Base station out-of-service prediction method and system

Abstract

The present application discloses a method and system for predicting the decommissioning of a base station. The method for predicting the decommissioning of a base station includes: obtaining historical data within a first preset time period of each base station; preprocessing the historical data; extracting alarm feature data with a similarity greater than a first threshold value from the preprocessed historical data; dividing the preprocessed historical data to obtain label information; constructing a prediction model based on the historical data within a preset number of days and the label information, wherein the preset number of days is the preset number of days before the decommissioning prediction is performed; inputting the historical data within the preset number of days of the base station to be predicted into the prediction model, and performing a decommissioning prediction for the base station to be predicted. The embodiment of the present application obtains a prediction model by processing historical data, which can transform base station maintenance from post-processing to pre-prediction processing, greatly improving the efficiency of base station accident processing, reducing the probability of base station decommissioning, and improving user experience.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method and system for predicting base station decommissioning. Background Art

[0002] When a network base station experiences a fault during operation, an alarm is generated, including some classified as out-of-service alarms. These alarms require the issuance of a fault ticket at various levels for operation and maintenance personnel to address. Out-of-service alarms indicate that a base station has been out of service. After a base station is out of service, users will be unable to connect to the network and communication will be disrupted.

[0003] The current approach is for the base station to generate a service outage alarm, indicating that the base station has been decommissioned. Only then does the operations department discover the problem and take a series of maintenance measures, including dispatching orders, troubleshooting, and repairs. However, from discovery to repair, users will be unable to use the network normally for a long time and cannot communicate, resulting in a poor user experience. Summary of the Invention

[0004] The embodiments of the present application provide a base station decommissioning prediction method and system to solve the problem in the prior art that it takes a long time from base station decommissioning to repair, during which time users cannot use the network and communicate, affecting the user experience.

[0005] In order to solve the above technical problems, this application is implemented as follows:

[0006] In a first aspect, a base station out-of-service prediction method is provided, the method comprising:

[0007] Obtaining historical data of each base station within a first preset time period;

[0008] Preprocessing the historical data;

[0009] Extracting alarm feature data having a similarity greater than a first threshold from the preprocessed historical data;

[0010] Performing segmentation processing on the pre-processed historical data to obtain tag information;

[0011] Building a prediction model based on historical data within a preset number of days and the marking information, wherein the preset number of days is a preset number of days before the decommissioning prediction is performed;

[0012] The historical data of the base station within a preset number of days is input into the prediction model to perform a service withdrawal prediction for the base station.

[0013] In a second aspect, a base station out-of-service prediction system is provided, the system comprising:

[0014] An acquisition module, configured to acquire historical data of each base station within a first preset time period;

[0015] A preprocessing module, used for preprocessing the historical data;

[0016] An extraction module, configured to extract, from the pre-processed historical data, alarm feature data having a similarity greater than a first threshold;

[0017] A division processing module, configured to perform division processing on the pre-processed historical data to obtain tag information;

[0018] A construction module, configured to construct a prediction model based on historical data within a preset number of days and the marking information, wherein the preset number of days is a preset number of days before a service withdrawal prediction is performed;

[0019] The prediction module is used to input the historical data of the base station within a preset number of days into the prediction model to predict the decommissioning of the base station.

[0020] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method according to the first aspect are implemented.

[0021] In an embodiment of the present application, historical data within a first preset time period of each base station is first obtained, the historical data is preprocessed, and then alarm feature data with a similarity greater than a first threshold value in the preprocessed historical data is extracted. The preprocessed historical data is then divided and processed to obtain tag information. A prediction model is constructed based on the historical data and tag information within a preset number of days. Finally, the historical data within a preset number of days of the base station to be predicted is input into the prediction model to predict the decommissioning of the base station to be predicted. By processing the historical data, an embodiment of the present application obtains a prediction model, which can transform base station maintenance from post-processing to pre-predictive processing, greatly improving the efficiency of base station accident processing, reducing the probability of base station decommissioning, and improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0023] Figure 1 This is a flowchart of a base station out-of-service prediction method provided by an embodiment of the present application;

[0024] Figure 2 Schematic diagram of a base station out-of-service prediction system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0026] The following is combined with Figure 1-2 , a base station decommissioning prediction method and system provided in an embodiment of the present application are described in detail through specific embodiments and application scenarios.

[0027] like Figure 1 As shown in FIG, a flow chart of a base station out-of-service prediction method provided by an embodiment of the present application. Figure 1 As shown, the base station out-of-service prediction method may include: the contents shown in S101 to S106.

