CN117113233A - Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method - Google Patents
Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method Download PDFInfo
- Publication number
- CN117113233A CN117113233A CN202311261935.0A CN202311261935A CN117113233A CN 117113233 A CN117113233 A CN 117113233A CN 202311261935 A CN202311261935 A CN 202311261935A CN 117113233 A CN117113233 A CN 117113233A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- node
- abnormal
- hierarchical
- energy
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 194
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 93
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000010276 construction Methods 0.000 title claims abstract description 12
- 238000010586 diagram Methods 0.000 claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 18
- 230000005856 abnormality Effects 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 15
- 230000005540 biological transmission Effects 0.000 claims description 9
- 239000002994 raw material Substances 0.000 description 35
- 230000005611 electricity Effects 0.000 description 25
- 230000008569 process Effects 0.000 description 25
- 238000001514 detection method Methods 0.000 description 9
- 238000004519 manufacturing process Methods 0.000 description 9
- 238000005259 measurement Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 4
- 238000009960 carding Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 235000012054 meals Nutrition 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000000547 structure data Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000011031 large-scale manufacturing process Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
-
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the technical field of hierarchical structure energy consumption anomaly identification, and discloses a hierarchical energy structure scene construction method and an energy consumption anomaly link tracing method, which are used for constructing a global hierarchical structure identification model of each energy consumption object in a factory, and specifically comprise the following steps: dividing the energy consumption objects according to a set hierarchical classification standard to form a tree-shaped hierarchical relationship; step two, extracting relevant energy consumption indexes of all the energy consumption objects, and establishing a directed non-weighted relation graph of all the energy consumption objects according to the energy consumption indexes; constructing a network diagram G; and thirdly, constructing an abnormal value recognition algorithm model and building an energy structure scene. According to the application, the weight relation of each energy consumption point is established according to the relation of each hierarchy, so that a complete structure diagram is formed, comprehensive and accurate relation tracing and abnormal link capturing are realized, the hierarchy energy structure analysis can be carried out, the links are flexibly disassembled and recombined, the expandability is strong, and the abnormal energy consumption point identification accuracy and analysis efficiency are improved.
Description
Technical Field
The application relates to the technical field of hierarchical energy consumption scene construction and tracing, in particular to a hierarchical energy structure scene construction method and an energy consumption abnormal link tracing method.
Background
In the production process of a factory, the energy consumption of each device in the factory needs to be monitored, so that the running state of the device is obtained. By analyzing the energy consumption, whether the corresponding equipment is abnormal or not can be judged, so that the abnormal equipment can be rapidly and timely interfered, corresponding measures are taken, and the running stability of the factory is improved.
However, the existing energy consumption abnormality identification method aims at single-point energy consumption abnormality, and for a scene with hierarchical energy structure data, accurate and timely abnormal state judgment cannot be given. Mainly represented by the following cases:
1. the top layer energy consumption is abnormal, and each energy consumption point of the lower layer is abnormal, so that the number of identification points is large, abnormal links cannot be accurately acquired, and the difficulty of identification and judgment is increased;
2. the top layer energy consumption is abnormal, but each energy consumption point of the lower layer is normal, and only abnormal points can be identified, so that only top layer energy consumption abnormal data can be obtained, but the top layer data are comprehensive energy consumption data, and specific abnormal points cannot be identified;
3. the top layer energy consumption is normal, each energy consumption point of the lower layer is normal, although no abnormality occurs at this time, the deviation value of each current energy consumption point cannot be analyzed, so that the information of the point position with the possibility of occurrence of the abnormality cannot be predicted, and the information is delayed;
4. the energy consumption abnormality judgment model of each layer cannot be automatically updated.
Therefore, the existing energy consumption anomaly identification method is complicated in identifying the structural anomaly of the hierarchical energy source, and can only react when the anomaly occurs at a single point, so that the existing identification method is low in accuracy and identification efficiency.
Disclosure of Invention
The application aims to provide a hierarchical energy structure scene construction method and an energy consumption abnormal link tracing method, which are used for solving the problems of complicated abnormal identification aiming at a hierarchical energy structure, low identification accuracy and low identification efficiency in the prior art.
In order to achieve the above purpose, the application adopts the following technical scheme that the method for constructing the hierarchical energy structure scene is used for constructing a hierarchical energy structure relation network diagram, can accurately acquire data information from the global point of view according to the energy consumption state of each hierarchy, comprehensively analyzes the deviation value of each hierarchy according to the constructed link relation, and pre-judges the abnormal state. The method specifically comprises the following steps:
dividing the energy consumption objects according to a set hierarchical classification standard to form a tree-shaped hierarchical relationship;
step two, extracting relevant energy consumption indexes of all the energy consumption objects, and establishing a directed non-weighted relation graph of all the energy consumption objects according to the energy consumption indexes; constructing a network diagram G;
and thirdly, constructing an abnormal value recognition algorithm model and building an energy structure scene.
