CN113239187A - Monitoring method based on multi-level industrial structure knowledge block division - Google Patents
Monitoring method based on multi-level industrial structure knowledge block division Download PDFInfo
- Publication number
- CN113239187A CN113239187A CN202110393994.8A CN202110393994A CN113239187A CN 113239187 A CN113239187 A CN 113239187A CN 202110393994 A CN202110393994 A CN 202110393994A CN 113239187 A CN113239187 A CN 113239187A
- Authority
- CN
- China
- Prior art keywords
- level
- sub
- statistics
- principal component
- residual
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 169
- 238000000034 method Methods 0.000 title claims abstract description 82
- 238000004519 manufacturing process Methods 0.000 claims abstract description 42
- 230000002159 abnormal effect Effects 0.000 claims abstract description 30
- 239000011159 matrix material Substances 0.000 claims description 85
- 230000008569 process Effects 0.000 claims description 25
- 238000010606 normalization Methods 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000000638 solvent extraction Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 11
- 238000000354 decomposition reaction Methods 0.000 claims description 7
- 238000012806 monitoring device Methods 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 10
- 238000009826 distribution Methods 0.000 description 10
- 238000000513 principal component analysis Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000009776 industrial production Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004807 localization Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000011144 upstream manufacturing Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000001311 chemical methods and process Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000013067 intermediate product Substances 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Abstract
The invention discloses a monitoring method based on multi-level industrial structure knowledge block division, which comprises the following steps: constructing a multi-level knowledge graph of an industrial process; sub-block division is carried out on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data are obtained according to the multi-level sub-blocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds; and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks. According to the method, a multi-level knowledge graph of an industrial process is constructed by using a small amount of simple expert field knowledge, subblock division is carried out by using the nodes of the multi-level knowledge graph as a basis, the principal component statistics and the residual error statistics are finally counted according to the multi-level subblocks, and the accurate positioning of abnormal nodes is realized through a contribution graph backtracking strategy.
Description
Technical Field
The invention relates to the technical field of industrial knowledge automation, in particular to a monitoring method based on multi-level industrial structure knowledge block division.
Background
With the great progress of modern communication, computer, sensor and other technologies, most modern large-scale process production processes have characteristics which are not possessed by the traditional processes. For example, numerous operating units, numerous monitored variables, and strongly coupled variable dependencies. These special characteristics make it increasingly difficult for the prior art to ensure the safety of large-scale industrial processes.
Existing related research shows that the distributed monitoring system established by using a blocking or dispersing monitoring mode can solve the safety of large-scale process industrial production, and the strategy generally obtains better monitoring performance than the traditional single monitoring method, such as improving abnormal state detection rate, reducing complexity of the monitoring system and the like. The key points for guaranteeing the performance of the distributed monitoring strategy are as follows: the production process block division has the capability of accurately describing abnormal working conditions; the running state of each partition block is quickly and sensitively reflected in the statistical index of the global abnormal state; the complex anomalies of the global state should be decoupleable for accurate localization of the local monitoring variables.
The existing block division method is mainly carried out by adopting a data-driven method, but various operation working condition data samples in an actual industrial system are difficult to obtain, or the data samples are not complete, so that the block division method based on the data drive is difficult to be suitable for monitoring the production process of the modern large-scale process.
Thus, there is still a need for improvement and development of the prior art.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide a monitoring method based on multi-level industrial structure knowledge block division, aiming at solving the problem that the factory-level process monitoring in the prior art is difficult to obtain various operation condition data samples in an actual industrial system, or the data samples are not complete enough, so that the block division method based on data driving is difficult to be applied to the monitoring of the modern large-scale process production process.
The technical scheme adopted by the invention for solving the problems is as follows:
in a first aspect, an embodiment of the present invention provides a monitoring method based on partitioning of a knowledge block of a multi-level industrial structure, where the method includes:
constructing a multi-level knowledge graph of an industrial process;
sub-block division is carried out on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data are obtained according to the multi-level sub-blocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds;
and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks.
In one implementation, wherein building the multi-level knowledge graph of the industrial process comprises:
acquiring an interested object set, a hierarchy parameter and domain expert knowledge; wherein the set of objects of interest is used to characterize features of the objects of interest; the hierarchy comprises a service layer, an index layer, a plant-level process layer, a sub-process layer and an equipment layer; the level parameters comprise control parameters, system state parameters, index data and market information data;
carrying out entity classification and entity relationship classification on the concerned object set and the hierarchy parameters to obtain an entity classification matrix and an entity relationship classification matrix;
and obtaining a multi-level knowledge graph of the industrial process according to the entity classification matrix and the entity relation classification matrix.
In one implementation, the multi-level sub-block statistics include a principal component statistics, a residual statistics;
the sub-block division of the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and the obtaining of the multi-level sub-block statistical data according to the multi-level sub-blocks comprises:
sub-block division is carried out on the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks;
performing normalization calculation on the entity classification matrix in the multi-level sub-blocks to obtain normalization parameters;
singular value decomposition is carried out on the normalization parameters to obtain a load matrix and a score matrix;
and obtaining principal component statistics and residual statistics according to the load matrix and the score matrix.
In one implementation, the obtaining principal component statistics and residual statistics according to the load matrix and the score matrix includes:
obtaining a principal component variable and a residual variable according to the load matrix and the score matrix;
and obtaining principal component statistics and residual error statistics according to the principal component variables and the residual error variables.
In one implementation, the multi-level sub-block statistic data includes a principal component statistical threshold and a residual component statistical threshold, the sub-block partitioning nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and obtaining multi-level sub-block statistic data according to the multi-level sub-blocks includes:
sub-block division is carried out on the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks;
performing normalization calculation on the entity classification matrix in the multi-level sub-blocks to obtain normalization parameters;
singular value decomposition is carried out on the normalization parameters to obtain a diagonal matrix;
and obtaining a principal component statistic threshold value and a residual error statistic threshold value according to the diagonal matrix.