[0028] In S101, historical data of each base station within a first preset time period is obtained.

[0029] Among them, the first preset time period is the data of a certain month or a certain year in the historical data, or the data from the current moment to the previous year, or the data of other time periods, which shall be based on actual application and is not limited in this embodiment.

[0030] In S102, the historical data is preprocessed.

[0031] In S103, alarm feature data having a similarity greater than a first threshold value is extracted from the pre-processed historical data.

[0032] In S104, the pre-processed historical data is divided to obtain tag information.

[0033] In S105 , a prediction model is constructed based on historical data and tag information within a preset number of days.

[0034] The preset number of days is the preset number of days before the decommissioning prediction is performed.

[0035] In S106, historical data of the base station to be predicted within a preset number of days is input into a prediction model, and a decommissioning prediction is performed on the base station to be predicted.

[0036] In an embodiment of the present application, historical data within a first preset time period of each base station is first obtained, the historical data is preprocessed, and then alarm feature data with a similarity greater than a first threshold value in the preprocessed historical data is extracted. The preprocessed historical data is then divided and processed to obtain tag information. A prediction model is constructed based on the historical data and tag information within a preset number of days. Finally, the historical data within a preset number of days of the base station to be predicted is input into the prediction model to predict the decommissioning of the base station to be predicted. By processing the historical data, an embodiment of the present application obtains a prediction model, which can transform base station maintenance from post-processing to pre-predictive processing, greatly improving the efficiency of base station accident processing, reducing the probability of base station decommissioning, and improving user experience.

[0037] In a possible implementation of the present application, preprocessing the historical data may include the following steps.

[0038] The alarm names in the historical data are encoded to obtain an alarm information coding table, which is used to associate the alarm information corresponding to the alarm name with the alarm coding table; the alarm corresponding to each base station is placed in a text in units of a second preset time period, and the alarm includes the base station identifier, base station name, alarm start time, alarm end time and alarm code corresponding to the alarm.

[0039] It is worth noting that the alarm name is text information. In order to convert it into data that is convenient for machines to recognize and learn, the alarm names in the historical data are encoded and a set of alarm information coding tables are established.

[0040] In the embodiment of the present application, the alarm information corresponding to the alarm name is associated with the alarm information coding table, so that the alarm name in the form of text information is converted into the corresponding alarm information code, which facilitates subsequent alarm information processing.

[0041] For example, an optical interface performance deterioration alarm is coded as A; an optical module failure alarm is coded as C; an interface abnormality alarm is coded as D, and so on.

[0042] After the base station alarm information is converted from text into a coded form, in order to unify the information of different alarm time points into one time dimension, the minimum time granularity can be defined as a second preset time period as a unit, for example, in days, and the alarms occurring at each base station are collected into a text by day.

[0043] After preprocessing base station alarm information, we can obtain a label indicating whether each base station is out of service for each day and the alarm information code string for that day. This alarm information code string can be used to extract relevant alarm information data for feature analysis. Furthermore, multiple alarms may occur before a base station is out of service, so the outage can be associated with the multiple days preceding it.

[0044] In a possible implementation of the present application, extracting alarm feature data having a similarity greater than a first threshold from pre-processed historical data may include the following steps.

[0045] According to the category of base station alarm information, the pre-processed historical data is used for alarm identification and frequency statistics; the outage data of any day is related to the data within the preset number of days before that day, and the number of days or times related to various faults within the preset number of days is counted.

[0046] The categories include network element interruption, transmission-related warning, power supply, remote radio unit fault warning, and base station equipment.

[0047] In this embodiment, all base station alarm information is classified by alarm category. Alarm code strings of corresponding categories are identified and their frequency is counted. This allows for the preliminary extraction of characteristic indicators for each day's base station alarm information. In addition to the aforementioned categories, other categories may include clock source, data configuration, and so on, which are not detailed in this embodiment.

[0048] In the embodiment of the present application, the goal of predicting base station outage is to use the historical alarm information data of the previous few days of the predicted base station to predict the base station outage situation in the next day. For example, the data of the previous week, that is, the historical alarm information data of the previous 7 days, can be used to predict the base station outage situation in the next day. A correlation can be established between the historical alarm data of the previous 7 days and the outage data of the next day. The number of occurrences of each base station is used as the alarm information record of the current base station each day. The historical number of households is combined into 7 days plus 1 day. The number of days with network element link interruption or cell network unavailability alarms within 7 days is counted, and the time from the last occurrence of the network element link interruption or cell network unavailability alarm is counted, the time interval between each network element link interruption or cell network unavailability improvement, the number of occurrences of each alarm type within 7 days, the total number of faults within 7 days, the total number of fault types within 7 days, the number of days with faults within 7 days, the average number of faults per day within 7 days, and so on.

[0049] Through the above analysis, the relationship between the base station decommissioning and the data a few days before the decommissioning, such as multiple types of alarms, can be determined.

[0050] Furthermore, the base station out-of-service prediction method may also include the following steps.

[0051] Based on the established relationship, multiple types of sub-device alarm information generated before a base station is decommissioned are obtained; multiple alarm information with a similarity greater than a first threshold is obtained from the multiple types of sub-device alarm information; and the same alarm information in the multiple alarm information is extracted as the alarm feature data of the base station decommissioning.

[0052] Based on the above analysis, a base station will generate multiple types of device alarms before a service outage occurs. This means that different alarms will appear before a base station goes out of service. For example, device A may generate an alarm. If this alarm is not resolved, device B may generate an alarm. If the alarm on device B is not resolved, device C may generate an alarm, and so on. This ultimately causes the base station to go out of service. Alarms A, B, and C may include sub-level alarms such as A1, A2, A3, ..., B1, B2, ..., C1, C2, ....

[0053] For example, if A1, B1, A2, B2, C1, A1, and C2 alarms appear on a certain base station, and A1, B2, and C1 appear in that order, the base station will most likely be out of service.

[0054] By using the maximum common subsequence calculation method involved in similarity calculation, the embodiment of the present application can calculate the maximum common subsequence of the fault alarm chain in the historical data of each base station and use it as the alarm feature data of whether the base station is out of service.

[0055] For example,

[0056] String 1: 1, 3, 5, 10, 7

[0057] String 2: 1, 4, 3, 9, 8, 5, 11, 10

[0058] Then the largest common subsequence of strings one and two is: 1, 3, 5, 10.

[0059] Then the occurrence of 1, 3, 5, and 10 can be used as alarm features of base station out of service.

[0060] Optionally, broader subsequences such as 1, 3, 5; 3, 5, 10; 5, 10, 1, etc., can also be used as important alarm features for base station out-of-service.

[0061] In a possible implementation of the present application, dividing the pre-processed historical data to obtain tag information may include the following steps.

[0062] The pre-processed historical data is normalized; the normalized data is cleaned, including backfilling missing values ​​and identifying outliers; the cleaned data is binary-classified according to whether the base station has been decommissioned, to obtain data marked with two different identifiers, where the two different identifiers mark the decommissioned base stations and the non-decommissioned base stations, respectively.

[0063] In this embodiment, in the initially constructed base station training set, since the base station training set data is composed of features composed of alarm information of different categories and listed together, in order to avoid the bias problem of the model caused by the absolute numerical size due to the expression method of the base station alarm features, this application first normalizes the historical data.

[0064] After the base station alarm information data is combined into 7+1, there may be a day without alarm data within 7 consecutive days. In order to avoid the occurrence of null values ​​and the resulting loss of data, in this embodiment, a 0-value backfill method is used to backfill the data to make it more complete.

[0065] In addition to the aforementioned missing data, data outliers may also occur. This embodiment can use a function, such as a regression function, to fit the data to smooth the data. Linear regression involves finding the best line that fits two attributes or variables, essentially using one attribute to predict the other. This allows for the timely identification of data outliers.

[0066] The purpose of the prediction model provided in this embodiment is to identify base stations that may be decommissioned in the future from historical data. Therefore, after backfilling the data and identifying outliers, labels are added to the data to identify network element link interruption and cell unavailability alarms in the alarm information coding string, and encode them as a first identifier, for example, encoded as mark 1, representing an identifier of a base station decommissioning, and the others are set to a second identifier, for example, encoded as mark 0, representing an identifier of a base station not decommissioning. Through the above processing, historical data can be divided into two types of samples, positive samples and negative samples.

[0067] The data after the above division can then be filtered to avoid a serious imbalance in the proportions of the two samples.

[0068] Furthermore, the processed historical data is divided into a training set and a verification set according to a preset ratio. The prediction model is trained with the training set, and the model is verified with the verification set.

[0069] During the training process, the LightGBM algorithm in the classification algorithm can be used to tune parameters, which can make the calculation faster.

[0070] The parameters include control parameters, such as the tree depth of the decision tree, regularization parameters, the minimum number of tree splits, etc., core parameters, such as model type, number of iterations, learning rate, etc., measurement parameters, etc.

[0071] Optionally, in this embodiment, after the prediction model is constructed, the k-fold cross-validation method can be used to evaluate the average score of the model so that the prediction model reaches a certain score, thereby making the prediction model more accurate.

[0072] Among them, k-fold cross validation reduces the variance by averaging the results of k different group training.

[0073] After cross-validation, the data processing scripts and models are solidified into the production scheduling, automatically completing the extraction of important base station alarm information and the prediction of base station decommissioning, making subsequent base station maintenance more efficient. Base stations can be maintained in a timely manner before they are decommissioned, avoiding or reducing the occurrence of base station decommissioning and improving user experience.

[0074] like Figure 2 As shown in FIG, a schematic diagram of a base station out-of-service prediction system provided by an embodiment of the present application is provided. Figure 2 As shown, the base station out-of-service prediction system may include: an acquisition module 201 , a pre-processing module 202 , an extraction module 203 , a division processing module 204 , a construction module 205 and a prediction module 206 .

[0075] Specifically, the acquisition module 201 is used to obtain historical data of each base station within a first preset time period; the preprocessing module 202 is used to preprocess the historical data; the extraction module 203 is used to extract alarm feature data whose similarity is greater than a first threshold in the preprocessed historical data; the division processing module 204 is used to divide the preprocessed historical data and obtain labeling information; the construction module 205 is used to construct a prediction model based on the historical number of households and labeling information within a preset number of days, wherein the prediction antenna is the preset number of days before the decommissioning prediction is performed; the prediction module 206 is used to input the historical data of the base station within the preset number of days into the prediction model to perform decommissioning prediction for the base station.

[0076] In summary of the embodiments of the present application, first, the acquisition module 201 acquires the historical data of each base station within the first preset time period, the preprocessing module 202 preprocesses the historical data, and then the extraction module 203 extracts the alarm feature data whose similarity is greater than the first threshold value in the preprocessed historical data, and then the division processing module 204 divides and processes the preprocessed historical data to obtain the marking information, and the construction module 205 constructs a prediction model based on the historical data and marking information within the preset number of days, and finally the prediction module 206 inputs the historical data within the preset number of days of the base station to be predicted into the prediction model, and predicts the decommissioning of the base station to be predicted. The embodiments of the present application obtain a prediction model by processing the historical data, which can transform the base station maintenance from post-processing to pre-prediction processing, greatly improving the efficiency of base station accident processing, reducing the probability of base station decommissioning, and improving user experience.

[0077] In a possible implementation of the present application, the pre-processing module 202 may be used to:

[0078] The alarm names in the historical data are encoded to obtain an alarm information coding table, which is used to associate the alarm information corresponding to the alarm name with the alarm coding table; the alarm corresponding to each base station is placed in a text in units of a second preset time period, and the alarm includes the base station identifier, base station name, alarm start time, alarm end time and alarm code corresponding to the alarm.

[0079] In a possible implementation of the present application, the extraction module may be used to:

[0080] According to the category of base station alarm information, the pre-processed historical data is used for alarm identification and frequency statistics. The categories include network element interruption category, transmission-related warning category, power category, radio frequency remote unit fault warning category and base station equipment category; a relationship is established between the out-of-service data of any day and the data within the preset number of days before the day, and the number of days or times related to various faults within the preset number of days is counted.

[0081] In a possible implementation of the present application, the extraction module may also be used to:

[0082] Based on the established relationship, multiple types of sub-device alarm information generated before a base station is decommissioned are obtained; multiple alarm information with a similarity greater than a first threshold is obtained from the multiple types of sub-device alarm information; and the same alarm information in the multiple alarm information is extracted as the alarm feature data of the base station decommissioning.

[0083] In a possible implementation of the present application, the division processing module may be used to:

[0084] The pre-processed historical data is normalized; the normalized data is cleaned, including backfilling missing values ​​and identifying outliers; the cleaned data is binary-classified according to whether the base station has been decommissioned, to obtain data marked with two different identifiers, where the two different identifiers mark the decommissioned base stations and the non-decommissioned base stations, respectively.

[0085] The functions of the base station out-of-service prediction system described in this application have been Figure 1 The method embodiment shown is described in detail, so for any details not fully described in this embodiment, please refer to the relevant descriptions in the aforementioned embodiments and will not be repeated here.

[0086] Optionally, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the various processes of the above-mentioned base station decommissioning prediction method embodiment are implemented, and the same technical effects are achieved. To avoid repetition, the details are not described here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0087] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0088] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0089] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. A method for predicting base station out-of-service, characterized in that: include: Obtaining historical data of each base station within a first preset time period; Preprocessing the historical data; Extracting alarm feature data having a similarity greater than a first threshold from the preprocessed historical data; Performing segmentation processing on the pre-processed historical data to obtain tag information; Building a prediction model based on historical data within a preset number of days and the marking information, wherein the preset number of days is a preset number of days before the decommissioning prediction is performed; Inputting historical data of a base station to be predicted within a preset number of days into the prediction model, and performing a service withdrawal prediction for the base station to be predicted; The extracting the alarm feature data having a similarity greater than a first threshold from the pre-processed historical data includes: Performing alarm identification and frequency statistics on the pre-processed historical data according to the categories of base station alarm information, wherein the categories include network element interruption category, transmission-related warning category, power category, remote radio unit failure warning category, and base station equipment category; Establish a relationship between the outage data of any day and the data within a preset number of days before the day, and count the days or times related to various failures within the preset number of days; According to the established relationship, multiple types of sub-device alarm information generated before a base station is decommissioned is obtained; Acquire multiple pieces of alarm information having a similarity greater than a first threshold from the multiple types of sub-device alarm information; The same alarm information among the multiple alarm information is extracted as the alarm feature data of the base station being out of service.

2. The method according to claim 1, characterized in that The preprocessing of the historical data includes: Encoding the alarm names in the historical data to obtain an alarm information encoding table, wherein the alarm information encoding table is used to associate the alarm information corresponding to the alarm name with the alarm encoding table; The alarm corresponding to each base station is placed in a text in units of a second preset time period, wherein the alarm includes the base station identifier corresponding to the alarm, the base station name, the alarm start time, the alarm end time and the alarm code.

3. The method according to claim 1, characterized in that The dividing and processing of the pre-processed historical data to obtain tag information includes: performing normalization processing on the preprocessed historical data; Performing data cleaning on the normalized data, wherein the data cleaning includes backfilling missing values ​​and identifying outliers; According to whether the base station is out of service, the data after the data cleaning process is subjected to binary classification processing to obtain data marked with two different identifiers, wherein the two different identifiers respectively mark the out-of-service base station and the non-out-of-service base station.

4. A base station out-of-service prediction system, characterized in that: include: An acquisition module, configured to acquire historical data of each base station within a first preset time period; A preprocessing module, used for preprocessing the historical data; An extraction module, configured to extract, from the pre-processed historical data, alarm feature data having a similarity greater than a first threshold; A division processing module, configured to perform division processing on the pre-processed historical data to obtain tag information; A construction module, configured to construct a prediction model based on historical data within a preset number of days and the marking information, wherein the preset number of days is a preset number of days before a service withdrawal prediction is performed; A prediction module, configured to input historical data of a base station within a preset number of days into the prediction model to predict the decommissioning of the base station; The extraction module is specifically configured to perform alarm identification and frequency statistics on the pre-processed historical data according to the categories of base station alarm information, wherein the categories include network element interruption category, transmission-related warning category, power category, remote radio unit failure warning category, and base station equipment category; Establish a relationship between the outage data of any day and the data within a preset number of days before the day, and count the days or times related to various failures within the preset number of days; According to the established relationship, multiple types of sub-device alarm information generated before a base station is decommissioned is obtained; Acquire multiple pieces of alarm information having a similarity greater than a first threshold from the multiple types of sub-device alarm information; The same alarm information among the multiple alarm information is extracted as the alarm feature data of the base station being out of service.

5. The system according to claim 4, characterized in that The pre-processing module is used to: Encoding the alarm names in the historical data to obtain an alarm information encoding table, wherein the alarm information encoding table is used to associate the alarm information corresponding to the alarm name with the alarm encoding table; The alarm corresponding to each base station is placed in a text in units of a second preset time period, wherein the alarm includes the base station identifier corresponding to the alarm, the base station name, the alarm start time, the alarm end time and the alarm code.

6. The system according to claim 4, characterized in that The division processing module is used to: performing normalization processing on the preprocessed historical data; Performing data cleaning on the normalized data, wherein the data cleaning includes backfilling missing values ​​and identifying outliers; According to whether the base station is out of service, the data after the data cleaning process is subjected to binary classification processing to obtain data marked with two different identifiers, wherein the two different identifiers respectively mark the out-of-service base station and the non-out-of-service base station.

Images