The principle and the advantages of the scheme are as follows:
when the abnormal energy consumption of the factory equipment is identified, in order to ensure that each energy consumption point can be identified in time, accurate judgment and quick response are realized, so that point-to-point single point identification is considered to be more accurate, and when the abnormal condition occurs, the position of single point detection is only required to be processed. There is thus an inherent thinking about the identification of energy consumption anomalies, and it is believed that in the identification of device energy consumption, the adoption of separate point-to-point detection is better than overall detection.
For the hierarchical energy structure, an inherent detection mode is adopted, so that the detection is considered to be performed in a point-to-point mode, and the effect is more accurate. Although the point-to-point detection has stronger pertinence, we neglect that for the hierarchical energy structure, the mutual influence relationship of each hierarchy and each energy consumption point in the hierarchy exists, and the single-point detection can only aim at a single body and can not judge the relevance of the single-point detection, so that the detection of the hierarchical energy structure is inaccurate, complex and low in efficiency.
Therefore, the application constructs a hierarchical energy structure relation graph according to the production units in the factory, thereby establishing the relevance among the energy consumption objects, and establishes the weight relation of the energy consumption points according to the hierarchical interrelationship, thereby forming a complete structure graph, realizing comprehensive and accurate relation tracing and abnormal link capturing, and improving the recognition accuracy and analysis efficiency of the abnormal energy consumption points.
Further, the class standard divides the energy consumption objects according to the physical level of the factory-process section-process unit; splitting energy consumption related elements which are nested in the same hierarchy layer by layer, and expressing the energy consumption related element set in each hierarchy layer as:
L＝{L n,i ,L n-1,j ,...,L 1,k }，
further, in the second step, the method further comprises the following sub-steps:
(1) Disassembling the energy consumption calculation formulas of all layers, and disassembling and expressing the energy consumption related elements of the same layer of energy consumption calculation formulas as
l N ＝{l N,1 ,L n,2 ,...,L n,i }；。
(2) The energy consumption object observation variable result L obtained by disassembly n，i With the lower layer node L n-1，j The connection constitutes a network node pair:
(L n,i ,L n-1,1 ,W n-1,1 )
(L n,i ,L n-1,2 ,W n-1,2 )
...
(L n,i ,L n-1,j ,W n-1,j )；
(3) Will L n,i Set as the initial node, L n-1,j Set as end node, W n-1,j Setting a weight value from an end node to a start node;
(4) And (3) traversing the steps (1) - (3) until the node pairs of all the levels are established, and forming a complete network graph G.
Further, the network graph G includes node pair information of all the energy consumption objects in the hierarchical structure and an initialization weight of each node pair, where the initialization weight is nullThe method comprises the steps of carrying out a first treatment on the surface of the Wherein the node pair information includes a start node L n,i End node L n-1,j And L n,i To L n-1,j Whether to communicate or not; w (W) n-1,j Set as end node L n-1,j To the start node L n,i Is used to initialize the weight value.
Further, in the third step, data of T time points of the energy consumption history of each level is collected, and the following formula is adopted to scale the feature with the robust statistical information of the outlier:
correspondingly, the application also provides a hierarchical energy consumption abnormal link tracing method which is applied to the constructed hierarchical energy structure scene. The method comprises the following steps:
step A, calculating the abnormality degree and state of each energy consumption index, and updating the abnormality degree and state into a directed non-weight relation diagram to form a directed weighted relation diagram;
step B, carrying out hierarchical ordering on the judged abnormal node pairs according to a set ordering rule;
and step C, calculating the shortest path of each abnormal node according to the obtained abnormal node pair, and storing and outputting the shortest path.
According to the scheme, according to the ordered abnormal node pairs, the abnormal state of each energy consumption object node can be intuitively obtained, each energy consumption abnormal link influencing the upper energy consumption is found out according to weight analysis, and effective correction measures are formulated according to the traced energy consumption abnormal links, so that the intervention efficiency of the abnormal state is improved.
Further, in step a, the degree of abnormality of each energy consumption index is output through a quantized value, and a state judgment result is output; the variable calculation mode of the abnormality degree adopts the following formula:
judging an abnormal state according to a formula 2, wherein the abnormal state judging formula is as follows:
further, in step a, the method further includes assigning an edge weight value to the upper node of the variable, where the edge weight value has a calculation formula as follows:
W n-1,j ＝|V' n-1,j i/1 equation 4.
Further, in step B, the ordering rules include a first ordering rule and a second ordering rule; the first ordering rule is to order different layers from top to bottom according to the layers; the second ordering rule is to order the same hierarchy according to the defined weight.
Further, in step C, the abnormal node data is stored as structure data, and trace back information is output; the traceability information comprises an abnormal node name, an abnormal node value, a node normal range, an analysis layer { n }, a reason node, an abnormal transmission path and an abnormal transmission length.
