CN110378371B - Energy consumption abnormity detection method based on average neighbor distance abnormity factor - Google Patents
Energy consumption abnormity detection method based on average neighbor distance abnormity factor Download PDFInfo
- Publication number
- CN110378371B CN110378371B CN201910503050.4A CN201910503050A CN110378371B CN 110378371 B CN110378371 B CN 110378371B CN 201910503050 A CN201910503050 A CN 201910503050A CN 110378371 B CN110378371 B CN 110378371B
- Authority
- CN
- China
- Prior art keywords
- neighbor distance
- data
- energy consumption
- average
- factor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 55
- 238000001514 detection method Methods 0.000 title claims abstract description 37
- 230000002159 abnormal effect Effects 0.000 claims abstract description 43
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000000034 method Methods 0.000 claims description 31
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 230000005856 abnormality Effects 0.000 claims description 5
- 230000001174 ascending effect Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 238000002360 preparation method Methods 0.000 claims 1
- 230000008569 process Effects 0.000 description 7
- 238000004519 manufacturing process Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000009194 climbing Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 2
- 238000001125 extrusion Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 229910021420 polycrystalline silicon Inorganic materials 0.000 description 1
- 229920005591 polysilicon Polymers 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Manufacturing & Machinery (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Public Health (AREA)
- Automation & Control Theory (AREA)
- Probability & Statistics with Applications (AREA)
- Water Supply & Treatment (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention provides an energy consumption abnormity detection method based on an average neighbor distance abnormity factor, which comprises the following steps: acquiring energy consumption data and converting the energy consumption data into alternating data; defining a time sequence characteristic value of the energy consumption data, and dividing the time sequence into subsequences which are respectively mapped to a four-dimensional characteristic space; respectively calculating average neighbor distance abnormal factors of the time subsequences in a four-dimensional feature space; processing the average neighbor distance abnormal factor of the subsequence to obtain an average neighbor distance abnormal factor of the time sequence; and calculating an average neighbor distance abnormal threshold according to the average neighbor distance abnormal factor, and judging whether the mode is abnormal or not. The energy consumption abnormity detection method based on the average neighbor distance abnormity factor eliminates the interference of mode abnormity detection, effectively improves the precision of mode abnormity detection, accurately positions the abnormity position, and effectively detects the abnormal data appearing in step-type and alternating performance consumption data.
Description
Technical Field
The invention relates to the technical field of energy consumption data anomaly detection and energy consumption state monitoring, in particular to an energy consumption anomaly detection method based on an average neighbor distance anomaly factor.
Background
With the development of industrial information technology, many high energy consuming enterprises have set up energy management systems to realize real-time acquisition of energy consumption data, with the purpose of optimizing energy operation and saving production cost. Therefore, high energy consumption enterprises also put higher requirements on real-time detection and monitoring of energy consumption data, and detection technologies with real-time performance and intelligence are needed. The production condition of the high energy consumption machine is complex, the long-term full load operation has high probability of abnormal energy consumption [1], [2], thereby generating a large amount of abnormal energy consumption data. In the production process, data of a plurality of high-energy consumption machines (such as a hydraulic machine and a polysilicon reduction furnace) has the characteristics of step, alternation, periodicity and the like. The abnormal energy consumption of the machine is accompanied by a great amount of energy loss and energy efficiency reduction, and even can cause shutdown and safety accidents which cannot be calculated, thereby influencing the normal production of the whole production line. Therefore, it is very important to develop a reliable, fast and automatic abnormal energy consumption detection technology. By using the new methods, production enterprises can monitor and process high-energy-consumption machines, avoid energy consumption loss and improve the energy efficiency of the machines. The pattern anomaly is a typical anomaly form in the energy consumption anomaly data, and the current challenge is to develop a reliable, rapid and automatic pattern anomaly detection technology.
At present, the abnormality detection is generally realized by manually inspecting and recording meter data, which has great hysteresis. And the energy consumption alarm threshold set through manual experience is a fixed and unchangeable threshold and cannot adapt to the requirement of real-time change of energy consumption data. The time series data abnormality detection method includes: distance-based anomaly detection, prediction-based anomaly detection, cluster-based anomaly detection, and the like [3] [4] [5]. The conventional anomaly detection methods are mainly classified into a local anomaly detection method and a global anomaly detection method.