In one implementation manner, the obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the multi-level subblock statistical data includes:
obtaining a global monitoring state and a local monitoring state of the multi-level sub-blocks according to the principal component statistic, the residual statistic, the principal component statistic threshold and the residual statistic threshold;
and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the global monitoring state and the local monitoring state.
In one implementation, the deriving the global monitor state and the local monitor state of the multi-level sub-block according to the principal component statistic and the residual statistic includes:
performing index calculation on the principal component statistic and the residual error statistic to obtain a plurality of global statistical indexes and a plurality of local statistical indexes;
and obtaining a whole local monitoring state and a local monitoring state at a plurality of moments according to the global statistical index and the local statistical index.
In one implementation, the obtaining the global monitor state and the local monitor state of the multi-level sub-block according to the principal component statistic and the residual statistic further includes:
when the principal component statistic is larger than a principal component statistic threshold value or the residual statistic is larger than a residual statistic threshold value, calculating a contribution graph according to the principal component statistic and the residual statistic; wherein the contribution graph is used to characterize the contribution rates of the principal component statistics and the residual statistics to the multi-hierarchy industrial structure knowledge block.
In an implementation manner, the obtaining the global monitoring result and the local monitoring result of the multi-level sub-block according to the global monitoring state and the local monitoring state includes:
when the global monitoring state is abnormal or the local monitoring state is abnormal, determining a cause variable according to the contribution graph, wherein the cause variable is used for representing the variable of a level parameter; the global monitoring state abnormity is that the global principal component statistic is larger than a global principal component statistic threshold or the global residual statistic is larger than a global residual statistic threshold; the local monitoring state is abnormal, namely the local principal component statistic is larger than the local principal component statistic threshold value or the local residual statistic is larger than the local residual statistic threshold value;
and calculating a global abnormal positioning result and a local abnormal positioning result of the reason variable by adopting an AND gate mode.
In a second aspect, an embodiment of the present invention further provides a monitoring device based on multi-level industrial structure knowledge block partitioning, where the device includes:
the multi-level knowledge graph building unit is used for building a multi-level knowledge graph of the industrial process;
the multi-level subblock statistical data unit is used for dividing the subblocks of the nodes in the multi-level knowledge graph to obtain multi-level subblocks and obtaining multi-level subblock statistical data according to the multi-level subblocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds;
and the global monitoring result and local monitoring result acquisition unit is used for acquiring the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks.
In a third aspect, an embodiment of the present invention further provides an intelligent terminal, including a memory, and one or more programs, where the one or more programs are stored in the memory, and configured to be executed by one or more processors, where the one or more programs include a monitoring method for performing multi-level industrial structure knowledge block partitioning.
In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, wherein instructions of the storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the monitoring method based on multi-hierarchy industrial structure knowledge block partitioning as described in any one of the above.
The invention has the beneficial effects that: the embodiment of the invention firstly constructs a multi-level knowledge graph of the industrial process; then, sub-block division is carried out on the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data are obtained according to the multi-level sub-blocks; the multi-level sub-block statistical data comprise principal component statistics, residual error statistics, principal component statistics thresholds and residual error statistics thresholds; finally, obtaining a global monitoring result and a local monitoring result of the multi-level subblocks according to the multi-level subblock statistical data; therefore, according to the method, the multi-level knowledge graph of the industrial process is constructed by using a small amount of simple expert field knowledge, the sub-blocks are divided by using the nodes of the multi-level knowledge graph as the basis, the principal component statistics and the residual error statistics are finally counted according to the multi-level sub-blocks, and the accurate positioning of the abnormal nodes is realized through a contribution graph backtracking strategy.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only some embodiments described in the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic diagram of an overall technical solution provided by an embodiment of the present invention.
Fig. 2 is a schematic flow chart of a monitoring method based on multi-level industrial structure knowledge block division according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of an implementation of a process of building a multi-level knowledge graph according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of another implementation of the building process of the multi-level knowledge graph according to the embodiment of the present invention.
Fig. 5 is a schematic diagram of a multi-level knowledge graph according to an embodiment of the present invention.
Fig. 6 is a diagram illustrating the calculation of statistics and the contribution of variables of the multi-level sub-blocks according to the embodiment of the present invention.
Fig. 7 is a schematic block diagram of a monitoring device based on multi-level industrial structure knowledge block division according to an embodiment of the present invention.
Fig. 8 is a schematic block diagram of an internal structure of an intelligent terminal according to an embodiment of the present invention.
Detailed Description
The invention discloses a monitoring method, a monitoring device, an intelligent terminal and a storage medium based on multi-level industrial structure knowledge block division, and in order to make the purpose, the technical scheme and the effect of the invention clearer and clearer, the invention is further described in detail by referring to the attached drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. As used herein, the term "and/or" includes all or any element and all combinations of one or more of the associated listed items.
It will be understood by those skilled in the art that, 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 this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The problem that a data-driven block partitioning method is difficult to be suitable for monitoring the modern large-scale process production process is caused by the fact that various operation condition data samples in an actual industrial system are difficult to obtain in the factory-level process monitoring in the prior art or the data samples are incomplete.
In order to solve the problems in the prior art, the embodiment provides a monitoring method based on multi-level industrial structure knowledge block division, a multi-level knowledge graph of an industrial process is constructed by using a small amount of simple expert field knowledge through the method, sub-block division is performed by using nodes of the multi-level knowledge graph as a basis, and finally, principal component statistics and residual statistics are calculated according to the multi-level sub-blocks, and accurate positioning of abnormal nodes is realized through a contribution graph backtracking strategy.
