CN110796366A - Quality difference cell identification method and device - Google Patents
Quality difference cell identification method and device Download PDFInfo
- Publication number
- CN110796366A CN110796366A CN201911031642.7A CN201911031642A CN110796366A CN 110796366 A CN110796366 A CN 110796366A CN 201911031642 A CN201911031642 A CN 201911031642A CN 110796366 A CN110796366 A CN 110796366A
- Authority
- CN
- China
- Prior art keywords
- kpi data
- poor
- cells
- cell
- historical
- 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
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000012549 training Methods 0.000 claims abstract description 39
- 238000009826 distribution Methods 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims description 47
- 238000001514 detection method Methods 0.000 claims description 37
- 238000010586 diagram Methods 0.000 claims description 13
- 230000009467 reduction Effects 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 description 38
- 238000004422 calculation algorithm Methods 0.000 description 14
- 238000010295 mobile communication Methods 0.000 description 9
- 238000012360 testing method Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 5
- 208000018910 keratinopathic ichthyosis Diseases 0.000 description 5
- 238000013507 mapping Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000007637 random forest analysis Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 238000003066 decision tree Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000003064 k means clustering Methods 0.000 description 3
- 238000000513 principal component analysis Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002955 isolation Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000011664 signaling Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design 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
- 230000003287 optical effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Theoretical Computer Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention provides a method and a device for identifying a poor cell, wherein the method comprises the following steps: acquiring KPI data of a cell to be identified; inputting the KPI data of the cell to be identified into a prediction model to obtain the quality difference problem type distribution probability of the cell to be identified; the prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers greater than or equal to 1; and displaying the distribution probability of the quality difference problem types of the cell to be identified. The method and the device for identifying the poor cell provided by the embodiment of the invention can improve the accuracy and efficiency of identifying the poor cell.
Description
Technical Field
The embodiment of the invention relates to the technical field of communication, in particular to a quality difference cell identification method and a quality difference cell identification device.
Background
A poor cell refers to a cell in which the quality of network communication has one or more poor problems, and the type of problem in poor cell relates to multiple aspects, such as radio coverage, interference, capacity, etc. At present, the problem type of the poor cell is mainly identified by a traditional network optimization (network optimization for short), and specifically, Key Performance Indicators (KPI) of each item of the cell are collected first. Then, the network optimization staff can set threshold values for each KPI respectively by means of expert knowledge and experience of the network optimization staff, and further judge the problem types of poor quality of the cell according to the threshold values of the KPIs and each KPI of the cell.
However, as the KPI dimension of a cell increases, the complexity of identifying the problem type of the poor cell increases, which results in low accuracy and efficiency in identifying the problem type of the poor cell by using the above method.
Disclosure of Invention
The embodiment of the invention provides a method and a device for identifying a poor cell, which are used for solving the problems of low accuracy and low efficiency in identifying the problem type of the poor cell.
In a first aspect, an embodiment of the present invention provides a method for identifying a poor cell, where the method includes:
and acquiring KPI data of the cell to be identified.
Inputting the KPI data of the cell to be identified into a prediction model to obtain the quality difference problem type distribution probability of the cell to be identified; the prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers greater than or equal to 1.
And displaying the distribution probability of the quality difference problem types of the cell to be identified.
Optionally, before the inputting the KPI data of the cell to be identified into the prediction model, the method further includes:
and training the prediction model by using the historical KPI data sets of X poor quality cells marked with poor quality problem types and the historical KPI data sets of Y non-poor quality cells marked with poor quality problem types.
Optionally, before the training of the prediction model, the method further includes:
KPI data of N sample cells in a historical preset time period are obtained, wherein N is an integer greater than or equal to 2.
Dividing KPI data of the N sample cells in a historical preset time period into a first data set and a second data set.
Training a detection model by using the first data set, and inputting a second data set into the trained detection model to obtain X poor quality cells and Y non-poor quality cells; the detection model is used for identifying whether the cell is a poor quality cell according to the KPI data.
And acquiring historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with poor quality problem types.
Optionally, before obtaining the historical KPI data sets of X poor cells marked with poor quality problem types and the historical KPI data sets of Y non-poor cells marked with no poor quality problem types, the method further includes:
generating KPI thermodynamic diagrams corresponding to the X quality difference cells by using the historical KPI data sets of the X quality difference cells; and/or the presence of a gas in the gas,
and clustering the historical KPI data sets of the X poor cells.
Optionally, before performing clustering processing on the historical KPI data sets of the X poor cells, the method further includes:
and performing dimension reduction processing on the historical KPI data sets of the X poor cells.
Optionally, the detection model is a self-encoder model.
Optionally, before dividing KPI data of the N sample cells within a historical preset time period into a first data set and a second data set, the method further includes:
and carrying out normalization processing on the KPI data of the N sample cells in a historical preset time period.
Optionally, before the inputting the KPI data of the cell to be identified into the prediction model, the method further includes:
and carrying out normalization processing on the KPI data of the cell to be identified.
In a second aspect, an embodiment of the present invention provides an apparatus for identifying a poor cell, where the apparatus includes:
the first acquisition module is used for acquiring KPI data of a cell to be identified;
the first processing module is used for inputting the KPI data of the cell to be identified into a prediction model to obtain the quality difference problem type distribution probability of the cell to be identified; the prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers greater than or equal to 1;
and the display module is used for displaying the distribution probability of the quality difference problem types of the cell to be identified.
Optionally, the apparatus further comprises:
and the second processing module is used for training the prediction model by using the historical KPI data sets of the X poor quality cells marked with the poor quality problem types and the historical KPI data sets of the Y non-poor quality cells marked with the poor quality problem types before the KPI data of the cell to be identified is input into the prediction model by the first processing module.
Optionally, the apparatus further comprises:
a second obtaining module, configured to obtain KPI data of N sample cells in a historical preset time period before the second processing module trains the prediction model, where N is an integer greater than or equal to 2;
the second processing module is further configured to divide KPI data of the N sample cells within a historical preset time period into a first data set and a second data set; training a detection model by using the first data set, and inputting the second data set into the trained detection model to obtain the X poor quality cells and the Y non-poor quality cells; acquiring historical KPI data sets of the X poor cells marked with the poor quality problem types and historical KPI data sets of the Y non-poor cells marked with the poor quality problem types;
and the detection model is used for identifying whether the cell is a poor cell or not according to the KPI data.
