 205. The mineral processing grinding classification process relates to the following treatment processes: supplementing water to the ore grinding pump pool 201, regulating the speed by the slurry pump 202, grinding and beating raw ores by the ball mill 203, and performing a flotation removal process by the hydrocyclone set 205, wherein the process parameters related in the process at least comprise the following parameters: the system comprises a pump pool liquid level LIT, a swirler feeding flow FIT, a swirler feeding concentration DIT, a swirler feeding pressure PIT and a swirler feeding pump frequency feedback SIC. The industrial process fault detection method provided by the embodiment can be used for monitoring the process parameters.
In one embodiment, the obtaining of the training data set comprises:
obtaining multi-dimensional historical data of normal working conditions;
and carrying out normalization calculation on the multi-dimensional historical data to obtain the training data set.
In this embodiment, the multidimensional historical data under the normal condition is determined according to an actual industrial process, for example, as shown in fig. 2, in the mineral processing and grinding classification process, the multidimensional historical data under the normal condition includes historical data such as a pump pool liquid level, a cyclone ore feeding flow rate, a cyclone ore feeding concentration, a cyclone ore feeding pressure, and a cyclone ore feeding pump frequency feedback.
In this embodiment, the multidimensional historical data may be normalized according to the mean and standard deviation of the multidimensional historical data. For example, if the length L =7000 of the pump pool liquid level data under the normal operating condition is used, then the mean value and the standard deviation of the 7000 pump pool liquid level data are calculated, the mean value is subtracted from each pump pool liquid level data to obtain a difference value, a quotient obtained by dividing the difference value by the standard deviation is used as the normalized pump pool liquid level corresponding to each pump pool liquid level data, and the normalized calculation processes of other parameters are the same and are not described herein.
In this embodiment, the process of performing convex hull PCA calculation on the training data set may adopt the existing convex hull PCA method for calculation, which is not described herein again. The first parameter group obtained by carrying out convex hull PCA calculation on the training data set at least comprises the number n of principal elements and a load matrixPConvex hull hyperplane normal vector matrixVAnd an offset vectorb。
Step S102, a second parameter group is obtained.
In this embodiment, the second parameter set at least includes a relaxation factor epsilon, a monitoring amount upper limit a, and a filter coefficient alpha. Specifically, the relaxation factor e, the upper limit a of the monitoring amount, and the filter coefficient α may be set according to an actual industrial process, for example, in the mineral processing grinding classification process, the relaxation factor e =1 × 10 is set -12 If the upper limit of the monitored amount is a =200, the monitoring threshold th = a/2=100, and the filter coefficient α = 0.95. In other practical industrial processes, the relaxation factor epsilon, the upper limit of the monitoring quantity A and the filter coefficient alpha can be set to be matched, but the condition that epsilon is required to be satisfied>0、A>0、0<α<1, are not limited herein.
And step S103, calculating a principal component vector according to the load matrix and the pre-acquired sample vector to be analyzed at the moment k.
In one embodiment, the obtaining of the sample vector to be analyzed includes:
collecting process data at the k moment, and performing normalization calculation on the process data at the k moment to obtain the sample vector to be analyzed, wherein the dimension number of the process data at the k moment is the same as the dimension number of the multi-dimensional historical data.
It should be noted that, collecting the process data at the time k may be understood as monitoring the production process on line, and collecting the data at the time k of the industrial process on line. The industrial process fault detection method provided by the embodiment can be applied to electronic equipment. In particular, control system data for an industrial process may be collected online. The electronic device can establish communication connection with the control system first, and process parameters at the current moment are collected from the control system. The Control System here may be a Distributed Control System (DCS).
For example, in the mineral processing grinding classification process shown in fig. 2, the k-time process data is collected, and at this time, the k-time process data includes the pump pool liquid level LIT, the cyclone ore feeding flow rate FIT, the cyclone ore feeding concentration DIT, the cyclone ore feeding pressure PIT, and the cyclone ore feeding pump frequency feedback SIC at the k-time.
In one embodiment, the real-time data may be normalized according to the mean and standard deviation of the multidimensional historical data. Specifically, the real-time data is subtracted from the mean value to obtain a difference value, and the difference value is divided by the standard deviation to obtain normalized real-time data.