Exemplary method
The embodiment provides a monitoring method based on multi-hierarchy industrial structure knowledge block division, and the method can be applied to an intelligent terminal for industrial knowledge automation. As shown in fig. 1, the method includes:
s100, constructing a multi-level knowledge graph of the industrial process;
because the data samples in the existing actual production process are few and the prior knowledge is difficult to obtain in the prior art, the invention constructs a multi-level knowledge graph of the industrial process on the basis of simple expert field knowledge, prepares for reasonable block division in the follow-up process, ensures the performance of a block or dispersion monitoring system and further ensures the safety of large-scale process industrial production. The multi-level knowledge graph is composed of a plurality of sheet triples (head H-relation r-tail T), wherein the head H and tail T can be regarded as physical entities or attribute values, as shown in fig. 3 (left), and the triples can be regarded as a fact representation in a space vector, and can be clearly described through a TransE embedding method, as shown in fig. 3 (right). However, there are many differences between the knowledge graph in an industrial process and the knowledge graph construction process generalized by the internet, and the data source in the industrial process comes from both the internet and the monitoring variable. If the tail T portion is the object of interest and, in fact, the object can be reflected by a number of monitored variables (a number of H nodes), as shown in fig. 4 (left), the relationship can be calculated from the binary symmetric adjacency (link) matrix a, which can be expressed as:
wherein, aklRepresenting the relationship between the kth tail node and the l head node, and which can be determined according to expert knowledge; the binary symmetric adjacency matrix A represents the relationship among a plurality of nodes in one level in the knowledge graph; the matrix a represents k nodes in the knowledge graph nodes, and each node can be described by l variables. Similarly, each knowledge graph node is composed of an object of interest and a plurality of monitored variables, each knowledge graph has a plurality of nodes, and each level has a plurality of knowledge graph nodes, as shown in fig. 4 (right) for a two-level knowledge graph. The multi-level knowledge graph of the industrial process can be regarded as a descriptive graph, which is one of the most appropriate and effective methods for describing the relationships among the multiple levels and can be extracted from metadata of numerous complex monitoring variables.
In order to obtain a multi-level knowledge graph, the building the multi-level knowledge graph of the industrial process comprises the following steps:
s101, acquiring an interested object set, a hierarchy parameter and field expert knowledge; wherein the set of objects of interest is used to characterize features of the objects of interest; the hierarchy comprises a service layer, an index layer, a plant-level process layer, a sub-process layer and an equipment layer; the level parameters comprise control parameters, system state parameters, index data and market information data;
s102, carrying out entity classification and entity relation classification on the concerned object set and the hierarchy parameters to obtain an entity classification matrix and an entity relation classification matrix;
s103, obtaining a multi-level knowledge graph of the industrial process according to the entity classification matrix and the entity relation classification matrix.
Specifically, an attention object set, a hierarchy, hierarchy parameters and domain expert knowledge are obtained firstly, wherein the attention object set is used for representing the characteristics of an attention object; the hierarchy comprises a service layer, an index layer, a plant-level process layer, a sub-process layer and an equipment layer; the level parameters comprise control parameters, system state parameters, index data and market information data; in one implementation, the calculation method of any node of the multi-level knowledge graph adopts the following formula:
wherein, ailRepresenting selected relevant monitoring variables or parameters or the like to describe the object of interest, i.e. the object of interest in the set of objects of interest. Xi,YvRespectively representing the features of the object of interest and the measurement observer. The above formula indicates that any knowledge graph node has a large number of measurement observers, and the monitoring variables between the knowledge graph nodes of different levels are overlapped and coupled, so that the coupling between the variables is favorably maintained. After obtaining the interested object set, the levels, the level parameters and the domain expert knowledge, entity classification and entity relationship classification are carried out on the interested object set and the level parameters to obtain an entity classification matrix and an entity relationship classification matrix, the named entity classification in the industrial production process mainly refers to the interested objects and related monitoring variables or parameters thereof, and the knowledge graph of each level contains a large amount of key market information, sensor measurement data, executors, raw material components, intermediate products, even intermediate statistical data and the like. In this embodiment, the interest object set and the hierarchy parameter may be subjected to entity classification using a rule-based method, for example: the rule-based named entity classification can be described as follows:
wherein, Xn×m,
Respectively representing the related principal component statistics T of the j level and the previous level in the ith knowledge graph node2And residual error statistics SPE;
respectively representing control parameters, system state data and index data;
market information data; m isiAnd i-1, 2,3,4 and 5 represent the dimension of the final characteristic variable. The high-precision entity can be obtained by the rule-based named entity classification method in the industrial process, and the high-precision entity is the guarantee of a high-performance multi-level knowledge graph. The difficulty with the named entity classification described above is accurately describing the entities and labeling the types of entities. In order to obtain a multi-level knowledge graph with higher accuracy, an embodiment of the present invention provides a "sitting-together" method, by which entity relationship classification is performed on the interest object set and the level parameters, for example: as shown in fig. 5 (a). FIG. 5(a) depicts five nodes of interest (1, 2,3,4, 5) and their corresponding monitored variables and parameters (v)1*，v2*，v3*，v4*，v5*). If the knowledge graph of the node 5 becomes the object of interest, the monitoring variables and other parameters reflecting the operating state of the node should include the knowledge graphs of the upstream and downstream nodes adjacent to the node. This means that the system characteristics of node 5 can be reflected by the numerous variables of nearby nodes (1, 2,3, 4) and their derivatives. The entity relationship classification matrix of the ith knowledge graph of the jth level can be represented as:
wherein the content of the first and second substances,
is a binary symmetric adjacency of the ith knowledge graph of the jth levelA matrix;
the binary adjacency matrix is constructed by the principal elements or residual statistics of the knowledge graph sub-blocks of the previous layer;
respectively are relationship matrixes among the entity, a plurality of control parameters, a system running state and index parameters;
a relationship matrix representing the objects of interest and the market information variables.
And obtaining a multi-level knowledge graph of the industrial process according to the entity classification matrix and the entity relation classification matrix. In a factory-level mass production process, a multi-level knowledge graph can be regarded as a higher-level knowledge organization form, the purpose of which can be summarized as integration, disambiguation, verification and update of heterogeneous data from different knowledge sources under a unified framework, and the knowledge framework is a primary key problem by integrating information, data, methods, experiences and human ideas, and the multi-level knowledge graph framework for a large-scale factory-level range comprises five levels in the embodiment: the system comprises a business layer, an index layer, a plant-level process layer, a sub-process layer and an equipment layer. As shown in fig. 5 (b). FIG. 5(c) shows the calculation and construction method of any node of each level knowledge graph. In this embodiment, the monitored variables or parameters of each node in the multi-level knowledge graph overlap, as opposed to the past independent co-distribution assumption. This independent co-distribution assumption is crucial to the conventional distribution anomaly monitoring method. The mechanism that the node state is influenced by the upstream and downstream node states is more consistent with the actual situation, and the distributed abnormal state monitoring is prepared.
Having obtained a multi-level knowledge map of an industrial process, the following steps can be performed as shown in FIG. 1: s200, dividing sub blocks of nodes in the multi-level knowledge graph to obtain multi-level sub blocks, and obtaining multi-level sub block statistical data according to the multi-level sub blocks; the statistical data of the multi-level sub-blocks comprise principal component statistics, residual error statistics, principal component statistics thresholds and residual error statistics thresholds; correspondingly, the sub-block division is performed on the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and the multi-level sub-block statistical data is obtained according to the multi-level sub-blocks, which includes the following steps:
s201, performing sub-block division on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks;
s202, carrying out normalization calculation on the entity classification matrixes in the multi-level sub-blocks to obtain normalization parameters;
s203, singular value decomposition is carried out on the normalization parameters to obtain a load matrix, a score matrix and a diagonal matrix;
and S204, obtaining multi-level sub-block statistical data according to the load matrix, the score matrix and the diagonal matrix.
Specifically, there are three types of block division in the prior art: sub-block division according to the physical entity distribution of the object of interest, sub-block division based on Principal Component Analysis (PCA), sub-block division based on an operation unit, sub-block division according to the physical entity distribution of the object of interest: the plant-level distributed modeling and monitoring framework is used for realizing block division by fusing process structures and big data knowledge. The distributed type includes two layers of meanings: spatially distributed modeling and layered monitoring for a plant-level process having a plurality of operational units; distributed parallel modeling of large process data with various functions. The framework divides the whole production process into factory level, block level and variable level according to the spatial distribution of data storage so as to realize the monitoring of the whole factory level production process. Wherein the block division is performed according to the distribution of the physical entities of the objects of interest. Sub-block division based on principal component of PCA decomposition: PCA finds the projection direction that best represents the raw data in the least mean square sense, and the k principal components are selected to be uncorrelated with each other. By constructing the sub-blocks on each irrelevant main component, the diversity requirement of sub-block division can be met; and the most relevant variables are selected on each irrelevant subblock, so that the precision requirement of each subblock model can be met. I.e., k +1 sub-blocks are constructed in total in the process data. And selecting the corresponding variable of each sub-block according to the contribution value of each variable to each main component, namely each sub-block. Similar studies have also constructed sub-block partitions based on independent principal component analysis. Sub-block partitioning based on operation units: aiming at the multi-unit plant-level chemical process, a local unit monitoring model needs to be established, the running state of a local unit is monitored and analyzed, and then the whole production process is monitored. According to the technology, the subblocks in the distributed monitoring system of the existing large-scale plant-level production process are divided reasonably by an expert knowledge and data driving method based on the physical distribution and production process data of the concerned monitored object. However, the drawbacks are two-fold: firstly, the actual industrial system is difficult to obtain a large number of strong and complete data samples of various operation conditions; secondly, the ambiguity and uncertainty of expert knowledge in an actual industrial system are strong, the complexity is high, the quantization and the conversion are difficult, and the relevance of monitoring variables can be split by improper multi-block division under the condition that an accurate model or priori knowledge is lacked in the existing block division method, so that the monitoring performance is reduced. In the embodiment of the invention, the nodes in the multi-level knowledge graph are divided into the sub-blocks to obtain the multi-level sub-blocks, and in practice, each node in the multi-level knowledge graph is used as a block node to divide the multi-level knowledge graph to obtain the multi-level sub-blocks. The distributed monitoring system with multi-sub-block division repeatedly divides a large number of monitoring variables into a plurality of sub-blocks, and then establishes a monitoring model for each sub-block to mine more process information, so that the monitoring effect is improved. After multi-level sub-blocks are obtained, carrying out normalization calculation on the entity classification matrixes in the multi-level sub-blocks to obtain normalization parameters; performing singular value decomposition on the normalization parameters to obtain a load matrix, a score matrix and a diagonal matrix; for example, the load matrix, the score matrix, and the diagonal matrix are calculated by the following calculation formulas:
wherein Λ ═ diag (λ)i，λi(i ═ 1, 2.., m)) is a diagonal matrix composed of each singular value. T is a score matrix and P is a load matrix.
And then obtaining multi-level sub-block statistical data according to the load matrix, the score matrix and the diagonal matrix. Correspondingly, the step of obtaining the statistical data of the multi-level subblocks according to the load matrix, the score matrix and the diagonal matrix comprises the following steps: obtaining a principal component variable and a residual variable according to the load matrix and the score matrix; obtaining principal component statistics and residual error statistics according to the principal component variables and the residual error variables; and obtaining a principal component statistic threshold value and a residual error statistic threshold value according to the diagonal matrix.
Specifically, principal component variables and residual variable are obtained according to the load matrix and the score matrix; for example: decomposing the data into principal component variables and residual variables by adopting a principal component analysis technology:
wherein T ∈ Rm×κ,P∈Rm×κRespectively a principal component scoring matrix and a load matrix corresponding to the principal component scoring matrix;
respectively representing a residual score matrix and a load matrix corresponding to the residual score matrix; e is an element of Rn×mRepresenting the residual matrix. k ≦ m represents the pivot number, whose value may be determined by the Cumulative Pivot Variance (CPV). Then, obtaining principal component statistic T according to the principal component variable and the residual variable2And residual error statistic SPE:
wherein t ═ xP represents the score vector for sample x; e ═ x-tPTRepresenting the residual vector of sample x. And then, according to the diagonal matrix, obtaining a principal component statistic threshold value and a residual statistic threshold value, for example: the diagonal matrix is Λ ═ diag (λ)i,λi(i ═ 1, 2.. times, m)), from the diagonal matrix, one can derive
h0＝1-2θ1θ3/3θ1 2And k is less than or equal to m, which represents the number of principal elements, and the principal element statistic threshold value and the residual error statistic threshold value are the following formulas:
wherein, Fκ,m-κ：αF distribution threshold representing k and m-k degrees of freedom with confidence level α; c. CαRepresenting a threshold value for the global distribution of the criteria at the confidence level alpha.
After obtaining the statistics of the sub-blocks at multiple levels, the following steps can be performed as shown in fig. 1: and step S300, obtaining a global monitoring result and a local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks. Correspondingly, in order to obtain the global monitoring result and the local monitoring result of the multi-level subblocks, the step of obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks comprises the following steps:
s301, obtaining a global monitoring state and a local monitoring state of the multi-level sub-block according to the principal component statistic, the residual statistic, the principal component statistic threshold and the residual statistic threshold;
s302, obtaining a global monitoring result and a local monitoring result of the multi-level subblocks according to the global monitoring state and the local monitoring state.
Specifically, a global monitoring state and a local monitoring state of the multi-level sub-block are obtained according to the principal component statistic, the residual statistic, the principal component statistic threshold and the residual statistic threshold; correspondingly, the step of obtaining the global monitoring state and the local monitoring state of the multi-level sub-blocks according to the principal component statistics and the residual error statistics comprises the following steps: performing index calculation on the principal component statistic and the residual statistic to obtain a plurality of global statistical indexes and a plurality of local statistical indexes; and obtaining the global monitoring state and the local monitoring state at a plurality of moments according to the global statistical index and the local statistical index.
Specifically, as shown in fig. 6, the principal component statistical measures and the residual statistical measures of the nodes in two adjacent hierarchical sub-blocks are used as monitoring variables, and the PCA technology, the independent principal component ICA analysis, the partial least squares analysis PLS, the typical correlation analysis CCA, and other technologies are used to calculate four global statistical indicators
As shown in fig. 4, the four statistical indexes are a principal component statistical index of a principal component variable of the principal component score matrix, a residual statistical index of the principal component variable, a principal component statistical index of the residual variable, and a residual statistical index of the residual variable, respectively. In addition to this, two local statistical indicators { T } can be obtained2And SPE, and then obtaining a global monitoring state and a local monitoring state at a plurality of moments according to the global statistical index and the local statistical index. For example, global and local operating conditions are derived for four cases that may monitor states: (1) global monitorable state and local monitorable state: any four global statistics are larger than the threshold value of the corresponding statistics, and any two local statistics are larger than the threshold value of the corresponding statistics; (2) global and local unmonitorable states: any four global statistics are larger than the threshold value of the corresponding statistics, and any two local statistics are not larger than the threshold value of the corresponding statistics; (3) global unmonitorable state and local monitorable state: any four global statistics are not greater than the threshold of the corresponding statistic, while any two local statistics are greater than the corresponding statisticA threshold value of the statistical quantity of (a); (4) global non-monitorable state and local non-monitorable state: any four global statistics are not greater than the threshold value of the corresponding statistics, while any two local statistics are not greater than the threshold value of the corresponding statistics. In addition, if the global or local statistic is larger than the threshold of the corresponding statistic, a contribution map is calculated according to the principal component statistic and the residual statistic, for example, a contribution map of each monitoring variable or parameter is calculated, and the formula is as follows:
wherein the content of the first and second substances,
respectively representing the principal component statistic T2And a contribution graph of the residual error statistic SPE;
D＝PTΛP，ξiexpress identity matrix ImCharacteristic values of the ith column.
And after the global monitoring state and the local monitoring state of the multi-level subblocks are obtained, obtaining a global monitoring result and a local monitoring result of the multi-level subblocks according to the global monitoring state and the local monitoring state. Correspondingly, the step of obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the global monitoring state and the local monitoring state comprises the following steps: when the global monitoring state is abnormal or the local monitoring state is abnormal, determining a cause variable according to the contribution graph, wherein the cause variable is used for representing a variable of a level parameter; the global monitoring state is abnormal, namely the global principal component statistic is larger than a global principal component statistic threshold value or the global residual statistic is larger than a global residual statistic threshold value; the local monitoring state is abnormal, namely the local principal component statistic is larger than the local principal component statistic threshold value or the local residual statistic is larger than the local residual statistic threshold value; and calculating a global abnormity positioning result and a local abnormity positioning result of the reason variable by adopting an AND gate mode.
Specifically, when the global monitoring state is abnormal or the local monitoring state is abnormal, a cause variable is determined according to the contribution graph, wherein the cause variable is used for representing a variable of a layer level parameter; the global monitoring state is abnormal, namely the global principal component statistic is larger than a global principal component statistic threshold value or the global residual statistic is larger than a global residual statistic threshold value; the local monitoring state anomaly is that the local principal component statistic is larger than the local principal component statistic threshold or the local residual statistic is larger than the local residual statistic threshold, for example, when an anomaly with unknown prior knowledge occurs, a variable or a parameter with a larger contribution rate can be used as a cause variable. And then calculating a global abnormity positioning result and a local abnormity positioning result of the cause variable by adopting an AND gate mode. For example, since multiple levels are involved, when the final monitoring variable is reached, the results of each monitoring variable at different levels are greatly different, and even result conflicts and the like. For this reason, the present embodiment adopts an and gate policy to implement the localization of the abnormal state.
Wherein R ispRepresenting an abnormal positioning result;
representing the abnormal reason positioning result of the ith node;
showing the abnormal result corresponding to the variable with larger contribution graph of each level after the abnormal of the ith node occurs, if
A value of 0 indicates that no abnormality occurred at any level,
a value of 1 indicates that at least one level of display exception occurred and it is necessary to trace back to this variable.
Exemplary device
As shown in fig. 7, an embodiment of the present invention provides a monitoring device based on partitioning of a knowledge block of a multi-level industrial structure, the device includes a multi-level knowledge graph building unit 401, a multi-level sub-block statistical data unit 402, and a global monitoring result and local monitoring result obtaining unit 403, where:
a multi-level knowledge graph building unit 401, configured to build a multi-level knowledge graph of an industrial process;
a multi-level sub-block statistical data unit 402, configured to perform sub-block division on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and obtain multi-level sub-block statistical data according to the multi-level sub-blocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds;
a global monitoring result and local monitoring result obtaining unit 403, configured to obtain a global monitoring result and a local monitoring result of the multi-level subblock according to the multi-level subblock statistical data.
In the embodiment, a multi-level knowledge graph constructing unit 401 constructs a multi-level knowledge graph of an industrial process; then, a multi-level sub-block statistical data unit 402 is used for sub-block division of the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data is obtained according to the multi-level sub-blocks; and finally, obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks by the global monitoring result and local monitoring result obtaining unit 403. The module is used for constructing a multi-level knowledge graph of an industrial process by using a small amount of simple expert field knowledge, sub-block division is carried out by taking the nodes of the multi-level knowledge graph as a basis, finally, principal component statistics and residual statistics are counted according to the multi-level sub-blocks, and the accurate positioning of abnormal nodes is realized by a contribution graph backtracking strategy.
Based on the above embodiment, the present invention further provides an intelligent terminal, and a schematic block diagram thereof may be as shown in fig. 8. The intelligent terminal comprises a processor, a memory, a network interface, a display screen and a temperature sensor which are connected through a system bus. Wherein, the processor of the intelligent terminal is used for providing calculation and control capability. The memory of the intelligent terminal comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the intelligent terminal is used for being connected and communicated with an external terminal through a network. The computer program is executed by a processor to implement a monitoring method based on multi-level industrial structure knowledge block partitioning. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen, and the temperature sensor of the intelligent terminal is arranged inside the intelligent terminal in advance and used for monitoring the operating temperature of internal equipment.
Those skilled in the art will appreciate that the schematic diagram of fig. 8 is only a block diagram of a part of the structure related to the present invention, and does not constitute a limitation to the intelligent terminal to which the present invention is applied, and a specific intelligent terminal may include more or less components than those shown in the figure, or combine some components, or have different component arrangements.
In one embodiment, an intelligent terminal is provided that includes a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for:
constructing a multi-level knowledge graph of an industrial process;
sub-block division is carried out on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data are obtained according to the multi-level sub-blocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds;
and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, databases, or other media used in embodiments provided herein may include non-volatile and/or volatile memory. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
In summary, the present invention discloses a monitoring method, device, intelligent terminal, and storage medium based on multi-level industrial structure knowledge block division, wherein the method comprises: constructing a multi-level knowledge graph of an industrial process; sub-block division is carried out on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data are obtained according to the multi-level sub-blocks; wherein the multi-level sub-block statistical data comprises principal component statistics, residual statistics, a principal component statistics threshold and a residual statistics threshold; and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks. According to the method, a multi-level knowledge graph of an industrial process is constructed by using a small amount of simple expert field knowledge, subblock division is carried out by using nodes of the multi-level knowledge graph as a basis, principal component statistics and residual error statistics are finally counted according to the multi-level subblocks, and the accurate positioning of abnormal nodes is realized through a contribution graph backtracking strategy.
Based on the above embodiments, the present invention discloses a monitoring method based on multi-level industrial structure knowledge block division, it should be understood that the application of the present invention is not limited to the above examples, and it is obvious to those skilled in the art that modifications and changes can be made according to the above description, and all such modifications and changes are intended to fall within the scope of the appended claims.
Claims (12)
1. A monitoring method based on multi-level industrial structure knowledge block division is characterized by comprising the following steps:
constructing a multi-level knowledge graph of an industrial process;
sub-block division is carried out on nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and multi-level sub-block statistical data are obtained according to the multi-level sub-blocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds;
and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks.
2. The monitoring method based on multi-hierarchy industrial structure knowledge block division as claimed in claim 1, wherein the building of the multi-hierarchy knowledge graph of the industrial process comprises:
acquiring an interested object set, a hierarchy parameter and domain expert knowledge; wherein the set of objects of interest is used to characterize features of the objects of interest; the hierarchy comprises a service layer, an index layer, a plant-level process layer, a sub-process layer and an equipment layer; the level parameters comprise control parameters, system state parameters, index data and market information data;
carrying out entity classification and entity relationship classification on the concerned object set and the hierarchy parameters to obtain an entity classification matrix and an entity relationship classification matrix;
and obtaining a multi-level knowledge graph of the industrial process according to the entity classification matrix and the entity relation classification matrix.