Optionally, the second processing module is further configured to generate KPI thermodynamic diagrams corresponding to X quality difference cells by using the historical KPI data sets of the X quality difference cells before acquiring historical KPI data sets of the X quality difference cells marked with quality difference problem types and historical KPI data sets of Y non-quality difference cells marked with or without quality difference problem types; and/or clustering the historical KPI data sets of the X poor cells.
Optionally, the second processing module is further configured to perform dimension reduction on the historical KPI data sets of the X poor cells before performing clustering processing on the historical KPI data sets of the X poor cells.
Optionally, the detection model is a self-encoder model.
Optionally, the second processing module is further configured to, before dividing the KPI data of the N sample cells in the historical preset time period into the first data set and the second data set, perform normalization processing on the KPI data of the N sample cells in the historical preset time period.
Optionally, the first processing module is further configured to perform normalization processing on the KPI data of the cell to be identified before the KPI data of the cell to be identified is input into a prediction model.
In a third aspect, an embodiment of the present invention further provides an apparatus for identifying a quality difference cell, including: at least one processor and memory;
the memory stores computer-executable instructions;
the at least one processor executing the computer-executable instructions stored by the memory causes the apparatus to perform the method of the first aspect.
In a fourth aspect, the embodiments of the present invention also provide a computer-readable storage medium, on which computer-executable instructions are stored, and when the computer-executable instructions are executed by a processor, the method of the first aspect is implemented.
According to the method and the device for identifying the poor quality cell, provided by the embodiment of the invention, the mapping relation between the KPI and the poor quality problem type distribution probability can be learned by using the prediction model obtained by training the historical KPI data sets of the X poor quality cells marked with the poor quality problem types and the historical KPI data sets of the Y non-poor quality cells marked with the poor quality problem types, so that the quality problem type distribution probabilities of the cell to be identified can be accurately identified by using the prediction model based on the KPI of the cell to be identified. The problem type that leads to this district poor quality that can accurate reflection of this distribution probability, even if net excel staff that experience is not enough, also can be through this each poor quality problem type distribution probability of waiting to discern the district, the poor quality problem type of this district of waiting to discern of quick and accurate understanding improves the rate of accuracy of discerning poor quality district problem type. In addition, after the trained prediction model is deployed, only the KPI data of the cell needs to be input into the prediction model, whether the cell has the problem of poor quality can be identified, and the efficiency of identifying the problem type of the poor cell is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a scene schematic diagram of a quality difference cell identification method according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of a method for identifying a quality difference cell according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart of acquiring a KPI data set according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a quality difference cell identification apparatus according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of another quality difference cell identification apparatus according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Conventional network optimization, particularly mobile communication network optimization, generally collects KPI data of each cell in a network through an Operation and Management Center (OMC). And setting a threshold value for each KPI by the network optimization staff according to the accumulated experience and expert knowledge. When one or more of the KPI data of a certain cell exceeds a threshold, it indicates that the cell has a quality problem, and then, the network optimization worker needs to determine the type of the quality problem existing in the cell by combining one or more KPI data of the cell. However, as the KPI dimension (which may also be referred to as the number of KPIs) increases (for example, the KPI dimension of the current network is about 30-60, and as the network technology evolves, the KPI dimension continues to increase), the difficulty of determining the quality problem type of the cell increases. The above method of positioning the quality problem types of the cell only by means of experience and expert knowledge is difficult to accurately position the quality problem types of the cell, and has the problems of high false alarm rate and low accuracy. On the other hand, the traditional network optimization mode is more and more difficult to adapt to the dynamic change of the network, and the efficiency of identifying the poor cell is difficult to meet the requirement.
In order to solve the technical problems, the invention can enable the prediction model to learn the mapping relation between KPIs and distribution probabilities of the quality difference problem types by using the prediction model obtained by training the historical KPI data sets of X quality difference cells marked with the quality difference problem types and the historical KPI data sets of Y non-quality difference cells marked with the quality difference problem types in advance, thereby accurately identifying the distribution probabilities of the quality difference problem types of the cells to be identified by using the prediction model, displaying the distribution probabilities of the quality difference problem types of the cells to be identified to network optimization workers and helping the network optimization workers to further implement network optimization work. By adopting the technical scheme of the invention, through the distribution probability of each quality difference problem type of the cell to be identified, even network-superior workers with insufficient experience can quickly and accurately acquire the quality difference problem type of the cell to be identified, and the accuracy of identifying the quality difference problem type can be improved. Meanwhile, after the trained prediction model is deployed, the KPI data of the cell are input into the prediction model every time, whether the cell has a poor quality problem can be identified, and the efficiency of identifying the poor quality cell problem type is improved.
The following describes in detail a technical solution of the method for identifying a poor cell according to the present invention with reference to several specific embodiments. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Fig. 1 is a schematic flow chart of a method for identifying a quality difference cell according to an embodiment of the present invention. The execution subject of the method for identifying poor quality cells in this embodiment may be a poor quality cell identification device, or may be an electronic device (e.g., a terminal, a server, etc.) integrated with a poor quality cell identification device, and fig. 1 illustrates an example in which the execution subject is an electronic device (e.g., an electronic device for short) integrated with a poor quality cell identification device. As shown in fig. 1, the method of the present invention may include:
and S11, acquiring KPI data of the cell to be identified.
The cell to be identified refers to a cell of a mobile communication system that is providing a service in a network, and the mobile communication system includes but is not limited to: narrowband Band-Internet of Things (NB-IoT), Global System for Mobile Communications (GSM), Enhanced Data rates for GSM Evolution (EDGE), Wideband Code Division Multiple Access (WCDMA), Code Division Multiple Access (Code Division Multiple Access, CDMA2000), Time Division-synchronous Code Division Multiple Access (Time Division-synchronous Code Division Multiple Access, TD-SCDMA), Long Term Evolution (Long Term Evolution, LTE), and next generation 5G Mobile communication systems, etc.