Drawings
Fig. 1 is a schematic flow chart of a first embodiment of the present application.
Fig. 2 is a schematic diagram of a structure of an object for dividing energy consumption according to a first embodiment of the present application.
Fig. 3 is a schematic diagram of a directional non-weighted relationship according to a first embodiment of the present application.
FIG. 4 is a flow chart of a method according to a second embodiment of the application.
Detailed Description
The following is a further detailed description of the embodiments:
example 1
As shown in fig. 1, a hierarchical energy structure scene construction method in this embodiment is used for constructing a comprehensive hierarchical energy structure relationship diagram according to production units of a factory, and establishing energy consumption links between all the hierarchies according to the relationship diagram, and quickly and accurately acquiring energy consumption information according to the links, so as to facilitate timely intervention measures.
The method specifically comprises the following steps:
s1, according to the determined energy consumption object and the set hierarchical classification standard, a hierarchical relationship is designed by using a network tree diagram.
The classification standard classifies the energy consumption objects according to a physical hierarchy dividing mode of the factory-process section-process unit, so that the energy consumption objects correspond to the physical units in factory production, and the readability of analysis results is improved.
Meanwhile, when multi-layer nested energy consumption related elements appear in the same-level energy consumption objects, the nested energy consumption related elements are split and independently divided into multiple layers until each energy consumption related element has no nesting relation. The method can cover comprehensive energy consumption objects, clearly and accurately divide the energy consumption objects, ensure that the energy consumption objects of each level are accurately divided in place, and facilitate the carding of accurate energy consumption link diagrams.
In this embodiment, as shown in fig. 2, according to the raw material grinding process of the factory, the electricity consumption of the raw material grinding process is divided into two primary energy consumption objects, namely the electricity consumption of the raw material grinding process and the total yield of raw materials. Judging whether nested energy consumption related elements exist in the primary energy consumption object, dividing the primary energy consumption object into secondary energy consumption objects according to the related elements, and dividing raw material process electricity into two secondary energy consumption objects, namely 1# raw material process electricity and 2# raw material process electricity; the total raw material yield is divided into two secondary energy consumption objects of No. 1 grinding yield and No. 2 grinding yield.
Judging whether the second-level energy consumption object has nested energy consumption related elements or not, and dividing the third-level energy consumption object according to the related elements, for example, dividing the electricity consumption of the 1# raw material grinding process into the public electricity consumption of the raw material system and the electricity consumption of the 1# grinding system; the electricity consumption of the 2# raw material grinding process is divided into the common electricity consumption of the raw material system and the electricity consumption of the 2# raw material grinding system, and the like until each energy consumption related element has no nesting relation.
Specifically, the expression of the energy consumption related element set in each hierarchy is as follows:
L＝{L n,i ,L n-1,j ,...,L 1,k }，
wherein L is n,i Observation variable name, L, of the ith energy consumption object representing the nth layer energy consumption n-1,j And the observation variable name of the jth energy consumption object representing the n-1 layer energy consumption.
If L n-1,j The observed variable of the energy consumption object is obtained by combining nested energy consumption related elements A, B, C, and then the energy consumption object variable L n-1,j The energy consumption related element variable A, B, C is left in the n-1 layer and is placed in the n-2 layer; similarly, if the observed variable L of the energy consumption related element n-2,A And then by nested related element A 1 、A 2 、A 3 The combination is obtained, the variable A is left at the n-2 th layer, A 1 、A 2 、A 3 Placed on the n-3 th layer. The tree-shaped hierarchical relation graph of each energy consumption object is formed, the combination of global multi-hierarchy structured energy consumption data and factory physical units is realized, the energy consumption related elements of each hierarchy are ensured to correspond to the physical units in production, the tree-shaped hierarchical relation graph is segmented layer by layer, the energy consumption evaluation result can be unified with the actual production units, responsibility is convenient to arrive at a point, and accurate rectifying and modifying measures are formulated according to specific reasons.
S2, establishing a directed non-weighted relation graph.
In this embodiment, according to the constructed energy consumption related element set formula, the related energy consumption index is extracted, and the energy consumption index relationship is combed according to the hierarchy, so as to form a directed non-weighted relationship diagram. The method comprises the following substeps:
s2.1, disassembling the energy consumption calculation formulas of all layers, and disassembling and expressing the energy consumption related elements of the energy consumption calculation formulas of all layers as
L＝{L n,i ,L n-1,j ,...,L 1,k }；
Wherein L is the energy consumption related element disassembly set of each layer energy consumption calculation formula, wherein
L n,i ＝{L n,1 ,L n,2 ,...,L n,i }，
L n,i Observation variable name, L, of the ith energy consumption object representing the nth layer energy consumption n-1,j And the observation variable name of the jth energy consumption object representing the n-1 layer energy consumption.
And (3) traversing and disassembling the energy consumption calculation formulas of all layers, and finding out the observation variables of all the energy consumption objects of the n-1 layer related to the n layer through the formula disassembly.