3. The monitoring method based on multi-hierarchy industrial structure knowledge block division according to claim 2, wherein the multi-hierarchy sub-block statistical data comprises principal component statistics, residual error statistics;
the sub-block division of the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks, and the obtaining of the multi-level sub-block statistical data according to the multi-level sub-blocks comprises the following steps:
sub-block division is carried out on the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks;
performing normalization calculation on the entity classification matrix in the multi-level sub-blocks to obtain normalization parameters;
singular value decomposition is carried out on the normalization parameters to obtain a load matrix and a score matrix;
and obtaining principal component statistics and residual error statistics according to the load matrix and the score matrix.
4. The monitoring method based on multi-hierarchy industrial structure knowledge block division according to claim 3, wherein the obtaining of principal component statistics and residual statistics according to the load matrix and the score matrix comprises:
obtaining a principal component variable and a residual variable according to the load matrix and the score matrix;
and obtaining principal component statistics and residual error statistics according to the principal component variables and the residual error variables.
5. The monitoring method based on multi-hierarchy industrial structure knowledge block division according to claim 2, wherein the multi-hierarchy sub-block statistical data comprises a principal component statistic threshold and a residual statistic threshold, the sub-block division of the nodes in the multi-hierarchy knowledge graph to obtain multi-hierarchy sub-blocks, and the obtaining multi-hierarchy sub-block statistical data according to the multi-hierarchy sub-blocks comprises:
sub-block division is carried out on the nodes in the multi-level knowledge graph to obtain multi-level sub-blocks;
performing normalization calculation on the entity classification matrix in the multi-level sub-blocks to obtain normalization parameters;
singular value decomposition is carried out on the normalization parameters to obtain a diagonal matrix;
and obtaining a principal component statistic threshold value and a residual error statistic threshold value according to the diagonal matrix.
6. The monitoring method based on multi-level industrial structure knowledge block division according to claim 5, wherein the obtaining of the global monitoring result and the local monitoring result of the multi-level sub-blocks according to the multi-level sub-block statistical data comprises:
obtaining a global monitoring state and a local monitoring state of the multi-level sub-blocks according to the principal component statistic, the residual statistic, the principal component statistic threshold and the residual statistic threshold;
and obtaining the global monitoring result and the local monitoring result of the multi-level subblocks according to the global monitoring state and the local monitoring state.
7. The monitoring method based on multi-hierarchy industrial structure knowledge block partitioning as claimed in claim 6, wherein said deriving global monitoring state and local monitoring state of multi-hierarchy sub-blocks according to the principal component statistics and the residual statistics comprises:
performing index calculation on the principal component statistic and the residual statistic to obtain a plurality of global statistical indexes and a plurality of local statistical indexes;
and obtaining the global monitoring state and the local monitoring state at a plurality of moments according to the global statistical index and the local statistical index.
8. The monitoring method based on multi-hierarchy industrial structure knowledge block partitioning as claimed in claim 6, wherein said deriving global monitoring state and local monitoring state of multi-hierarchy sub-blocks according to the principal component statistics and the residual statistics further comprises:
when the principal component statistic is larger than a principal component statistic threshold value or the residual statistic is larger than a residual statistic threshold value, calculating a contribution graph according to the principal component statistic and the residual statistic; wherein the contribution graph is used to characterize the contribution rates of the principal component statistics and the residual statistics to the multi-hierarchy industrial structure knowledge block.
9. The monitoring method based on multi-hierarchy industrial structure knowledge block division according to claim 8, wherein the obtaining of the global monitoring result and the local monitoring result of the multi-hierarchy sub-blocks according to the global monitoring state and the local monitoring state comprises:
when the global monitoring state is abnormal or the local monitoring state is abnormal, determining a cause variable according to the contribution graph, wherein the cause variable is used for representing a variable of a level parameter; the global monitoring state is abnormal, namely the global principal component statistic is larger than a global principal component statistic threshold value or the global residual statistic is larger than a global residual statistic threshold value; the local monitoring state is abnormal, namely the local principal component statistic is larger than the local principal component statistic threshold value or the local residual statistic is larger than the local residual statistic threshold value;
and calculating a global abnormity positioning result and a local abnormity positioning result of the reason variable by adopting an AND gate mode.
10. A monitoring device based on multi-level industrial structure knowledge block partitioning, the device comprising:
the multi-level knowledge graph building unit is used for building a multi-level knowledge graph of the industrial process;
the multi-level subblock statistical data unit is used for dividing the subblocks of the nodes in the multi-level knowledge graph to obtain multi-level subblocks and obtaining multi-level subblock statistical data according to the multi-level subblocks; wherein the multi-level sub-block statistics comprise principal component statistics, residual statistics, principal component statistics thresholds, and residual statistics thresholds;
and the global monitoring result and local monitoring result acquisition unit is used for acquiring a global monitoring result and a local monitoring result of the multi-level subblocks according to the statistical data of the multi-level subblocks.
11. An intelligent terminal comprising a memory, and one or more programs, wherein the one or more programs are stored in the memory, and wherein the one or more programs being configured to be executed by the one or more processors comprises instructions for performing the method of any of claims 1-9.