The local anomaly detection method excessively emphasizes small local change, so that higher false detection rate and lower expandability are caused; the global anomaly detection method neglects local slight anomaly, so that the report missing rate is high. Their detection efficiency is low and adaptability is poor due to lack of proper optimization [6]. Therefore, the prior art cannot effectively detect abnormal data occurring in the step-wise and alternating performance consumption data.
Disclosure of Invention
The invention provides an energy consumption abnormity detection method based on an average neighbor distance abnormity factor, aiming at overcoming the technical defect that the existing energy consumption abnormity detection method cannot effectively detect abnormal data appearing in jump-type and alternating performance consumption data.
In order to solve the technical problems, the technical scheme of the invention is as follows:
an energy consumption anomaly detection method based on average neighbor distance anomaly factors comprises the following steps:
s1: acquiring energy consumption data, and converting the energy consumption data into alternating data by a rain flow counting method;
s2: defining a time sequence characteristic value of the energy consumption data, and dividing the time sequence into subsequences which are respectively mapped to a four-dimensional characteristic space;
s3: respectively calculating average neighbor distance abnormal factors of the time subsequences in a four-dimensional feature space;
s4: processing the average neighbor distance abnormal factor of the subsequence to obtain an average neighbor distance abnormal factor of the time sequence;
s5: and calculating an average neighbor distance abnormal threshold according to the average neighbor distance abnormal factor, and judging whether the mode is abnormal or not.
Wherein, the step S1 specifically includes the following steps:
s11: collecting energy consumption data to obtain a time sequence X = (X) of the energy consumption data 1 ,x 2 ,...x n ) The period length CL, the length m of the sliding window and the weight factor lambda of the abnormal characteristic are calculated;
s12: converting the energy consumption data into alternating data by a rain flow counting method, wherein the specific calculation formula is as follows:
(X i -X i-1 )(X i -X i+1 )<0i=2,3,4,...,M-1;
wherein X i Energy consumption data collected on site; and the end points of the data segment are peak-valley points, filtering is carried out according to the standard of a peak-valley sequence, namely a PV sequence, all non-peak-valley points are deleted, and the interference data of an ascending slope and a descending slope are deleted to obtain alternating data.
Wherein, the step S2 specifically includes the following steps:
s21: time series X = (X) according to sliding window length m 1 ,x 2 ,...x n ) Intercepting into a plurality of subsequences;
s22: defining four characteristic values of the time subsequence, specifically including:
the subsequence height h is defined as:
h=x max -x min ;
variance σ of the subsequence 2 Is defined as:
the maximum adjacent distance m of a subsequence is defined as:
m=max(|x i -x i-1 |),(i=2,3,4...n);
In the step S3, energy data are clustered into 2 types by adopting a K-Means clustering method, and then average neighbor distance abnormal factors of the 2 types of time sequences are respectively calculated by adopting an average neighbor distance abnormal factor MNNDAF algorithm.
The MNNDAF algorithm specifically comprises the following steps:
s31: setting pointAnd pointInto a four-dimensional feature spaceAt an arbitrary point of (2), the Euclidean distance of the feature spaceExpressed as:
s32: defining the average nearest neighbor distance MNND as the average value of the distances between the point p and all the adjacent points, and recording the k-nearest neighbor distance of the point p as k-dist (p); let k be an element of N +,the average nearest neighbor distance MNND for point p is defined as:
s33: is provided withThe eigenvalue subspaces of the four eigenvalues of point C are each C h (h 1 ,h 2 ,...,h n ),C σ (σ 1 ,σ 2 ,...,σ n ),C m (m 1 ,m 2 ,...,m n ) Separately calculating the point c in the four-dimensional feature spaceAnd the average nearest neighbor distance in the feature value subspace, MNND, MNNDh (c),MNND σ (c) and MNNDm (c), so far, the mean neighbor distance anomaly factor for point c is:
MNNDAF(c)=MNND(c)+max{X};
Wherein, the step S4 specifically includes the following steps:
s41: interpolating the average neighbor distance anomaly factor value of the subsequence in the 2 types of energy data;
s42: normalizing the interpolated energy data;
s43: and reordering the energy data after the normalization processing according to the sequence of the original data to obtain the average neighbor distance abnormal factor of the time sequence.