KPI data of a cell is used to evaluate the operational status of the network of the cell, including but not limited to at least one of the following KPIs: radio Resource Control (RRC) connection User number, RRC maximum connection User number, RRC establishment attempt number, downlink average Discontinuous Transmission (DTX) duty ratio, downlink per Transmission Time Interval (TTI) scheduling User number, Uplink per TTI scheduling User number, downlink per TTI activation User number, Uplink per TTI activation User number, downlink traffic flow, downlink signaling flow, Uplink signaling flow, downlink User rate, Uplink User rate, random access success rate, random access attempt number, User terminal (User Equipment, UE) power limitation ratio, Uplink average DTX duty ratio, Uplink Physical Resource Block (PRB) utilization ratio, downlink PRB utilization ratio, Channel Quality identification (Channel Quality Indication, CQI) ratio of 6 or less, average CQI, Physical Uplink Shared Channel (Physical Shared Channel) ratio, and RRC establishment attempt number, PUSCH) Interference, Physical Uplink Control Channel (PUCCH) Interference, uplink average path loss path lososs, uplink path lososs below 130dB ratio, PUSCH average Signal To Interference Plus noise ratio (Signal To Interference Plus noise ratio, SINR), PUCCH average SINR, PUCCH SINR below 0dB ratio, Control Channel Element (CCE) average load, PUSCH SINR below-2 dB ratio, master board average load, master board load greater than 90% ratio. It should be understood how many dimensions, there are items of KPI data for a cell. Taking the example that the KPI data of a cell includes 26 KPI data, 26-dimensional KPI data of the cell may also be referred to.
Fig. 2 is a schematic view of a scenario of a quality-difference cell identification method according to an embodiment of the present invention, and as shown in fig. 2, an operation management center may periodically collect KPI data of cells, for example, collect KPI data of cell 1, cell 2, and cell 3 in the figure every one hour, and store the KPI data in the operation management center. The electronic device may obtain KPI data for the cell to be identified from an operations management center. Taking the cell to be identified as the cell 1 as an example, the electronic device may read KPI data of the cell 1 from the operation management center.
As a possible implementation manner, network optimization workers can also collect KPI data of the cell to be identified through user declaration, Call Quality dialing Test (CQT), Drive Test (DT), and the like, and after the KPI data is input into the KPI database of the cell to be processed, the electronic device reads the KPI data of the cell to be identified from the KPI database of the cell to be processed; or, the network optimization staff directly inputs KPI data of the cell to be identified into the electronic equipment, and the like.
And S12, inputting the KPI data of the cell to be identified into a prediction model to obtain the quality difference problem type distribution probability of the cell to be identified.
The prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers which are more than or equal to 1. The type of the poor quality problem referred to herein refers to a specific type of the poor quality problem existing in the network, for example: high uplink load, high downlink load, high uplink and downlink load, coverage class, PUSCH interference, PUCCH interference, PUSCH & PUCCH interference, high load uplink interference, and the like.
The types of the quality problems marked by the historical KPI data sets of the X quality cells are as follows: the type of the quality difference problem of the cell is characterized by the historical KPI data set of the cell, and the type of the quality difference problem of the cell can be obtained by analyzing the historical KPI data of the cell by a network optimization worker according to experience and expert knowledge of the network optimization worker. For example, a high uplink load problem exists in one of the history KPI data sets of the X poor cells, and the type of the poor cell existing in the history KPI data set of the cell may be marked as a high uplink load. Similarly, the fact that the type of the quality difference problem exists in the historical KPI data sets of the Y non-quality difference cells means that the historical KPI data set of the cell represents that the cell has no quality difference problem.
The prediction model is obtained by training historical KPI data sets of X poor quality cells marked with poor quality problem types and historical KPI data sets of Y non-poor quality cells marked with no poor quality problem types. Through the training mode, the prediction model can learn the mapping relation between the KPI and the quality difference problem type distribution probability, so that the quality difference problem type distribution probability of the cell to be identified can be accurately identified by using the prediction model. When the probability of a certain poor quality problem type is higher, the probability that the problem type causes the poor quality of the cell is higher. When the probability of a certain poor quality problem type is lower, the probability that the problem type causes the poor quality of the cell is lower. Therefore, through the distribution probability of each quality problem type of the cell to be identified, even network optimization workers with insufficient experience can quickly and accurately acquire the quality problem type of the cell to be identified. It should be understood that the distribution probability of the quality difference problem types of the cell to be identified as referred to herein means that the probability of each quality difference problem type of the cell to be identified is obtained, for example, the probability of a high uplink load of a certain cell to be identified is 60%.
And S13, displaying the distribution probability of the quality difference problem types of the cell to be identified.
The distribution probability of the types of the poor quality problems of the cell to be identified is displayed in a text mode, a graph mode and/or a table mode and is provided for network optimization workers, the network optimization workers can conveniently, quickly and intuitively obtain the distribution probability of the types of the poor quality problems of the cell to be identified, and then the network optimization workers can quickly and accurately position the types of the poor quality problems of the cell to be identified, so that the network optimization work can be pertinently performed on the cell, and the poor quality problems of the cell are improved.
The method provided by the embodiment of the present invention is described below by way of an example with reference to the example shown in fig. 2:
firstly, a network optimization worker can collect 26-dimensional KPI data of 24 hours in the previous day of a cell 1 through an OMC (operation, management and control), namely the KPI data obtained from the cell 1 is 24 × 26-dimensional KPI data. Secondly, the net optimization staff can input the 24 × 26 dimensional KPI data into the electronic device, so that the electronic device obtains the probability of each quality problem type of the cell 1 based on a prediction model. The probability of this cell 1 at each type of quality problem is assumed to be as follows: 60% of high uplink load, 15% of high downlink load, 20% of high uplink and downlink load, 10% of coverage, 0% of PUSCH interference, 8% of PUCCH interference, 4% of PUSCH and PUCCH interference and 30% of high load uplink interference. Then, the electronic device may show the quality difference problem type distribution probability of the cell 1 to the network optimization worker in a text, graph and/or table manner, so that the network optimization worker performs the next network optimization work on the cell 1 based on the quality difference problem distribution probabilities, so as to improve the quality difference problem of the cell.
According to the method and the device for identifying the poor quality cell, the KPI data of the cell to be identified are acquired and input into the prediction model, so that the distribution probability of the poor quality problem types of the cell to be identified is obtained and displayed to network optimization workers to execute the next network optimization work. The method comprises the steps that a prediction model obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with poor quality problem types is used, the mapping relation between KPI and poor quality problem type distribution probability can be learned, and therefore the poor quality problem type distribution probability of a cell to be identified can be accurately identified by the prediction model. When the probability of a certain poor quality problem type is higher, the probability that the problem type causes the poor quality of the cell is higher. When the probability of a certain poor quality problem type is lower, the probability that the problem type causes the poor quality of the cell is lower. Therefore, through the distribution probability of each quality difference problem type of the cell to be identified, even network optimization workers with insufficient experience can quickly and accurately acquire the quality difference problem type of the cell to be identified, and the accuracy rate of identifying the quality difference problem type is improved. In addition, after the trained prediction model is deployed, only the KPI data of the cell needs to be input into the prediction model, whether the cell has the problem of poor quality can be identified, and the efficiency of identifying the problem type of the poor cell is improved.
As described in the foregoing embodiment, the prediction model according to the foregoing embodiment is obtained by training using the historical KPI data sets of X poor quality cells to which the poor quality problem types are labeled and the historical KPI data sets of Y non-poor quality cells to which the poor quality problem types are labeled. The historical KPI data sets of the X poor cells and the historical KPI data sets of the Y non-poor cells may be obtained, for example, by extracting the historical KPI data sets of the cells for which the types of the poor problems have been determined and the historical KPI data sets of the common cells from the historical KPI database, and then performing poor problem type labeling on the historical KPI data of the cells by a network optimization engineer.
As a possible implementation manner, the historical KPI data sets of X poor cells marked with poor quality problem types and the historical KPI data sets of Y non-poor cells marked with poor quality problem types may also be obtained as follows:
fig. 3 is a schematic flowchart of acquiring a KPI data set according to an embodiment of the present invention. The main execution body of the method may be the electronic device executing the quality difference cell identification method shown in fig. 2, or may be other devices. The following embodiments all describe the present embodiment by taking the electronic device that executes the method for identifying a poor quality cell shown in fig. 2 as an example. In addition, in order to facilitate understanding of the embodiment of the present invention, the embodiment is exemplified by 26-dimensional KPI data of 70000 cells in 10 days in a certain urban mobile communication network, which is collected by a network optimization worker. I.e. N equals 70000. As shown in fig. 2, the method may include:
s21, KPI data of N sample cells in a historical preset time period are obtained, wherein N is an integer greater than or equal to 2.
The network optimization staff collects 26-dimensional KPI data of 70000(N is 70000) cells in a certain urban mobile communication network for 10 days and 24 hours, namely 10X 24X 26-dimensional KPI data. For the above KPI data, a plurality of 26-dimensional KPI data can be extracted continuously from the KPI data as a training sample, for example, 26-dimensional KPI data for 1 day is extracted from 10 days, that is, 24 × 26-dimensional KPI data is used as a training sample, and the network optimization worker inputs the 24 × 26-dimensional KPI data of the 70000 cells into the electronic device. Alternatively, the net optimization staff may input all 10 × 24 × 26 dimensional KPI data of the 70000 cell into the electronic device.
And S22, dividing KPI data of the N sample cells in a historical preset time period into a first data set and a second data set.
The 24 × 26 dimensional KPI data of the 70000 cells are randomly divided into a first data set and a second data set, and the division ratio may be any ratio, for example, the division ratio may be an average division ratio, that is, the first data set and the second data set each account for 50%, that is, the 24 × 26 dimensional KPI data of the 35000 cells. For example, the division may be performed at a ratio of 6:4, i.e., the first data set is 24 × 26 dimensional KPI data of 42000 cells, and the second data set is 24 × 26 dimensional KPI data of 28000 cells.
S23, training the detection model by using the first data set.
The first data set is partitioned into a training set and a test set, and the partition ratio is not limited, for example, the partition ratio of the training set and the test set may be 7:3, and may be 8: 2. And inputting the training set into the detection model for training, then evaluating the detection model by adopting the test set, and converging the model when the loss of the detection model is lower than a certain preset threshold value.
The detection model may be, for example, an Isolation Forest model (Isolation Forest Models), a graph model (Graphic Models), a self-encoder model, a Variational self-encoder (VAE) model, or the like. One possible implementation may be trained using an auto-encoder (AE) model, which is a class of artificial neural networks used in semi-supervised and unsupervised learning, which is effectively a pair of interconnected neural networks, including an encoder and a decoder. The encoder neural network transforms the input data into a smaller, more compact encoded representation in the implicit space, while the decoder restores this encoding back to the original input data. But the representation in implicit space here is not continuous, making differences and perturbations therein difficult to accomplish. Since the self-encoder is a prior art, it is not described herein in detail.
Another possible implementation may use a Variational Auto Encoder (VAE) model. The design idea and architecture of the variational self-encoder also have two structures of an encoder and a decoder, and the process of reconstructing input data by compressing and decompressing the input data is also used for trying to make the output and the input of the neural network the same. But it has quite different characteristics from the self-encoder, its implicit space is designed as a continuous distribution for random sampling and interpolation, and therefore can have a similar regularization effect to prevent overfitting. Since the variational self-encoder is prior art, it is not described here in detail.
Taking the example that the first data set includes 24 × 26 KPI data of 35000 cells, the 35000 cell 24 × 26 KPI data may be split according to a ratio of 8:2 to obtain a training set of 28000 cells and a test set of 7000 cells, the training set is input to train the detection model, and the test set is used to evaluate the trained detection model until the loss of the detection model is lower than a certain threshold. It will be appreciated that the lower the loss, the better the convergence obtained by the detection model, and in this case the higher the accuracy with which the detection model calculates the cell reconstruction error.
And S24, inputting the second data set into the trained detection model to obtain X poor quality cells and Y non-poor quality cells. The detection model is used for identifying whether the cell is a poor quality cell according to the KPI data.
And inputting 24 × 26-dimensional KPI data of cells in the second data set 35000 into a trained detection model, and calculating the reconstruction error of each cell in the second data set.
The reconstruction error is a mean square error between the input data x and the output data y of the detection model, and can be represented by the following formula (1):
wherein x is input 24 × 26-dimensional KPI data, y is 24 × 26-dimensional KPI data reconstructed by the detection model, n is a dimension of x or y, where n is 24 × 26, loss is a reconstruction error, and i is an ith coordinate of x or y.