12. A non-transitory computer-readable storage medium, wherein instructions in the storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the method of any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110393994.8A CN113239187B (en) | 2021-04-13 | Monitoring method based on multi-level industrial structure knowledge block division |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110393994.8A CN113239187B (en) | 2021-04-13 | Monitoring method based on multi-level industrial structure knowledge block division |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113239187A true CN113239187A (en) | 2021-08-10 |
| CN113239187B CN113239187B (en) | 2024-05-14 |
Family
ID=
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115307669A (en) * | 2022-10-11 | 2022-11-08 | 蘑菇物联技术（深圳）有限公司 | Method, apparatus, and medium for detecting abnormal sensor of system under test |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020072882A1 (en) * | 2000-03-23 | 2002-06-13 | Uwe Kruger | Multivariate statistical process monitors |
| US20090132450A1 (en) * | 2007-11-21 | 2009-05-21 | N4 Llc | Systems and methods for multivariate influence analysis of heterogenous mixtures of categorical and continuous data |
| US20090299695A1 (en) * | 2008-05-29 | 2009-12-03 | General Electric Company | System and method for advanced condition monitoring of an asset system |
| WO2012090492A1 (en) * | 2010-12-28 | 2012-07-05 | 株式会社 東芝 | Process monitoring and diagnosis system |
| CN102880151A (en) * | 2012-10-11 | 2013-01-16 | 浙江大学 | Double-layer data model-driven plant-level chemical process monitoring method |
| CN103389701A (en) * | 2013-07-15 | 2013-11-13 | 浙江大学 | Plant-level process fault detection and diagnosis method based on distributed data model |
| 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 |
| CN108762228A (en) * | 2018-05-25 | 2018-11-06 | 江南大学 | A kind of multi-state fault monitoring method based on distributed PCA |
| US20180330300A1 (en) * | 2017-05-15 | 2018-11-15 | Tata Consultancy Services Limited | Method and system for data-based optimization of performance indicators in process and manufacturing industries |
| CN109507972A (en) * | 2018-12-19 | 2019-03-22 | 中国计量大学 | Industrial processes fault monitoring method based on layer-stepping non-gaussian monitoring algorithm |
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020072882A1 (en) * | 2000-03-23 | 2002-06-13 | Uwe Kruger | Multivariate statistical process monitors |
| US20090132450A1 (en) * | 2007-11-21 | 2009-05-21 | N4 Llc | Systems and methods for multivariate influence analysis of heterogenous mixtures of categorical and continuous data |
| US20090299695A1 (en) * | 2008-05-29 | 2009-12-03 | General Electric Company | System and method for advanced condition monitoring of an asset system |
| WO2012090492A1 (en) * | 2010-12-28 | 2012-07-05 | 株式会社 東芝 | Process monitoring and diagnosis system |
| CN102880151A (en) * | 2012-10-11 | 2013-01-16 | 浙江大学 | Double-layer data model-driven plant-level chemical process monitoring method |
| CN103389701A (en) * | 2013-07-15 | 2013-11-13 | 浙江大学 | Plant-level process fault detection and diagnosis method based on distributed data model |
| 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 |
| US20180330300A1 (en) * | 2017-05-15 | 2018-11-15 | Tata Consultancy Services Limited | Method and system for data-based optimization of performance indicators in process and manufacturing industries |
| CN108762228A (en) * | 2018-05-25 | 2018-11-06 | 江南大学 | A kind of multi-state fault monitoring method based on distributed PCA |
| CN109507972A (en) * | 2018-12-19 | 2019-03-22 | 中国计量大学 | Industrial processes fault monitoring method based on layer-stepping non-gaussian monitoring algorithm |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115307669A (en) * | 2022-10-11 | 2022-11-08 | 蘑菇物联技术（深圳）有限公司 | Method, apparatus, and medium for detecting abnormal sensor of system under test |
| CN115307669B (en) * | 2022-10-11 | 2023-01-10 | 蘑菇物联技术（深圳）有限公司 | Method, apparatus, and medium for detecting abnormal sensor of system under test |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112202736B (en) | Communication network anomaly classification method based on statistical learning and deep learning | |
| Lin et al. | An improved statistical approach for consistency test in AHP | |
| Zhu et al. | Information concentrated variational auto-encoder for quality-related nonlinear process monitoring | |
| Shi | In-process quality improvement: Concepts, methodologies, and applications | |
| Li et al. | A diagnostic procedure for high-dimensional data streams via missed discovery rate control | |
| Chen et al. | Detection and analysis of real-time anomalies in large-scale complex system | |
| Wang et al. | Decentralized plant-wide monitoring based on mutual information-Louvain decomposition and support vector data description diagnosis | |
| Liu et al. | Long–short-term memory encoder–decoder with regularized hidden dynamics for fault detection in industrial processes | |
| Feng et al. | Traffic data recovery from corrupted and incomplete observations via spatial-temporal trpca | |
| Liu et al. | Anomaly detection of high-frequency sensing data in transportation infrastructure monitoring system based on fine-tuned model | |
| Shao et al. | A comprehensive G-Lz-ADC effectiveness evaluation model for the single communication satellite system in the context of poor information | |
| Zhou et al. | Multi-expert attention network for long-term dam displacement prediction | |
| Nandanoori et al. | Data-driven operator theoretic methods for phase space learning and analysis | |
| Solomentsev et al. | Diagnostics programs efficiency analysis in operation system of radioelectronic eguipment | |
| Jin et al. | Transfer learning with quantile regression | |
| CN113239187B (en) | Monitoring method based on multi-level industrial structure knowledge block division | |
| CN113239187A (en) | Monitoring method based on multi-level industrial structure knowledge block division | |
| Yan et al. | A variable‐selection control chart via penalized likelihood and Gaussian mixture model for multimodal and high‐dimensional processes | |
| CN105117606A (en) | Method for determining element fault probability change tendency | |
| Tian et al. | Random sampling-arithmetic mean: A simple method of meteorological data quality control based on random observation thought | |
| Yang et al. | A semi-supervised feature contrast convolutional neural network for processes fault diagnosis | |
| CN115549082A (en) | Power load prediction method based on load mining and LSTM neural network | |
| CN114779739A (en) | Fault monitoring method for industrial process under cloud edge end cooperation based on probability map model | |
| Declerck et al. | State estimation and detection of changes in time interval models | |
| Zhou et al. | Similarity based robust probability latent variable regression model and its kernel extension for process monitoring |
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 |