The specific process of the normalization processing in step S42 is:
establishing a mapping from c to c' such that L of c 2 Norm 1, then:
In step S5, a specific calculation formula of the average neighbor distance anomaly threshold δ is:
wherein alpha is a mode abnormity coefficient set according to actual conditions, and if MNNDAF is larger than delta, the time series of the energy consumption data is judged to be mode abnormity.
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
the invention provides an energy consumption anomaly detection method based on average neighbor distance anomaly factors, which adopts a rain flow counting method to eliminate the interference of mode anomaly detection; improving the accuracy of mode anomaly detection by adopting a K-Means clustering method, and accurately positioning the anomaly position; the MNNDAF algorithm is adopted to accurately detect the mode abnormity of the energy data time sequence, and the abnormal data appearing in the step-type and alternating performance consumption data are effectively detected by the method.
Drawings
FIG. 1 is a schematic flow diagram of the process;
FIG. 2 is a schematic time series diagram of raw energy data;
FIG. 3 is a schematic diagram of a time series of energy data processed by a rain flow counting method;
fig. 4 is a schematic diagram of a pattern abnormality detection result.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the present embodiments, certain elements of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described with reference to the drawings and the embodiments.
Example 1
As shown in fig. 1, an energy consumption anomaly detection method based on an average neighbor distance anomaly factor includes the following steps:
s1: acquiring energy consumption data, and converting the energy consumption data into alternating data by a rain flow counting method;
s2: defining a time sequence characteristic value of the energy consumption data, and dividing the time sequence into subsequences which are respectively mapped to a four-dimensional characteristic space;
s3: respectively calculating average neighbor distance abnormal factors of the time subsequences in a four-dimensional feature space;
s4: processing the average neighbor distance abnormal factor of the subsequence to obtain an average neighbor distance abnormal factor of the time sequence;
s5: and calculating an average neighbor distance abnormal threshold according to the average neighbor distance abnormal factor, and judging whether the mode is abnormal.
More specifically, the step S1 specifically includes the following steps:
s11: acquiring energy consumption data to obtain a time series X = (X) of the energy consumption data 1 ,x 2 ,...x n ) The period length CL, the length m of the sliding window and the weight factor lambda of the abnormal characteristic are calculated;
s12: converting the energy consumption data into alternating data by a rain flow counting method, wherein the specific calculation formula is as follows:
(X i -X i-1 )(X i -X i+1 )<0i=2,3,4,...,M-1;
wherein, X i Energy consumption data collected on site; and the end points of the data segment are peak-valley points, filtering is carried out according to a peak-valley sequence, namely the standard of a PV sequence, all non-peak-valley points are deleted, and the interference data of an ascending slope and a descending slope are deleted to obtain alternating data.
More specifically, the step S2 specifically includes the following steps:
s21: time series X = (X) according to sliding window length m 1 ,x 2 ,...x n ) Intercepting into a plurality of subsequences;
s22: defining four characteristic values of the time subsequence, specifically including:
the subsequence height h is defined as:
h=x max -x min ;
variance σ of the subsequence 2 Is defined as:
the maximum neighbor distance m of a subsequence is defined as:
m=max(|x i -x i-1 |),(i=2,3,4...n);
More specifically, in the step S3, the energy data are clustered into 2 classes by using a K-Means clustering method, and then the average neighbor distance anomaly factors of the 2 classes of time series are respectively calculated by using an average neighbor distance anomaly factor mnnda algorithm.
More specifically, the mnnda algorithm specifically includes the following steps:
s31: setting pointAnd pointIn a four-dimensional feature spaceThe euclidean distance of the feature space is then expressed as:
s32: defining the average nearest neighbor distance MNND as the average value of the distances between the point p and all the near neighbor points, and recording the k-nearest neighbor distance of the point p as k-dist (p); let k be equal to N + ,The average nearest neighbor distance MNND for point p is defined as:
s33: is provided withThe feature value subspaces of the four feature values of point C are C h (h 1 ,h 2 ,...,h n ),C σ (σ 1 ,σ 2 ,...,σ n ),C m (m 1 ,m 2 ,...,m n ) Separately computing the point c in a four-dimensional feature spaceAnd the average nearest neighbor distance in the feature value subspace, MNND, MNNDh (c),MNND σ (c) and MNNDm (c), so far, the average neighbor distance anomaly factor for point c is:
MNNDAF(c)=MNND(c)+max{X};
More specifically, the step S4 specifically includes the following steps:
s41: interpolating the average neighbor distance anomaly factor value of the subsequence in the 2 types of energy data;
s42: normalizing the interpolated energy data;
s43: and reordering the energy data after the normalization processing according to the sequence of the original data to obtain the average neighbor distance abnormal factor of the time sequence.
