CN115828118A - Air conditioner abnormity identification method based on machine learning - Google Patents
Air conditioner abnormity identification method based on machine learning Download PDFInfo
- Publication number
- CN115828118A CN115828118A CN202211479669.4A CN202211479669A CN115828118A CN 115828118 A CN115828118 A CN 115828118A CN 202211479669 A CN202211479669 A CN 202211479669A CN 115828118 A CN115828118 A CN 115828118A
- Authority
- CN
- China
- Prior art keywords
- data
- abnormal
- air conditioner
- time
- model
- 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
Images
Classifications
-
- 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
- Y02B—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO BUILDINGS, e.g. HOUSING, HOUSE APPLIANCES OR RELATED END-USER APPLICATIONS
- Y02B30/00—Energy efficient heating, ventilation or air conditioning [HVAC]
- Y02B30/70—Efficient control or regulation technologies, e.g. for control of refrigerant flow, motor or heating
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of air conditioner control, and provides an air conditioner abnormity identification method based on machine learning for accurately and timely finding out air conditioner abnormity, which comprises the following steps: step 1, acquiring air conditioner operation data and operation environment data; step 2, preprocessing and sequencing data to obtain time sequence data; step 3, carrying out frequency mixing processing on the time sequence data to unify the time granularity of the data; step 4, clustering the data obtained in the step 3 to obtain abnormal data clusters; step 5, performing service relocation on the abnormal data cluster according to the on-off state of the air conditioner; and 6, constructing a KMBOD model, and performing exception identification on the data acquired in the step 5 to obtain a corresponding exception category. The air conditioner abnormity can be accurately and timely found by adopting the steps.
Description
Technical Field
The invention relates to the field of air conditioner control, in particular to an air conditioner abnormity identification method based on machine learning.
Background
With the continuous development of information communication, the normal operation of a communication base station is an important guarantee for normal production and life of the society, and an air conditioner in the base station plays an important role in guaranteeing the normal operation of the base station. At present, most base stations are in an unattended state, the discovery of the abnormity of the air conditioner is originated from the regular inspection of the air conditioner and the fault code when the air conditioner breaks down, and the larger information hysteresis exists, so that the discovery and the identification of the abnormity of the air conditioner are more passive, the abnormity of the air conditioner cannot be discovered in time, and the timely elimination of the fault is influenced; on the other hand, the temperature, the operation data of the air conditioner, and the information of the environment outside the base station are not fully utilized, and the function of assisting the air conditioner in troubleshooting cannot be exerted.
Disclosure of Invention
In order to accurately and timely find out the air conditioner abnormity, the application provides an air conditioner abnormity identification method based on machine learning.
The technical scheme adopted by the invention for solving the problems is as follows:
the air conditioner abnormity identification method based on machine learning comprises the following steps:
step 1, acquiring air conditioner operation data and operation environment data;
step 2, preprocessing and sequencing data to obtain time sequence data;
step 3, carrying out frequency mixing processing on the time sequence data to unify the time granularity of the data;
step 4, clustering the data obtained in the step 3 to obtain abnormal data clusters;
step 5, performing service relocation on the abnormal data cluster according to the on-off state of the air conditioner;
and 6, constructing a KMBOD model, and performing exception identification on the data acquired in the step 5 to obtain a corresponding exception category.
For convenience of data processing, step 1 acquires structured data when acquiring data.
Further, the data preprocessing in step 2 includes deduplication, review and verification.