In this embodiment, the principal component vector may be obtained by multiplying the sample vector to be analyzed by the load matrix. Specifically, step S103 includes:
calculating the pivot vector according to the following formula;
wherein,
representing the vector of the pivot element(s),
representing the sample vector to be analyzed, P representing the load matrix.
And step S104, determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor.
In the present embodiment, in order to prevent the normal data adjacent to the convex hull hyperplane from being erroneously determined as the failure data, a relaxation factor epsilon is introduced to appropriately expand the convex hull space. Vector of principal elements
And multiplying the vector matrix V of the convex hull hyperplane method to obtain a first vector, multiplying the relaxation factors epsilon and I to obtain a second vector, wherein I represents a vector with all elements of 1, and adding the first vector, the second vector and the offset vector b to obtain a final reference vector.
Specifically, step S104 includes:
determining the final reference vector according to the following formula;
wherein w (k) represents the final reference vector, t (k) represents the pivot vector, V represents the convex hull hyperplane normal vector matrix, b represents the offset vector, epsilon represents the relaxation factor, epsilon >0, I represents a vector with elements of 1, epsilon I represents a relaxation vector, wherein the number of elements of the relaxation vector and the offset vector is the same.
And step S105, acquiring the element maximum value of the final reference vector, and determining the monitoring amount at the k moment according to the element maximum value and the monitoring amount upper limit.
In this embodiment, the maximum value of the elements of the final reference vector may be determined according to the following formula:
wherein,
representing element maximum values, t (k) representing the principal component vector, V representing the convex hull hyperplane normal vector matrix, b representing the offset vector, epsilon representing the relaxation factor, I representing a vector with elements of 1, epsilon I representing a relaxation vector, and max representing the element maximum value of the selected reference vector W (k).
In this embodiment, the maximum value of the element can be passed
And (4) calculating the monitoring quantity of the industrial process at the k moment, wherein the calculated monitoring quantity is not more than the upper limit A of the monitoring quantity.
Specifically, the determining the monitoring amount at the k moment according to the maximum value of the element and the upper limit of the monitoring amount includes:
calculating the k moment monitoring quantity according to the following formula;
wherein t (k) represents the monitored quantity at the time k, a represents the upper limit of the monitored quantity, and d (k) represents the maximum value of the elements of the reference vector. It should be noted that, in the above formula, the upper limit a of the monitoring amount is greater than 0, in the above formula, e represents a natural constant,
if the value of (a) is greater than 1, then T (k) will be less than A.
Therefore, the monitoring amount can not exceed the upper limit of the monitoring amount, and the monitoring amount can be conveniently and visually displayed on a human-computer interaction interface.
And step S106, filtering the monitored quantity at the k moment according to the filter coefficient to obtain a filtered value at the k moment.
In this embodiment, the filter value at time k may be calculated according to the monitored quantity at time k, the filter value at time k-1, and the filter coefficient. The filter value at the moment k-1 is the filter value at the last moment of the moment k in the filter value time sequence, the time interval between the moments k and k-1 can be a monitoring period, and the monitoring period can be set as required, for example, the monitoring period is 1 second, the filter value at the moment k is the filter value at 1 point, 20 minutes and 30 seconds, the filter value at the moment k-1 can be understood as the filter value at 1 point, 20 minutes and 29 seconds in the filter value time sequence, and the filter value at the moment k +1 can be understood as the filter value at 1 point, 20 minutes and 31 seconds in the filter value time sequence. The monitoring period may also have other values, and is not limited herein.
In one embodiment, step S106 includes: the filtering processing is performed on the monitored quantity at the k moment according to the filter coefficient to obtain a filter value at the k moment, and the method comprises the following steps:
calculating the k moment filtering value according to the following formula;
wherein,
represents the filtered value at the time instant k,
representing the filtered value at time k-1,
represents the monitoring amount at the k time, and α represents the filter coefficient.
In this embodiment, the calculated filter values may be sorted according to a time sequence to obtain a filter value time sequence, so that the filter value at the later k time may be calculated based on the filter value at the k-1 time.