More specifically, the specific process of the normalization process in step S42 is:
establishing a mapping from c to c' such that L of c 2 Norm 1, then:
More specifically, in step S5, the specific calculation formula of the average neighbor distance anomaly threshold δ is as follows:
wherein alpha is a mode abnormity coefficient set according to actual conditions, and if MNNDAF is larger than delta, the time series of the energy consumption data is judged to be mode abnormity.
In the specific implementation process, the method adopts a rain flow counting method to eliminate the interference of mode anomaly detection; improving the accuracy of mode anomaly detection by adopting a K-Means clustering method, and accurately positioning the anomaly position; the MNNDAF algorithm is adopted to accurately detect the mode abnormity of the energy data time sequence, and the abnormal data appearing in the step-type and alternating performance consumption data are effectively detected by the method.
Example 2
More specifically, on the basis of embodiment 1, the proposed anomaly detection method is implemented by MATLAB programming, and the compiling tool is: MATLAB R2018a, runtime environment: windows 7 or above, hardware: client computer CPU3.3G, and memory 4,0M.
In the specific implementation process, firstly, the collected energy data is imported, and relevant parameters are adjusted, specifically including: cycle length CL =100; setting a classification number, namely the number of the centers of the initial clusters n =2; sliding window length m =5; normalized coefficientThe mode anomaly coefficient α =1.4.
As shown in fig. 2, there are 5 periods in the original energy data time series diagram. The black oval marks the three mode anomalies that need to be detected. Data at the positions of the 'climbing' and the 'descending' and data of 'trial extrusion' before the 'climbing' are interference data, and the interference data need to be cleared before detection.
The time sequence of the energy data processed by the rain flow counting method is shown in fig. 3, and it can be seen that interference data at the position of the trial extrusion before the climbing, the downhill and the climbing are removed by the processing of the rain flow counting method; the data segment of the load forward is translated upwards by a distance d, so that the distinguishing degree of no-load operation and load operation is effectively improved, and a K-Means clustering method is used for classification; as can be seen from fig. 4, the anomaly detection method provided by the present invention correctly detects the mode anomalies at the "no-load operation" and the "load operation".
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
[1]Li L,Huang H,Zhao F,et al.Operation scheduling of multi–hydraulic press system for energy consumption reduction[J].Journal of Cleaner Production,2017,165.
[2]Gao M,Li X,Huang H,et al.Energy–saving Methods for Hydraulic Presses Based on Energy Dissipation Analysis[J].Procedia Cirp,2016,48:331–335.
[3]Ahmed M,Mahmood A N,Maher M J.Heart Disease Diagnosis Using Co–clustering[M]//Scalable Information Systems.Springer International Publishing,2014:61–70.
[4]Cecílio I M,Ottewill J R,Pretlove J,et al.Nearest neighbors method for detecting transient disturbances in process and electromechanical systems[J].Journal of Process Control,2014,24(9):1382–1393.
[5]Drugman T.Using mutual information in supervised temporal event detection:Application to cough detection[J].Biomedical Signal Processing&Control,2014,10(1):50–57.
[6]Ren H,Ye Z,Li Z.Anomaly detection based on a dynamic Markov model[J].Information Sciences,2017,411.