Further, in the step 3, an EM algorithm is adopted to perform frequency mixing processing on the data, specifically:
step 31, determining the granularity [ t ] of the target time 1 ,t 2 ,t 3 ,…,t i ]；
And 32, performing data expansion on data except the target time granularity by using an EM algorithm: let the time to be subjected to mixing be t' 1 ,t' 2 ,t' 3 ,…,t' j ]The corresponding data is [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]The data expansion step is as follows:
based on [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]And the EM algorithm estimates that the model parameter is theta 1 ，
According to theta 1 Estimate [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]At [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is
According to [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]And
estimating and updating model parameters to θ 2 ，
According to theta 2 Estimating and updating [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]At [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is
According to [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]And
estimating and updating model parameters to θ 3 ，
According to theta 3 Estimating and updating data, iterating until model parameters converge, and finally estimating [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]At [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is [ Q' 21 ,Q' 22 ,Q' 23 ,…,Q' 2i ]；
Step 33, the expanded data and the target time granularity [ t ] 1 ,t 2 ,t 3 ,…,t i ]And fusing original data.
Further, the step 4 clusters the data by using the DBSCAN model, specifically:
step 41, constructing model training data by using historical data, wherein the model training data comprises a training set and a test set;
step 42, training the DBSCAN model by using the training data;
and 42, classifying the data obtained in the step 3 by using the constructed model to obtain an abnormal data cluster.
Further, the step 4, after acquiring the abnormal data cluster, further includes: time sequencing is carried out on data in the abnormal data cluster; and extracting the data segment of the abnormal starting moment and the time step W before and after the abnormal starting moment according to the time-sequenced abnormal data cluster, and reconstructing the abnormal time slice.
Further, the step 6 of constructing the KMBOD model specifically comprises the following steps:
step 61, calculating the similarity DTW (X, Y) between the abnormal time slices by using a dynamic time warping algorithm;
step 62, performing KMBOD clustering based on the similarity DTW (X, Y):
randomly selecting K abnormal time slices as central points, calculating the similarity between the remaining abnormal time slices and each central point, classifying the abnormal time slices and the most similar central points into one class to form K abnormal time slice clusters, and calculating the total loss E of the current cluster;
e is the sum of the similarity of all non-central abnormal time slices of each cluster and the central time slice of the cluster, K is the number of central points, C i As an abnormal time slice cluster, O i Is C i P is C i A non-center point of (1);
for each center point O and non-center point P, the following steps are performed:
exchanging roles of O and P, reclustering P as a central point, and calculating the total loss E after clustering; if the total loss is increased, the role exchange is not carried out, and if the total loss is decreased, the role exchange is carried out; repeating the above steps until the total loss E is not reduced any more;
step 63, obtaining an optimal KMBOD model: and calculating the contour coefficients of the clustering results under different K values, wherein the K value corresponding to the maximum contour coefficient is the optimal clustering number, and the corresponding model is the optimal KMBOD model.
Further, the method also comprises a step 7 of outputting the abnormal category.
Compared with the prior art, the invention has the beneficial effects that: by carrying out frequency mixing processing on the data and utilizing a data adding algorithm EM to add the data with large time granularity, the time granularity is reduced, the time interval of each dimension of data is unified, the time sequence of the data is more complete, and the subsequent abnormal positioning effect is more ideal.
Through DBSCAN clustering, an abnormal data cluster containing abnormal data is preliminarily positioned, then service relocation is carried out on abnormal fragments, abnormal time slices are effectively screened out, a KMBOD model is directly constructed by the abnormal time slices and abnormal category detection is carried out, the problem of unbalance of positive and negative samples is effectively avoided, the model construction cost is directly reduced, and the detection efficiency and accuracy are improved.
Abnormal category detection is carried out on abnormal time slice data through a KMBOD algorithm, the algorithm is clustered on the basis of fully considering time sequence data similarity, the problem that the similarity of the same type of abnormal time slices with time translation cannot be effectively depicted is solved through DTW (the same type of abnormal time slices of the air conditioner have time translation due to the fact that the same type of abnormal duration may be different, and the similarity between the same type of abnormal time slices cannot be effectively measured), and the accuracy and the effectiveness of the model are improved.
Drawings
Fig. 1 is a flowchart of an air conditioner abnormality identification method based on machine learning.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following 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.