And S107, judging whether the filtering value at the moment k is greater than a pre-acquired monitoring threshold, wherein the monitoring threshold is determined according to the upper limit of the monitoring amount.
In this embodiment, if the filtered value at time k is greater than the monitoring threshold, step S108 is executed. And if the filtering value at the moment k is less than or equal to the monitoring threshold, indicating that the production at the moment k is not abnormal, acquiring real-time data at the next moment, and performing the next period monitoring calculation of the industrial process.
It should be added that the monitoring threshold may be determined according to the following formula;
where th denotes a monitoring threshold value, and a denotes an upper limit of the monitoring amount.
And S108, if the filtered value at the moment k is greater than the monitoring threshold, determining that a production fault exists at the moment k.
In this embodiment, data such as the k moment monitoring amount and the k moment filtering value can be displayed on a human-computer interaction interface, a user can conveniently and directly check real-time data, and after a production fault is determined to exist at the k moment, an alarm is further triggered, and a warning sound can be sent by an audible alarm to achieve the purpose of quick reminding. And after the production fault exists at the moment k, a fault output instruction can be generated, and the fault output instruction controls triggering alarm. In addition, an alarm signal can be generated and transmitted to the control system, so that the control system can adopt a corresponding fault avoidance strategy according to the alarm signal.
In an embodiment, if the filtered value at time k is less than or equal to the monitoring threshold, it is determined that production at time k is normal.
Referring to fig. 3, fig. 3 is another schematic flow chart of the industrial process fault detection method according to the embodiment of the present disclosure. Specifically, the method for detecting faults in the industrial process shown in fig. 3 is applied to a mineral processing, grinding and classifying process, and the process parameters of the mineral processing, grinding and classifying process include 5 parameters, namely, a pump pool liquid level, a cyclone ore feeding flow rate, a cyclone ore feeding concentration, a cyclone ore feeding pressure and a cyclone ore feeding pump frequency feedback. The method is characterized in that historical data of pump pool liquid level, cyclone ore feeding flow, cyclone ore feeding concentration, cyclone ore feeding pressure and cyclone ore feeding pump frequency feedback in the mineral processing ore grinding classification process under normal working conditions are collected, for example, the length L =7000 of the collected normal working condition data is required.
Fig. 3 shows a method for detecting a fault in an industrial process according to an embodiment of the present application, which includes the following steps:
step S301, off-line training is carried out to obtain n, P, V and b.
In step S301, normalization calculation is performed on the collected historical data of the normal operating condition to obtain a training data set, and convex hull PCA calculation is performed on the training data set. Obtaining the number n =2 of the principal elements to obtain a load matrix P, wherein the P belongs to R 5×2 Obtaining the convex hull hyperplane normal vector matrix V and the offset vector b, V ∈ R 17×2 、b∈R 17 . In P ∈ R 5×2 In, R 5×2 And the real number matrix represents 5 rows and 2 columns, 5 is determined by the quantity of 5 process parameters including pump pool liquid level, cyclone ore feeding flow, cyclone ore feeding concentration, cyclone ore feeding pressure and cyclone ore feeding pump frequency feedback, and 2 is determined by the number n =2 of principal elements. In V ∈ R 17×2 In, R 17×2 A real matrix representing 17 rows and 2 columns, 17 being calculated from convex hull PCA, 2 being determined by the number of principal elements n = 2. In b ∈ R 17 In, R 17 A vector of real numbers representing 17 rows and 1 columns, 17 being calculated from convex hull PCA.
In this embodiment, the mean and the standard deviation according to the historical data may be normalized, and the specific calculation process is referred to the related contents, which is not described herein again.
Step S302, initializing, setting epsilon, A and alpha, and calculating a monitoring threshold th.
Specifically, the relaxation factor e =1 × 10 is set -12 If the upper limit of the monitored amount is a =200, the monitoring threshold th = a/2=100, and the filter coefficient α = 0.95.
Step S303, collecting data at the time k from the control system, normalizing the data at the time k, and obtaining a principal component vector t (k).
In this embodiment, the time data may be normalized according to the mean and standard deviation of the historical data to obtain a sample vector to be analyzed
Calculating a principal component vector
And k represents the time of k in the time series,
representing the principal component vector at time k. In that
In, R 5 And the real number matrix represents 5 rows and 1 columns, and 5 is determined by the quantity of 5 process parameters including pump pool liquid level, cyclone ore feeding flow, cyclone ore feeding concentration, cyclone ore feeding pressure and cyclone ore feeding pump frequency feedback. In that
In, R 2 A real number matrix of 2 rows and 1 column is represented, 2 being determined by the number of pivot elements n = 2.
Step S304, d (k) is calculated.
In this embodiment, d (k) is calculated as follows;
wherein,
representing the maximum value of elements of a reference vector, t (k) representing the principal component vector, V representing the convex hull hyperplane normal vector matrix, b representing the offset vector, epsilon representing the relaxation factor, I representing the element as a 1 vector, epsilon I representing the relaxation vector, and max representing the maximum value of elements of the selected reference vector W (k). Wherein the relaxation vector ε I ∈ R 5 In the relaxation vector ε I ∈ R 5 And 5 is determined by the quantity of 5 process parameters of pump pool liquid level, cyclone ore feeding flow, cyclone ore feeding concentration, cyclone ore feeding pressure and cyclone ore feeding pump frequency feedback.
In step S305, the monitored quantity T (k) is calculated.
Specifically, the monitored quantity t (k) is calculated according to the following formula;
wherein t (k) represents the monitored quantity at the time k, a represents the upper limit of the monitored quantity, and d (k) represents the maximum value of the element. It should be noted that, in the above formula, the upper limit a of the monitoring amount is greater than 0, in the above formula, e represents a natural constant,
if the calculation result of (b) is a value greater than 1, T (k) is less than A and satisfies
。
In step S306, a filtered value is calculated and stored.
Specifically, the filter value at the time k is calculated according to the following formula;
wherein,
represents the filtered value at the time instant k,
representing the filtered value at time k-1,
represents the monitoring amount at the k time, and α represents the filter coefficient.
Step S307, it is determined whether the filtered value is greater than the monitoring threshold.
In this embodiment, if the filtered value is less than or equal to the monitoring threshold, the process returns to step S303, and if the filtered value is greater than the monitoring threshold, the process returns to step S308.
And step S308, writing the alarm signal back to the control system.
In the specific implementation process, if the industrial process fault detection method continues, the process may continue to step S303, otherwise, the process ends.
Through the industrial process fault detection method provided by the embodiment, online abnormal fault monitoring can be performed on the mineral processing ore grinding classification process, and fault early warning prompts can be provided for operators in time. The monitoring quantity is displayed in real time, and the value is not higher than 200, so that an operator can conveniently observe the timing chart of the monitoring quantity based on a human-computer interface picture of the discrete control system.
Referring to fig. 4, fig. 4 is a timing chart of filtering amount provided by the embodiment of the present application, in which fig. 4 shows a monitoring threshold line L1 and a filtering amount curve L2, a vertical axis of the monitoring threshold line L1 corresponds to 100, a filtering amount of a front portion of the filtering amount curve L2 from 0 to 400 is less than or equal to 100, which indicates that the control system is in a normal state, a filtering amount of a rear portion of the filtering amount curve L2 after 400 is greater than 100, which indicates that the control system is in a fault state, and an intersection point D of the monitoring threshold line L1 and the filtering amount curve L2 is a transition point between the fault state and the normal state. The user can observe through the filtering capacity time sequence diagram conveniently, whether breaks down in the production process is judged based on the filtering capacity time sequence diagram, the operator can be greatly facilitated to carry out quantitative evaluation on the current production condition, and convenience is brought to monitoring and operation.
In the method for detecting the industrial process fault, convex hull PCA calculation is performed on a pre-acquired training data set to obtain a first parameter set, and a second parameter set is obtained; calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k; determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor; acquiring the element maximum value of the final reference vector, and determining the monitoring amount at the k moment according to the element maximum value and the monitoring amount upper limit; filtering the monitored quantity at the k moment according to the filter coefficient to obtain a filtered value at the k moment; judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, wherein the monitoring threshold value is determined according to the upper limit of the monitoring amount; and if the filtering value at the moment k is greater than the monitoring threshold value, determining that the production fault exists at the moment k. The monitoring quantity of the principal component convex hull detection is constructed on the basis of the convex hull PCA calculation, the monitoring quantity can be limited within a certain range, the monitoring quantity is subjected to filtering processing to obtain a filtering value, whether a fault occurs in the production process is judged based on the filtering value, the quantitative evaluation of the current production condition by an operator is facilitated, and convenience is brought to monitoring and operation.
Example 2
In addition, the embodiment of the disclosure provides an industrial process fault detection device.
Specifically, as shown in fig. 5, the industrial process fault detection apparatus 500 includes:
the first calculation module 501 is configured to perform convex hull PCA calculation on a pre-acquired training data set to obtain a first parameter set, where the first parameter set at least includes a load matrix, a convex hull hyperplane normal vector matrix, and an offset vector;
a first obtaining module 502, configured to obtain a second parameter set, where the second parameter set at least includes a relaxation factor, a monitoring amount upper limit, and a filter coefficient;
a second calculating module 503, configured to calculate a principal component vector according to the load matrix and a pre-obtained sample vector to be analyzed at time k;
a determining module 504, configured to determine a final reference vector according to the pivot vector, the convex hull hyperplane normal vector matrix, the offset vector, and the relaxation factor;
a second obtaining module 505, configured to obtain an element maximum value of the final reference vector, and determine a monitoring amount at the k time according to the element maximum value and the monitoring amount upper limit;
a filtering processing module 506, configured to perform filtering processing on the monitored quantity at the time k according to the filtering coefficient to obtain a filtering value at the time k;
a judging module 507, configured to judge whether the filter value at the time k is greater than a pre-obtained monitoring threshold, where the monitoring threshold is determined according to the upper limit of the monitoring amount;
and a fault capture module 508, configured to determine that a production fault exists at the time k if the filtered value at the time k is greater than the monitoring threshold.
In an embodiment, the second obtaining module 505 is further configured to calculate the monitoring amount at the time k according to the following formula;
wherein t (k) represents the monitored quantity at the time k, a represents the upper limit of the monitored quantity, and d (k) represents the maximum value of the element.
In an embodiment, the filtering processing module 506 is further configured to calculate the filtered value at time k according to the following formula;
wherein,
represents the filtered value at the time instant k,
representing the filtered value at time k-1,
represents the monitoring amount at the k time, and α represents the filter coefficient.
In an embodiment, the determining module 504 is further configured to determine the final reference vector according to the following formula;
wherein w (k) represents the final reference vector, t (k) represents the pivot vector, V represents the convex hull hyperplane normal vector matrix, b represents the offset vector, epsilon represents the relaxation factor, and I represents a vector whose elements are all 1, wherein the number of elements of the relaxation vector and the offset vector is the same.
In an embodiment, the second calculating module 503 is further configured to calculate the pivot vector according to the following formula;
wherein,
representing the vector of the pivot element(s),
representing the sample vector to be analyzed, P representing the load matrix.
In an embodiment, the first calculating module 501 is further configured to obtain multidimensional historical data of normal operating conditions;
carrying out normalization calculation on the multi-dimensional historical data to obtain the training data set;
the second calculating module 503 is further configured to collect process data at a time k, and perform normalization calculation on the process data at the time k to obtain the sample vector to be analyzed, where the number of dimensions of the process data at the time k is the same as the number of dimensions of the multidimensional historical data.
In an embodiment, the fault capture module 508 is further configured to determine that the production is normal at time k if the filtered value at time k is less than or equal to the monitoring threshold.
The industrial process fault detection apparatus 500 provided in this embodiment can implement the industrial process fault detection method provided in embodiment 1, and is not described herein again to avoid repetition.