Unbalanced training samples in machine learning can result in the training model emphasizing classes with a higher number of samples, while "ignoring" classes with a lower number of samples. Because the number of the common cells in the mobile communication network is far greater than the number of the poor quality cells, when data of the unbalanced common cells and the poor quality cells are input into the self-encoder neural network for training, the detection model mainly learns the characteristics of the data of the common cells. In the distribution of the reconstruction errors, the reconstruction errors of the common cells are obviously smaller than those of the poor quality cells, so that the poor quality cells and the non-poor quality cells can be distinguished by setting a reconstruction error threshold.
The reconstruction error threshold is usually set with reference to a box map of the reconstruction error, for example, an upper limit value of the box map is set as the reconstruction error threshold. At this time, the cells with the reconstruction errors larger than or equal to the reconstruction error threshold are judged as poor quality cells, and the cells with the reconstruction errors smaller than the reconstruction error threshold are judged as non-poor quality cells, namely X poor quality cells and Y non-poor quality cells. In order to improve the accuracy of the prediction model, the numbers of X poor quality cells and Y non-poor quality cells may be taken to be the same value, that is, X equals Y. For example, when there are 950 poor cells and 34050 non-poor cells, X may be 950 (all poor cells) and Y may be 950 (all non-poor cells).
S25, obtaining the historical KPI data set of X poor quality cells marked with poor quality problem types and the historical KPI data set of Y non-poor quality cells marked with no poor quality problem types.
For example, after obtaining X poor quality cells and Y non-poor quality cells, the electronic device may output historical KPI data sets of the X poor quality cells and historical KPI data sets of the Y non-poor quality cells, so that a network optimization worker may obtain the poor quality problem types of the X poor quality cells based on the data sets, and label 24 × 26 dimensional KPI data of 950 poor quality cells and 24 × 26 dimensional KPI data of 950 non-poor quality cells by a manual labeling manner. Then, the network optimization staff can input the historical KPI data sets of the X poor quality cells marked with the poor quality problem types and the historical KPI data sets of the Y non-poor quality cells marked with the poor quality problem types into the electronic equipment.
Optionally, as a possible implementation manner, when the above-described manner is adopted to obtain the historical KPI data sets of X poor quality cells and the historical KPI data sets of Y non-poor quality cells marked with the poor quality problem types, before dividing the KPI data of N sample cells in the historical preset time period into the first data set and the second data set, the KPI data of the N sample cells in the historical preset time period may be further normalized. Because different KPI data may have different dimensions and dimension units, the KPI data are normalized to eliminate the dimension difference between KPIs, so that the KPI data make the same contribution, and the reliability of the detection model is improved. If the KPI data of the N sample cells in the historical preset time period are normalized, the KPI data of the cell to be identified should be normalized at the same time.
Optionally, as a possible implementation manner, before the step S25, a KPI thermodynamic diagram corresponding to X quality difference cells may also be generated by using the historical KPI data sets of the X quality difference cells.
Through generating the KPI thermodynamic diagrams corresponding to the X quality difference cells by the historical KPI data sets of the X quality difference cells, each KPI data can be visually checked more intuitively, and the expert can conveniently locate the quality difference problem types of the quality difference cells.
Optionally, as a possible implementation manner, before the step S25, a clustering process may be further performed on the historical KPI data sets of the X poor cells. As the similarity between KPI data sets in the same cluster is greater than the similarity between KPI data sets in different clusters, poor cells of similar problem type are grouped together. Therefore, by clustering, the number of types of the quality problems existing in the quality poor cell can be obtained, for example: 9 quality difference problem types such as high load, uplink interference and the like in the quality difference cell can be obtained through clustering. In another possible implementation, the type of the quality problem of the poor cell may be obtained directly based on the experience of the network optimization staff.
The clustering algorithm referred to herein may be, for example, any of the existing clustering algorithms, such as the K-Means clustering algorithm. The K-Means clustering algorithm is to randomly select K objects as initial clustering centers, then calculate the distance between each object and each seed clustering center, and assign each object to the nearest clustering center. The cluster centers and the objects assigned to them represent a cluster. The cluster center of a cluster is recalculated for each sample assigned based on the objects existing in the cluster. This process will be repeated until some termination condition is met. The K-Means clustering algorithm is prior art and will not be described herein.
Optionally, if the number of KPI dimensions of the historical KPI data sets of the X poor quality cells is large, the historical KPI data sets of the X poor quality cells may be subjected to dimension reduction processing based on the requirement of the clustering algorithm on the KPI dimensions before the KPI data sets are subjected to clustering processing, so as to improve the clustering effect of the KPI data sets.
The dimension reduction can be any dimension reduction algorithm in the prior art. One possible implementation is to use a principal component analysis algorithm. The Principal Component Analysis (PCA) method is a linear dimensionality reduction technique. The dimensionality reduction result is a linear combination of the original data features and the maximized sample variance, and new features are ensured to be independent of each other as much as possible, namely, the linear combination of the first few features which contribute more to the deviation is taken.
Another possible implementation is to use a local linear Embedding algorithm, where Local Linear Embedding (LLE) is an unsupervised nonlinear dimension reduction method, which obtains the global structure of data by combining the local attributes of data points, and the local attributes are composed of linear combinations of data points and their neighbors, the main idea is to use linear local structures to represent global nonlinear structures, and the local neighbors of data points overlap to represent the overall geometry of data, e.g. 24X 26 dimensional data before dimension reduction becomes 10-50 dimensional data after dimension reduction.
In the above method, the step of converting the KPI thermodynamic diagram and the step of clustering may also be performed simultaneously.
After the historical KPI data sets of X poor cells marked with poor quality problem types and the historical KPI data sets of Y non-poor cells marked with poor quality problem types are obtained in any of the above manners, the prediction model may be trained using the historical KPI data sets of X poor cells marked with poor quality problem types and the historical KPI data sets of Y non-poor cells marked with poor quality problem types.
One possible implementation may use a convolutional neural network algorithm to train to obtain the prediction model. The convolutional neural network is a hierarchical model, and high-level semantic information is extracted from an original data input layer by layer through layer-by-layer stacking of a series of operations such as convolutional operation, converging operation, nonlinear activation function mapping and the like, and is abstracted layer by layer. Finally, the last layer of the convolutional neural network formalizes its target task (classification, regression, etc.) as an objective function (also called cost function, loss function). By calculating the error or loss between the predicted value and the true value, the error or loss is fed back layer by layer from the last layer by means of a back propagation algorithm, parameters of each layer are updated, and the parameters are fed forward again after being updated, and the steps are repeated until the network model is converged, so that the purpose of model training is achieved. Since convolutional neural networks are prior art, they are not described in detail here.