Examples
As shown in fig. 1, the method for identifying abnormal air conditioner based on machine learning includes:
s1, acquiring air conditioner operation data and operation environment data, wherein the air conditioner operation data comprises air conditioner operation data such as temperature and humidity data, air conditioner on-off states, air conditioner current and voltage power and the like in a base station machine room and outdoor temperature and humidity data; for convenience of data processing, the present embodiment acquires structured data at the time of data acquisition, and assumes that there are data F1 and F2 with two time granularities, where the time granularities are F1 and F2, respectively, and the time sequence F1 is [ t ] in time sequence 1 ,t 2 ,t 3 ,…,t i ]A total of i time points, each time point having P fields P 1 ,P 2 ,P 3 ,…,P p ]The data sequence F2 is sequentially [ t' 1 ,t' 2 ,t' 3 ,…,t' j ]A total of j time points, each time point having a Q field [ Q ] 1 ,Q 2 ,Q 3 ,…,Q q ]The data structures of the two data should be as follows:
and S2, preprocessing and sequencing the data to obtain time sequence data, wherein in the embodiment, the preprocessing of the data mainly comprises duplicate removal, examination and verification. The duplicate removal means that whether the same time tag t appears for multiple times in the data is detected, if the same time tag t appears for multiple times, only one piece of data is reserved, and the rest part is deleted; the examination refers to checking whether the data is missing or not according to the time continuity; the check refers to checking whether the data is in a specified range, is invalid or not and is null or not according to a related data protocol. In order to improve the data integrity, missing values and invalid values can be filled to obtain cleaned time sequence data.
And step S3, performing mixing processing on the data obtained in the step S2: data expansion of data F2 using EM algorithm, using known t' 1 ,t' 2 ,t' 3 ,…,t' j ]Data corresponding to the time, estimate [ t ] 1 ,t 2 ,t 3 ,…,t i ]The time granularity of the data corresponding to each time is reduced from F2 to F1, and the data is subjected to multidimensional data fusion with the data F1, for example: mixing certain field in F2, and utilizing EM algorithm to process according to t' 1 ,t' 2 ,t' 3 ,…,t' j ]Corresponding known data [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]Estimating the model parameter theta 1 Estimating Q from the model parameters 2 At a time [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value
Then according to the known data [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]And the data estimated in the previous step
Estimating and updating model parameters to θ 2 According to the model parameter theta 2 Estimating and updating the time of day t 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is
Then according to the known data [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]And the data estimated in the previous step
Estimating and updating model parameters to θ 3 According to the model parameter theta 3 Estimating and updating data, iterating until the model parameters are converged, and finally estimating a field Q 2 At a time [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value[Q' 21 ,Q' 22 ,Q' 23 ,…,Q' 2i ]. After the data expansion is carried out on the F2, the data expansion is carried out on the F2 and the multi-dimensional data fusion is carried out on the F1, and the fused data structure is as follows:
[{t 1 :[P 1 ,P 2 ,P 3 ,…,P p ,Q' 1 ,Q' 2 ,Q' 3 ,…,Q' q ]},
{t 2 :[P 1 ,P 2 ,P 3 ,…,P p ,Q' 1 ,Q' 2 ,Q' 3 ,…,Q' q ]},
{t 3 :[P 1 ,P 2 ,P 3 ,…,P p ,Q' 1 ,Q' 2 ,Q' 3 ,…,Q' q ]},
…
{t i :[P 1 ,P 2 ,P 3 ,…,P p ,Q' 1 ,Q' 2 ,Q' 3 ,…,Q' q ]}]
and S4, training a DBSCAN clustering model by using the historical data, clustering the data obtained in the step S3 by using the trained DBSCAN clustering model, and calibrating abnormal data to obtain abnormal data clusters containing abnormal periods of the air conditioner. Processing the historical data in steps S2-S3 to obtain a data set H (which data are known to be abnormal data) which can be used for training, and dividing the data set H into a training set Train and a Test set Test; training by using a training set Train and a Test set Test to obtain a DBSCAN clustering model; classifying the data obtained in the step S3 by using the model obtained by training to obtain n abnormal data clusters { D 1 ,D 2 ,D 3 ,…,D n The data within the anomalous data cluster is time-out-of-order, e.g. D 1 And D 2 Examples of data within are as follows:
and S5, sequencing the data in each abnormal data cluster according to the time field t.