The industrial process fault detection device provided by this embodiment performs convex hull PCA calculation on a pre-acquired training data set to obtain a first parameter set, and obtains a second parameter set; calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k; determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor; acquiring the element maximum value of the final reference vector, and determining the monitoring amount at the k moment according to the element maximum value and the monitoring amount upper limit; filtering the monitored quantity at the k moment according to the filter coefficient to obtain a filtered value at the k moment; judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, wherein the monitoring threshold value is determined according to the upper limit of the monitoring amount; and if the filtering value at the moment k is greater than the monitoring threshold value, determining that the production fault exists at the moment k. The monitoring quantity of the principal component convex hull detection is constructed on the basis of the convex hull PCA calculation, the monitoring quantity can be limited within a certain range, the monitoring quantity is subjected to filtering processing to obtain a filtering value, whether a fault occurs in the production process is judged based on the filtering value, the quantitative evaluation of the current production condition by an operator is facilitated, and convenience is brought to monitoring and operation.
Example 3
Furthermore, an embodiment of the present disclosure provides an electronic device, including a memory and a processor, where the memory stores a computer program, and the computer program executes the industrial process fault detection method provided in embodiment 1 when running on the processor.
The electronic device provided in this embodiment can implement the industrial process fault detection method provided in embodiment 1, and is not described herein again to avoid repetition.
Example 4
In addition, the present application also provides a computer readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the industrial process fault detection method provided in embodiment 1.
In this embodiment, the computer-readable storage medium may be a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk.
The computer-readable storage medium provided in this embodiment can implement the industrial process fault detection method provided in embodiment 1, and is not described herein again to avoid repetition.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present application.
While the present embodiments have been described with reference to the accompanying drawings, it is to be understood that the invention is not limited to the precise embodiments described above, which are meant to be illustrative and not restrictive, and that various changes may be made therein by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A method of industrial process fault detection, the method comprising:
performing convex hull PCA calculation on a pre-acquired training data set to obtain a first parameter set, wherein the first parameter set at least comprises a load matrix, a convex hull hyperplane normal vector matrix and an offset vector;
acquiring a second parameter group, wherein the second parameter group at least comprises a relaxation factor, a monitoring quantity upper limit and a filter coefficient;
calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k;
determining a final reference vector according to the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector and the relaxation factor;
acquiring the element maximum value of the final reference vector, and determining the monitoring amount at the k moment according to the element maximum value and the monitoring amount upper limit;
filtering the monitored quantity at the k moment according to the filter coefficient to obtain a filtered value at the k moment;
judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, wherein the monitoring threshold value is determined according to the upper limit of the monitoring amount;
and if the filtering value at the moment k is greater than the monitoring threshold value, determining that the production fault exists at the moment k.
2. The method of claim 1, wherein determining the amount of monitoring at time k based on the element maximum and the upper limit of the amount of monitoring comprises:
calculating the k moment monitoring quantity according to the following formula;
wherein t (k) represents the monitored quantity at the time k, a represents the upper limit of the monitored quantity, and d (k) represents the maximum value of the elements of the reference vector.
3. The method of claim 1, wherein the filtering the monitored amount at time k according to the filter coefficient to obtain a filtered value at time k comprises:
calculating the k moment filtering value according to the following formula;
wherein,
represents the filtered value at the time instant k,
representing the filtered value at time k-1,
represents the monitoring amount at the k time, and α represents the filter coefficient.
4. The method of claim 1, wherein determining a final reference vector based on the principal component vector, the convex hull hyperplane normal vector matrix, the offset vector, and the relaxation factor comprises:
determining the final reference vector according to the following formula;
wherein w (k) represents the final reference vector, t (k) represents the pivot vector, V represents the convex hull hyperplane normal vector matrix, b represents the offset vector, epsilon represents the relaxation factor, epsilon >0, I represents a vector whose elements are all 1, epsilon I represents a relaxation vector, wherein the number of elements of the relaxation vector and the offset vector is the same.
5. The method according to claim 1, wherein the calculating a principal component vector according to the load matrix and a pre-obtained sample vector to be analyzed at time k comprises:
calculating the pivot vector according to the following formula;
wherein,
representing the vector of the pivot element(s),
representing the sample vector to be analyzed, P representing the load matrix.
6. The method of claim 1, wherein the obtaining of the training data set comprises:
obtaining multi-dimensional historical data of normal working conditions;
carrying out normalization calculation on the multi-dimensional historical data to obtain the training data set;
the obtaining of the sample vector to be analyzed comprises:
collecting process data at the k moment, and performing normalization calculation on the process data at the k moment to obtain the sample vector to be analyzed, wherein the dimension number of the process data at the k moment is the same as the dimension number of the multi-dimensional historical data.