Claims (4)
1. An energy consumption anomaly detection method based on an average neighbor distance anomaly factor is characterized by comprising the following steps:
s1: acquiring energy consumption data, and converting the energy consumption data into alternating data by a rain flow counting method;
s2: defining a time sequence characteristic value of the energy consumption data, and dividing the time sequence into subsequences which are respectively mapped to a four-dimensional characteristic space;
s3: respectively calculating average neighbor distance abnormal factors of the time subsequences in a four-dimensional feature space;
s4: processing the average neighbor distance abnormal factor of the subsequence to obtain an average neighbor distance abnormal factor of the time sequence;
s5: calculating an average neighbor distance abnormal threshold according to the average neighbor distance abnormal factor, and judging whether a mode is abnormal or not;
the step S1 specifically includes the steps of:
s11: acquiring energy consumption data to obtain a time series X = (X) of the energy consumption data 1 ,x 2 ,...x n ) The period length CL, the length m of the sliding window and the weight factor lambda of the abnormal characteristic are calculated;
s12: converting the energy consumption data into alternating data by a rain flow counting method, wherein the specific calculation formula is as follows:
(X i -X i-1 )(X i -X i+1 )<0
wherein i =2,3,4., M-1,X i Energy consumption data collected on site; the end points of the data segments are peak-valley points, filtering is carried out according to the standard of a peak-valley sequence, namely a PV sequence, all non-peak-valley points are deleted, and interference data of an ascending slope and a descending slope are deleted to obtain alternating data;
the step S2 specifically includes the following steps:
s21: time series X = (X) according to sliding window length m 1 ,x 2 ,...x n ) Intercepting a plurality of subsequences;
s22: defining four characteristic values of the time subsequence, specifically including:
the subsequence height h is defined as:
h=x max -x min ;
variance σ of the subsequence 2 Is defined as:
the maximum adjacent distance m of a subsequence is defined as:
m=max(|x i -x i-1 |),(i=2,3,4...n);
s23: mapping temporal subsequences to four-dimensional feature spaceThe preparation method comprises the following steps of (1) performing;
s3, clustering the energy data into 2 types by adopting a K-Means clustering method, and calculating average neighbor distance abnormal factors of 2 types of time sequences respectively by adopting an average neighbor distance abnormal factor MNNDAF algorithm;
the MNNDAF algorithm specifically comprises the following steps:
s31: setting pointAnd pointInto a four-dimensional feature spaceThe euclidean distance of the feature space is then expressed as:
s32: defining the average nearest neighbor distance MNND as the average value of the distances between the point p and all the near neighbor points, and recording the k-nearest neighbor distance of the point p as k-dist (p); let k be an element of N + ,The average nearest neighbor distance MNND for point p is defined as:
s33: is provided withThe feature value subspaces of the four feature values of point C are C h (h 1 ,h 2 ,...,h n ),C σ (σ 1 ,σ 2 ,...,σ n ),C m (m 1 ,m 2 ,...,m n ) Separately calculating the point c in the four-dimensional feature spaceAnd the average nearest neighbor distance in the feature value subspace, MNND, MNNDh (c),MNND σ (c) and MNNDm (c), so far, the mean neighbor distance anomaly factor for point c is:
MNNDAF(c)=MNND(c)+max{X};
2. The method according to claim 1, wherein the step S4 specifically includes the following steps:
s41: interpolating the average neighbor distance anomaly factor value of the subsequence in the 2 types of energy data;
s42: normalizing the interpolated energy data;
s43: and reordering the energy data after the normalization processing according to the sequence of the original data to obtain the average neighbor distance abnormal factor of the time sequence.
4. The method for detecting energy consumption anomaly based on average nearest neighbor distance anomaly factor according to claim 3, wherein in said step S5, the specific calculation formula of the average nearest neighbor distance anomaly threshold δ is:
wherein α is a mode abnormality coefficient set according to an actual condition, and if MNNDAF > δ, the time series of the energy consumption data is determined as a mode abnormality.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910503050.4A CN110378371B (en) | 2019-06-11 | 2019-06-11 | Energy consumption abnormity detection method based on average neighbor distance abnormity factor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910503050.4A CN110378371B (en) | 2019-06-11 | 2019-06-11 | Energy consumption abnormity detection method based on average neighbor distance abnormity factor |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110378371A CN110378371A (en) | 2019-10-25 |
CN110378371B true CN110378371B (en) | 2022-12-16 |
Family
ID=68250136