Step S6, difference according to time sequenceAnd extracting data segments near the abnormal starting occurrence time of the constant data cluster, and reconstructing abnormal time slices. The earliest time in the abnormal data cluster is the abnormal occurrence time t s With t s As a starting point, sliding forward and backward respectively according to time for a fixed time step W, and taking a time t s-W To t s+W The data in between as a new exceptional time slice, as follows:
s7, carrying out service relocation on the abnormal segment: the abnormal time slice obtained in the step S6 comprises an air conditioner on-off state field flag; detecting an air conditioner switch state field flag, wherein 0 represents a closing state, 1 represents an opening state, and if the condition that the field flag in the time slice is not equal to 0 is detected, reserving the abnormal time slice.
S8, constructing a KMBOD model to identify the abnormal time slice according to the abnormal category: training a KMBOD model by using historical abnormal time slice data, and calculating the similarity between abnormal time slices by using a dynamic time warping algorithm, such as calculating the similarity of an abnormal time slice X and an abnormal time slice Y (with m fields and n time points), wherein X and Y are as follows:
the similarity of X and Y is equal to the sum of the similarity of each corresponding field, namely:
calculating the similarity of each field time sequence corresponding to the field X k And Y k First, an n × n matrix D is constructed, the matrix elements D ij ＝dist(x ki ,y kj )＝|x ki -y kj L, where dist (x) ki ,y kj ) For Euclidean distance, then D is searched out in matrix D by using dynamic programming 11 To d nn Shortest path of (length L) min ) From the upper left corner element D of the matrix D 11 Starting right lower corner element d nn Searching, wherein one of the three directions of right, lower right and lower right can be selected as the next searching direction at each element, and the current path length = the previous path length + the size of the current element, and d is set 11 To any point d ij Has a shortest path length of L min (i, j) starting with L min (1,1)＝d 11 And (3) recursion can be performed:
L min (i,j)＝min{L min (i-1,j),L min (i,j-1),L min (i-1,j-1)}+d ij
the length of the shortest path is X k And Y k The sum of the similarity of each corresponding field is the similarity DTW (X, Y) of the abnormal time slices X and Y, and the smaller the similarity is, the more similar the description is; performing KMBOD clustering by taking the similarity as the distance between samples, randomly selecting K abnormal time slices as central points (central abnormal time slices), calculating the similarity between the remaining abnormal time slices and each central point, classifying the abnormal time slices and the most similar central points into one class to form K abnormal time slice clusters, and calculating the total loss E of the current clustering, wherein the formula is as follows:
e is the sum of the similarity of all non-central abnormal time slices of each cluster and the central time slice of the cluster, K is the number of central points, C i As an abnormal time slice cluster, O i Is C i P is C i Is not a central point. For each center point O and non-center point P, the following steps are performed:
1) Exchanging roles of O and P, reclustering P as a central point, and calculating the total loss E after clustering;
2) If the total loss is increased, the role exchange is not carried out, and if the total loss is decreased, the role exchange is carried out;
3) Steps 1 and 2 are repeated until the total loss E is no longer reduced.
And calculating contour coefficients of clustering results under different K values, wherein the K value corresponding to the maximum contour coefficient is the optimal clustering number, and the corresponding value is the optimal KMBOD model. And (4) carrying out abnormal type identification on the abnormal time slices in the step (S7) by using the constructed model to obtain corresponding abnormal types, and outputting the abnormal types in the step (S9).
The KMBOD model is an improvement of a K-media model, the traditional K-media is clustered according to Euclidean distance or Manhattan distance, in order to adapt time sequence data for clustering, the distance is replaced by the similarity between time sequence data calculated according to Dynamic Time Warping (DTW), and the KMBOD model is named here.
Claims (8)
1. The air conditioner abnormity identification method based on machine learning is characterized by comprising the following steps:
step 1, acquiring air conditioner operation data and operation environment data;
step 2, preprocessing and sequencing data to obtain time sequence data;
step 3, carrying out frequency mixing processing on the time sequence data to unify the time granularity of the data;
step 4, clustering the data obtained in the step 3 to obtain abnormal data clusters;
step 5, performing service relocation on the abnormal data cluster according to the on-off state of the air conditioner;
and 6, constructing a KMBOD model, and performing exception identification on the data acquired in the step 5 to obtain a corresponding exception category.
2. The machine learning-based air conditioner abnormality recognition method according to claim 1, wherein the step 1 acquires structured data when acquiring data.
3. The machine learning-based air conditioner abnormality identification method according to claim 1, characterized in that the data preprocessing in the step 2 includes deduplication, review and verification.
4. The machine learning-based air conditioner anomaly identification method according to claim 1, wherein in the step 3, the data are subjected to frequency mixing processing by adopting an EM algorithm, specifically:
step 31, determining the granularity [ t ] of the target time 1 ,t 2 ,t 3 ,…,t i ]；
And 32, performing data expansion on data except the target time granularity by using an EM algorithm: let the time to be subjected to mixing be t' 1 ,t' 2 ,t' 3 ,…,t j ']The corresponding data is [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]The data expansion step is as follows:
based on [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]And the EM algorithm estimates that the model parameter is theta 1 ，
According to theta 1 Estimate [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]At [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is
According to [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]And
estimating and updating model parameters to θ 2 ，
According to theta 2 Estimating and updating [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]At [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is
According to [ Q ] 21 ,Q 22 ,Q 23 ,…,Q 2j ]And
estimating and updating model parameters to θ 3 ，
According to theta 3 Estimating and updating data, iterating until model parameters converge, and finally estimating [ Q 21 ,Q 22 ,Q 23 ,…,Q 2j ]At [ t ] 1 ,t 2 ,t 3 ,…,t i ]Corresponding value is [ Q' 21 ,Q' 22 ,Q' 23 ,…,Q' 2i ]；
Step 33, the expanded data and the target time granularity]t 1 ,t 2 ,t 3 ,…,t i ]And fusing original data.
5. The machine learning-based air conditioner anomaly identification method according to claim 1, wherein in the step 4, a DBSCAN model is used to cluster data, specifically:
step 41, building model training data by using historical data, wherein the model training data comprises a training set and a test set;
step 42, training the DBSCAN model by using the training data;
and 42, classifying the data obtained in the step 3 by using the constructed model to obtain an abnormal data cluster.
6. The machine learning-based air conditioner anomaly identification method according to claim 1, wherein the step 4 further comprises, after acquiring the anomaly data cluster: time sequencing is carried out on data in the abnormal data cluster; and extracting the data segment of the abnormal starting moment and the time step W before and after the abnormal starting moment according to the time-sequenced abnormal data cluster, and reconstructing the abnormal time slice.
7. The machine learning-based air conditioner anomaly identification method according to claim 1, wherein the step 6 of constructing the KMBOD model comprises the following specific steps:
step 61, calculating the similarity DTW (X, Y) between the abnormal time slices by using a dynamic time warping algorithm;
step 62, performing KMBOD clustering based on the similarity DTW (X, Y):
randomly selecting K abnormal time slices as central points, calculating the similarity between the remaining abnormal time slices and each central point, classifying the abnormal time slices and the most similar central points into one class to form K abnormal time slice clusters, and calculating the total loss E of the current cluster;
e is the sum of the similarity of all non-central abnormal time slices of each cluster and the central time slice of the cluster, K is the number of central points, C i As an abnormal time slice cluster, O i Is C i P is C i A non-center point of (1);
for each center point O and non-center point P, the following steps are performed:
exchanging roles of O and P, reclustering P as a central point, and calculating the total loss E after clustering; if the total loss is increased, the role exchange is not carried out, and if the total loss is decreased, the role exchange is carried out; repeating the above steps until the total loss E is not reduced any more;
step 63, obtaining an optimal KMBOD model: and calculating the contour coefficients of the clustering results under different K values, wherein the K value corresponding to the maximum contour coefficient is the optimal clustering number, and the corresponding model is the optimal KMBOD model.