S2.2, the observed variable results of the energy consumption object obtained by disassembling in the S2.1 form a network node pair:
(L n,i ,L n-1,1 ,W n-1,1 )
(L n,i ,L n-1,2 ,W n-1,2 )
...
(L n,i ,L n-1,j ,W n-1,j )
in which W is n-1,j The abnormal degree of the observed variable of the j-th energy consumption related element of the n-1 layer energy consumption is represented.
S2.3, L n,i Set as the initial node, L n-1,j Set as end node, W n-1,j Set as the weight value from the end node to the start node.
S2.4, traversing the steps S2.1-S2.3 until the establishment of node pairs of all levels is completed, and forming a complete network graph G, wherein the network graph G comprises node pair information of all energy consumption objects in a level structure and an initialization weight of each node pair, wherein the empty edge weight is an end node to start node initialization weight, and the initialization weight is null; wherein the node pair information includes a start node L n,i End node L n-1,j And L n,i To L n-1,j Whether to communicate or not; w (W) n-1,j Set as end node L n-1,j To the start node L n,i Is used to initialize the weight value. And assignment is required to be carried out according to the node abnormality degree in the later period so as to represent the relation strength between node pairs, thereby influencing the representation and the prediction result of the nodes. In this embodiment, the directional non-weighted relationship diagram shown in fig. 3 is formed through calculation. By means of the directed non-weight relation diagram, all energy consumption measuring points of the top layer and the lower layer can be communicated, and the abnormal states of the energy consumption objects of each layer can be conveniently checked by sequencing and outputting the energy consumption points according to the established weight relation diagram and the layers.
S3, constructing an outlier recognition algorithm model.
And (3) modeling by using the data of the last T time points of the history according to the energy consumption indexes of each level of the carding in the step (S2) through an outlier recognition algorithm. Specifically, the modeling step mainly includes the following sub-steps;
and S3.1, collecting data of T time points of the energy consumption history of each level.
S3.2 scaling the features using the following formula for outlier robust statistics
Wherein V 'is' i Is the normalized value of the sample, V i Representing the real-time value of the sample. media is the median of the sample tgun data and IQR (InterQuartile Range) is the quartile range of the sample tgun data. The median value is subtracted from the equation and the data is scaled according to the quantile range (default is IQR, the quartile range). IQR is the range between the 1 st quartile (25% th quartile) and the 3 rd quartile (75% th).
And (3) respectively centering and scaling each characteristic by calculating the relevant statistics of the samples in the training set. As shown in table 1 below: when the historical time t=6388 of the raw material grinding process electricity consumption is 15.155, 16.426, 17.863, respectively, and the 25%, 50%, 75% quantiles of the raw material grinding process electricity consumption are 15.155, 16.426, 17.863, respectively, medium=16.426, iqr= 17.863-15.155 = 2.708 in the formula 1, and the real time value V of the raw material grinding process electricity consumption at the i-th time is i ＝18.412，
Then
At the moment, centering the electricity consumption of the raw material grinding process is media= 16.426, and scaling result V 'of the electricity consumption real-time value of the raw material grinding process at the ith moment' i =0.733. Similar to the above case, other variables need to store the centering value and the quarter bit distance IQR in practical application, and then normalize the real-time data by using the "transformation" method of formula 1. Normalization of data sets is a common requirement for many machine-learning estimators. Typically by removing the average and scalingTo the unit variance. However, outliers tend to negatively impact the sample mean/variance. In this case, the median and quartile range of the above formula will generally give better results, thus constructing a more accurate outlier recognition algorithm model. The built model can automatically collect the change data of various equipment, various processes, raw material properties and the like, so that the built model is automatically updated in time, the real-time update of the overall energy consumption abnormality degree evaluation model of each node is realized, and the method is suitable for the change of the staged production condition.
In this embodiment, the median of each energy consumption node can be obtained through the calculation of S3 as shown in the following table:
TABLE 1
In the table, for ten energy consumption object node variables, the history t=6388 time sample points are used, and the median statistic is obtained. Therefore, a hierarchical energy structure scene is constructed, and abnormal data identification and analysis of each hierarchical energy object are facilitated.
In this embodiment, the energy consumption objects are classified according to the plant-plant section-process units, so that a comprehensive and accurate hierarchical relationship graph is constructed, the constructed hierarchical relationship is more attached to the actual production units of the plant, and the transmission path and the sequencing of the energy consumption anomalies can be given from the global point of view. Meanwhile, according to the divided hierarchical structure, the energy consumption object analysis of a designated hierarchy can be met, flexible disassembly and recombination of links are met, and the expandability is stronger.
Meanwhile, the historical data such as equipment, process, raw material properties and the like can be accurately and automatically acquired according to the established model through the established relation diagram, and further the data can be updated on time or in quantity or according to set triggering, so that real-time updating is realized, different analysis and evaluation objects are met, and the recognition accuracy of the abnormal recognition model on each energy consumption object is improved.