In another possible implementation manner, a prediction model can be obtained by training with a random forest algorithm. The random forest is an ensemble learning algorithm based on decision tree, which comprises a plurality of decision trees trained by a bootstrapping aggregation algorithm (Bagging) ensemble learning technology, and when a sample to be classified is input, the final classification result is voted and determined by the output result of a single decision tree. The random forest has good tolerance to noise and abnormal values, and has good expandability and parallelism to the high-dimensional data classification problem. Since random forests are prior art, no further description is given here.
After the prediction model is trained in the above manner, the prediction model can be used to execute the quality difference cell identification method shown in fig. 1, so as to assist network optimization workers to quickly and accurately know the quality difference problem type of the cell to be identified, and improve the accuracy and efficiency of identifying the quality difference cell problem type.
It should be understood that, when the electronic device executing the quality difference cell identification method shown in fig. 1 is a server in a kubernets cluster, a deep learning model may be deployed on the server in a Docker manner as the aforementioned prediction model, and the quality difference cell identification method may be executed. Because the deep learning model has a large number of application components and dependent components, and is packaged into a container Docker, the method can realize more lightweight virtualization and is convenient and rapid to deploy, thereby obviously reducing the time cost and the labor cost of deployment.
In addition, the embodiments of the methods described above all use a cell in a mobile communication system as an example, and how to identify the type of the poor quality problem is described and introduced, and those skilled in the art can understand that the methods described above are also applicable to any network with network quality, such as a video monitoring network, a street lamp network, and the like.
Those of ordinary skill in the art will understand that: all or a portion of the steps of implementing the above-described method embodiments may be performed by hardware associated with program instructions. The program may be stored in a computer-readable storage medium. When executed, the program performs steps comprising the method embodiments described above; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.
Fig. 4 is a schematic structural diagram of a quality difference cell identification apparatus according to an embodiment of the present invention, and as shown in fig. 4, the apparatus may include a first obtaining module 11, a first processing module 12, and a display module 13.
Wherein the content of the first and second substances,
a first obtaining module 11, configured to obtain KPI data of a cell to be identified.
The first processing module 12 is configured to input KPI data of the cell to be identified into the prediction model, so as to obtain a quality difference problem type distribution probability of the cell to be identified; the prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers which are more than or equal to 1.
And the display module 13 is configured to display the distribution probability of the quality problem types of the cell to be identified.
With continued reference to fig. 4, optionally, in some embodiments, the apparatus further comprises a second processing module 22.
Wherein the content of the first and second substances,
the second processing module 22 is configured to train the prediction model by using the historical KPI data sets of X poor cells marked with poor quality problem types and the historical KPI data sets of Y non-poor cells marked with no poor quality problem types before the KPI data of the cell to be identified is input into the prediction model by the first processing module 12.
With continued reference to fig. 4, optionally, in some embodiments, the apparatus further comprises a second acquisition module 21.
Wherein the content of the first and second substances,
a second obtaining module 21, configured to obtain KPI data of N sample cells in a historical preset time period before training the prediction model, where N is an integer greater than or equal to 2.
In this implementation, the second processing module 22 is further configured to divide KPI data of the N sample cells within a historical preset time period into a first data set and a second data set; training a detection model by using a first data set, and inputting a second data set into the trained detection model to obtain X poor quality cells and Y non-poor quality cells; and acquiring historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with poor quality problem types. The detection model is used for identifying whether the cell is a poor cell or not according to the KPI data. For example, the detection model is an auto-encoder model.
Optionally, the second processing module 22 is further configured to generate KPI thermodynamic diagrams corresponding to X quality difference cells by using the historical KPI data sets of the X quality difference cells before acquiring the historical KPI data sets of the X quality difference cells marked with the quality difference problem type and the historical KPI data sets of the Y non-quality difference cells marked with the quality difference problem type; and/or clustering the historical KPI data sets of the X poor cells.
Optionally, the second processing module 22 is further configured to perform dimension reduction on the historical KPI data sets of the X poor cells before performing clustering processing on the historical KPI data sets of the X poor cells.
Optionally, the second processing module 22 is further configured to, before dividing the KPI data of the N sample cells in the historical preset time period into the first data set and the second data set, perform normalization processing on the KPI data of the N sample cells in the historical preset time period. In this implementation, the first processing module 12 is further configured to perform normalization processing on the KPI data of the cell to be identified before the KPI data of the cell to be identified is input into the prediction model.
The poor cell identification apparatus provided in the embodiment of the present invention may execute the actions of the electronic device in the foregoing method embodiment, and the implementation principle and technical effect thereof are similar, and are not described herein again.
Fig. 5 is a schematic structural diagram of another quality difference cell identification apparatus according to an embodiment of the present invention, as shown in fig. 5, the apparatus includes: a memory 301 and at least one processor 302.
The processor 302 is configured to implement the method for identifying a poor cell in the embodiment of the present invention when the program instruction is executed, and the specific implementation principle may refer to the foregoing embodiment, which is not described herein again.
The poor cell identifying means may further comprise an input/output interface 303.
The input/output interface 303 may include a separate output interface and input interface, or may be an integrated interface that integrates input and output. The output interface is used for outputting data, the input interface is used for acquiring input data, the output data is a general name output in the method embodiment, and the input data is a general name input in the method embodiment.
The present application further provides a readable storage medium, in which an execution instruction is stored, and when the execution instruction is executed by at least one processor of the poor cell identification apparatus, when the execution instruction is executed by the processor, the poor cell identification method in the above embodiment is implemented.
The present application also provides a program product comprising execution instructions stored in a readable storage medium. The at least one processor of the poor quality cell identification apparatus may read the execution instruction from the readable storage medium, and the execution of the execution instruction by the at least one processor causes the poor quality cell identification apparatus to implement the poor quality cell identification method provided in the above various embodiments.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (18)
1. A method for identifying a quality difference cell, the method comprising:
acquiring KPI data of a cell to be identified;
inputting the KPI data of the cell to be identified into a prediction model to obtain the quality difference problem type distribution probability of the cell to be identified; the prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers greater than or equal to 1;
and displaying the distribution probability of the quality difference problem types of the cell to be identified.
2. The method of claim 1, wherein before inputting the KPI data of the cell to be identified into a predictive model, further comprising:
and training the prediction model by using the historical KPI data sets of the X poor cells marked with the poor quality problem types and the historical KPI data sets of the Y non-poor cells marked with the poor quality problem types.
3. The method of claim 2, wherein prior to training the predictive model, further comprising:
acquiring KPI data of N sample cells in a historical preset time period, wherein N is an integer greater than or equal to 2;
dividing KPI data of the N sample cells in a historical preset time period into a first data set and a second data set;
training a detection model by using the first data set, and inputting the second data set into the trained detection model to obtain the X poor quality cells and the Y non-poor quality cells; the detection model is used for identifying whether the cell is a poor cell or not according to the KPI data;
and acquiring the historical KPI data sets of the X poor cells marked with the poor quality problem types and the historical KPI data sets of the Y non-poor cells marked with the poor quality problem types.
4. The method according to claim 3, wherein before obtaining the historical KPI data sets of the X poor quality cells marked with the poor quality problem type and the historical KPI data sets of the Y non-poor quality cells marked with the poor quality problem type, the method further comprises:
generating KPI thermodynamic diagrams corresponding to the X quality difference cells by using the historical KPI data sets of the X quality difference cells; and/or the presence of a gas in the gas,
and clustering the historical KPI data sets of the X poor cells.
5. The method according to claim 4, wherein before clustering the historical KPI data sets of the X poor quality cells, further comprising:
and performing dimension reduction processing on the historical KPI data sets of the X poor cells.
6. The method of any of claims 3-5, wherein the detection model is a self-encoder model.
7. The method according to any of claims 3-5, wherein before dividing KPI data for the N sample cells over a historical preset time period into a first data set and a second data set, the method further comprises:
and carrying out normalization processing on the KPI data of the N sample cells in a historical preset time period.
8. The method according to claim 7, wherein before entering KPI data for the cell to be identified into a predictive model, the method further comprises:
and carrying out normalization processing on the KPI data of the cell to be identified.
9. An apparatus for identifying a quality difference cell, the apparatus comprising:
the first acquisition module is used for acquiring KPI data of a cell to be identified;
the first processing module is used for inputting the KPI data of the cell to be identified into a prediction model to obtain the quality difference problem type distribution probability of the cell to be identified; the prediction model is obtained by training historical KPI data sets of X poor cells marked with poor quality problem types and historical KPI data sets of Y non-poor cells marked with no poor quality problem types, wherein X and Y are integers greater than or equal to 1;
and the display module is used for displaying the distribution probability of the quality difference problem types of the cell to be identified.
10. The apparatus of claim 9, further comprising:
and the second processing module is used for training the prediction model by using the historical KPI data sets of the X poor quality cells marked with the poor quality problem types and the historical KPI data sets of the Y non-poor quality cells marked with the poor quality problem types before the KPI data of the cell to be identified is input into the prediction model by the first processing module.
11. The apparatus of claim 10, further comprising:
a second obtaining module, configured to obtain KPI data of N sample cells in a historical preset time period before the second processing module trains the prediction model, where N is an integer greater than or equal to 2;
the second processing module is further configured to divide KPI data of the N sample cells within a historical preset time period into a first data set and a second data set; training a detection model by using the first data set, and inputting the second data set into the trained detection model to obtain the X poor quality cells and the Y non-poor quality cells; acquiring historical KPI data sets of the X poor cells marked with the poor quality problem types and historical KPI data sets of the Y non-poor cells marked with the poor quality problem types;
and the detection model is used for identifying whether the cell is a poor cell or not according to the KPI data.
12. The apparatus of claim 11,
the second processing module is further configured to generate KPI thermodynamic diagrams corresponding to X quality difference cells by using the historical KPI data sets of the X quality difference cells before acquiring historical KPI data sets of the X quality difference cells marked with quality difference problem types and historical KPI data sets of Y non-quality difference cells marked with quality difference problem types; and/or clustering the historical KPI data sets of the X poor cells.
13. The apparatus of claim 12,
the second processing module is further configured to perform dimension reduction processing on the historical KPI data sets of the X poor cells before performing clustering processing on the historical KPI data sets of the X poor cells.
14. The apparatus of any one of claims 11-13, wherein the detection model is a self-encoder model.
15. The apparatus according to any one of claims 11-13,
the second processing module is further configured to, before dividing the KPI data of the N sample cells in the historical preset time period into the first data set and the second data set, perform normalization processing on the KPI data of the N sample cells in the historical preset time period.
16. The apparatus of claim 15,
the first processing module is further configured to perform normalization processing on the KPI data of the cell to be identified before the KPI data of the cell to be identified is input into a prediction model.
17. An apparatus for identifying a quality difference cell, comprising: at least one processor and memory;
the memory stores computer-executable instructions;
the at least one processor executing the computer-executable instructions stored by the memory causes the apparatus to perform the method of any of claims 1-8.