8. The machine learning-based air conditioner abnormality recognition method according to any one of claims 1 to 7, characterized by further comprising step 7 of outputting abnormality categories.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211479669.4A CN115828118B (en) | 2022-11-24 | 2022-11-24 | Air conditioner abnormality identification method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211479669.4A CN115828118B (en) | 2022-11-24 | 2022-11-24 | Air conditioner abnormality identification method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115828118A true CN115828118A (en) | 2023-03-21 |
| CN115828118B CN115828118B (en) | 2023-06-16 |
Family
ID=85530944
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211479669.4A Active CN115828118B (en) | 2022-11-24 | 2022-11-24 | Air conditioner abnormality identification method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115828118B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190130017A1 (en) * | 2017-11-01 | 2019-05-02 | Mad Street Den, Inc. | Method and System for Efficient Clustering of Combined Numeric and Qualitative Data Records |
| US20190261204A1 (en) * | 2016-10-28 | 2019-08-22 | Nanjing Howso Technology Co., Ltd | Method and system for abnormal value detection in lte network |
| US10554665B1 (en) * | 2019-02-28 | 2020-02-04 | Sailpoint Technologies, Inc. | System and method for role mining in identity management artificial intelligence systems using cluster based analysis of network identity graphs |
| CN112150209A (en) * | 2020-06-19 | 2020-12-29 | 南京理工大学 | Construction method of CNN-LSTM time sequence prediction model based on clustering center |
| CN112329868A (en) * | 2020-11-10 | 2021-02-05 | 西安电子科技大学 | CLARA clustering-based manufacturing and processing equipment group energy efficiency state evaluation method |
| US20210081833A1 (en) * | 2019-09-18 | 2021-03-18 | International Business Machines Corporation | Finding root cause for low key performance indicators |
| CN112989332A (en) * | 2021-04-08 | 2021-06-18 | 北京安天网络安全技术有限公司 | Abnormal user behavior detection method and device |
| CN114880384A (en) * | 2022-07-11 | 2022-08-09 | 杭州宇谷科技有限公司 | Unsupervised two-wheeled electric vehicle charging time sequence abnormity detection method and system |
| CN115358306A (en) * | 2022-08-11 | 2022-11-18 | 浙江工业大学 | Non-supervision BGP abnormity detection method and system based on graph structure |
-
2022
- 2022-11-24 CN CN202211479669.4A patent/CN115828118B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190261204A1 (en) * | 2016-10-28 | 2019-08-22 | Nanjing Howso Technology Co., Ltd | Method and system for abnormal value detection in lte network |
| US20190130017A1 (en) * | 2017-11-01 | 2019-05-02 | Mad Street Den, Inc. | Method and System for Efficient Clustering of Combined Numeric and Qualitative Data Records |
| US10554665B1 (en) * | 2019-02-28 | 2020-02-04 | Sailpoint Technologies, Inc. | System and method for role mining in identity management artificial intelligence systems using cluster based analysis of network identity graphs |
| US20210081833A1 (en) * | 2019-09-18 | 2021-03-18 | International Business Machines Corporation | Finding root cause for low key performance indicators |
| CN112150209A (en) * | 2020-06-19 | 2020-12-29 | 南京理工大学 | Construction method of CNN-LSTM time sequence prediction model based on clustering center |
| CN112329868A (en) * | 2020-11-10 | 2021-02-05 | 西安电子科技大学 | CLARA clustering-based manufacturing and processing equipment group energy efficiency state evaluation method |
| CN112989332A (en) * | 2021-04-08 | 2021-06-18 | 北京安天网络安全技术有限公司 | Abnormal user behavior detection method and device |
| CN114880384A (en) * | 2022-07-11 | 2022-08-09 | 杭州宇谷科技有限公司 | Unsupervised two-wheeled electric vehicle charging time sequence abnormity detection method and system |