Generally, for larger production factories, the number of measuring points of energy consumption objects is large, in order to improve the identification accuracy of each energy consumption node, a point-to-point single-point measurement mode is adopted to refine the measurement number, reduce the data volume of a single measurement node, improve single-point pertinence, further ensure the measurement accuracy, and a mode of globally constructing the measurement node is not adopted. The application adopts a hierarchical division and layer-by-layer energy consumption object weight distribution mode to reasonably distribute each energy consumption object, and the relation among each energy consumption node is represented by a directed non-weight relation graph, so that the data quantity is reduced when the global relation graph is constructed, the hierarchical relation of each energy consumption object is definitely divided, the global node arrangement is ensured, the effect of multi-hierarchy structured link tracing from the global is realized, the problem that the global node structuring is difficult to realize is further overcome, and better model support is provided for realizing abnormal link tracing from the global.
Example two
In this embodiment, a method for performing link tracing on the energy consumption anomalies of each energy consumption object by applying the hierarchical energy structure scene model constructed in the first embodiment is also provided. As shown in fig. 4, the method mainly comprises the following steps:
and step 1, calculating the abnormality degree and state of each energy consumption index, and updating the abnormality degree and state into a directed non-weighted relation graph to form a directed weighted relation graph. And outputting the abnormal degree quantized value and the state judgment result of each energy consumption index of each level. And updating the abnormal degree quantized value of each energy consumption index into the directed non-weighted relation diagram established in the step S2. Specifically, the method comprises the following substeps:
step 1.1, calculating L according to equation 1 n-1,j The degree of abnormality of the variables is expressed as follows:
wherein V 'is' n-1,j Is L n-1,j Variable real-time value V n-1,j The degree of abnormality quantization value calculated according to equation 2. media L n-1，j Is L n-1，j Median, IQRL, of variable history T time point data n-1，j Is L n-1，j The variable history T is the quartile range of the time point data.
Step 1.2, judging L according to the following formula n-1，j Whether the variable is abnormal or not.
Wherein t is n-1，j Is a variable L n-1，j K is a variable L n-1，j Whether or not the threshold is normal. In this embodiment, values of a plurality of K are circulated, and the maximum value k=1 is appropriate corresponding to the actual requirement of the field abnormal energy consumption abnormality expert experience early warning.
Step 1.3, imparting L n-1 ， j The edge weight value to the upper node adopts the following formula:
W n-1，j ＝|V' n-1，j i/1 equation 4
Wherein W is n-1，j Is a variable L n-1，j To the edge weight value of the upper node.
Step 1.4, traversing the node L in the network graph G, and using the formulas 2-4 to obtain the degree of abnormality, the judgment result and the weight value of the upper-layer edge node of each energy consumption node, in this embodiment, the judgment statistical result of the abnormal value is shown in the following table,
TABLE 2
| count | mean | td | min | 25％ | 50％ | 75％ | max | |
| Raw material grinding process electricity consumption trans | 6388 | 5.672103 | 298.0028 | -6.06461 | -0.469356 | 0 | 0.530644 | 22483.78 |
| Raw material grinding process electricity consumption_job | 6388 | 0.140106 | 0.347125 | 0 | 0 | 0 | 0 | 1 |
| Electricity-trans for raw material procedure | 6388 | 0.012126 | 17.53359 | -2.25407 | -0.858448 | 1.48E-16 | 0.141552 | 1048.323 |
| Raw material process electricity consumption_job | 6388 | 0.09737 | 0.296484 | 0 | 0 | 0 | 0 | 1 |
| Total yield of raw meal_trans | 6388 | -0.23927 | 0.576818 | -1.89111 | -0.764444 | 0 | 0.235556 | 9.037778 |
| Total yield of raw meal_job | 6388 | 0.079211 | 0.270089 | 0 | 0 | 0 | 0 | 1 |
| Electric trans for 1# raw material grinding process | 6388 | 1.966057 | 253.0007 | -3177.12 | -0.843797 | 0 | 0.156203 | 19968.55 |
| Process electricity consumption_plug of 1# raw material grinding | 6388 | 0.239512 | 0.426819 | 0 | 0 | 0 | 0 | 1 |
| Yield of 1# mill_trans | 6388 | -0.52834 | 1.159849 | -2.94937 | -0.778481 | 0 | 0.221519 | 13.77215 |
| Yield of mill 1 _ j uge | 6388 | 0.232624 | 0.422537 | 0 | 0 | 0 | 0 | 1 |
| 2# mill yield_trans | 6388 | -0.4551 | 0.970074 | -2.41081 | -0.816216 | 0 | 0.183784 | 9.897297 |
| Yield of 2# mill_joint | 6388 | 0.233876 | 0.423327 | 0 | 0 | 0 | 0 | 1 |
| Common electricity_trans of raw material system | 6388 | -0.0587 | 0.838543 | -4.75918 | -0.543147 | 0 | 0.456853 | 26.3819 |