18. A computer-readable storage medium having computer-executable instructions stored thereon which, when executed by a processor, implement the method of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911031642.7A CN110796366A (en) | 2019-10-28 | 2019-10-28 | Quality difference cell identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911031642.7A CN110796366A (en) | 2019-10-28 | 2019-10-28 | Quality difference cell identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110796366A true CN110796366A (en) | 2020-02-14 |
Family
ID=69441556
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911031642.7A Pending CN110796366A (en) | 2019-10-28 | 2019-10-28 | Quality difference cell identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110796366A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113660687A (en) * | 2021-08-17 | 2021-11-16 | 中国联合网络通信集团有限公司 | Network difference cell processing method, device, equipment and storage medium |
| CN113965445A (en) * | 2020-07-02 | 2022-01-21 | 中国移动通信集团山东有限公司 | Method and device for positioning quality difference root cause, computer equipment and storage medium |
| CN114827951A (en) * | 2022-04-24 | 2022-07-29 | 重庆长安汽车股份有限公司 | Vehicle network quality analysis method and system based on vehicle terminal and storage medium |
| CN115065606A (en) * | 2022-05-31 | 2022-09-16 | 中移（杭州）信息技术有限公司 | Home wide difference analysis method, device, equipment and storage medium |
| CN114449541B (en) * | 2020-10-30 | 2023-10-27 | 中国移动通信集团浙江有限公司 | Abnormal perception cell positioning method, device, equipment and readable medium |
| WO2024093364A1 (en) * | 2022-10-31 | 2024-05-10 | 中兴通讯股份有限公司 | Recognition method for poor-quality cell, and electronic device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140006325A1 (en) * | 2012-06-28 | 2014-01-02 | International Business Machines Corporation | Detecting anomalies in real-time in multiple time series data with automated thresholding |
| KR101888683B1 (en) * | 2017-07-28 | 2018-08-14 | 펜타시큐리티시스템 주식회사 | Method and apparatus for detecting anomaly traffic |
| CN109005556A (en) * | 2018-07-24 | 2018-12-14 | 武汉虹信技术服务有限责任公司 | A kind of 4G network quality optimization method and system based on user bill |
| CN109214501A (en) * | 2017-06-29 | 2019-01-15 | 北京京东尚科信息技术有限公司 | The method and apparatus of information for identification |
| CN109495920A (en) * | 2017-09-13 | 2019-03-19 | 中国移动通信集团设计院有限公司 | A kind of cordless communication network feature portrait method, equipment and computer program product |
| JP2019117464A (en) * | 2017-12-26 | 2019-07-18 | 株式会社Ｋｉｓ | Abnormality monitoring system, abnormality monitoring method and program |
-
2019
- 2019-10-28 CN CN201911031642.7A patent/CN110796366A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140006325A1 (en) * | 2012-06-28 | 2014-01-02 | International Business Machines Corporation | Detecting anomalies in real-time in multiple time series data with automated thresholding |
| CN109214501A (en) * | 2017-06-29 | 2019-01-15 | 北京京东尚科信息技术有限公司 | The method and apparatus of information for identification |
| KR101888683B1 (en) * | 2017-07-28 | 2018-08-14 | 펜타시큐리티시스템 주식회사 | Method and apparatus for detecting anomaly traffic |
| CN109495920A (en) * | 2017-09-13 | 2019-03-19 | 中国移动通信集团设计院有限公司 | A kind of cordless communication network feature portrait method, equipment and computer program product |
| JP2019117464A (en) * | 2017-12-26 | 2019-07-18 | 株式会社Ｋｉｓ | Abnormality monitoring system, abnormality monitoring method and program |
| CN109005556A (en) * | 2018-07-24 | 2018-12-14 | 武汉虹信技术服务有限责任公司 | A kind of 4G network quality optimization method and system based on user bill |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113965445A (en) * | 2020-07-02 | 2022-01-21 | 中国移动通信集团山东有限公司 | Method and device for positioning quality difference root cause, computer equipment and storage medium |
| CN113965445B (en) * | 2020-07-02 | 2023-10-27 | 中国移动通信集团山东有限公司 | Positioning method and device for quality difference root cause, computer equipment and storage medium |
| CN114449541B (en) * | 2020-10-30 | 2023-10-27 | 中国移动通信集团浙江有限公司 | Abnormal perception cell positioning method, device, equipment and readable medium |
| CN113660687A (en) * | 2021-08-17 | 2021-11-16 | 中国联合网络通信集团有限公司 | Network difference cell processing method, device, equipment and storage medium |
| CN113660687B (en) * | 2021-08-17 | 2023-07-04 | 中国联合网络通信集团有限公司 | Network difference cell processing method, device, equipment and storage medium |
| CN114827951A (en) * | 2022-04-24 | 2022-07-29 | 重庆长安汽车股份有限公司 | Vehicle network quality analysis method and system based on vehicle terminal and storage medium |
| CN115065606A (en) * | 2022-05-31 | 2022-09-16 | 中移（杭州）信息技术有限公司 | Home wide difference analysis method, device, equipment and storage medium |
| CN115065606B (en) * | 2022-05-31 | 2023-10-27 | 中移（杭州）信息技术有限公司 | Home-wide mass spectrometry method, device, equipment and storage medium |
| WO2024093364A1 (en) * | 2022-10-31 | 2024-05-10 | 中兴通讯股份有限公司 | Recognition method for poor-quality cell, and electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110796366A (en) | Quality difference cell identification method and device | |
| CN111177095B (en) | Log analysis method, device, computer equipment and storage medium | |
| EP3798846B1 (en) | Operation and maintenance system and method | |
| CN109522304B (en) | Abnormal object identification method and device and storage medium | |
| CN110147722A (en) | A kind of method for processing video frequency, video process apparatus and terminal device | |
| CN112529204A (en) | Model training method, device and system | |
| CN112529023A (en) | Configured artificial intelligence scene application research and development method and system | |
| CN112183758A (en) | Method and device for realizing model training and computer storage medium | |
| US20210073669A1 (en) | Generating training data for machine-learning models | |
| CN111782491B (en) | Disk failure prediction method, device, equipment and storage medium | |
| CN111242171A (en) | Model training, diagnosis and prediction method and device for network fault and electronic equipment | |
| CN111935140B (en) | Abnormal message identification method and device | |
| CN113642809A (en) | Power consumption prediction method and device, computer equipment and storage medium | |
| CN113660687B (en) | Network difference cell processing method, device, equipment and storage medium | |
| CN111163482A (en) | Data processing method, device and storage medium | |
| CN112925964A (en) | Big data acquisition method based on cloud computing service and big data acquisition service system | |
| CN114640669A (en) | Edge calculation method and device | |
| CN115438190B (en) | Power distribution network fault auxiliary decision knowledge extraction method and system | |
| CN111950753A (en) | Scenic spot passenger flow prediction method and device | |
| CN114490303B (en) | Fault root cause determination method and device and cloud equipment | |
| CN116595049A (en) | Ship intelligent navigation virtual test scene generation method, system and storage medium | |
| CN115883392A (en) | Data perception method and device of computing power network, electronic equipment and storage medium | |
| CN111428886B (en) | Method and device for adaptively updating deep learning model of fault diagnosis | |
| CN110658769B (en) | Unit monitoring system and method | |
| CN115941433A (en) | Network slice performance optimization guarantee method and system, storage medium and electronic equipment |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200214 |