7. The method of claim 1, further comprising:
and if the filtering value at the moment k is less than or equal to the monitoring threshold, determining that the production at the moment k is normal.
8. An industrial process fault detection apparatus, characterized in that the apparatus comprises:
the first calculation module is used for performing convex hull PCA calculation on a pre-acquired training data set to obtain a first parameter set, wherein the first parameter set at least comprises a load matrix, a convex hull hyperplane normal vector matrix and an offset vector;
a first obtaining module, configured to obtain a second parameter set, where the second parameter set at least includes a relaxation factor, a monitoring amount upper limit, and a filter coefficient;
the second calculation module is used for calculating a principal component vector according to the load matrix and a pre-acquired sample vector to be analyzed at the moment k;
a determining module, configured to determine a final reference vector according to the pivot vector, the convex hull hyperplane normal vector matrix, the offset vector, and the relaxation factor;
the second obtaining module is used for obtaining the element maximum value of the final reference vector and determining the monitoring quantity at the k moment according to the element maximum value and the monitoring quantity upper limit;
the filtering processing module is used for carrying out filtering processing on the k moment monitoring quantity according to the filtering coefficient to obtain a k moment filtering value;
the judgment module is used for judging whether the filtering value at the moment k is larger than a pre-acquired monitoring threshold value, and the monitoring threshold value is determined according to the upper limit of the monitoring amount;
and the fault capturing module is used for determining that a production fault exists at the moment k if the filtered value at the moment k is greater than the monitoring threshold value.
9. An electronic device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, performs the industrial process fault detection method of any of claims 1 to 7.
10. A computer-readable storage medium, characterized in that it stores a computer program which, when run on a processor, performs the industrial process fault detection method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210776318.3A CN114841295B (en) | 2022-07-04 | 2022-07-04 | Industrial process fault detection method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210776318.3A CN114841295B (en) | 2022-07-04 | 2022-07-04 | Industrial process fault detection method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114841295A true CN114841295A (en) | 2022-08-02 |
| CN114841295B CN114841295B (en) | 2022-09-13 |
Family
ID=82575187
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210776318.3A Active CN114841295B (en) | 2022-07-04 | 2022-07-04 | Industrial process fault detection method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114841295B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106940808A (en) * | 2017-04-28 | 2017-07-11 | 宁波大学 | A kind of fault detection method based on modified Principal Component Analysis Model |
| CN107111309A (en) * | 2014-09-10 | 2017-08-29 | 西门子能源有限公司 | Utilize the combustion gas turbine failure predication of supervised learning method |
| US20180136019A1 (en) * | 2016-11-11 | 2018-05-17 | Chevron U.S.A. Inc. | Fault detection system utilizing dynamic principal components analysis |
| CN111368428A (en) * | 2020-03-04 | 2020-07-03 | 山东科技大学 | Sensor precision reduction fault detection method based on monitoring second-order statistics |
| CN111752147A (en) * | 2020-05-28 | 2020-10-09 | 山东科技大学 | Multi-working-condition process monitoring method with continuous learning capability and improved PCA (principal component analysis) |
| CN113359665A (en) * | 2021-05-31 | 2021-09-07 | 上海工程技术大学 | Industrial process fault detection method and system based on weighted key pivot |
| CN114398800A (en) * | 2022-01-27 | 2022-04-26 | 矿冶科技集团有限公司 | Fault diagnosis method and device for crusher system |
-
2022
- 2022-07-04 CN CN202210776318.3A patent/CN114841295B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107111309A (en) * | 2014-09-10 | 2017-08-29 | 西门子能源有限公司 | Utilize the combustion gas turbine failure predication of supervised learning method |
| US20180136019A1 (en) * | 2016-11-11 | 2018-05-17 | Chevron U.S.A. Inc. | Fault detection system utilizing dynamic principal components analysis |
| CN106940808A (en) * | 2017-04-28 | 2017-07-11 | 宁波大学 | A kind of fault detection method based on modified Principal Component Analysis Model |
| CN111368428A (en) * | 2020-03-04 | 2020-07-03 | 山东科技大学 | Sensor precision reduction fault detection method based on monitoring second-order statistics |
| CN111752147A (en) * | 2020-05-28 | 2020-10-09 | 山东科技大学 | Multi-working-condition process monitoring method with continuous learning capability and improved PCA (principal component analysis) |
| CN113359665A (en) * | 2021-05-31 | 2021-09-07 | 上海工程技术大学 | Industrial process fault detection method and system based on weighted key pivot |
| CN114398800A (en) * | 2022-01-27 | 2022-04-26 | 矿冶科技集团有限公司 | Fault diagnosis method and device for crusher system |
Non-Patent Citations (8)
| Title |
|---|
| BALIGH MNASSRI: "Fault Detection and Diagnosis Based on PCA and a New Contribution Plots", 《PROCEEDINGS OF THE 7TH IFAC SYMPOSIUM ON FAULT DETECTION, SUPERVISION AND SAFETY OF TECHNICAL PROCESSES》 * |
| EVAN L. RUSSELL: "Fault detection in industrial processes using canonical variate analysis and dynamic principal component analysis", 《CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS》 * |
| ZHOU BO: "Process monitoring of iron-making process in a blast furnace with PCA-based methods", 《CONTROL ENGINEERING》 * |
| 张世荣,林虞: "结合小波去噪的PCA工业过程传感器故障检测", 《自动化与仪表》 * |
| 张瀚文: "基于高斯混合模型的MWPCA高炉异常监测算法", 《化工学报》 * |
| 朱雄卓: "数据驱动的高炉炼铁过程故障监测研究综述", 《冶金自动化》 * |
| 王波: "基于规则和MWCHPCA的高炉异常炉况监测系统", 《017中国自动化大会（CAC2017）暨国际智能制造创新大会（CIMIC2017）论文集》 * |
| 祝朋艳: "基于核凸非负矩阵分解算法的故障检测方法", 《计算机应用与软件》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114841295B (en) | 2022-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100489870C (en) | Method and multidimensional system for statistical process control | |
| CN113369979B (en) | Online monitoring numerically controlled lathe cooling system based on cloud computing | |
| CN101470426B (en) | Fault detection method and system | |
| US20150346066A1 (en) | Asset Condition Monitoring | |
| CN109977146B (en) | Fault diagnosis method and device and electronic equipment | |
| CN116502925B (en) | Digital factory equipment inspection evaluation method, system and medium based on big data | |
| CN114861120A (en) | Flotation froth grade calculation method, device, electronic equipment and medium | |
| CN110000705B (en) | Detection method and system for abnormal machining operation of mill and related components | |
| CN116665401A (en) | Accident prevention alarm system for chemical production | |
| CN116430231A (en) | Battery abnormality monitoring method, device, server, medium and charging and battery changing system | |
| CN114841295B (en) | Industrial process fault detection method and device, electronic equipment and storage medium | |
| CN117112336B (en) | Intelligent communication equipment abnormality detection method, equipment, storage medium and device | |
| CN112631258B (en) | Fault early warning method for key indexes of industrial process | |
| CN113673600A (en) | Industrial signal abnormity early warning method, system, storage medium and computing equipment | |
| CN117372377A (en) | Broken line detection method and device for monocrystalline silicon ridge line and electronic equipment | |
| CN102330608B (en) | Systems and methods for determining steady state conditions in a gas turbine | |
| CN116740885A (en) | Smoke flame alarm method and device, electronic equipment and storage medium | |
| CN111458630A (en) | Fault diagnosis method for high-voltage circuit breaker | |
| CN115310475A (en) | Fault diagnosis method, device, equipment, system and storage medium of crusher | |
| CN110543830B (en) | Motion detection method, motion detection device, and storage medium | |
| CN114490829A (en) | Fault identification method, device, equipment and storage medium | |
| CN116502957B (en) | Product production quality monitoring method, system and readable storage medium | |
| JPH06281544A (en) | Plant monitor and diagnostic apparatus and abnormality indication judgment method | |
| CN104766094B (en) | A kind of recognition methods of video monitoring flame | |
| JPH06274784A (en) | Plant monitoring diagnostic device and its abnormal sign identification method |
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 |