| Common electricity of raw material system_job | 6388 | 0.119912 | 0.324884 | 0 | 0 | 0 | 0 | 1 |
| Electric-trans for No. 1 grinding system | 6388 | 2.024779 | 256.0939 | -3215.71 | -0.850932 | 0 | 0.149068 | 20212.79 |
| 1# grinding system electricity consumption_job | 6388 | 0.237007 | 0.42528 | 0 | 0 | 0 | 0 | 1 |
| Electric trans for 2# raw material grinding process | 6388 | -0.15581 | 28.79946 | -3.12209 | -0.861285 | 1.83E-16 | 0.138715 | 2299.805 |
| Process electricity consumption_plug of 2# raw material mill | 6388 | 0.235285 | 0.42421 | 0 | 0 | 0 | 0 | 1 |
| 2# raw material grinding system electricity-trans | 6388 | -0.12313 | 29.04585 | -3.01381 | -0.841459 | -1.85E-16 | 0.158541 | 22319.664 |
| Electricity consumption_plug of 2# raw material grinding system | 6388 | 0.233719 | 0.423229 | 0 | 0 | 0 | 0 | 1 |
In table 2, the abnormal value determination results are obtained for 10 energy consumption node variables in the first embodiment using the historical t=6388 time sample pointsStatistics, wherein the variable ending with ×_trans is the calculation result statistics of outlier robustness for the ×variable according to formula 2; the variables ending with ×_j are used to make a threshold determination as to whether the variables are normal according to equation 3. The weight of all energy consumption node variables to the upper edge is given by equation 4, where V' n-1，j Calculated =.x_trans.
In this embodiment, the energy consumption abnormality degree is represented according to the relative position relationship between the real-time energy consumption data and the historical energy consumption data distribution, so that the energy consumption can be monitored in real time, and even if the energy consumption monitoring point does not exceed the set threshold value, the link analysis prediction can be performed according to the evaluation of the abnormality degree, so that the real-time online evaluation can be achieved.
And step 2, carrying out hierarchical ordering on the judged abnormal nodes according to a set ordering rule. The first ordering rule is as follows: for different levels, the levels are ordered from top to bottom, i.e. from the first level energy consumption object downwards. The second ordering rule is: and for the same hierarchy, sorting the abnormal degrees according to the defined weight, and outputting all abnormal node pair information.
Wherein t is selected according to formula 3 n-1，j All abnormal nodes of =1, then L for different levels according to the first ordering rule n-1，j Sorting variables, e.g. L n-1，j ＞L n-2，j ＞L n-3，j At the same level, the energy consumption objects are ordered according to a second ordering rule, such as W n-1，j ＞W n-1，k ＞W n-1，m . And (3) orderly arranging all the abnormal nodes according to the ordering rule, and ensuring the clear display of the abnormal nodes.
In this embodiment, the result of the hierarchical ordering of the abnormal nodes is shown in the following table,
TABLE 3 Table 3
| node_levle | node_name | node_value | node_trans | node_juge | node_trans_abs |
| 0 | Electricity consumption in raw material grinding process | 6.787 | -3.558796382 | 1 | 3.558796382 |
| 1 | Total yield of raw materials | 2129 | 2.837777778 | 1 | 2.837777778 |
| 2 | Yield of No. 1 mill | 1110 | 4.075949367 | 1 | 4.075949367 |
| 2 | Yield of No. 2 mill | 1019 | 3.097297297 | 1 | 3.097297297 |
As can be seen from table 3, the raw material grinding process power consumption node_job=1 of the node_level=0 layer was abnormal. The lower node causing the above index abnormality is the total yield of raw meal of node_1evel=1 layer. Continuing the lower layer trace back, the lower layer indicators of the total applied raw meal yield are mill yield No. 1 and mill yield No. 2, wherein the mill yield No. 1 is ordered before mill yield No. 1 because the degree of abnormality of mill yield no_trans_abs=4.08 is greater than the degree of abnormality of mill yield No. 2_trans_abs=3.10. And thus, the ordering position relation of each abnormal node is obtained.
And step 3, calculating the shortest path of each abnormal node according to the obtained abnormal nodes and storing the shortest paths.
Specifically, according to the directed weighted relation diagram established in the step 1, the ordered abnormal node pairs obtained in the step 2 are calculated one by one. And if the node pairs can be communicated and the initial node is not a leaf node, outputting the traceability information. The traceability information comprises an abnormal node name, an abnormal node value, a node normal range, an analysis layer { n }, a reason node, an abnormal transmission path and an abnormal transmission length, and is stored as structural data. The analysis results are stored according to the hierarchical structure, so that the analysis results are conveniently presented by using the structured tree diagram, anomalies in the network are more intuitively and clearly found, and the analysis results are convenient to inquire and trace.
In this embodiment, according to the ordered abnormal node pairs, the abnormal state of each energy consumption object node can be intuitively obtained, each energy consumption abnormal link influencing the upper layer energy consumption is found out according to weight analysis, and effective correction measures are formulated according to the traced energy consumption abnormal links, so that the intervention efficiency on the abnormal state is improved.
For large-scale production enterprises, the number of energy consumption measuring points is huge, and especially when energy consumption data with a hierarchical relationship are summarized layer by layer, the more upwards, the energy consumption data are calculated through a bottom layer energy consumption measuring point formula, and are not physical measuring points. Therefore, the abnormal point cannot be found accurately due to the fact that the abnormal degree of the upper layer is large because the abnormal point is generally influenced by the energy consumption data of each node of the lower layer. According to the application, in the energy consumption analysis process, the transmission path with abnormal energy consumption can be restored through link tracing, so that the lower-layer measuring point with the largest influence on the upper-layer energy consumption is found out, and targeted processing is performed. And realizing the trace back analysis of the multi-level structured energy consumption abnormal full link.
Meanwhile, when the lower energy consumption is in a normal value range but is in a slightly higher state, the lower energy consumption does not exceed an energy consumption alarm threshold, but the upper energy consumption is abnormal due to the error accumulation effect of each level. At the moment, according to the designed weight relation diagram, the contribution degree of lower energy consumption to upper energy consumption is quantitatively evaluated, so that the main and secondary influence energy consumption points of the lower layer and the ordering condition of the abnormal transmission paths can be determined, the position of the energy consumption abnormality can be accurately positioned, and effective measures can be taken.
The foregoing is merely exemplary of the present application, and specific technical solutions and/or features that are well known in the art have not been described in detail herein. It should be noted that, for those skilled in the art, several variations and modifications can be made without departing from the technical solution of the present application, and these should also be regarded as the protection scope of the present application, which does not affect the effect of the implementation of the present application and the practical applicability of the patent. The protection scope of the present application is subject to the content of the claims, and the description of the specific embodiments and the like in the specification can be used for explaining the content of the claims.
Claims (10)
1. The method for constructing the hierarchical energy structure scene is characterized by comprising the following steps of: the method is used for constructing a global hierarchical structure identification model of each energy consumption object in the factory and specifically comprises the following steps of:
dividing the energy consumption objects according to a set hierarchical classification standard to form a tree-shaped hierarchical relationship;
step two, extracting relevant energy consumption indexes of all the energy consumption objects, and establishing a directed non-weighted relation graph of all the energy consumption objects according to the energy consumption indexes; constructing a network diagram G;
and thirdly, constructing an abnormal value recognition algorithm model and building an energy structure scene.
2. The hierarchical energy structure scene construction method according to claim 1, wherein: classifying the energy consumption objects according to the physical level of the factory-process section-process unit by the classification standard; splitting energy consumption related elements which are nested in the same hierarchy layer by layer, and expressing the energy consumption related element set in each hierarchy layer as:
l＝{l N,i ，L n-1,j ，...，L 1,k }。
3. the hierarchical energy structure scene construction method according to claim 1, wherein: in the second step, the method further comprises the following sub-steps:
(1) Disassembling the energy consumption calculation formulas of all layers, and disassembling and expressing energy consumption elements in the same layer of energy consumption calculation formulas as
L n ＝{L n,1 ，L n，2 ,...,L n,i }；
(2) The energy consumption object observation variable result L obtained by disassembly n,i With the lower layer node L n-1 The connection constitutes a network node pair:
(L n,i ,L n-1,1 ,W n-1,1 )
(L n,i ,L n-1,2 ,W n-1,2 )
...
(L n,i ,L n-1,j ,W n-1,j )；
(3) Will L n,i Set as the initial node, L n-1,j Set as end node, W n-1,j Setting a weight value from an end node to a start node;
(4) And (3) traversing the steps (1) - (3) until the node pairs of all the levels are established, and forming a complete network graph G.
4. A hierarchical energy structure scene construction method as claimed in claim 3, wherein: the network diagram G comprises node pair information of all energy consumption objects in a hierarchical structure and an initialization weight of each node pair, wherein the initialization weight is null; wherein the node pair information includes a startNode L n,i End node L n-1,j And L n,i To L n-1,j Whether to communicate or not; w (W) n-1,j Set as end node L n-1,j To the start node L n,i Is used to initialize the weight value.
5. The hierarchical energy structure scene construction method according to claim 1, wherein: and thirdly, collecting data of T time points of the energy consumption history of each level, and scaling the characteristics by adopting the following formula to carry out robust statistical information on the abnormal values:
6. the hierarchical energy consumption abnormal link tracing method is characterized by comprising the following steps of: the scene construction method applied to any of the above claims 1-5, comprising the steps of:
step A, calculating the abnormality degree and state of each energy consumption index, and updating the abnormality degree and state into a directed non-weight relation diagram to form a directed weighted relation diagram;
step B, carrying out hierarchical ordering on the judged abnormal node pairs according to a set ordering rule;
and step C, calculating the shortest path of each abnormal node according to the obtained abnormal node pair, and storing and outputting the shortest path.
7. The hierarchical energy consumption anomaly link tracing method of claim 6, wherein: in the step A, the abnormal degree of each energy consumption index is output through a quantized value, and a state judgment result is output; the variable calculation mode of the abnormality degree adopts the following formula:
judging an abnormal state according to a formula 2, wherein the abnormal state judging formula is as follows:
8. the hierarchical energy consumption anomaly link tracing method of claim 7, wherein: in the step A, the method further comprises the step of endowing the variables with edge weight values of upper nodes, wherein the calculation formula of the edge weight values is as follows:
W n-1,j ＝|V′ n-1,j i/1 equation 4.
9. The hierarchical energy consumption anomaly link tracing method of claim 6, wherein: in step B, the ordering rules include a first ordering rule and a second ordering rule; the first ordering rule is to order different layers from top to bottom according to the layers; the second ordering rule is to order the same hierarchy according to the defined weight.
10. The hierarchical energy consumption anomaly link tracing method of claim 6, wherein: in the step C, the abnormal node data is stored as structural data, and trace back information is output; the traceability information comprises an abnormal node name, an abnormal node value, a node normal range, an analysis layer { n }, a reason node, an abnormal transmission path and an abnormal transmission length.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311261935.0A CN117113233A (en) | 2023-09-27 | 2023-09-27 | Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311261935.0A CN117113233A (en) | 2023-09-27 | 2023-09-27 | Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117113233A true CN117113233A (en) | 2023-11-24 |
Family
ID=88800230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311261935.0A Pending CN117113233A (en) | 2023-09-27 | 2023-09-27 | Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117113233A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117763484A (en) * | 2024-02-21 | 2024-03-26 | 新奥数能科技有限公司 | Energy use abnormality diagnosis method and device based on enterprise energy use space |
-
2023
- 2023-09-27 CN CN202311261935.0A patent/CN117113233A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117763484A (en) * | 2024-02-21 | 2024-03-26 | 新奥数能科技有限公司 | Energy use abnormality diagnosis method and device based on enterprise energy use space |
| CN117763484B (en) * | 2024-02-21 | 2024-05-14 | 新奥数能科技有限公司 | Energy use abnormality diagnosis method and device based on enterprise energy use space |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113572625B (en) | Fault early warning method, early warning device, equipment and computer medium | |
| CN113779496A (en) | Power equipment state evaluation method and system based on equipment panoramic data | |
| Chien et al. | A data mining approach for analyzing semiconductor MES and FDC data to enhance overall usage effectiveness (OUE) | |
| Chang et al. | Clustering and visualization of failure modes using an evolving tree | |
| CN113762329A (en) | Method and system for constructing state prediction model of large rolling mill | |
| CN117113233A (en) | Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method | |
| CN110757510B (en) | Method and system for predicting remaining life of robot | |
| CN111310139A (en) | Behavior data identification method and device and storage medium | |
| CN114273977A (en) | MES-based cutter wear detection method and system | |
| CN113343587A (en) | Flow abnormity detection method for electric power industrial control network | |
| CN116485020B (en) | Supply chain risk identification early warning method, system and medium based on big data | |
| CN115664038A (en) | Intelligent power distribution operation and maintenance monitoring system for electrical safety management | |
| CN117808166A (en) | Chemical industry safety automation detection monitoring system of clouding PLC | |
| CN112711912B (en) | Air quality monitoring and alarming method, system, device and medium based on cloud computing and machine learning algorithm | |
| CN113283546B (en) | Furnace condition abnormity alarm method and system of heating furnace integrity management centralized control device | |
| Wu et al. | Imaging feature-based clustering of financial time series | |
| CN117216702A (en) | Power transformation equipment parameter processing method and device and electronic equipment | |
| CN117171157A (en) | Clearing data acquisition and cleaning method based on data analysis | |
| CN116030955B (en) | Medical equipment state monitoring method and related device based on Internet of things | |
| Chen et al. | Big data analytic for multivariate fault detection and classification in semiconductor manufacturing | |
| Saravanakumar et al. | An efficient ensemble model for forecasting time series analysis of crypto currency data using machine learning | |
| KR20210126378A (en) | Real-time sliding window based anomaly detection system for multivariate data generated by manufacturing equipment | |
| CN114896228B (en) | Industrial data stream cleaning model and method based on filtering rule multistage combination optimization | |
| CN115330362A (en) | Engineering progress data processing method and system | |
| CN112836570B (en) | Equipment abnormality detection method utilizing Gaussian noise |
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 |