Base station antenna damage assessment method and device and medium

Abstract

The invention provides a base station antenna damage assessment method and device and a medium, and relates to the technical field of communication. The method comprises the following steps: acquiring pre-disaster performance index data and post-disaster performance index data of a cell according to a plurality of preset indexes; according to the pre-disaster performance index data of the cell and the post-disaster performance index data of the cell, obtaining a significant difference between the pre-disaster performance index data and the post-disaster performance index data of each index; according to the pre-disaster performance index data of the cell and the post-disaster performance index data of the cell, acquiring the pre-disaster and post-disaster variation amplitude of each index; and obtaining a disaster influence score of each index according to the significant difference and the change amplitude, and obtaining a comprehensive evaluation score of the base station antenna damage caused by the disaster according to the disaster influence score of each index. According to the method, the significant difference and the variation amplitude of the multiple indexes before and after the disaster are calculated, the disaster influence of the antenna is evaluated from the multiple indexes, the damage of the antenna is evaluated by integrating the disaster influence of the multiple indexes, and the accuracy of quantitative evaluation of the damage of the base station antenna is improved.

Description

Technical Field

[0001] This application relates to at least the field of communication technology, and in particular to a method, apparatus and medium for assessing damage to base station antennas. Background Technology

[0002] In the field of mobile communication network operation and maintenance, the damage to the physical condition of base station antennas (such as tilt and azimuth angles) caused by extreme weather disasters such as typhoons is one of the main reasons for the sharp decline in network performance. Quickly and accurately identifying damaged antennas and quantifying their impact after a disaster is crucial for efficiently allocating operation and maintenance resources and shortening network recovery time.

[0003] Disadvantages of existing technology:

[0004] 1. The assessment results are easily affected by interference and have a high misjudgment rate: Existing technologies rely on the absolute change of a single performance indicator (such as traffic or RSRP mean) for judgment, which fails to effectively distinguish between systemic performance degradation caused by typhoons and random fluctuations caused by non-disaster factors such as daily business fluctuations and changes in user behavior. This results in a high false positive or false negative rate in the assessment results and a lack of reliability.

[0005] 2. Lack of unified and objective quantitative standards: Existing assessment methods can usually only give a binary conclusion of "damaged" or "undamaged", which cannot scientifically quantify and classify the severity of antenna damage. There is a lack of unified assessment standards that can be compared horizontally in different regions and scenarios, which is not conducive to the accurate and prioritized allocation of post-disaster resources.

[0006] 3. The evaluation process is fragmented and lacks a systematic closed loop: Existing technologies rely on isolated data analysis or experience-based judgments, failing to form a standardized, closed-loop processing flow from data collection, model analysis, on-site verification to optimization execution, making it difficult to achieve large-scale, efficient promotion and application within the operator system. Summary of the Invention

[0007] To address the aforementioned shortcomings, this application provides a base station antenna damage assessment method, apparatus, and medium to solve the following technical problem: how to accurately assess antenna damage.

[0008] In a first aspect, this application provides a method for assessing damage to a base station antenna, the method comprising:

[0009] Based on multiple preset indicators, obtain pre-disaster performance index data and post-disaster performance index data of the community;

[0010] Based on the pre-disaster and post-disaster performance index data of the community, obtain the significant differences of each index before and after the disaster.

[0011] Based on the pre-disaster and post-disaster performance index data of the community, obtain the change range of each index before and after the disaster.

[0012] The disaster impact score for each indicator is obtained based on the significance of the difference and the magnitude of the change. The comprehensive assessment score of the damage to the base station antenna caused by the disaster is obtained based on the disaster impact score of each indicator.

[0013] Furthermore, based on multiple preset indicators, pre-disaster performance index data and post-disaster performance index data of the community are obtained, specifically including:

[0014] Choose four dimensions: coverage antenna physical pointing, signal quality, user perception, and network stability. Select at least one indicator for each dimension, including:

[0015] In the dimension of coverage antenna physical pointing, select the TA coverage distance indicator; in the dimension of signal quality, select the two indicators: RSRP>-112 sampling point percentage and CQI>7 sampling point percentage; in the dimension of user perception, select the traffic indicator; and in the dimension of network stability, select the number of handovers within the system indicator.

[0016] Select the pre-disaster baseline period, the post-disaster assessment period, and the target base station cell. Obtain the pre-disaster performance index data and post-disaster performance index data of each indicator of the target base station cell from the operator's network management system or MR data platform on a daily basis, and clean outliers from the data.

[0017] Furthermore, based on the pre-disaster and post-disaster performance index data of the community, the significant differences of each index before and after the disaster were obtained, specifically including:

[0018] A two-sample t-test was performed on the pre-disaster and post-disaster performance index data of each community for each indicator to obtain the p-value of the pre-disaster and post-disaster performance index data of each community. The p-value was compared with the preset p-value threshold to obtain the score representing the significant difference between the pre-disaster and post-disaster performance index data of each indicator.

[0019] Alternatively, perform a Bayesian factor test on the pre-disaster and post-disaster performance data of each indicator to obtain the BF value of the pre-disaster and post-disaster performance data of each indicator. Compare the BF value with a preset BF value threshold to obtain a score representing the significant difference between the pre-disaster and post-disaster performance of each indicator.

[0020] Furthermore, a two-sample t-test was performed on the pre-disaster and post-disaster performance data of each community for each indicator to obtain the p-values. These p-values ​​were then compared with a preset p-value threshold to obtain a score representing the significant difference between pre- and post-disaster performance for each indicator. Specifically, this included:

[0021] Obtain the average of the pre-disaster performance index data and the post-disaster performance index data for each community. and ,variance and and sample size and ;

[0022] Calculation based on variance Calculate the sample degrees of freedom based on the sample size. , Based on the sample degrees of freedom, consult the F-distribution table to determine whether the variances of the pre-disaster performance index data and the post-disaster performance index data of each community are unequal.

[0023] If the variances are homogeneous, use the traditional Student's t-test to calculate the pooled standard deviation. ,calculate Based on the sample degrees of freedom, consult the two-tailed t-distribution table to obtain the p-values ​​for the pre-disaster and post-disaster performance indicators of each community.

[0024] If the variances are unequal, use the corrected Welch t test to calculate... Calculate and adjust the degrees of freedom Based on the adjusted degrees of freedom, consult the two-sided t-distribution table to obtain the p-values ​​of the pre-disaster performance index data and post-disaster performance index data of each community.

[0025] If the p-value of the k-th indicator is less than 0.01, then the significance score of the k-th indicator before and after the disaster, S_test_k, is equal to 2.

[0026] If the p-value for the k-th indicator is 0.01 ≤ p < 0.05, then the significance score for the k-th indicator before and after the disaster, S_test_k = 1.

[0027] If the p-value of the k-th indicator is ≥0.05, then the significance score of the k-th indicator before and after the disaster is S_test_k=0.

[0028] Furthermore, Bayesian factor tests are performed on the pre-disaster and post-disaster performance data of each cell for each indicator to obtain the BF value. The BF value is then compared with a preset BF value threshold to obtain a score representing the significant difference between pre- and post-disaster performance for each indicator. Specifically, this includes:

[0029] Obtain the average of the pre-disaster performance index data and the post-disaster performance index data for each community. and ,variance and and sample size and ;

[0030] Calculate the common mean Calculate the pre-disaster performance index data of each community under the H0 hypothesis that "both sets of data come from the same normal distribution". and community post-disaster performance index data likelihood value ;

[0031] Calculate the difference between the means Calculate the likelihood values ​​of the pre-disaster and post-disaster performance indicators of each community under the H1 hypothesis that "there is a difference between the means of the two sets of data". , For a given difference The probability density of time data for The prior distribution;

[0032] Calculate Bayesian factors ;

[0033] If the kth indicator If the score is greater than 10, then the significance score of the k-th indicator before and after the disaster, S_test_k, is 2.

[0034] If the k-th indicator is 3 < If the score is ≤10, then the significance score of the k-th indicator before and after the disaster, S_test_k = 1.

[0035] If the kth indicator If the score is ≤3, then the score of the significant difference between the pre-disaster and post-disaster values ​​for the k-th indicator is S_test_k=0.

[0036] Furthermore, based on the pre-disaster and post-disaster performance index data of the community, the magnitude of change of each index before and after the disaster is obtained, specifically including:

[0037] Calculate the pre-disaster average value X_before_k of the pre-disaster performance index data of the k-th index in the community;

[0038] Calculate the post-disaster average value X_after_k of the post-disaster performance index data of the k-th index in the community;

[0039] If X_before_k ≠ 0, calculate the change rate of the k-th indicator before and after the disaster: ChangeRate_k = |(X_after_k - X_before_k) / X_before_k|.

[0040] If X_before_k=0, calculate the change rate of the k-th indicator before and after the disaster: ChangeRate_k= | (X_after_k - X_before_k) / preset value|.

[0041] Furthermore, based on the significance and magnitude of change, a disaster impact score is obtained for each indicator. Based on the disaster impact score for each indicator, a comprehensive assessment score for the damage to base station antennas caused by the disaster is obtained, specifically including:

[0042] Calculate the disaster impact score of the k-th indicator: S_final_k = S_test_k × ChangeRate_k;

[0043] The comprehensive assessment score S_total for base station antenna damage caused by the disaster is calculated as follows: S_total = ∑ k (W_k×S_final_k), where W_k is the comprehensive evaluation weight of the k-th indicator.

[0044] Furthermore, the method also includes:

[0045] Compare S_total with a preset S_total threshold to obtain the degree of cell antenna damage, including:

[0046] Severely damaged: S_total ≥ 6.0,

[0047] Significant damage: 4.0 ≤ S_total < 6.0

[0048] Slight or no effect: S_total < 4.0;

[0049] Based on the extent of antenna damage in the community, a priority list for antenna repair within the affected area is generated.

[0050] Secondly, this application provides a base station antenna damage assessment device, the device comprising:

[0051] The data module is used to obtain pre-disaster performance index data and post-disaster performance index data of the community based on multiple preset indicators.

[0052] The significance module, connected to the data module, is used to obtain the significance difference of each indicator before and after the disaster based on the pre-disaster performance index data and the post-disaster performance index data of the community.

[0053] The variation magnitude module, connected to the data module, is used to obtain the variation magnitude of each indicator before and after the disaster based on the pre-disaster performance index data and the post-disaster performance index data of the community.

[0054] The scoring module, connected to the significance module and the change range module, is used to obtain the disaster impact score for each indicator based on the significance difference and the change range, and to obtain the comprehensive assessment score of the damage to the base station antenna caused by the disaster based on the disaster impact score of each indicator.

[0055] Thirdly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the base station antenna damage assessment method described above.

[0056] This application provides a method, apparatus, and medium for assessing base station antenna damage. By calculating the significant differences and changes in multiple indicators before and after a disaster, the method assesses the impact of the disaster on the antenna by combining statistical significance and engineering changes from multiple indicators. The method comprehensively assesses the damage to the antenna by considering the impact of the disaster on multiple indicators, thereby improving the accuracy of quantitative assessment of base station antenna damage. Attached Figure Description

[0057] Figure 1 This is a flowchart of a base station antenna damage assessment method according to an embodiment of this application;

[0058] Figure 2 This is a schematic diagram of the structure of a base station antenna damage assessment device according to an embodiment of this application;

[0059] Figure 3 This is a flowchart of another base station antenna damage assessment method according to an embodiment of this application;

[0060] Figure 4 This is a schematic diagram of the structure of a computer-readable storage medium according to an embodiment of this application;

[0061] Figure 5 This is a schematic diagram of the structure of a computer device according to an embodiment of this application. Detailed Implementation

[0062] To enable those skilled in the art to better understand the technical solution of this application, the embodiments of this application will be further described in detail below with reference to the accompanying drawings.

[0063] It is understood that the specific embodiments and accompanying drawings described herein are merely for explaining this application and are not intended to limit this application.

[0064] It is understood that, without conflict, the various embodiments and features in the embodiments of this application can be combined with each other.

[0065] It is understood that, for ease of description, only the parts relevant to this application are shown in the accompanying drawings, while parts unrelated to this application are not shown in the drawings.

[0066] It is understood that each module or unit involved in the embodiments of this application may correspond to only one entity structure, or may be composed of multiple entity structures, or multiple modules or units may be integrated into one entity structure.

[0067] It is understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of this application may occur in a different order than that marked in the accompanying drawings.

[0068] It is understood that the flowcharts and block diagrams of this application illustrate the possible architecture, functions, and operations of systems, apparatuses, devices, and methods according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, unit, program segment, or code, containing executable instructions for implementing the specified function. Furthermore, each block or combination of blocks in the block diagrams and flowcharts may be implemented using a hardware-based device to implement the specified function, or using a combination of hardware and computer instructions.

[0069] It is understood that the modules and units involved in the embodiments of this application can be implemented by software or by hardware. For example, the modules and units can be located in the processor.

[0070] Example 1:

[0071] like Figure 1 As shown, this application provides a base station antenna damage assessment method, the method comprising:

[0072] S1. Based on multiple preset indicators, obtain pre-disaster performance index data and post-disaster performance index data of the community.

[0073] S2. Based on the pre-disaster performance index data and post-disaster performance index data of the community, obtain the significant differences of each index before and after the disaster.

[0074] S3. Based on the pre-disaster performance index data and post-disaster performance index data of the community, obtain the change range of each index before and after the disaster.

[0075] S4. Obtain the disaster impact score for each indicator based on the significance difference and the magnitude of change, and obtain the comprehensive assessment score for the damage to the base station antenna caused by the disaster based on the disaster impact score for each indicator.

[0076] In this embodiment, the provided method calculates the significant differences and magnitudes of change of multiple indicators before and after a disaster, assesses the impact of the disaster on the antenna by combining statistical significance and engineering change magnitude from multiple indicators, and comprehensively assesses antenna damage by integrating the impact of the disaster on multiple indicators, thereby improving the accuracy of quantitative assessment of base station antenna damage. Figure 1 The method shown is applicable to, for example, Figure 2 The apparatus shown.

[0077] Specifically, this embodiment provides a quantitative assessment method for typhoon-induced base station antenna damage based on a multi-index significance and rate of change product model.

[0078] Currently, the industry generally uses the following methods to assess the damage to base station antennas caused by weather disasters, but all of them have significant shortcomings:

[0079] 1. Manual road testing and passive troubleshooting driven by user complaints

[0080] This is the most traditional and widely used method, which operators mainly rely on after a typhoon:

[0081] Manual drive testing: Maintenance personnel carry professional equipment and conduct on-site tests based on experience or areas with frequent complaints. They measure signal indicators such as RSRP (Reference Signal Receiving Power) and SINR (Signal to Interference plus Noise Ratio) to determine whether the coverage is abnormal.

[0082] User complaint analysis: Collect user complaints about "no service" and "poor signal" through the customer service system to pinpoint the problem areas.

[0083] shortcoming:

[0084] Extremely inefficient and narrow coverage: Manual road testing is time-consuming and labor-intensive, and cannot achieve comprehensive and rapid coverage when the disaster area is wide and the number of stations is large.

[0085] Highly subjective and often delayed: Relying on staff experience and complaint feedback, problems are often only discovered after they have seriously affected the user experience, making it impossible to conduct proactive and forward-looking assessments.

[0086] 2. Simple threshold judgment based on a single key performance indicator (KPI).

[0087] Some operators or research institutions are attempting to use performance data from network management systems for automated preliminary screening. Common practices include:

[0088] Monitor single indicators such as cell traffic, average RSRP, or access success rate before and after a typhoon.

[0089] Set a fixed threshold for the decrease (e.g., a decrease in traffic exceeding 30%), and if the traffic exceeds this threshold, it is considered "potentially damaged".

[0090] shortcoming:

[0091] High false positive rate and poor anti-interference ability: Single indicators are easily affected by non-typhoon factors, such as holiday effects, large-scale events, daily business fluctuations, and interference from neighboring areas, resulting in a large number of "false positive" or "false negative" results.

[0092] Lack of quantification and standardization: It can only give a binary judgment of "yes / no", and cannot distinguish whether the antenna is slightly offset or severely damaged. The threshold is difficult to unify in different regions and scenarios, and it lacks universality.

[0093] 3. Anomaly detection based on time series prediction models

[0094] Some cutting-edge research attempts to incorporate machine learning methods, such as ARIMA (Autoregressive Integrated Moving Average) and LSTM (Long Short-Term Memory) time series models.

[0095] Using historical data to train a model, predict the normal values ​​of indicators for a certain period after a typhoon;

[0096] Anomalies are determined by comparing the residuals (errors) between the actual and predicted values.

[0097] shortcoming:

[0098] The models are complex and difficult to implement: They have high requirements for data quality and stability, and the model training and parameter tuning process is complex and computationally expensive, making it difficult to quickly deploy and scale up in the operator's existing production systems.

[0099] Poor interpretability and insufficient stability: The model is a "black box," and its judgment logic is not transparent to front-line operations and maintenance personnel, which is not conducive to trust and adoption. In scenarios with sparse data or sudden events, the prediction effect is unstable.

[0100] In summary, existing technologies are either inefficient, have a high error rate, or are too complex to be implemented effectively. Their core flaw lies in the lack of a standardized method that can effectively withstand the disruptions caused by daily business fluctuations and quantify the extent of the impact.

[0101] This embodiment addresses the problems existing in the above-mentioned prior art by innovatively proposing an evaluation system based on a model of "specific indicator combination + statistical significance test + weighted variation amplitude". It aims to provide a precise, efficient, and quantifiable solution, and a method for evaluating the impact of typhoons on base station antennas, including:

[0102] 1. A precise identification method for interference resistance is provided: by constructing a multi-dimensional index combination and statistical test model, interference from non-typhoon factors is effectively filtered out, and base station antennas physically offset by typhoons are accurately identified.

[0103] 2. Establish a quantifiable evaluation and grading standard: By combining statistical significance with the magnitude of indicator changes, the degree of antenna damage can be numerically scored and graded, providing an objective and unified scientific basis for the priority allocation of post-disaster maintenance resources.

[0104] 3. Establish a standardized closed-loop application process: Design a complete and replicable operation process from data input, model calculation, score output to on-site execution to ensure that the evaluation results can efficiently and directly guide front-line operation and maintenance work and improve the overall efficiency of network post-disaster recovery.

[0105] More specifically, this embodiment proposes a quantitative assessment method for the performance damage of base station antennas after a typhoon based on a two-sample t-test and multi-dimensional index variation amplitude. The core of this method lies in the assessment model of "specific index combination + statistical significance test + variation amplitude weighting" and its systematic application process. Its core technical features are as follows:

[0106] 1. Construct a "combination of typhoon impact sensitive indicators"

[0107] Breaking away from the traditional single RSRP metric model, this system combines five metrics—Timing Advance (TA) coverage distance, percentage of RSRP > -112 sampling points, traffic volume (GB), percentage of CQI (Channel Quality Indicator) > 7 sampling points, and number of intra-system handovers—into an evaluation framework. This framework covers four dimensions: antenna physical pointing, signal quality, user perception, and network stability, enabling comprehensive capture of typhoon impacts. The correspondence between dimensions and metrics is shown in Table 1 below.

[0108] Table 1: Correspondence between Dimensions and Indicators

[0109]

[0110] 2. Introduce a two-sample t-test as a significance judgment mechanism.

[0111] Unlike the crude methods of traditional "difference comparison" or "threshold judgment", this study uses data from the same period before and after the typhoon (for example, when the typhoon makes landfall on August 3, data from July 4-10 before the typhoon and August 4-10 after the typhoon) to conduct a two-sample t-test, and uses the p-value as the core basis for judging whether the change in the indicator is significant.

[0112] The P-value of the two-sample t-test is the probability of observing a difference (or a more extreme difference) in the means of the two groups of data, assuming that "there is no significant difference between the means of the two groups of samples" (i.e., the null hypothesis H0) is true. The judgment standard is the industry standard (usually α=0.05 as the threshold, and P<0.05 is considered to be significant).

[0113] This method can effectively eliminate interference from natural business fluctuations. By scientifically quantifying the boundary between "natural fluctuations" and "typhoon-induced anomalies," it can accurately distinguish between the two types of changes, fundamentally solving the problem of high misjudgment rate of typhoon impact by existing technologies.

[0114] 3. Construct a two-factor scoring model based on "significance + magnitude of change".

[0115] A quantitative evaluation model is constructed: First, the T-test score (generated by mapping from the P-value, for example, with scores increasing from 0 to 10 as the smaller the P-value, the stronger the significance) is multiplied by the magnitude of the indicator change (measured by the rate of change; for example, if a signal indicator decreases by 30% after a typhoon compared to before, then the rate of change is 30%) to obtain the final score for each core indicator (such as RSRP compliance rate, user traffic, number of switching times, etc.). This score avoids the bias of "only looking at significance and ignoring actual changes" and also eliminates the problem of "only looking at the magnitude of change and ignoring random fluctuations," achieving dual verification of "statistical significance + engineering significance" at the single indicator level.

[0116] Subsequently, an overall assessment is achieved through "cell-level total score aggregation": For a single communication cell, the final scores of all core indicators such as coverage, signal, traffic, and stability are calculated one by one. Then, the weights of each indicator's impact on the cell's communication function (such as the signal quality indicator having a higher weight than non-core auxiliary indicators) are combined and weighted, or the average of the final scores of all indicators is taken according to business needs. The scattered individual indicator scores are integrated into a "cell-level total score" that can comprehensively reflect the overall damage situation of the cell.

[0117] Based on the total score at the community level, the degree of damage can be accurately quantified and classified. For example: a total score > 6 points indicates severe damage, requiring the initiation of an emergency repair process; a total score 4 < 6 points indicates significant impact, requiring inclusion in the key optimization list; and a total score ≤ 4 points indicates minor or no damage, requiring only routine monitoring.

[0118] This model is the first to deeply integrate statistical significance (T-test score) with the magnitude of engineering changes (|rate of change|), forming a complete closed loop from single-index calculation to community-level total score aggregation. It completely solves the blind spots in traditional assessments where "the changes are large but not necessarily significant (may be random fluctuations)" or "the changes are only significant but small (no practical engineering value)", greatly improving the scientific nature and practicality of community typhoon damage assessment.

[0119] 4. Establish a standardized and reproducible evaluation process.

[0120] From data collection, parameter comparison, indicator calculation, model scoring to outputting a priority list, a closed-loop automated process is formed, requiring no manual intervention. It is suitable for rapid deployment in different regions and under different typhoon intensities, solving the shortcomings of existing technologies that "rely on expert experience and cannot be standardized and promoted".

[0121] The core of this embodiment is to provide a standardized operating procedure for accurately identifying and quantifying the impact of typhoons on the physical state of base station antennas by analyzing base station performance data after a typhoon disaster.

[0122] In one embodiment, S1, based on multiple preset indicators, pre-disaster performance index data and post-disaster performance index data of the cell are obtained, specifically including:

[0123] Choose four dimensions: coverage antenna physical pointing, signal quality, user perception, and network stability. Select at least one indicator for each dimension, including:

[0124] In the dimension of coverage antenna physical pointing, select the TA coverage distance indicator; in the dimension of signal quality, select the two indicators: RSRP>-112 sampling point percentage and CQI>7 sampling point percentage; in the dimension of user perception, select the traffic indicator; and in the dimension of network stability, select the number of handovers within the system indicator.

[0125] Select the pre-disaster baseline period, the post-disaster assessment period, and the target base station cell. Obtain the pre-disaster performance index data and post-disaster performance index data of each indicator of the target base station cell from the operator's network management system or MR data platform on a daily basis, and clean outliers from the data.

[0126] In this embodiment, as Figure 3 As shown, the method includes:

[0127] Step 1: Data Acquisition and Preprocessing

[0128] Objective: To obtain raw data for evaluation and ensure data quality.

[0129] Operation details:

[0130] 1. Define the time window:

[0131] a. Pre-disaster baseline period: A period of stable network operation is selected before the typhoon makes landfall, usually 7 calendar days before the typhoon. Data during this period will serve as the baseline for "normal conditions".

[0132] b. Post-disaster assessment period: This period selects the time after the typhoon has passed, when the network begins to recover but before large-scale human intervention has been carried out, typically from the first to the third calendar day after the typhoon makes landfall. Data during this period will reflect the actual impact of the typhoon.

[0133] 2. Collect performance data: Extract daily granular data of the following five key performance indicators for the target base station cell within the two time windows mentioned above from the operator's network management system or MR (Measurement Report) data platform:

[0134] TA coverage distance (unit: meters)

[0135] RSRP > -112 sampling point ratio (unit: %)

[0136] Total flow rate of the residential area (unit: GB)

[0137] Proportion of CQI>7 sampling points (unit: %)

[0138] Number of system handovers (unit: times)

[0139] 3. Data preprocessing: Clean the collected data and remove outliers caused by non-typhoon factors such as equipment failure or data cutover (e.g., data for a certain day is 0 or null) to ensure that the data used for analysis truly reflects the impact of the typhoon.

[0140] Inter-step relationships: This step is the foundation of the entire evaluation process, providing input for subsequent statistical analysis.

[0141] In one embodiment, S2, based on the pre-disaster performance index data and post-disaster performance index data of the community, the significant differences between the pre-disaster and post-disaster performance of each index are obtained, specifically including:

[0142] A two-sample t-test was performed on the pre-disaster and post-disaster performance index data of each community for each indicator to obtain the p-value of the pre-disaster and post-disaster performance index data of each community. The p-value was compared with the preset p-value threshold to obtain the score representing the significant difference between the pre-disaster and post-disaster performance index data of each indicator.

[0143] Alternatively, perform a Bayesian factor test on the pre-disaster and post-disaster performance data of each indicator to obtain the BF value of the pre-disaster and post-disaster performance data of each indicator. Compare the BF value with a preset BF value threshold to obtain a score representing the significant difference between the pre-disaster and post-disaster performance of each indicator.

[0144] In this embodiment, the method for performing the significance test of the indicator can be a two-sample t-test score or a Bayesian factor test.

[0145] The two-sample t-test is a statistical method used to compare whether there are significant differences in the means of two independent sample populations. The core is to determine whether the differences between the two sets of data are caused by random fluctuations or real differences, and it is one of the core tools for comparing two sets of data in data analysis. Its core logic is to first assume that "there is no difference in the means of the two sets of data" (the null hypothesis), calculate the means, standard deviations, and sample sizes of the two sets of data to obtain the t-statistic, and then determine the "probability of the null hypothesis being true (p-value)" based on the t-statistic: if the p-value is extremely small, it means that the possibility of "no difference in the two means" is extremely low, and it can be considered that the difference truly exists.

[0146] The two-sample t-test can be replaced by the Bayesian factor test: take the conventional non-informative prior for the prior distribution, calculate the Bayesian factor BF of the two sets of indicator data before and after the typhoon; if BF > 10, it is considered that there is a significant difference and 2 points are assigned; 3 < BF ≤ 10, 1 point is assigned; BF ≤ 3, 0 points are assigned. The subsequent steps remain unchanged. This method can also exclude natural fluctuations.

[0147] In one implementation, perform a two-sample t-test on the pre-disaster performance indicator data and post-disaster performance indicator data of each cell for each indicator, obtain the p-values of the pre-disaster performance indicator data and post-disaster performance indicator data of each cell for each indicator, and compare the p-values with a preset p-value threshold to obtain a significance difference score representing before and after each indicator, specifically including:

[0148] Obtain the means of the pre-disaster performance indicator data and post-disaster performance indicator data of each cell for each indicator and 、variance and and sample size and ;

[0149] Calculate according to the variance, calculate the sample degrees of freedom , , check the F-distribution table according to the sample degrees of freedom to determine whether the variances of the pre-disaster performance indicator data and post-disaster performance indicator data of each cell for each indicator are heterogeneous;

[0150] If the variances are homogeneous, use the traditional Student's t-test to calculate the pooled standard deviation , calculate , check the two-sided test t-distribution table according to the sample degrees of freedom to obtain the p-values of the pre-disaster performance indicator data and post-disaster performance indicator data of each cell for each indicator,

[0151] If the variances are heterogeneous, use the corrected Welch t-test to calculate , calculate the adjusted degrees of freedom Based on the adjusted degrees of freedom, consult the two-sided t-distribution table to obtain the p-values ​​of the pre-disaster performance index data and post-disaster performance index data of each community.

[0152] If the p-value of the k-th indicator is less than 0.01, then the significance score of the k-th indicator before and after the disaster, S_test_k, is equal to 2.

[0153] If the p-value for the k-th indicator is 0.01 ≤ p < 0.05, then the significance score for the k-th indicator before and after the disaster, S_test_k = 1.

[0154] If the p-value of the k-th indicator is ≥0.05, then the significance score of the k-th indicator before and after the disaster is S_test_k=0.

[0155] In this embodiment, as Figure 3 As shown, the method includes:

[0156] Step 2: Significance test of the indicator (calculation of T-test score)

[0157] Objective: To scientifically determine whether there are statistically significant differences in each indicator before and after a disaster, in order to mitigate the interference of daily business fluctuations.

[0158] Operation details:

[0159] 1. For the five core indicators of coverage, signal, traffic, and stability, the seven data points of the "pre-disaster baseline period" (e.g., one week before the typhoon landfall) and the seven data points of the "post-disaster assessment period" (e.g., one week after the typhoon landfall) are regarded as two independent sample groups, providing a standardized data basis for the subsequent two-sample t-test.

[0160] 2. Perform a two-sample t-test on the two sets of data (it is recommended to use the heteroscedasticity t-test, i.e., Welch's t-test, because it does not assume that the variances of the two sets of data are equal, which is more in line with the actual situation). Apply the two-sample t-test once for each of the five indicators, and then summarize the scores.

[0161] 3. Obtain the p-value of the T-test. The core of the t-distribution table is the correspondence between "degrees of freedom (df)" and "significance level (α)". In a two-tailed test, α represents the "sum of the probabilities of the rejection regions on both sides".

[0162] 4. Based on the preset p-value threshold, convert the p-value into a T-test score (S_T-test):

[0163] If the p-value is < 0.01, then S_T-test = 2 (indicating a highly significant difference).

[0164] If 0.01 ≤ p-value < 0.05, then S_T-test = 1 (indicating a significant difference).

[0165] If the p-value is ≥ 0.05, then S_T-test = 0 (indicating no significant difference).

[0166] Inter-step relationship: The output of this step (S_T-test) is one of the two inputs for calculating the final score of a single metric. It addresses the question of whether the change is real.

[0167] The following is a comparison of the results of a two-sample t-test using the CQI index of a certain community as an example. The results of the t-test algorithm are as follows:

[0168] Data period and sample size

[0169] Each set of data represents 15 days of CQI sampling (July 1st - 15th, August 1st - 15th, n1=n2=15), and the data is shown in Table 2 below:

[0170] Table 2: Example Table of Pre-Disaster and Post-Disaster Data

[0171]

[0172] Calculate the mean of the two sets of data

[0173] July average:

[0174]

[0175] August average:

[0176]

[0177] Calculate the variance of the two sets of data

[0178]

[0179] Test for homogeneity of variance (F-test)

[0180]

[0181] Degrees of freedom: The sample size for both groups is 15, therefore df1 = df2 = 15 - 1 = 14

[0182] Conclusion: Referring to the F-distribution table, F... 0.05 (14,14)=2.48. Since F=5.21>2.48, the variances are unequal and Welch T test (adjusting degrees of freedom) is required.

[0183] Calculate the Welch t-statistic and adjusted degrees of freedom

[0184]

[0185]

[0186] Determine the P-value and score

[0187] Table 3 below shows a commonly used t-distribution table (two-tailed test, excerpted with core degrees of freedom and critical values), including key data needed for subsequent P-value determination:

[0188] Table 3: Excerpt of the t-distribution table

[0189]

[0190] Based on the community's test data (t-statistic = 79.59, adjusted degrees of freedom df = 19), the logic for determining the range of p-values ​​through table lookup is as follows:

[0191] Define the core parameters:

[0192] The t-statistic for this cell is 79.59 with 19 degrees of freedom, corresponding to the row "df=19" in the table.

[0193] Compare the critical value with the t-statistic:

[0194] When α=0.001 (two-tailed test, i.e., P=0.001), the critical t-value corresponding to df=19 is 3.883;

[0195] The t-statistic (79.59) of this community is much greater than 3.883, indicating that the actual observed "difference between the two means" is more extreme than the "critical difference when α=0.001".

[0196] Derivation of the range of P values:

[0197] The larger the t-statistic, the smaller the p-value (the p-value is the probability of the current or more extreme difference occurring).

[0198] Since t=79.59 > t 0.001 (19) = 3.883, therefore the P-value < α = 0.001 (accurate calculation using statistical tools yields P ≈ 1.2 × 10⁻⁶). -25 (approaching 0).

[0199] Scoring: Statistically significant differences are correlated with scores. Significant differences are scored as 2 points, insignificant differences as 0 points, and the middle range as 1 point. The judgment criteria are shown in Table 4 below:

[0200] Table 4: Criteria for Judging p-values

[0201]

[0202] According to the judgment criteria, a p-value < 0.01 is considered "significant difference in indicators" and the score is 2.

[0203] Similarly, the p-values ​​of the other four indicators of the community can be obtained.

[0204] In one implementation, a Bayesian factor test is performed on the pre-disaster and post-disaster performance data of each indicator to obtain the BF value of the pre-disaster and post-disaster performance data of each indicator. The BF value is then compared with a preset BF value threshold to obtain a score representing the significant difference between the pre-disaster and post-disaster performance of each indicator. Specifically, this includes:

[0205] Obtain the average of the pre-disaster performance index data and the post-disaster performance index data for each community. and ,variance and and sample size and ;

[0206] Calculate the common mean Calculate the pre-disaster performance index data of each community under the H0 hypothesis that "both sets of data come from the same normal distribution". and community post-disaster performance index data likelihood value ;

[0207] Calculate the difference between the means Calculate the likelihood values ​​of the pre-disaster and post-disaster performance indicators of each community under the H1 hypothesis that "there is a difference between the means of the two sets of data". , For a given difference The probability density of time data for The prior distribution;

[0208] Calculate Bayesian factors ;

[0209] If the kth indicator If the score is greater than 10, then the significance score of the k-th indicator before and after the disaster, S_test_k, is 2.

[0210] If the k-th indicator is 3 < If the score is ≤10, then the significance score of the k-th indicator before and after the disaster, S_test_k = 1.

[0211] If the kth indicator If the score is ≤3, then the score of the significant difference between the pre-disaster and post-disaster values ​​for the k-th indicator is S_test_k=0.

[0212] In this embodiment, the Bayesian factor (denoted as BF) 10 The CQI strength (HQI) is a core indicator in Bayesian statistics used to quantify the "strength of evidence." It is used to compare the explanatory power of the alternative hypothesis H1 (there is a difference in CQI data before and after the typhoon) and the null hypothesis H0 (there is no difference in CQI data before and after the typhoon) on the observed data. The core formula is:

[0213]

[0214] Take the CQI sampling points of a certain community as an example, as shown in Table 5 below:

[0215] Table 5: Examples of sample size, mean, and variance for the pre-disaster and post-disaster data sets.

[0216]

[0217] Step 1: Calculate the likelihood value P(data|H0) under H0.

[0218] H0 assumes that "the two sets of data come from the same normal distribution," and the likelihood value is the joint probability density of the two sets of data with a common mean μ≈50484.87, which is expressed by the formula:

[0219]

[0220] To simplify the calculation, the natural logarithm of the likelihood value is taken (the logarithmic function is monotonically increasing and does not change the relative magnitude of the result):

[0221]

[0222]

[0223] Step 2: Calculate the likelihood value P(data|H1) under H1.

[0224]

[0225]

[0226] Step 3: Calculate the Bayesian factor BF 10

[0227]

[0228] Result: The neighborhood's BF 10 ≈10 3960 It is far greater than the threshold for determining "extremely strong evidence" (BF). 10 The value >100 indicates that the observed data strongly supports the finding that "there are differences in CQI data before and after the typhoon." This conclusion is consistent with the two-sample t-test (P<0.001), but the advantage of the Bayesian factor is that:

[0229] The T-test can only determine whether to reject H0, while the BF test... 10 The quantifiable "strength of evidence supporting H1" (10 3960 This means that "H1's explanatory power for the data is 10 times that of H0". 3960 times");

[0230] Engineering decisions that are more adapted to communication networks (such as when determining whether an emergency repair of a cell is needed, the size of the BF can be used to distinguish between "significant differences" and "extremely significant differences").

[0231] The aforementioned alternative achieves the core objective of "quantitatively distinguishing between natural fluctuations and anomalies caused by typhoons, outputting the level of damage and recovery priority," and can serve as a supplement to the main solution as an external technical approach.

[0232] However, this alternative approach has slightly higher computational complexity and requires prior estimation of a priori parameters. For the Bayesian factor test (an alternative to the two-sample t-test) of data before and after the typhoon, the most critical prior parameters are: the prior distribution type of the effect size (mean difference) (e.g., Cauchy distribution vs. normal distribution); the scaling parameter of the effect size distribution (e.g., r for Cauchy distribution, σ for normal distribution); and the prior distribution parameters of the standard deviations of the two sets of data (e.g., α and β for inverse gamma distribution, or the scaling parameter for semi-Cauchy distribution). Among these, the scaling parameter of the effect size is the most crucial to estimate in advance or set according to domain conventions (default values ​​can be used in uninformed scenarios, such as r=0.707 for Cauchy distribution), directly affecting the calculation results of the Bayesian factor.

[0233] In one embodiment, S3, based on the pre-disaster performance index data and post-disaster performance index data of the community, obtain the change range of each index before and after the disaster, specifically including:

[0234] Calculate the pre-disaster average value X_before_k of the pre-disaster performance index data of the k-th index in the community;

[0235] Calculate the post-disaster average value X_after_k of the post-disaster performance index data of the k-th index in the community;

[0236] If X_before_k ≠ 0, calculate the change rate of the k-th indicator before and after the disaster: ChangeRate_k = |(X_after_k - X_before_k) / X_before_k|.

[0237] If X_before_k=0, calculate the change rate of the k-th indicator before and after the disaster: ChangeRate_k= | (X_after_k - X_before_k) / preset value|.

[0238] In this embodiment, as Figure 3 As shown, the method includes:

[0239] Step 3: Calculation of the magnitude of indicator change

[0240] Objective: To quantify the severity of changes in each indicator before and after the disaster.

[0241] Operation details:

[0242] 1. Calculate the average value of each indicator in the "pre-disaster baseline period" (X_before).

[0243] 2. Calculate the average value (X_after) of each indicator during the "post-disaster assessment period".

[0244] 3. Calculate the absolute value of the change in this indicator (ChangeRate):

[0245] ChangeRate = | (X_after - X_before) / X_before |

[0246] Note: For "number of switching times within the system", if X_before is 0, you can use the absolute difference |X_after - X_before| or set a minimum value as the denominator.

[0247] Inter-step relationship: The output of this step (ChangeRate) is another input for calculating the final score of a single metric. It addresses the question of "how much change there is".

[0248] In one implementation, S4, the disaster impact score for each indicator is obtained based on the significance difference and the magnitude of change, and a comprehensive assessment score for the damage to the base station antenna caused by the disaster is obtained based on the disaster impact score for each indicator, specifically including:

[0249] Calculate the disaster impact score of the k-th indicator: S_final_k = S_test_k × ChangeRate_k;

[0250] The comprehensive assessment score S_total for base station antenna damage caused by the disaster is calculated as follows: S_total = ∑ k (W_k×S_final_k), where W_k is the comprehensive evaluation weight of the k-th indicator.

[0251] In this embodiment, as Figure 3 As shown, the method includes:

[0252] Step 4: Calculation of final score for individual indicators

[0253] Objective: To combine "significance" and "severity" to obtain a comprehensive evaluation score for each indicator.

[0254] Operation details:

[0255] Multiply the S_T-test obtained in step two by the ChangeRate obtained in step three to obtain the final score (S_final_i) of this indicator:

[0256] S_final_i = S_T-test × ChangeRate

[0257] For example, if the S_T-test = 2 and ChangeRate = 0.35 for the "TA coverage distance" indicator of a certain community, then S_final_i = 2 × 0.35 = 0.7.

[0258] Inter-step relationship: This step correlates the outputs of the first two steps to generate a quantified score with practical significance.

[0259] Step 5: Calculation and Classification of the Community's Total Score

[0260] Objective: To summarize the scores of all indicators and conduct a comprehensive assessment and classification of the overall damage level of the community.

[0261] Operation details:

[0262] 1. Sum the final scores of the five indicators (S_final_1 to S_final_5) to obtain the total evaluation score (S_total) for the base station cell:

[0263] S_total = S_final_1 + S_final_2 + S_final_3 + S_final_4 + S_final_5

[0264] 2. Classify the communities according to their total scores:

[0265] Severely damaged: S_total ≥ 6.0

[0266] Significant damage: 4.0 ≤ S_total < 6.0

[0267] Slight or no effect: S_total < 4.0

[0268] Inter-step relationships: This step elevates the evaluation results of individual indicators to the community level and outputs the final, actionable evaluation conclusions.

[0269] In one embodiment, the method further includes:

[0270] Compare S_total with a preset S_total threshold to obtain the degree of cell antenna damage, including:

[0271] Severely damaged: S_total ≥ 6.0,

[0272] Significant damage: 4.0 ≤ S_total < 6.0

[0273] Slight or no effect: S_total < 4.0;

[0274] Based on the extent of antenna damage in the community, a priority list for antenna repair within the affected area is generated.

[0275] In this embodiment, as Figure 3 As shown, the method includes:

[0276] Step Six: Application of Evaluation Results (Closed-Loop Optimization)

[0277] Objective: To translate the assessment results into practical actions and guide on-site operation and maintenance.

[0278] Operation details:

[0279] 1. Sort all cells to be evaluated from highest to lowest according to their S_total score.

[0280] 2. Generate a "Typhoon-Affected Antenna Repair Priority List", which includes: cell ID, total score, damage level, and suggested processing time.

[0281] 3. Distribute the list to the front-line operation and maintenance teams, and guide them to develop the optimal survey and adjustment routes according to the scores, prioritizing the handling of "severely damaged" communities.

[0282] 4. After on-site adjustments, data can be collected again to verify the effectiveness of the evaluation model, and the model parameters (such as p-value thresholds and grading standards) can be fine-tuned and optimized based on feedback.

[0283] Relationship between steps: This step is the ultimate manifestation of the value of this embodiment, connecting the data analysis results with the physical world operation to form a closed loop.

[0284] The process in this embodiment follows the order of the above six steps. First, the data is collected and cleaned. Then, the T-test score and change range of each indicator are calculated. Next, the two are multiplied to obtain the single indicator score. Then, the scores of the five single indicators are added together to obtain the total score of the cell and classified. Finally, the classification results guide the on-site operation and maintenance, forming a closed loop.

[0285] Specific implementation example demonstration:

[0286] Scenario: After a typhoon makes landfall in a certain region, an assessment is conducted on 5,807 base station cells along the coast.

[0287] Operation process:

[0288] 1. Data collection: Extract data on 5 indicators for 7 days before the typhoon (July 22-July 28) and 3 days after the typhoon (August 2-August 4).

[0289] 2. Model calculation: Perform steps two through five above for each indicator of each cell.

[0290] 3. Output Results: The system outputs a list containing the scores and grades of 282 communities (score ≥ 6 points).

[0291] The calculation process for community A is as follows:

[0292] TA coverage distance: S_T-test = 2, ChangeRate = 0.4 → S_final = 0.8

[0293] RSRP > -112 sampling point ratio: S_T-test = 2, ChangeRate = 0.5 → S_final = 1.0

[0294] Total cell traffic: S_T-test = 2, ChangeRate = 0.6 → S_final = 1.2

[0295] Proportion of CQI>7 sampling points: S_T-test = 1, ChangeRate = 0.3 → S_final = 0.3

[0296] System switch count: S_T-test = 0, ChangeRate = 0.2 → S_final = 0.0

[0297] S_total = 0.8 + 1.0 + 1.2 + 0.3 + 0.0 = 3.3 → Slight or no effect.

[0298] 4. Application: The list showed 72 cells with a score ≥8.0, classified as "severely damaged." The operations and maintenance team prioritized these 72 cells for antenna adjustments. On-site verification confirmed that 67 of these cells did indeed have significant antenna downtilt or azimuth offset issues. After adjustments, network metrics significantly recovered, with an accuracy rate reaching 93.06%.

[0299] This embodiment, targeting the specific scenario of "typhoon disaster," constructs a three-in-one evaluation system integrating "indicators, models, and processes." Its key technical points are as follows:

[0300] 1. Specific Multidimensional Indicator Combination for Typhoon Scenarios: This paper proposes a standardized combination of five performance indicators—"TA Coverage Distance," "Ratio of RSRP > -112 Sampling Points," "Total Cell Traffic (GB)," "Ratio of CQI > 7 Sampling Points," and "Number of Handovers Within the System"—specifically designed to assess the physical impact of typhoons on base station antennas. From five interrelated dimensions—"Physical Coverage Shrinkage," "Signal Degradation for Edge Users," "Traffic Loss," "Channel Quality Degradation," and "Network Handover Anomalies"—this combination collectively captures the characteristics of antenna offset caused by typhoons, enabling the implementation of indicator combinations and their collaborative diagnostic logic for specific scenarios.

[0301] 2. A Two-Dimensional Quantitative Assessment Model Based on "Statistical Significance × Change Magnitude": This model constructs a core assessment model by multiplying the p-value of a two-sample t-test (used to determine whether an indicator change is statistically significant to resist daily fluctuations) with the absolute value of the magnitude of the indicator change (used to quantify the severity of the change). This model goes beyond simple "difference comparison" or "threshold judgment," achieving two-dimensional quantification of the impact through the product of "significance score" and "magnitude." The assessment results not only identify problems but also scientifically differentiate the severity and urgency of problems, providing a precise basis for resource allocation.

[0302] 3. Closed-loop optimization process centered on evaluation scores: A complete, standardized application process driven by the "total cell score" was designed: from data collection -> index calculation -> model scoring -> score ranking -> on-site inspection -> optimization execution -> effect verification. This tightly integrates the abstract evaluation model with specific operational actions (such as "prioritizing cells with scores ≥ 6"), forming an executable, replicable, and measurable closed-loop system, solving the problem of disconnect between evaluation and execution in existing technologies.

[0303] Other alternative methods can be used to obtain significant differences, such as:

[0304] 1. End-to-end score prediction based on a machine learning classification model: The t-test and magnitude multiplication are eliminated, and a Lightweight Gradient Boosting Tree (LightGBM) is used instead. It directly uses the five-dimensional index ΔX / X_before as input features, outputting 0-1 probabilities, which are then mapped to integer scores of 0-6. The model is trained offline using historical typhoon labeled data, eliminating the need for p-value calculation during online inference. This approach can automatically learn non-linear weights, but requires sufficient pre-accumulation of labeled samples, and its interpretability is lower than that of statistical tests.

[0305] 2. A comprehensive deviation scheme based on principal component analysis (PCA) and Euclidean distance: After standardizing the five-dimensional indicators, PCA is used for dimensionality reduction, retaining the first two principal components. The centroid Euclidean distance D of the samples before and after the typhoon in the two-dimensional principal component space is calculated, and D is compared with the historical statistical 95th percentile threshold: if D > 95th percentile, 2 points are awarded; 80-95th percentile, 1 point is awarded; < 80th percentile, 0 points are awarded. Subsequent aggregation and classification steps are the same as the main scheme. This alternative does not require assumptions about the distribution, but it requires a large amount of data, and the physical meaning of the principal components is not as intuitive as the original indicators.

[0306] 3. Significance Identification Based on Time Series Anomaly Detection (SH-ESD): Each indicator is serialized continuously before and after the typhoon, and an outlier is detected using the Seasonal-Hybrid ESD algorithm. If an outlier falls within the typhoon's passage window and has a significance level of α=0.05, the corresponding indicator receives 2 points. The scoring rules for other indicators are consistent with the main scheme. This method can utilize longer historical data, but it has strict requirements for data continuity, and missing values ​​need to be imputed.

[0307] Other methods can be used to select indicators. For example, in emergency scenarios, only the TA coverage distance, the percentage of sampling points with RSRP > -112, and the three-dimensional flow rate indicators can be selected, with the remaining steps the same as the main scheme; the total score threshold is adjusted accordingly to S_total ≥ 3 and ≥ 5. This alternative sacrifices some detection accuracy in exchange for a reduction in computational load of approximately 40%, making it suitable for resource-constrained situations or situations requiring only coarse screening.

[0308] The implementation of the method in this embodiment may combine the following general techniques:

[0309] Data Acquisition: The five key indicators required, namely “TA coverage distance”, “RSRP”, “traffic”, “CQI”, and “number of handovers”, can be directly obtained through the operator’s existing network management system (such as Huawei U2000, ZTE NetNumen) or MR (Measurement Report) data platform, and the data source is stable and reliable.

[0310] Algorithm Implementation: The two-sample t-test is one of the most basic and mature hypothesis testing methods in statistics. Its calculation logic is clear and can be efficiently implemented in a few lines of code using common data analysis tools such as Python (SciPy library), R language or SPSS, with extremely low computational resource consumption.

[0311] Process execution: The entire evaluation process (data extraction -> T-test calculation -> score summary -> sorting output) has a simple logic and can be encapsulated into an independent Python script or SQL stored procedure. Frontline operations and maintenance personnel can operate it after simple training.

[0312] This embodiment of the solution can be designed to be lightweight and have standard interfaces, allowing for seamless integration into the operator's existing IT infrastructure.

[0313] Independent modular deployment: It can be deployed as an independent data analysis module in the operator's big data platform or network performance analysis system, and connect with upstream data sources and downstream work order systems through API or database interfaces;

[0314] Standardized output: The evaluation results are output in a standard format of "Community ID + Total Score + Priority", which can be directly imported into the operation and maintenance work order system to automatically generate survey and optimization tasks, thus realizing a closed loop of evaluation and execution.

[0315] This solution has been successfully validated in real-world scenarios, and the performance data strongly supports its value:

[0316] Pilot Case: Following a typhoon in a certain region, this solution was applied to evaluate 5807 coastal base station cells. The model successfully identified 281 problematic cells (score ≥ 6 points), of which 72 cells were high-priority problematic with scores ≥ 8 points. Field verification confirmed that 67 of the 72 high-priority problematic cells did indeed exhibit increased antenna downtilt angles; after adjustment, network performance significantly improved, achieving an accuracy rate of 93.06%.

[0317] Quantitative results: After antenna adjustments were made to the identified high-priority cells, the average RSRP of the relevant cells increased by 12 dB, the total cell traffic recovery rate reached 94.2%, and the number of user complaints about "poor signal" decreased by 83%.

[0318] Efficiency Improvement: Compared with traditional manual inspection, on-site inspection efficiency is increased by 95%, and the waste of operation and maintenance resources is reduced by 74.5%, which verifies the great value of this solution in improving post-disaster recovery efficiency.

[0319] The method in this embodiment establishes criteria for judging whether an antenna is affected, enabling rapid assessment of damaged antennas and the extent of their impact after a typhoon. It then develops a plan to restore the antennas according to the optimal route, allowing the affected antennas to be restored to pre-disaster levels as soon as possible.

[0320] Example 2:

[0321] like Figure 2 As shown, this application provides a base station antenna damage assessment device, the device comprising:

[0322] The data module is used to obtain pre-disaster performance index data and post-disaster performance index data of the community based on multiple preset indicators.

[0323] The significance module, connected to the data module, is used to obtain the significance difference of each indicator before and after the disaster based on the pre-disaster performance index data and the post-disaster performance index data of the community.

[0324] The variation magnitude module, connected to the data module, is used to obtain the variation magnitude of each indicator before and after the disaster based on the pre-disaster performance index data and the post-disaster performance index data of the community.

[0325] The scoring module, connected to the significance module and the change range module, is used to obtain the disaster impact score for each indicator based on the significance difference and the change range, and to obtain the comprehensive assessment score of the damage to the base station antenna caused by the disaster based on the disaster impact score of each indicator.

[0326] In one embodiment, the data module specifically includes:

[0327] The indicator selection unit is used to select from four dimensions: coverage antenna physical pointing, signal quality, user perception, and network stability. At least one indicator must be selected for each dimension, including:

[0328] In the dimension of coverage antenna physical pointing, select the TA coverage distance indicator; in the dimension of signal quality, select the two indicators: RSRP>-112 sampling point percentage and CQI>7 sampling point percentage; in the dimension of user perception, select the traffic indicator; and in the dimension of network stability, select the number of handovers within the system indicator.

[0329] The data acquisition unit, connected to the indicator selection unit, is used to select the pre-disaster baseline period, the post-disaster assessment period, and the target base station cell. It obtains the pre-disaster performance indicator data and post-disaster performance indicator data of each indicator of the target base station cell in the pre-disaster baseline period and the post-disaster assessment period from the operator's network management system or MR data platform on a daily basis, and cleans outliers from the data.

[0330] In one embodiment, the saliency module specifically includes:

[0331] The two-sample t-test unit is used to perform a two-sample t-test on the pre-disaster performance index data and post-disaster performance index data of each indicator, obtain the p-value of the pre-disaster performance index data and post-disaster performance index data of each indicator, and compare the p-value with the preset p-value threshold to obtain the score representing the significant difference between the pre-disaster and post-disaster performance of each indicator.

[0332] Alternatively, a Bayesian test unit can be used to perform Bayesian factor tests on the pre-disaster and post-disaster performance data of each indicator, obtain the BF value of the pre-disaster and post-disaster performance data of each indicator, and compare the BF value with a preset BF value threshold to obtain a score representing the significant difference between the pre-disaster and post-disaster performance of each indicator.

[0333] In one embodiment, the two-sample t-test unit specifically includes:

[0334] The mean-variance unit is used to obtain the mean of the pre-disaster performance index data and the post-disaster performance index data of each community. and ,variance and and sample size and ;

[0335] The F query unit, connected to the mean and variance unit, is used to calculate based on the variance. Calculate the sample degrees of freedom based on the sample size. , Based on the sample degrees of freedom, consult the F-distribution table to determine whether the variances of the pre-disaster performance index data and the post-disaster performance index data of each community are unequal.

[0336] The t-test cell, connected to the F-query cell, is used to calculate the pooled standard deviation using the traditional Student's t-test if the variances are homogeneous. ,calculate Based on the sample degrees of freedom, consult the two-tailed t-distribution table to obtain the p-values ​​for the pre-disaster and post-disaster performance indicators of each community.

[0337] Furthermore, if the variances are unequal, use the corrected Welch t test to calculate... Calculate and adjust the degrees of freedom Based on the adjusted degrees of freedom, consult the two-sided t-distribution table to obtain the p-values ​​of the pre-disaster performance index data and post-disaster performance index data of each community.

[0338] The significance score unit, connected to the t-test unit, is used to determine the significance score of the k-th indicator before and after the disaster if the p-value of the k-th indicator is < 0.01, in which case S_test_k = 2.

[0339] If the p-value for the k-th indicator is 0.01 ≤ p < 0.05, then the significance score for the k-th indicator before and after the disaster, S_test_k = 1.

[0340] Furthermore, if the p-value of the k-th indicator is ≥0.05, then the significance score of the k-th indicator before and after the disaster, S_test_k = 0.

[0341] In one embodiment, the Bayesian test unit specifically includes:

[0342] The mean-variance unit is used to obtain the mean of the pre-disaster performance index data and the post-disaster performance index data of each community. and ,variance and and sample size and ;

[0343] The H0 hypothesis cell, connected to the mean and variance cell, is used to calculate the common mean. Calculate the pre-disaster performance index data of each community under the H0 hypothesis that "both sets of data come from the same normal distribution". and community post-disaster performance index data likelihood value ;

[0344] The H1 hypothesis cell, connected to the mean-variance cell, is used to calculate the mean difference. Calculate the likelihood values ​​of the pre-disaster and post-disaster performance indicators of each community under the H1 hypothesis that "there is a difference between the means of the two sets of data". , For a given difference The probability density of time data for The prior distribution;

[0345] The BF calculation unit, connected to the H0 hypothesis unit and the H1 hypothesis unit, is used to calculate the Bayesian factor. ;

[0346] The significance score unit, connected to the BF calculation unit, is used to determine the significance score of the k-th indicator. If the score is greater than 10, then the score of the significant difference between the pre-disaster and post-disaster values ​​for the k-th indicator is S_test_k = 2.

[0347] And, if the k-th index 3 < If the score is ≤10, then the significance score of the k-th indicator before and after the disaster, S_test_k = 1.

[0348] And, if the k-th indicator If the score is ≤3, then the score of the significant difference between the k-th indicator before and after the disaster is S_test_k = 0.

[0349] In one embodiment, the variation range module specifically includes:

[0350] The pre-disaster average unit is used to calculate the pre-disaster average value X_before_k of the pre-disaster performance index data of the k-th index in the community.

[0351] The post-disaster average unit is used to calculate the post-disaster average value X_after_k of the post-disaster performance index data of the k-th index in the community.

[0352] The rate of change unit, connected to the pre-disaster average unit and the post-disaster average unit, is used to calculate the change rate of the k-th indicator before and after the disaster if X_before_k ≠ 0: ChangeRate_k = |(X_after_k - X_before_k) / X_before_k|.

[0353] If X_before_k=0, calculate the change rate of the k-th indicator before and after the disaster: ChangeRate_k= |(X_after_k - X_before_k) / preset value|.

[0354] In one embodiment, the scoring module specifically includes:

[0355] The disaster impact score unit is used to calculate the disaster impact score of the k-th indicator, S_final_k = S_test_k × ChangeRate_k;

[0356] The comprehensive evaluation scoring unit, connected to the disaster impact scoring unit, is used to calculate the comprehensive evaluation score S_total = ∑_{i=1}^{n- ... k (W_k×S_final_k), where W_k is the comprehensive evaluation weight of the k-th indicator.

[0357] In one embodiment, the apparatus further includes:

[0358] The damage grading unit, connected to the comprehensive evaluation score unit, is used to compare S_total with a preset S_total threshold to obtain the degree of cell antenna damage, including:

[0359] Severely damaged: S_total ≥ 6.0,

[0360] Significant damage: 4.0 ≤ S_total < 6.0

[0361] Slight or no effect: S_total < 4.0;

[0362] The repair list unit, connected to the damage classification unit, is used to output a priority list of antenna repairs within the disaster area based on the degree of damage to the cell antennas.

[0363] Example 3:

[0364] like Figure 4 As shown, Embodiment 3 of this application provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it implements the base station antenna damage assessment method as described in Embodiment 1.

[0365] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program units, or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory or other memory technologies, CD-ROM (Compact Disc Read-Only Memory), DVD or other optical disc storage, cartridges, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer.

[0366] like Figure 5 As shown, this application can also provide a computer device, including a memory and a processor. The memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the base station antenna damage assessment method as described in Embodiment 1. This computer device can be the base station antenna damage assessment device as described in Embodiment 2.

[0367] The memory is connected to the processor. The memory can be flash memory, read-only memory or other types of memory. The processor can be a central processing unit or a microcontroller.

[0368] Embodiments 1-3 of this application provide a method, apparatus, and medium for assessing base station antenna damage. By calculating the significant differences and changes in multiple indicators before and after a disaster, the impact of the disaster on the antenna is assessed by combining statistical significance and engineering change magnitude from multiple indicators. The damage to the antenna is assessed by comprehensively considering the impact of the disaster on multiple indicators, thereby improving the accuracy of quantitative assessment of base station antenna damage.

[0369] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of this application, and this application is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of this application, and these modifications and improvements are also considered to be within the scope of protection of this application.

Claims

1. A method of base station antenna impairment assessment, the method comprising: The method comprises: According to the preset multiple indicators, the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data are obtained; According to the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data, the significant difference of each indicator before and after the disaster is obtained; According to the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data, the change amplitude of each indicator before and after the disaster is obtained; According to the significant difference and the change amplitude, the disaster impact score of each indicator is obtained, and according to the disaster impact score of each indicator, the comprehensive evaluation score of the base station antenna damage caused by the disaster is obtained.

2. The method of claim 1, wherein, According to the preset multiple indicators, the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data are obtained, specifically comprising: Selecting four dimensions of coverage antenna physical pointing, signal quality, user perception and network stability, selecting at least one indicator in each dimension, including: Selecting TA coverage distance indicator in coverage antenna physical pointing dimension, selecting RSRP>-112 sampling point proportion and CQI>7 sampling point proportion two indicators in signal quality dimension, selecting traffic indicator in user perception dimension, and selecting system inner handover times indicator in network stability dimension; Selecting pre-disaster reference period, post-disaster evaluation period and target base station cell, obtaining the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator of the target base station cell in the pre-disaster reference period and the post-disaster evaluation period from the network management system or the MR data platform of the operator in day granularity, and cleaning the abnormal values of the data.

3. The method according to claim 1 or 2, characterized in that, According to the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data, the significant difference of each indicator before and after the disaster is obtained, specifically comprising: Performing double-sample T test on the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator to obtain the p value of the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator, comparing the p value with the preset p value threshold to obtain the significant difference score representing each indicator before and after the disaster; Or, performing Bayesian factor test on the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator to obtain the BF value of the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator, comparing the BF value with the preset BF value threshold to obtain the significant difference score representing each indicator before and after the disaster.

4. The method of claim 3, wherein, Performing double-sample T test on the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator to obtain the p value of the cell pre-disaster performance indicator data and the cell post-disaster performance indicator data of each indicator, comparing the p value with the preset p value threshold to obtain the significant difference score representing each indicator before and after the disaster, specifically comprising: Obtaining the mean of the cell pre-disaster performance index data and the cell post-disaster performance index data of each index and variance and and sample size and ; According to the variance calculation , according to the sample size calculation sample degrees of freedom , , according to the sample degrees of freedom F distribution table, judge whether the variance of each index of cell pre-disaster performance index data and cell post-disaster performance index data is not uniform; If the variances are equal, use the traditional Student's t-test to calculate the combined standard deviation , calculate , according to the sample degrees of freedom, look up the two-sided test t-distribution table to obtain the p-value of the pre-disaster performance index data and the post-disaster performance index data of each index, If the variance is not equal, use the corrected Welch t test to calculate , calculate the adjusted degrees of freedom , according to the adjusted degrees of freedom, look up the two-sided test t distribution table to obtain the p value of the cell pre-disaster performance index data and the cell post-disaster performance index data of each index; If the p value of the kth indicator is less than 0.01, the significant difference score S_test_k of the kth indicator before and after the disaster is 2, If the p value of the kth indicator is 0.01 to less than 0.05, the significant difference score S_test_k of the kth indicator before and after the disaster is 1, If the p value of the kth indicator is greater than or equal to 0.05, the significant difference score S_test_k of the kth indicator before and after the disaster is 0.

5. The method of claim 3, wherein, performing Bayesian factor test on the cell pre-disaster performance index data and the cell post-disaster performance index data of each index, obtaining the BF value of the cell pre-disaster performance index data and the cell post-disaster performance index data of each index, and comparing the BF value with a preset BF value threshold to obtain a significant difference score representing the pre-disaster and post-disaster of each index, specifically including: obtaining the mean of the cell pre-disaster performance index data and the cell post-disaster performance index data of each index and variance and and sample size and ; Compute common mean , Compute the likelihood value of the pre-disaster performance indicator data of each index under the H0 hypothesis "the two groups of data come from the same normal distribution" and the post-disaster performance indicator data of each index of each index ; calculate the mean difference , calculate the likelihood value of the cell pre-disaster performance index data and the cell post-disaster performance index data of each index under the H1 hypothesis "there is a difference between the means of the two groups of data" , the probability density of the data when the given difference , is the prior distribution of ​ Computing bayesian factors ; If the kth indicator is > 10, then the kth indicator's significance difference score S_test_k = 2, If 3 < kth index < 10, then the significant difference score S_test_k = 1 for the kth index before and after the disaster. If the kth indicator's ≤ 3, then the kth indicator's pre-disaster and post-disaster significance difference score S_test_k = 0.

6. The method according to claim 4 or 5, characterized in that, According to the cell pre-disaster performance index data and the cell post-disaster performance index data, the change range of each index before and after the disaster is obtained, specifically including: calculating the pre-disaster average value X_before_k of the cell pre-disaster performance index data of the kth index; calculating the post-disaster average value X_after_k of the cell post-disaster performance index data of the kth index; if X_before_k≠0, calculating the change range ChangeRate_k of the kth index before and after the disaster = | (X_after_k - X_before_k) / X_before_k |, if X_before_k=0, calculating the change range ChangeRate_k of the kth index before and after the disaster = | (X_after_k - X_before_k) / preset value |.

7. The method of claim 6, wherein, According to the significant difference and the change range, the disaster impact score of each index is obtained, and according to the disaster impact score of each index, the comprehensive evaluation score of the base station antenna damage caused by the disaster is obtained, specifically including: calculating the disaster impact score S_final_k of the kth index = S_test_k × ChangeRate_k. The comprehensive evaluation score S_total of the disaster caused damage to the base station antenna is calculated as S_total =∑ k (W_k x S_final_k), where W_k is the comprehensive evaluation weight of the kth index.

8. The method of claim 7, wherein, The method further includes: comparing S_total with a preset S_total threshold to obtain the cell antenna damage degree, including: severe damage: S_total ≥ 6.0, greater damage: 4.0 ≤ S_total < 6.0, slight or no impact: S_total < 4.0; According to the cell antenna damage degree, an antenna repair priority list in the disaster range is output.

9. A base station antenna damage assessment apparatus, comprising: The device includes: a data module for obtaining cell pre-disaster performance index data and cell post-disaster performance index data according to a plurality of preset indexes; a significance module connected with the data module, for obtaining the significant difference of each index before and after the disaster according to the cell pre-disaster performance index data and the cell post-disaster performance index data; a change range module connected with the data module, for obtaining the change range of each index before and after the disaster according to the cell pre-disaster performance index data and the cell post-disaster performance index data; a scoring module connected with the significance module and the change range module, for obtaining the disaster impact score of each index according to the significant difference and the change range, and obtaining the comprehensive evaluation score of the base station antenna damage caused by the disaster according to the disaster impact score of each index.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, when the computer program is run by the processor, the base station antenna damage evaluation method of any one of claims 1-8 is realized. The computer readable storage medium stores a computer program, when the computer program is run by the processor, the base station antenna damage evaluation method of any one of claims 1-8 is realized.

Images