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910503050.4A Active CN110378371B (en) | 2019-06-11 | 2019-06-11 | Energy consumption abnormity detection method based on average neighbor distance abnormity factor |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110378371B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112381181B (en) * | 2020-12-11 | 2022-10-04 | 桂林电子科技大学 | Dynamic detection method for building energy consumption abnormity |
CN112966017B (en) * | 2021-03-01 | 2023-11-14 | 北京青萌数海科技有限公司 | Abnormal subsequence detection method for indefinite length in time sequence |
CN113048807B (en) * | 2021-03-15 | 2022-07-26 | 太原理工大学 | Air cooling unit backpressure abnormity detection method |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108435819A (en) * | 2018-05-29 | 2018-08-24 | 广东工业大学 | A kind of aluminum section extruder energy consumption method for detecting abnormality |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005108607A1 (en) * | 2004-05-09 | 2005-11-17 | Technion Research & Development Foundation Ltd. | Compositions and methods for treating disorders associated with abnormal phosphate metabolism |
CN102706563A (en) * | 2012-06-14 | 2012-10-03 | 哈尔滨工业大学 | Detection method for neighbor abnormities of gas turbine |
CN106330624B (en) * | 2016-11-07 | 2019-08-06 | 国网江苏省电力公司南京供电公司 | A kind of Power Information Network Traffic anomaly detection method |
US11621969B2 (en) * | 2017-04-26 | 2023-04-04 | Elasticsearch B.V. | Clustering and outlier detection in anomaly and causation detection for computing environments |
JP6824121B2 (en) * | 2017-07-14 | 2021-02-03 | 株式会社東芝 | State detection device, state detection method and program |
CN107818135B (en) * | 2017-09-26 | 2020-02-04 | 广东电网有限责任公司电力调度控制中心 | Voronoi diagram electric power big data abnormality detection method based on gray correlation method |
-
2019
- 2019-06-11 CN CN201910503050.4A patent/CN110378371B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108435819A (en) * | 2018-05-29 | 2018-08-24 | 广东工业大学 | A kind of aluminum section extruder energy consumption method for detecting abnormality |
Non-Patent Citations (3)
Title |
---|
基于时间序列和聚类的挤压机能耗异常检测研究;曾利云 等;《机电工程技术》;20180929;第32-36页 * |
塔机疲劳剩余寿命预测***研究;肖冬桂;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20140615;第C029-406页 * |
港口起重机金属结构疲劳寿命分析与评价;路世青 等;《物流工程三十年技术创新发展之道》;20101001;第74-76页 * |
Also Published As
Publication number | Publication date |
---|---|
CN110378371A (en) | 2019-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110362608B (en) | Rain flow counting method and local anomaly factor-based energy consumption anomaly detection method | |
CN110378371B (en) | Energy consumption abnormity detection method based on average neighbor distance abnormity factor | |
CN110263846B (en) | Fault diagnosis method based on fault data deep mining and learning | |
CN111275288B (en) | XGBoost-based multidimensional data anomaly detection method and device | |
CN111353482A (en) | LSTM-based fatigue factor recessive anomaly detection and fault diagnosis method | |
CN110309886B (en) | Wireless sensor high-dimensional data real-time anomaly detection method based on deep learning | |
CN110895526A (en) | Method for correcting data abnormity in atmosphere monitoring system | |
CN117290802B (en) | Host power supply operation monitoring method based on data processing | |
CN112766342A (en) | Abnormity detection method for electrical equipment | |
CN113344133B (en) | Method and system for detecting abnormal fluctuation of time sequence behaviors | |
BR112021013786A2 (en) | METHOD AND SYSTEM TO DETECT ANOMALIES OR PREMATURE INDICATIONS OF EQUIPMENT FAILURE | |
CN112173636B (en) | Method for detecting faults of belt conveyor carrier roller by inspection robot | |
CN116304766A (en) | Multi-sensor-based quick assessment method for state of switch cabinet | |
CN109241870B (en) | Coal mine underground personnel identity identification method based on gait identification | |
CN115220396B (en) | Intelligent monitoring method and system for numerical control machine tool | |
CN113526282B (en) | Method, device, medium and equipment for diagnosing medium and long-term aging faults of elevator | |
CN116335925B (en) | Data enhancement-based intelligent regulation and control system for underground coal mine emulsification pump station | |
CN116404186B (en) | Power lithium-manganese battery production system | |
CN117556377B (en) | Multi-source data fusion processing method for station automation terminal | |
CN114417971A (en) | Electric power data abnormal value detection algorithm based on K nearest neighbor density peak clustering | |
CN116662111A (en) | Intelligent network management control system for train | |
CN110688969A (en) | Video frame human behavior identification method | |
CN107862866B (en) | Data noise point detection method based on average offset translation | |
CN116383747A (en) | Anomaly detection method for generating countermeasure network based on multi-time scale depth convolution | |
CN116085290A (en) | Sliding window thermal imaging-based fan thermal fault detection method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |