Method and system for identifying offline reason of communication base station under natural disaster

Abstract

The invention discloses a method and system for identifying an offline reason of a communication base station under a natural disaster, and the method comprises the steps: obtaining and preprocessing multi-source data, and obtaining the preprocessed multi-source data; the multi-source data comprises real-time meteorological information, communication base station off-line and on-line information and power supply tower off-line and on-line information; constructing a base station-iron tower association relationship network based on the association relationship between the communication base station and the antenna and the association relationship between the antenna and the power supply iron tower; according to a historical offline rule of the communication base station, recognizing regular base station offline caused by the energy-saving protection strategy, and obtaining a base station energy-saving offline recognition result; and based on the multi-source data, the association relationship network and the identification result, multi-dimensional features are extracted, feature vectors are constructed, and a pre-trained machine learning classification model is utilized to identify the offline reason of the base station. According to the method, the base station offline reason identification time under natural disasters is effectively shortened, the identification accuracy is improved, and an accurate data basis is provided for formulating a communication base station maintenance scheme.

Description

Technical Field

[0001] This application relates to the fields of communication assurance and intelligent operation and maintenance technology, and in particular to a method and system for identifying the causes of communication base station offline under natural disasters. Background Technology

[0002] Mobile communication networks are a critical infrastructure of modern society, and their stable operation is essential for daily life, production, and emergency disaster relief. As the core nodes of the network, the online status of communication base stations directly determines the availability of regional communication services. When major natural disasters occur (such as rainstorms, typhoons, and earthquakes), they are often accompanied by large-scale power outages, damage to transmission cables, and physical damage to base station equipment, resulting in a large number of base stations going offline and communication services being interrupted.

[0003] Currently, when a base station goes offline, operators typically rely on alarm information from the Network Management System (NMS). However, these alarms often only reflect the "base station offline" status and cannot directly and accurately reveal the underlying cause (e.g., whether it's due to water damage from heavy rainfall, a break in the associated power supply tower, or the triggering of nighttime energy-saving strategies). To determine the specific cause, it often requires dispatching technicians to the disaster site for physical investigation and testing, a time-consuming and labor-intensive process that delays optimal repair decisions and actions, prolongs communication outages, and severely impacts disaster relief command and public communication.

[0004] Therefore, there is an urgent need for a technical solution that can quickly, remotely, and automatically identify the reasons for base station offline after a disaster, in order to support accurate maintenance and scheduling and improve the emergency recovery capability of communication networks. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this application is to provide a method and system for identifying the causes of communication base station offline under natural disasters, which can shorten the fault diagnosis time and improve the accuracy of identifying the causes of communication base station offline.

[0006] To achieve the above objectives, this application provides a method for identifying the causes of communication base station offline under natural disasters, including: Multi-source data is acquired and preprocessed to obtain preprocessed multi-source data. The multi-source data includes real-time meteorological information, online / offline information of communication base stations, and online / offline information of power supply towers. Based on the correlation between communication base stations and antennas, and between antennas and power supply towers, a base station-tower correlation network is constructed. According to the historical offline patterns of communication base stations, regular base station offline caused by energy-saving protection strategies is identified, and a base station energy-saving offline identification result is obtained. Based on the multi-source data, the correlation network, and the identification result, multi-dimensional features are extracted and feature vectors are constructed. A pre-trained machine learning classification model is used to identify the reasons for base station offline.

[0007] Furthermore, the step of acquiring and preprocessing multi-source data to obtain preprocessed multi-source data also includes: performing time alignment and regional aggregation on the acquired multi-source data; and calculating the base station offline rate and tower offline rate for each region.

[0008] Furthermore, the step of performing time alignment and regional aggregation on the acquired multi-source data further includes: the regional aggregation is performed on a district / county basis; the meteorological information includes rainfall data, and the regional aggregation includes segmenting the rainfall information according to the district / county administrative division vector data and calculating the rainfall statistics of each district / county at each time point.

[0009] Furthermore, the step of constructing a base station-tower association network based on the association relationship between the communication base station and the antenna, and between the antenna and the power supply tower, further includes: obtaining a first association relationship between the communication base station and its connected antenna; obtaining a second association relationship between the antenna and the power supply tower, wherein the second association relationship is determined based on network topology and / or spatial location; and determining the power supply dependency relationship between the communication base station and the power supply tower based on the first association relationship and the second association relationship, thereby forming an association network.

[0010] Furthermore, the step of identifying regular base station offline events caused by energy-saving protection strategies based on the historical offline patterns of communication base stations and obtaining energy-saving identification results further includes: for the target base station, statistically analyzing the start time, end time, and offline duration of its historical offline events; if the historical offline events meet the conditions of occurring regularly within a specific time period each day and having a stable offline duration, then it is determined that the base station has an energy-saving offline mode, and the reason for its offline events during that time period is energy-saving protection.

[0011] Furthermore, in the step of extracting multidimensional features and constructing feature vectors based on the multi-source data, the correlation relationship, and the energy-saving identification results, and using a pre-trained machine learning classification model to identify the causes of base station offline, the extraction of multidimensional features includes: extracting static features reflecting historical behavior patterns in the region; extracting time-series features reflecting recent dynamic changes; and extracting correlation features reflecting the interaction between different data sequences. The correlation features include the Pearson correlation coefficient between the base station offline rate sequence and the tower offline rate sequence, as well as the temporal relationship of the increase in their offline rates.

[0012] Furthermore, the step of extracting multi-dimensional features and constructing feature vectors based on the multi-source data, the correlation relationships, and the energy-saving identification results, and using a pre-trained machine learning classification model to identify the causes of base station offline, further includes: acquiring historically verified and labeled base station interruption events and their causes to form a training sample set; training the machine learning classification model based on the training sample set and the extracted multi-dimensional features; and when an abnormal base station offline rate is detected, extracting the current feature vector in real time and inputting it into the trained model to output the classification result of the offline cause.

[0013] Furthermore, the machine learning classification model is a gradient boosting decision tree model.

[0014] To achieve the above objectives, this application also provides a system for identifying the causes of communication base station offline under natural disasters. The system is configured to execute the method for identifying the causes of communication base station offline under natural disasters as described above, including: a data acquisition and preprocessing module, responsible for connecting to various data sources and receiving, cleaning, converting, and storing streaming or batch data; an association relationship construction module, used to call the association relationship data between communication base stations and antennas, and between antennas and power supply towers in a resource database, perform association relationship calculations between communication base stations and power supply towers, and store the calculation results in a graph database or relational database; an energy-saving offline identification module, used to identify regular base station offline events caused by energy-saving protection strategies and update the base station energy-saving mode model library; and a feature extraction and cause identification module, used to extract features in real time based on the data stored in each database and load models to predict the features.

[0015] Furthermore, the feature extraction and cause identification module includes: a feature extraction unit for extracting static features, temporal features, and correlation features; and a model inference unit for inputting the features extracted by the feature extraction unit into a pre-trained machine learning classification model to obtain and output the identification result of the offline cause.

[0016] To achieve the above objectives, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program stored in the memory to implement the method for identifying the causes of communication base station offline under natural disasters as described above. To achieve the above objectives, this application also provides a computer-readable storage medium storing a computer program, which is loaded and executed by a processor to implement the method for identifying the causes of communication base station offline under natural disasters as described above. The method for identifying the causes of communication base station offline under natural disasters provided by this application, through multi-source data analysis and machine learning classification models, can effectively distinguish different causes of base station offline, providing accurate data basis for formulating targeted maintenance plans.

[0017] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing this application. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the present application and form part of the specification. Together with the embodiments of the present application, they serve to explain the present application but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a method for identifying the causes of communication base station offline under natural disasters according to an embodiment of this application; Figure 2 This is a schematic diagram of simultaneous rainfall and offline rate according to an embodiment of this application; Figure 3 This is a schematic diagram of a base station outage event report according to an embodiment of this application; Figure 4 This is a schematic diagram of a communication base station offline cause identification system according to an embodiment of this application; Figure 5 This is a schematic diagram of an electronic device structure according to an embodiment of this application. Detailed Implementation

[0019] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application.

[0020] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0021] The term "comprising" and its variations as used in this application are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0022] It should be noted that the terms "first" and "second" may be used in this application only to distinguish different devices, components or parts, and are not used to define the order of functions performed by these devices, components or parts or their interdependence.

[0023] It should be noted that the terms "one" and "more" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless explicitly stated otherwise in the context, they should be understood as "one or more". "More" should be understood as two or more.

[0024] This application discloses a method for identifying the causes of communication base station offline under natural disasters, comprising: acquiring and preprocessing multi-source data to obtain preprocessed multi-source data; the multi-source data includes real-time meteorological information, online / offline information of communication base stations, and online / offline information of power supply towers; constructing a base station-tower association network based on the association relationship between communication base stations and antennas, and between antennas and power supply towers; identifying regular base station offline caused by energy-saving protection strategies based on the historical offline patterns of communication base stations, and obtaining base station energy-saving offline identification results; extracting multi-dimensional features and constructing feature vectors based on the multi-source data, the association network, and the identification results, and using a pre-trained machine learning classification model to identify the causes of base station offline. This method addresses the problems of high difficulty and long cycle in identifying the causes of communication base station offline under disaster scenarios by introducing a method for identifying the causes of base station offline under natural disasters based on meteorological information, online / offline information of communication base stations and power supply towers. While avoiding the risks of on-site verification by technicians in natural disaster scenarios, it effectively shortens the time for identifying the causes of base station offline, provides accurate basis for formulating communication base station maintenance plans, can significantly improve base station maintenance efficiency, and thus shorten communication interruption time.

[0025] Example 1 Figure 1 The following is a flowchart of a method for identifying the causes of communication base station offline under natural disasters according to an embodiment of this application. Figure 1 The embodiments of this application are described in detail.

[0026] Step 101: Acquire multi-source data and preprocess it.

[0027] This step includes: Step 1011: Acquire multi-source data, including real-time meteorological information, online / offline information of communication base stations, and online / offline information of power supply towers. Among these: The meteorological information is obtained from the meteorological bureau's API or data center and contains at least high spatiotemporal resolution gridded rainfall data.

[0028] The online / offline information of the communication base station is obtained in real time from the operator's network management system (NMS) or network management center (OMC), and includes fields such as base station ID, district / county, online status (0 / 1), and status update time.

[0029] The online / offline information of the power supply towers is collected from the tower company's monitoring system or power and environment monitoring system, and includes fields such as tower ID, geographical location, power supply status, and status update time.

[0030] In some exemplary embodiments, this step also includes screening rainfall disasters that have a significant impact on base station offline as research objects for meteorological disasters causing base station offline.

[0031] Step 1012: Perform preprocessing on the multi-source data, including time alignment, regional aggregation, offline rate calculation, and data visualization (optional).

[0032] The time alignment refers to the use of UTC timestamps for all data, and aggregation and resampling according to a preset time granularity (e.g., 5 minutes), with missing values ​​handled by forward padding or interpolation.

[0033] The aforementioned regional aggregation refers to loading vector data (GIS vector boundary files) of district and county administrative divisions. For meteorological data, spatial analysis techniques (such as zonal statistics) are used to calculate the statistical values ​​of each district and county at each time granularity, such as the maximum rainfall, average rainfall, and minimum rainfall in the past 12 hours, forming the district and county rainfall statistics table as shown in Table 1. Table 1 Rainfall data for a certain district / county

[0034] The offline rate calculation involves counting the number of base stations and towers that are offline in each district / county at each time granularity, and then dividing them by the total number of base stations and towers in that district / county (which needs to be obtained from the resource database in advance) to obtain the base station offline rate and tower offline rate of that district / county at that time. Base station offline rate = ∑(number of offline base stations in the district / county) ÷ ∑(total number of base stations in the district / county) Tower offline rate = ∑(number of offline towers in districts and counties) ÷ ∑(total number of towers in districts and counties).

[0035] The data visualization includes plotting the processed rainfall sequence, base station offline rate sequence, and tower offline rate sequence on the same time axis, such as... Figure 2 As shown (simultaneous rainfall and offline rate), it intuitively demonstrates the changing trends and potential correlations of the three over time.

[0036] Step 102: Construct a network of relationships between base stations and towers.

[0037] In the embodiments of this application, base station offline is often caused by the interruption of the power supply tower to which it supplies power. Therefore, clarifying the power supply dependency between the base station and the tower is crucial. This step constructs a network of relationships in the following way: Step 1021: Query the base station-antenna attribution table from the network resource database (each communication base station connects to one or more antennas).

[0038] Step 1022: Retrieve the physical connection table of antennas and towers from the tower resource database (each antenna is installed on one tower). Simultaneously, for antennas lacking clear topology records, perform spatial proximity analysis (e.g., buffer analysis) using their latitude and longitude coordinates and the latitude and longitude coordinates of the towers. If the antenna coordinates fall within a preset distance range (e.g., 50 meters) of a tower, then establish a connection.

[0039] Based on the two relationship tables mentioned above, the final base station-tower relationship table is derived by using antennas as intermediate nodes and performing relationship connection operations. This base station-tower relationship table records the power supply towers (which may be one or more) that each base station depends on. This relationship network is static or quasi-static and can be updated periodically.

[0040] In some exemplary embodiments, the base station-antenna attribution table is obtained based on the communication base station and antenna topology diagram. The antenna-tower physical connection table is obtained based on the antenna and tower topology diagram, and the antenna-tower association is supplemented through spatial association relationships. Through these antenna-to-communication base station association relationships and antenna-to-tower association relationships, the base station and tower are associated, thereby forming an association network.

[0041] Step 103: Identify offline status caused by energy-saving protection.

[0042] To save energy, operators often shut down some base stations for energy conservation during off-peak hours (such as at night). This type of offline activity is recurring and needs to be distinguished from sudden offline activities caused by disasters. The steps include: Step 1031: For each base station to be analyzed, extract all offline events from its historical status log and record the start time, end time, and duration of each event. For example, for an independent communication base station, construct an offline event statistics table to count the start time, end time, and offline duration of offline events, forming the offline event statistics table shown in Table 2.

[0043] Table 2 Base Station Offline Events

[0044] Step 1032: Perform pattern analysis to identify regular base station offline events caused by energy-saving protection strategies. Specifically, check whether offline events are concentrated in specific time periods each day (e.g., 2:00 AM to 4:00 AM daily); calculate the mean and variance of the duration of these events. If the variance is small (e.g., standard deviation less than 15 minutes), it indicates that the offline duration is stable.

[0045] If a communication base station meets both the conditions of occurring at regular intervals and having a stable duration, it is marked as having an energy-saving offline mode. When the base station goes offline during its energy-saving period, the cause is primarily inferred to be energy-saving protection rather than a fault.

[0046] In some exemplary implementations, by identifying regular offline events of communication base stations, if the offline events mostly occur at the same time and the offline duration is roughly the same (the difference is within a preset threshold range), it is determined to be a base station energy-saving offline event (regular base station offline caused by energy-saving protection strategy).

[0047] Step 104: Use a pre-trained machine learning classification model to identify the reasons for the base station going offline.

[0048] The foundation of this step is training a model that automatically categorizes base station offline events into predefined cause categories based on current and historical data features. For example, at the district / county level, it constructs time-series trends of meteorological information, base station offline events, and tower offline events, extracts trend features, constructs feature vectors, and uses a pre-trained machine learning classification model to identify the causes of base station offline events in a given area.

[0049] This step specifically includes: Step 1041: Extract static features (reflecting the inherent attributes and historical behavior patterns of the district / county) from the district / county's historical (e.g., the past 3 months) base station offline and tower offline data. These features include whether the base station is offline for energy saving, the time period of the base station's offline for energy saving (e.g., '02:00-04:00'), the number of times the base station has been offline in history, the historical average daily base station offline rate, whether the tower is offline for energy saving, the time period of the tower's offline for energy saving, the number of times the tower has been offline in history, and the historical average daily tower offline rate.

[0050] Step 1042: Extract time-series features (reflecting recent dynamic changes) from the current time and a retrospective period (e.g., the last 24 hours) of meteorological information, base station offline data, and tower offline data for the district / county. These features include the maximum rainfall in the last 24 hours, the rainfall trend (which can be represented by the slope of a linear fit), the maximum rainfall intensity (rainfall per unit time), the maximum base station offline rate, the increase in the maximum base station offline rate in the last 24 hours, and the average base station offline rate in the last 24 hours. Similarly, extract features such as the maximum tower offline rate, the increase in the maximum tower offline rate, and the average tower offline rate.

[0051] Step 1043: Characterize the interaction relationships between different data sequences (using Pearson correlation coefficient in this embodiment) to obtain correlation features. Specifically, the Pearson correlation coefficient is used to characterize the pairwise interactions between meteorological information, base station offline status, and tower offline status at the current time and over a period of time in the district / county. Based on the determination that there is a correlation (high Pearson correlation coefficient), further analysis is conducted on the order of the two at key time points of sudden increases in offline rates (i.e., the temporal relationship of the rise in offline rates). For example, the Pearson correlation coefficient between the 24-hour rainfall sequence and the base station offline rate sequence, the Pearson correlation coefficient between the 24-hour base station offline rate sequence and the tower offline rate sequence, and the temporal relationship between the sudden increases in base station offline rate and tower offline rate (e.g., if the increase in tower offline rate precedes the increase in base station offline rate, it is encoded as 1, indicating that power supply was interrupted first).

[0052] Step 1044: Perform sample annotation. Collect historical base station outage event reports from the maintenance work order system that have been ultimately confirmed on-site by technical personnel, such as... Figure 3 As shown, each report includes: the warning district / county, the base station and / or tower outage rate, and the reason for the offline status.

[0053] In some exemplary implementations, each report also includes: time, number of affected base stations, and confirmed root cause (e.g., "flooding of the equipment room due to heavy rain," "power outage to the associated tower," "equipment hardware failure," etc.). These causes are categorized into several main labels, such as: "weather disaster," "power outage," "energy-saving plan," "equipment failure," and "other." The static and temporal characteristics of historical events are paired with their corresponding cause labels to form an labeled training sample set.

[0054] Step 1045: Perform model training and deployment. This step includes: 1) Choose a classification algorithm, such as XGBoost, LightGBM, or Random Forest. These algorithms can effectively handle structured features and have good learning capabilities for complex relationships between features. Among them, XGBoost and LightGBM are both gradient boosting decision tree algorithms, which are popular tools in the field of machine learning for processing tabular data (such as feature vectors).

[0055] 2) Divide the labeled sample set into training set and test set in proportion (e.g., 7:3).

[0056] 3) Train the model using the training set and adjust hyperparameters (such as tree depth, learning rate, etc.) through cross-validation to optimize model performance.

[0057] 4) Evaluate the model on the test set. Common metrics include accuracy, precision, recall, F1 score, and confusion matrix.

[0058] 5) After the model training is completed, save the model file and integrate it into the online recognition system.

[0059] Step 1046: Perform online recognition using the trained model.

[0060] In the embodiments of this application, when a natural disaster occurs and an abnormally high offline rate of base stations in a certain district or county is detected (when the deviation of the base station offline rate from the historical period of the same period in the region or a preset threshold exceeds a set threshold, i.e., the offline rate abnormality judgment is triggered), the identification process of this application is automatically triggered. Steps 101-103 are executed in real time, and the features described in step 104 are extracted for the district or county in real time and a feature vector is constructed. The feature vector is input into the deployed machine learning model. The model outputs the probability distribution of each cause category. The category with the highest probability is taken as the identification result of the main cause of the current base station offline in the district or county, and further drill-down analysis can be performed on the specific affected base stations and their possible causes (combining the correlation in step 102 and the energy-saving identification in step 103). The identification result (such as "XX district or county, the main cause of the current base station offline: power outage (related to large-scale offline of iron towers)") is pushed to the operation and maintenance command center to provide decision support for dispatching repair teams and preparing corresponding spare parts.

[0061] The method in this application integrates multi-dimensional real-time data such as meteorological data, base station status data, and tower status data, and introduces artificial intelligence algorithms to achieve automated and intelligent identification of the causes of base station offline under natural disasters, and has the following significant advantages: (1) Rapid response: No need to wait for manual on-site verification, the cause analysis results can be output within minutes after the data is complete, which greatly shortens the fault location time.

[0062] (2) Accurate identification: Through multi-source data analysis and machine learning classification models, it is possible to effectively distinguish different reasons that cause base stations to go offline, providing accurate data basis for formulating targeted maintenance plans.

[0063] (3) Ensuring safety: It reduces the need for technicians to work on-site in dangerous disaster environments and lowers the risk to personal safety.

[0064] (4) Improve efficiency: assist maintenance personnel in prioritizing the handling of faults with a wide impact and clear root causes, optimize the allocation of emergency repair resources, accelerate the overall network recovery speed, and ensure the communication lifeline during disasters.

[0065] Example 2 In the embodiments of this application, a system for identifying the causes of communication base station offline under natural disasters is provided, employing the method described in Embodiment 1. This system can be deployed in the cloud or on a carrier's private cloud platform.

[0066] Figure 4As shown in Figure 4, the system 300 for identifying the causes of communication base station offline during natural disasters, according to an embodiment of this application, includes a data acquisition and preprocessing module 201, a correlation construction module 202, an energy-saving offline identification module 203, and a feature extraction and cause identification module 204. The data acquisition and preprocessing module 201 is responsible for connecting to various data source APIs (pulling or receiving raw data from external systems) to perform streaming or batch data access, cleaning, transformation, and storage (which can be stored in a time-series database or a big data platform). The input of the correlation construction module 202 is connected to the data acquisition and preprocessing module 201, and its output is connected to the input of the feature extraction and cause identification module 204. It is used to call a resource database, perform correlation calculations, and store the results in a graph database or a relational database. The input of the energy-saving offline identification module 203 is connected to the data acquisition and preprocessing module 201, and the output is connected to the feature extraction and cause identification module 204. It is used to periodically (e.g. daily) analyze the historical status of base stations, identify regular base station offline events caused by energy-saving protection strategies, and update the base station energy-saving mode library (marking which base stations will go offline due to energy saving during which time periods). This is the key input for constructing static features such as "whether the base station is offline due to energy saving" and eliminating false judgments.

[0067] The feature extraction and cause identification module 204, as the core processing engine, is scheduled when a fault occurs and serves as the system's intelligent decision-making endpoint, outputting the final service result. It comprises a feature extraction unit and a model inference unit. The feature extraction unit extracts features from stored data; the model inference unit loads the model and makes predictions based on the features. This module can output the results to upper-layer application systems (such as operations dashboards or work order systems) via API or message queues.

[0068] Example 3 In embodiments of this application, an electronic device is also provided. Figure 5 This is a schematic diagram of the electronic device structure according to an embodiment of this application, such as... Figure 5 As shown, the electronic device of this application includes a processor 301 and a memory 302, wherein, The memory 302 stores a computer program, which, when read and executed by the processor 301, performs the steps described above in the embodiment of the method for identifying the cause of communication base station offline under natural disasters.

[0069] Example 4 In the embodiments of this application, a computer-readable storage medium is also provided, which stores a computer program, wherein the computer program is configured to execute the steps in the embodiments of the method for identifying the cause of communication base station offline under natural disasters as described above when it is run.

[0070] In this embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0071] It will be understood by those skilled in the art that the above descriptions are merely preferred embodiments of this application and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for identifying the causes of communication base station offline under natural disasters, characterized in that, Includes the following steps: Acquire and preprocess multi-source data to obtain preprocessed multi-source data; the multi-source data includes real-time meteorological information, online / offline information of communication base stations, and online / offline information of power supply towers. Based on the relationship between communication base stations and antennas, and between antennas and power supply towers, a base station-tower relationship network is constructed. Based on the historical offline patterns of communication base stations, regular base station offline events caused by energy-saving protection strategies are identified, and the base station energy-saving offline identification results are obtained. Based on the multi-source data, the correlation network, and the identification results, multi-dimensional features are extracted and feature vectors are constructed. A pre-trained machine learning classification model is then used to identify the reasons for the base station's offline status.

2. The method for identifying the causes of communication base station offline under natural disasters according to claim 1, characterized in that, The step of acquiring and preprocessing multi-source data to obtain preprocessed multi-source data further includes: Perform time alignment and regional aggregation on the acquired multi-source data; Calculate the base station offline rate and tower offline rate for each region.

3. The method for identifying the causes of communication base station offline under natural disasters according to claim 2, characterized in that, The steps of performing time alignment and regional aggregation on the acquired multi-source data further include: The regional aggregation is performed on a district / county basis; the meteorological information includes rainfall data, and the regional aggregation includes segmenting rainfall information according to district / county administrative division vector data and calculating the rainfall statistics for each district / county at each time point.

4. The method for identifying the causes of communication base station offline under natural disasters according to claim 1, characterized in that, The step of constructing a base station-tower association network based on the association between communication base stations and antennas, and between antennas and power supply towers, further includes: Obtain the first correlation between the communication base station and its connected antenna; Obtain a second association between the antenna and the power supply tower, the second association being determined based on network topology and / or spatial location; Based on the first and second association relationships, the power supply dependency relationship between the communication base station and the power supply tower is determined, forming an association network.

5. The method for identifying the causes of communication base station offline under natural disasters according to claim 1, characterized in that, The step of identifying regular base station offline events caused by energy-saving protection strategies based on the historical offline patterns of communication base stations, and obtaining energy-saving identification results, further includes: For the target base station, the start time, end time, and offline duration of its historical offline events are recorded. If the historical offline events meet the conditions of occurring regularly within a specific time period each day and having a stable offline duration, then the base station is determined to have an energy-saving offline mode, and the reason for its offline status during that time period is energy-saving protection.

6. The method for identifying the causes of communication base station offline under natural disasters according to claim 1, characterized in that, In the step of extracting multidimensional features and constructing feature vectors based on the multi-source data, the correlation relationships, and the energy-saving identification results, and using a pre-trained machine learning classification model to identify the causes of base station offline, the extraction of multidimensional features includes: extracting static features reflecting historical behavior patterns in the region; extracting temporal features reflecting recent dynamic changes; and extracting correlation features reflecting the interaction relationships between different data sequences. The correlation features include the Pearson correlation coefficient between the base station offline rate sequence and the tower offline rate sequence, as well as the temporal relationship of the order in which their offline rates rise.

7. The method for identifying the causes of communication base station offline under natural disasters according to claim 1, characterized in that, The step of extracting multi-dimensional features and constructing feature vectors based on the multi-source data, the correlation relationships, and the energy-saving identification results, and using a pre-trained machine learning classification model to identify the reasons for base station offline, further includes: acquiring historically verified and labeled base station interruption events and their causes to form a training sample set; training the machine learning classification model based on the training sample set and the extracted multi-dimensional features; and when an abnormal base station offline rate is detected, extracting the current feature vector in real time and inputting it into the trained model to output the classification result of the offline cause.

8. The method for identifying the causes of communication base station offline under natural disasters according to claim 7, characterized in that, The machine learning classification model is a gradient boosting decision tree model.

9. A system for identifying the causes of communication base station offline under natural disasters, the system being configured to perform the method for identifying the causes of communication base station offline under natural disasters as described in any one of claims 1 to 8, characterized in that, include: The data acquisition and preprocessing module is responsible for connecting to various data sources and receiving, cleaning, transforming and storing streaming or batch data; The association construction module is used to call the association data between communication base stations and antennas, and between antennas and power supply towers in the resource database, perform the association calculation between communication base stations and power supply towers, and store the calculation results in a graph database or a relational database. The energy-saving offline identification module is used to identify regular base station offline events caused by energy-saving protection strategies and update the base station energy-saving mode database; the feature extraction and cause identification module is used to extract features in real time based on the data stored in each database and load the model to predict the features.

10. The system for identifying the causes of communication base station offline under natural disasters according to claim 9, characterized in that, The feature extraction and cause identification module further includes: a feature extraction unit for extracting static features, temporal features, and correlation features; and a model inference unit for inputting the features extracted by the feature extraction unit into a pre-trained machine learning classification model to obtain and output the offline cause identification result.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor is used to execute the computer program stored in the memory to implement the method for identifying the cause of communication base station offline under natural disasters as described in any one of claims 1 to 8.

12. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is loaded and executed by a processor to implement the method for identifying the cause of communication base station offline under natural disasters as described in any one of claims 1 to 8.

Images