| CN115358306A (en) * | 2022-08-11 | 2022-11-18 | 浙江工业大学 | Non-supervision BGP abnormity detection method and system based on graph structure |
Non-Patent Citations (3)
| Title |
|---|
| ERICH SCHUBERT 等: "Fast and eager k-medoids clustering: O(k) runtime improvement of the PAM, CLARA, and CLARANS algorithms", ELSEVIER, pages 1 - 19 * |
| 孙敏杰 等: "基于用户行为聚类的人物角色量化模型创建实证研究", 现代图书情报技术, pages 21 - 26 * |
| 李亚玲 等: "改进K-means算法在风电异常数据的识别研究", 计算机时代, pages 12 - 14 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115828118B (en) | 2023-06-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022110557A1 (en) | Method and device for diagnosing user-transformer relationship anomaly in transformer area | |
| CN109740648B (en) | Method and device for identifying abnormal data of power load and computer equipment | |
| JP2019521443A (en) | Cell annotation method and annotation system using adaptive additional learning | |
| US20150219530A1 (en) | Systems and methods for event detection and diagnosis | |
| US20230300159A1 (en) | Network traffic anomaly detection method and apparatus, and electronic apparatus and storage medium | |
| CN112084237A (en) | Power system abnormity prediction method based on machine learning and big data analysis | |
| CN108304567B (en) | Method and system for identifying working condition mode and classifying data of high-voltage transformer | |
| WO2020001311A1 (en) | Method for detecting interference, apparatus, device, and storage medium | |
| CN111398886A (en) | Detection method and system for automatically detecting online abnormity of epitope of assembly line | |
| CN110942099A (en) | Abnormal data identification and detection method of DBSCAN based on core point reservation | |
| CN112836720B (en) | Building operation and maintenance equipment abnormality diagnosis method, system and computer readable storage medium | |
| Chen et al. | A whole building fault detection using weather based pattern matching and feature based PCA method | |
| US20230385699A1 (en) | Data boundary deriving system and method | |
| CN110837874A (en) | Service data abnormity detection method based on time series classification | |
| Netzer et al. | Intelligent anomaly detection of machine tools based on mean shift clustering | |
| CN114417971A (en) | Electric power data abnormal value detection algorithm based on K nearest neighbor density peak clustering | |
| CN115021679A (en) | Photovoltaic equipment fault detection method based on multi-dimensional outlier detection | |
| CN111465093A (en) | Fingerprint positioning method and device, electronic equipment and computer storage medium | |
| CN114139624A (en) | Method for mining time series data similarity information based on integrated model | |
| CN114781779A (en) | Unsupervised energy consumption abnormity detection method and device and storage medium | |
| CN115828118B (en) | Air conditioner abnormality identification method based on machine learning | |
| CN116244600A (en) | Method, system and equipment for constructing GIS intermittent discharge mode identification model | |
| CN116400168A (en) | Power grid fault diagnosis method and system based on depth feature clustering | |
| CN113255771B (en) | Fault diagnosis method and system based on multi-dimensional heterogeneous difference analysis | |
| Salvato et al. | An adaptive immune based anomaly detection algorithm for smart WSN deployments |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Wu Linchun Inventor after: Liu Yutong Inventor after: Hou Dong Inventor before: Wu Linchun Inventor before: Liu Yutong Inventor before: Hou Dong Inventor before: Liu Yang |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |