Intelligent operation and maintenance 5G power management system and management method

Abstract

This invention relates to the field of power operation and maintenance management technology, and discloses an intelligent 5G power management system and management method. The method includes: acquiring and normalizing four-dimensional key performance indicators of multiple base stations within a target area to construct base station health credit values; generating a regionalized hexagonal grid risk heat map based on the geographical coordinates and health credit values ​​of the base stations; when a base station fails and causes coverage holes, selecting candidate compensation base stations in the neighborhood and allocating compensation power according to credit priority; implementing an active rotation strategy for base stations that have completed compensation tasks; identifying potentially risky base stations and identifying maintenance intentions, and pushing maintenance tasks in conjunction with the risk heat map; this invention can effectively suppress the phenomenon of "fault contagion", extend the life of the power system, and reduce operation and maintenance costs.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power operation and maintenance management, more specifically, the present application relates to an intelligent operation and maintenance 5G power management system and management method. BACKGROUND

[0002] With the large-scale deployment of 5G networks, the operation and maintenance management of base station power systems is facing unprecedented challenges. In the prior art, the 5G base station operation and maintenance management system and method disclosed in the Chinese patent application with publication number CN114418714A focuses on the compliance control of operation and maintenance data, intelligently verifies business data such as electricity billing through a multi-dimensional audit rule model, and constructs a decision support framework based on standardized processes, optimizing the credibility and processing efficiency of operation and maintenance data. The Chinese patent with authorized publication number CN119205070B discloses a method and system for operation and maintenance management of 5G communication base station equipment state, aiming at equipment state management, constructing a degradation model and operation and maintenance decision model based on historical data, realizing intelligent scheduling of equipment maintenance resources through fault point identification and maintenance plan optimization, and improving the matching accuracy of equipment state and maintenance demand.

[0003] However, the prior art lacks quantitative evaluation and fine management mechanism of base station health status. In actual operation and maintenance scenarios, when a base station is offline due to failure, the cloud platform can automatically adjust the power of surrounding base stations for compensation, but does not reasonably allocate according to the health status of the base station, which may accelerate the aging of base stations with poor health status due to additional load, and thus trigger new failures. After each failure occurs, the allocation of compensation tasks is only based on the geographical location and load capacity of the base station, without fully considering the health credit status of the base station and the regional risk distribution, ultimately forming a "failure transmission" phenomenon, which accelerates the overall deterioration of the base station power system in the region, and the maintenance cost grows exponentially. At the same time, the existing technology lacks precision in identifying potential risk base stations and pushing maintenance tasks, and cannot dynamically adjust and intelligently schedule according to real-time risk heat maps and engineer maintenance intentions. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the present application provides an intelligent operation and maintenance 5G power management system and management method, which realizes intelligent operation and maintenance management of base station power systems by constructing base station health credit value, generating regionalized hexagonal grid risk heat map, credit priority compensation allocation, active rest strategy and precise maintenance task pushing. The present application can effectively suppress the "failure transmission" phenomenon, prolong the service life of the power system, reduce the operation and maintenance cost, improve the operation and maintenance efficiency and network reliability, and provide strong guarantee for the stable operation of 5G base stations.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] Intelligent operation and maintenance methods for 5G power management include:

[0007] Obtain and normalize the four-dimensional key performance indicators of multiple base stations within the target area, and construct a quantified base station health credit value for all base stations;

[0008] Based on the geographic coordinates and health credit values ​​of all base stations, a regionalized hexagonal grid risk heat map is generated.

[0009] When a base station is detected to have a coverage hole due to a fault, candidate compensation base stations are selected in the neighborhood of the faulty base station. Based on the base station health credit value of the candidate compensation base stations, credit-based compensation allocation is performed.

[0010] For candidate compensation base stations that have completed the power compensation task, an active rotation strategy is implemented;

[0011] Identify potential risky base stations within the target area, identify maintenance intentions, and when a maintenance intention is identified, push maintenance tasks based on a regionalized hexagonal grid risk heat map.

[0012] Furthermore, the method for constructing the base station health credit value quantified for all base stations includes: determining the state of the key performance indicator sequence and calculating the integral increase or decrease; based on the calculated integral increase or decrease, accumulating it within a rolling time window ΔT to generate the final base station health credit value.

[0013] Furthermore, the method for determining the state of the key performance indicator sequence and calculating the integral increase or decrease includes:

[0014] Set the steady-state threshold α for the k-th dimension key performance indicator. k With stress threshold β k Where k is the index variable for key performance indicators, k = 1, 2, 3, 4; β k >α k ;

[0015] If the sequence value of the k-th dimension key performance indicator is less than or equal to the corresponding steady-state threshold α k If the k-th key performance indicator is in a steady state, then the contribution of the k-th key performance indicator to health credit is positively accumulated by +ω; ω is the basic unit of integration.

[0016] If the sequence value of the k-th key performance indicator is greater than or equal to the corresponding stress threshold β k If the key performance indicator of dimension k is in a state of stress, the contribution of the key performance indicator of dimension k to health credit is reduced by -ω / 2.

[0017] If the sequence value of the k-th key performance indicator is greater than α k And less than β kIf the condition is not met, it is considered a transitional state, the integral is 0, and credit neutrality is maintained.

[0018] Furthermore, the method for generating a regionalized hexagonal grid risk heatmap includes:

[0019] Construct a grid system consisting of multiple regular hexagonal grid cells that divides the entire target area;

[0020] The base station health credit value of each base station is projected onto the regular hexagonal grid cell in which it is located, and the average health of the grid is calculated.

[0021] Calculate the grid base station density, combine the grid average health and the grid base station density to construct a risk intensity function, and obtain the risk intensity value of the grid cell;

[0022] Based on the risk intensity value of the grid cells, a regionalized hexagonal grid risk heat map is generated.

[0023] Furthermore, in the risk intensity function, the risk intensity value is the dependent variable, and the grid average health and grid base station density are the independent variables. The risk intensity value is negatively correlated with the grid average health and positively correlated with the grid base station density.

[0024] Furthermore, the method for selecting candidate compensation base stations within the neighborhood of a faulty base station includes: defining all base stations within the neighborhood as candidate base stations; traversing all candidate base stations; and if the base station health credit value of the i'th candidate base station is higher than a safety threshold and the risk intensity value of the grid cell where the candidate base station is located is lower than a high-risk intensity threshold Ψ. red If i' is added as a candidate compensation base station to the candidate compensation base station set, then i' is the index variable of the candidate base station.

[0025] Furthermore, the proactive rotation strategy is as follows: candidate compensation base stations participating in the compensation task are defined as participating base stations; after the compensation task is completed, rotation status tags are written for all participating base stations and they are defined as rotation base stations, and the rotation base stations are removed from the candidate compensation base station set; during the rotation period, the load of the rotation base stations is proactively reduced.

[0026] Furthermore, the method for identifying potential risky base stations within the target area is as follows:

[0027] Obtain the historical sequence of base station health credit values ​​for each base station within the target area, and use a pre-built base station health credit prediction model to obtain the future preset time ΔT. p The sequence of base station health credit prediction values ​​for each base station within the range, if the base station health credit prediction value in the future ΔT p If the base station is below the preset maintenance threshold for a period of time, it will be marked as a potentially risky base station.

[0028] Furthermore, the method for performing maintenance intent recognition is as follows:

[0029] Identify the target base station from the maintenance team list, set the grid cell containing the target base station as the target grid, and set the geometric center of the target grid as the target point;

[0030] Within a preset time window, the location coordinates of engineers sorted by time are periodically collected;

[0031] For every two consecutive position coordinate points within the time window, a motion vector representing the engineer's actual movement direction is calculated; and for each motion vector, with its starting position coordinate point as the starting point and the target point as the ending point, the target vector corresponding to the motion vector is calculated.

[0032] Calculate the angle between each motion vector and its corresponding target vector; count the number of instances where the angle is less than 90 degrees within the time window, and record them as the effective number of moves; divide the effective number of moves by the total number of motion vectors calculated within the time window to obtain the convergence ratio.

[0033] When the proximity ratio is greater than a preset intent threshold, it is determined that the engineer has the intent to maintain the target base station.

[0034] Furthermore, the method for pushing maintenance tasks when a maintenance intention is identified, in conjunction with a regionalized hexagonal grid risk heatmap, includes:

[0035] All identified potential risk base stations are prioritized according to the risk intensity value of their respective grid cells to form a maintenance team list;

[0036] If a maintenance intent is detected, a task information package is pushed out in conjunction with the regionalized hexagonal grid risk heat map. The task information package includes maintenance task information and emergency risk task information for the target base station. If no maintenance intent is detected, no task information package is pushed out to the engineer.

[0037] Furthermore, the method for generating a regionalized hexagonal grid risk heatmap based on the risk intensity value of the grid cells includes:

[0038] When Ψ j <Ψ yellow When Ψ is reached, the j-th grid cell is marked as green. yellow ≤Ψ j <Ψ red When Ψ is reached, the j-th grid cell is marked in yellow. j ≥Ψ red When the j-th grid cell is marked in red, Ψ j Let Ψ be the risk intensity value of the j-th grid cell.yellow For low-risk intensity threshold, Ψ red This is the high-risk intensity threshold.

[0039] Furthermore, the sudden risk task information refers to the situation where, in the regionalized hexagonal grid risk heat map, when a non-target grid changes from green to red or from yellow to red within a preset time Δt from the current moment, if the distance between the engineer's real-time location and the geometric center of the non-target grid is less than a preset distance threshold, then base station maintenance information within the non-target grid is pushed to the engineer.

[0040] Furthermore, while writing the rest period status tag to all participating base stations, a countdown timer T for the remaining rest time is started. r Only when countdown timer T r The rest status label will automatically expire when the value is reset to zero.

[0041] Furthermore, the method for constructing a grid system consisting of multiple regular hexagonal grid cells that divides the entire target area includes: obtaining the GPS coordinates of each base station to generate a set of GPS coordinate points of all base stations within the target area, determining the optimal grid side length L of the regular hexagonal grid cells based on the set of GPS coordinate points of all base stations, and constructing a grid system consisting of multiple regular hexagonal grid cells.

[0042] Furthermore, the method for constructing a grid system consisting of multiple regular hexagonal grid cells that divides the entire target area also includes: calculating the geometric centroid of the target area using the Wei algorithm, and iteratively solving to minimize the sum of the weighted distances from each base station coordinate to the geometric centroid. The initial point of the iteration is selected as the center of the region boundary box, and the iteration terminates when the coordinate deviation between two iterations is less than a preset accuracy threshold. The region boundary box refers to the smallest rectangular boundary that can completely enclose the set of GPS coordinate points of all base stations within the target area.

[0043] A smart operation and maintenance 5G power management system, used to implement the above-mentioned smart operation and maintenance 5G power management method, the system comprising:

[0044] Health credit construction module: used to acquire and normalize the four-dimensional key performance indicators of multiple base stations in the target area, and construct the quantified health credit value of all base stations;

[0045] Heatmap generation module: Generates a regionalized hexagonal grid risk heatmap based on the geographic coordinates and health credit values ​​of all base stations;

[0046] Compensation module: When a coverage hole is detected due to a base station going offline due to a fault, candidate compensation base stations are selected in the neighborhood of the faulty base station. Based on the base station health credit value of the candidate compensation base stations, credit-priority compensation allocation is performed.

[0047] Rotation module: Used to implement an active rotation strategy for candidate compensation base stations that have completed power compensation tasks;

[0048] Maintenance task push module: used to identify potential risky base stations within the target area, identify maintenance intentions, and push maintenance tasks when a maintenance intention is identified, combined with a regionalized hexagonal grid risk heat map.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] This invention achieves precise assessment of the health status of base station power systems by constructing a quantified base station health credit value, providing a reliable quantitative basis for subsequent coordinated scheduling. The generation of a regionalized hexagonal grid risk heat map combines the dispersed health status of base stations with geographical distribution characteristics, enabling visualization and precise location of regional risks, providing spatial-dimensional decision support for risk prevention and power optimization. The priority compensation allocation mechanism based on health credit values ​​during faults ensures that base stations with better health status take priority in compensation tasks, reducing the risk of faults caused by excessive load on base stations with poor health. Simultaneously, it avoids excessive energy consumption and reduces energy waste caused by inefficient operation of base stations with poor health. The proactive rotation strategy directly reduces energy consumption during rotation periods by reducing the load and restoring the status of base stations after compensation, avoiding energy surges and accelerated equipment aging caused by accumulated compensation tasks. The potential risk identification and maintenance task push mechanism enables early intervention and targeted maintenance of risks, effectively preventing base stations from entering inefficient operation due to health deterioration and reducing ineffective energy consumption at the source. By combining maintenance intent recognition with risk heatmaps to push maintenance tasks, precise maintenance of potentially risky base stations is achieved, avoiding prolonged inefficient energy consumption caused by fault delays. The overall solution effectively breaks the chain reaction of faults, significantly suppresses "fault contagion," extends the overall lifespan of the base station power system in the area, and reduces operation and maintenance costs. Through refined power control strategies, such as dynamic power allocation, load reduction, and early maintenance of inefficient equipment, energy utilization efficiency is improved, achieving energy-saving effects, further reducing energy costs in operation, and improving operation and maintenance efficiency and network reliability. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart of the intelligent operation and maintenance 5G power management method of the present invention;

[0053] Figure 2 This is a flowchart of the method for generating a regionalized hexagonal grid risk heat map according to the present invention;

[0054] Figure 3 This is a schematic diagram illustrating the principle of determining whether an engineer has the intention to maintain the target base station according to the present invention.

[0055] Figure 4 This is a functional block diagram of the intelligent operation and maintenance 5G power management system in this invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Example 1

[0058] Please see Figure 1 As shown, this embodiment provides an intelligent operation and maintenance 5G power management method, including:

[0059] Step S10: Obtain and normalize the four-dimensional key performance indicators of multiple base stations within the target area, and construct the quantified base station health credit value of all base stations.

[0060] Further, step S10 includes:

[0061] Step S11: Obtain and normalize the four-dimensional key performance indicators of multiple base stations within the target area to form a standardized key performance indicator sequence.

[0062] Specifically, through a management system deployed on a cloud platform, four core parameters of each base station's power system are continuously collected via an IoT gateway interface: State of Health (SOH), reflecting the ratio of the battery's current capacity to its nominal capacity (e.g., a new battery has an SOH of 1.0, which may drop to 0.7 after aging); Mean Time Between Failures (MTBF), the average duration of continuous normal operation of the base station over the past 12 months, expressed in hours; Load Utilization, the percentage of current load power to rated power (e.g., a base station with a rated power of 500W and a current load of 300W has a utilization rate of 60%); and Power Module Temperature Stress Range, the difference between the highest and lowest temperatures over a certain period, expressed in degrees Celsius, reflecting the damage caused by temperature fluctuations to the module. Because the sensor ranges of different manufacturers' equipment vary, the raw data needs to be processed using a minimum-maximum normalization method. After this processing, all four dimensions of data are mapped to the [0,1] interval, forming a standardized sequence that can be directly compared. Step S11 solves the problem of incomparability of traditional cross-device data, providing a unified benchmark for the performance indicators of base stations from different manufacturers, and providing consistent input for subsequent health assessments.

[0063] Step S12: Determine the status of the key performance indicator sequence and calculate the integral increase or decrease.

[0064] Further, step S12 includes:

[0065] Step S121: Set the steady-state threshold α for the k-th dimension key performance indicator. k With stress threshold β k Where k is the index variable for key performance indicators, k = 1, 2, 3, 4; β k >α k ;

[0066] Step S122: If the sequence value of the k-th key performance indicator is less than or equal to the corresponding steady-state threshold α k If the k-th key performance indicator is in a steady state, then the contribution of the k-th key performance indicator to health credit is positively accumulated by +ω; ω is the basic unit of integration.

[0067] Step S123: If the sequence value of the k-th key performance indicator is greater than or equal to the corresponding stress threshold β k If the key performance indicator of dimension k is in a state of stress, the contribution of the key performance indicator of dimension k to health credit is reduced by -ω / 2.

[0068] Step S124, if the sequence value of the k-th key performance indicator is greater than α k And less than β k If the condition is not met, it is considered a transitional state, the integral is 0, and credit neutrality is maintained.

[0069] Specifically, the steady-state threshold αk and stress threshold β k The determination of the steady-state threshold α needs to be based on historical operating data of the base station power system within the target area and technical manuals provided by equipment manufacturers, combined with correlation analysis of key performance indicators and fault occurrences in long-term operation and maintenance records. Specifically, it is necessary to collect key performance indicator data and corresponding fault records for at least three complete operation and maintenance cycles of all similar base stations within the target area, and determine the steady-state threshold α by statistically analyzing the fluctuation range of key performance indicators under normal operating conditions. k That is, the typical upper limit value of the key performance indicator during normal equipment operation is selected as α. k Stress threshold β k Based on the probability distribution of failures occurring in the short term after abnormal fluctuations in key performance indicators, the key performance indicator value corresponding to a significant increase in the failure probability is selected as β. k , ensure β k Within the abnormal fluctuation range of key performance indicators and greater than α k If the sequence value of the k-th key performance indicator is less than or equal to the corresponding steady-state threshold α... k If the key performance indicator of that dimension is in a steady state, then the contribution of that dimension to health credit is positively accumulated to +ω, where ω is the basic unit of integration. It is determined by the following steps: collect the four-dimensional indicators and corresponding fault records of base stations in the target area for the past three operation and maintenance cycles, use multiple linear regression to fit the correlation between the increase or decrease of the integral and the fault probability, and use "make the probability of no fault in steady state ≥95%" as a constraint to solve for the optimal value of ω, which is usually in the range of 0.05-0.2, for example, 0.1.

[0070] The positive accumulation design here is based on the positive impact of steady-state operation on the health of the base station power system. When the indicator is in a steady state, it indicates that the equipment is operating stably and there is no significant deterioration trend. By positively accumulating ω, the health credit value can be increased cumulatively with the stable operating time, objectively reflecting the long-term stable operation health status of the equipment. If the sequence value of the k-th dimension key performance indicator is greater than or equal to the corresponding stress threshold β... k If the value of the k-th dimension is in a stress state, its contribution to health credit is reduced by -ω / 2, which, using the example above, is -0.05. This asymmetric, mild penalty mechanism is used because the critical performance indicator reaching the stress threshold may be caused by a brief network disturbance, not by substantial degradation of the equipment itself. Setting the deduction to half of the positive accumulation amount helps identify potential anomalies while preventing drastic fluctuations in the health credit value due to momentary volatility, ensuring the stability of the credit assessment. If the sequence value of the k-th critical performance indicator is greater than α... k And less than β kIf the condition is abnormal, it is considered a transitional state, and the score is 0 to maintain credit neutrality. The transitional state is designed to distinguish between normal operating fluctuations and abnormal symptoms. The fluctuations in the indicators within this range are within the normal operating range of the equipment and have not yet reached the stress state that may cause a failure. Therefore, no increase or decrease in the score is made, so that the health credit value remains stable within this range, further improving the robustness of credit assessment and avoiding misjudgment due to oversensitivity to normal fluctuations.

[0071] The positive accumulation mechanism in step S12 during steady state enables the health credit value to reflect the cumulative effect of long-term stable operation of the equipment, providing a quantitative basis for assessing the equipment's health trend. The asymmetric deduction mechanism under stress effectively identifies potential risks while filtering out the impact of instantaneous disturbances, reducing unnecessary fluctuations in the credit value, and is more consistent with the actual characteristics of equipment operation than the symmetric penalty mechanism. The neutral processing in the transition state keeps the credit assessment stable within the normal fluctuation range, avoiding overreaction to minor fluctuations and improving the reliability of the assessment results. These designs collectively achieve accurate determination of the status of key performance indicators, laying the foundation for the subsequent cumulative calculation of the health credit value. Without this step, the health credit value cannot accurately reflect the actual health status of the equipment, potentially leading to misjudgments of the equipment's health trend, which in turn affects the subsequent selection of compensation base stations and power allocation, weakening the overall solution's suppression effect on "fault contagion."

[0072] Step S13: Based on the calculated integral increase or decrease, the accumulation is carried out within the rolling time window ΔT to generate the final base station health credit value, and high-risk base stations are identified based on the base station health credit value.

[0073] The choice of ΔT is related to the characteristics of base station faults. For example, a 24-hour period can be used to cover the load fluctuation of a complete daily cycle. The integral values ​​of all data points within the time window are summed for each ΔT period to obtain the total health credit value Ω. i Ω i Let Ωi be the base station health credit value of the i-th base station. To prevent the credit value from decreasing indefinitely, when the calculated result of Ωi is less than or equal to 0, the system automatically sets it to 0 and adds a high-risk base station electronic flag to base station i to trigger an alarm. For example, if a base station is in a stress state for several consecutive days and accumulates a score of -0.5, then Ωi will be set to 0 and an alarm will be triggered.

[0074] The selection of four-dimensional indicators in step S10 covers battery status, reliability, load, and thermal stress, comprehensively reflecting the health status of the base station power supply. This solves the problem of traditional evaluation indicators being singular, such as the one-sidedness caused by focusing only on battery voltage. The dual-threshold design reduces the impact of instantaneous disturbances, such as a sudden temperature rise within 5 minutes, on the credit value by dividing it into steady-state, transient, and stress states, thus reducing the false judgment rate compared to the single-threshold method. The rolling time window accumulation allows the health credit value to reflect long-term trends rather than instantaneous states. For example, when a base station occasionally experiences a stress state but remains in a steady state overall, Ω... iIt remains at a high value to ensure the stability of the assessment; when Ω i An alarm is triggered when the score is less than or equal to 0, providing a clear basis for subsequent selection of candidate compensation base stations and preventing base stations with poor health from participating in compensation, thus reducing the risk of fault contagion from the source. The zero-score design in the transition state makes the credit value change smoother, making it easier for the system to capture potential deterioration trends, such as when the index gradually migrates from a steady state to a stress state, the score growth slows down; the asymmetric penalty mechanism suppresses short-term fluctuations while accumulating deductions for continuous stress states, ensuring the sensitivity of the assessment while avoiding overreaction. Without step S10, subsequent scheduling would lack quantitative basis, which may lead to base stations with poor health being selected for compensation, exacerbating fault contagion and making it impossible to achieve the risk control objectives of the entire scheme.

[0075] Step S20: Generate a regionalized hexagonal grid risk heat map based on the geographical coordinates and health credit values ​​of all base stations.

[0076] Please see Figure 2 As shown, step S20 further includes:

[0077] Step S21: Construct a mesh system consisting of multiple regular hexagonal mesh cells that divide the entire target area;

[0078] Step S22: Project the base station health credit value of each base station to the corresponding hexagonal grid cell and calculate the average health of the grid.

[0079] Step S23: Calculate the grid base station density, combine the grid average health and the grid base station density to construct a risk intensity function, and obtain the risk intensity value of the grid cell;

[0080] Step S24: Generate a regionalized hexagonal grid risk heat map based on the risk intensity value of the grid cells.

[0081] Specifically, step S21 constructs a grid system consisting of multiple regular hexagonal grid cells that divide the entire target area. The process involves obtaining the GPS coordinates of each base station to generate a set of GPS coordinate points for all base stations within the target area. Based on this set, the optimal side length L of the regular hexagonal grid cells is determined, and a grid system consisting of multiple regular hexagonal grid cells is constructed. The geographical coordinates of the base stations are their GPS coordinates. Based on the set of GPS coordinate points for all base stations within the area, the geometric centroid O of the target area is calculated using the Weiss algorithm. The Weiss algorithm iteratively solves the problem to minimize the sum of weighted distances from each base station coordinate to the geometric centroid. The initial iteration point can be the center of the area's bounding box. The iteration terminates when the coordinate deviation between two iterations is less than a preset accuracy threshold. This ensures that the centroid O represents the central position of the base station distribution within the area. The preset accuracy threshold is set based on the measurement accuracy of the base station coordinates within the target area and the accuracy requirements of subsequent grid division. It must ensure that the iteration results reflect the true central trend of the base station distribution. Generally, a small value sufficient to distinguish subtle changes in the centroid position is chosen; for example, for meter-level accuracy based on GPS coordinates, a value within the range of 1 to 10 meters can be set. The region boundary box refers to the smallest rectangular boundary that can completely enclose the set of GPS coordinate points of all base stations within the target region. It is determined by calculating the maximum and minimum longitude, maximum and minimum latitude of all base station GPS coordinates in the set. Simultaneously, the spatial distribution characteristics of the GPS coordinate point set are analyzed to fit the optimal hexagonal grid side length L. Specifically, the average distance between all base stations is calculated, and combined with the typical coverage radius range of the base stations, a side length value is selected as L that ensures the number of base stations contained in each grid falls within a preset reasonable range. The preset reasonable range is determined based on the total number of base stations in the region and the number of target grids. The number of target grids is set based on the principle of clearly distinguishing areas with different base station distribution densities. Using the centroid O as the origin, a non-overlapping, seamless regular hexagonal grid system covering the entire operation and maintenance area is generated using side length L. Each grid is assigned a unique coordinate index (u, v). The coordinate index assignment rule is to start from the centroid grid (0, 0) and increase or decrease sequentially along the horizontal and 60-degree angular directions to ensure the continuity of coordinate indices between adjacent grids. u is the horizontal axial index, and v is the 60° axial index.

[0082] Step S22 projects the base station health credit value of each base station to its corresponding hexagonal grid cell and calculates the average health score of the grid. Specifically, based on the GPS coordinates of each base station, it assigns the base station to the corresponding hexagonal grid cell of the grid system. This is achieved through a coordinate mapping algorithm, which calculates the distance between the base station's GPS coordinates and the center coordinates of each grid cell. The grid cell with the smallest distance is the grid cell to which the base station belongs. The arithmetic mean of the base station health credit values ​​of all base stations within each hexagonal grid cell is calculated to obtain the average health score of the corresponding grid cell. If there is no base station in the grid cell, the average health score of the grid is temporarily set as a preset benchmark value, which is obtained by interpolation based on the average health scores of adjacent grid cells.

[0083] The calculation method for grid base station density involves dividing the number of base stations within a grid by the geometric area of ​​each grid cell to obtain the base station density per unit area. In the risk intensity function, the risk intensity value is the dependent variable, while the average grid health and grid base station density are the independent variables. The risk intensity value is negatively correlated with the average grid health, meaning the lower the average grid health, the higher the risk intensity value. Conversely, the risk intensity value is positively correlated with the grid base station density, meaning the higher the grid base station density, the higher the risk intensity value. This comprehensively reflects the impact of the health status and distribution density of base stations within a region on the overall risk. For example, the risk intensity function uses... Among them, Ψ(M) j ,κ j ) represents the risk intensity, M j Let κ be the average health of the j-th grid cell. j Let M be the grid base station density of the j-th grid cell, where ∈ is the minimum value, for example, 0.01, to avoid M j The calculation is abnormal when =0, w1 is M j The weighting coefficients, w2 is κ j The weighting coefficients satisfy w1 + w2 = 1, with max(·) representing the maximum value. This function implies that the lower the average health of the grid and the denser the base stations, the higher the potential risk intensity. The weighting coefficients are determined based on the influence of the average health of the grid and the density of base stations on the probability of fault propagation in historical operation and maintenance data. The standardized regression coefficients of these two factors are calculated through multiple linear regression analysis and used as the initial values ​​of w1 and w2, respectively. Iterative verification and adjustment are then performed to minimize the fitting error between the risk intensity value and the actual fault occurrence frequency. All grid cells are scanned in real time, and the risk intensity value of each grid cell is calculated using the risk intensity function.

[0084] Please refer to Table 1, which shows the correspondence between risk intensity thresholds and heatmap colors. The method for generating a regionalized hexagonal grid risk heatmap includes: setting a low risk intensity threshold Ψ. yellow and high-risk intensity threshold Ψ red Ψyellow and Ψ red The determination is based on the statistical correlation between risk intensity values ​​and fault propagation range in historical risk events. The maximum risk intensity value that makes the fault incidence rate in low-risk areas lower than a preset value is selected as Ψ. yellow The minimum risk intensity value that makes the probability of fault propagation in high-risk areas higher than the preset value is selected as Ψ. red And Ψ red Greater than Ψ yellow When Ψ j <Ψ yellow When Ψ is reached, the j-th grid cell is marked as green, representing a low-risk area with a failure rate of [missing information]. yellow ≤Ψ j <Ψ red When Ψ is reached, the j-th grid cell is marked in yellow, indicating a medium-risk area; when Ψ j ≥Ψ red At that time, the j-th grid cell is marked in red as a high-risk area with a high probability of fault propagation, requiring priority attention; among them, Ψ j Let be the risk intensity value of the j-th grid cell.

[0085] Table 1. Correspondence between risk intensity thresholds and heatmap colors.

[0086]

[0087] Compared to traditional square grids, the hexagonal grid system in step S20 provides richer spatial connection paths for each grid with six adjacent grids, reducing path breaks during cross-grid retrieval. This makes the risk assessment more closely match the circular characteristics of base station signal coverage and reduces coverage errors caused by grid shape. The geometric centroid calculated by the Wei algorithm serves as the origin, giving the grid system spatial symmetry and preventing index imbalance caused by excessive concentration of base stations in a certain area, ensuring the balance of risk assessment. The calculation of the grid average health transforms discrete base station health credit values ​​into regionalized data, realizing the shift from single-point assessment to area-based assessment. This provides a data foundation for macro-risk analysis and, in conjunction with the base station health credit values ​​generated in step S10, ensures that the health status assessment has both individual precision and regional breadth, improving the comprehensiveness of risk assessment. The risk intensity function integrates the grid average health and grid base station density, overcoming the one-sidedness of single-factor assessment. For example, if a grid has a low average health but sparse base stations, its risk intensity value will not be too high. Conversely, even if the average health is slightly higher in dense areas, the risk intensity may still increase due to the density contribution, which is more consistent with actual fault propagation patterns. The risk heatmap visually displays the risk distribution through color, enabling maintenance personnel to quickly locate high-risk areas. This provides spatial decision-making support for the compensation base station selection in step S30 and the maintenance scheduling in step S50. For example, when selecting candidate compensation base stations, base stations within high-risk grids can be prioritized for exclusion to avoid exacerbating the risk. The coordinate index (u,v) of the regular hexagonal grid enables precise spatial addressing, facilitating subsequent data management and spatial indexing, and improving the efficiency of risk data retrieval. When combined with the neighborhood selection in steps S20 and S30, the multi-adjacent characteristic of the regular hexagonal grid cells makes the selection range of candidate base stations more reasonable, reducing the omission of candidate base stations due to grid boundaries and improving the effectiveness of the compensation scheme. Without step S20, regional risks cannot be visualized and quantified, and subsequent compensation scheduling and maintenance decisions will lack spatial dimension support. This may lead to base stations in high-risk areas being selected as compensation targets, exacerbating fault contagion and making it difficult to achieve the overall risk control objectives of the scheme.

[0088] Step S30: When a base station is detected to have a coverage hole due to a fault, candidate compensation base stations are selected in the neighborhood of the faulty base station. Based on the base station health credit value of the candidate compensation base stations, credit-priority compensation allocation is performed.

[0089] Further, step S30 includes:

[0090] Step S31: Obtain the average output power P of the most recent rolling time window ΔT before the base station goes offline. n According to the average output power P n Calculate the coverage void value σ;

[0091] Coverage holes refer to areas where the signal strength in a base station's original coverage area falls below the required communication quality level after the base station ceases operation due to a fault. The size of this area is related to the output power of the faulty base station, its coverage range, and the distribution density of surrounding base stations. Average output power P n The calculation is the arithmetic mean of the base station's output power within the rolling time window ΔT. ΔT is consistent with the rolling time window in step S13 to ensure the uniformity of the data statistical period. When calculating the coverage hole σ, a coverage attenuation coefficient ρ is introduced. The determination of the coverage attenuation coefficient ρ needs to be based on the geographical location characteristics of the offline base station, obtained by analyzing the compensation effect data of surrounding base stations after historical base station failures in the area: For base stations located at the edge of the target area, due to the low density of surrounding base stations and the small signal coverage overlap area, compensation is more difficult, so the ρ value needs to be set to a larger value, such as 1.5; for base stations located in the center of the target area, the surrounding base stations are densely distributed, the signal coverage overlap area is larger, and compensation is less difficult, so the ρ value is set to a smaller value, such as 1.1. The formula for calculating the coverage hole value σ is σ = P n ×ρ, the coverage hole value represents the power gap that needs to be shared by surrounding base stations, and its size varies with P. n The power distribution increases with the increase of ρ, ensuring that subsequent power allocation can specifically fill coverage gaps.

[0092] Step S32: Based on the base station health credit value of the faulty base station and the risk intensity value of the grid cell where the faulty base station is located, a set of candidate compensation base stations is selected in the neighborhood of the faulty base station.

[0093] The neighborhood of a faulty base station is defined as a circular area centered on the faulty base station with a preset radius r. The radius r is determined by referring to the typical signal coverage radius of the base station, combined with the average distribution density of base stations within the target area, and is obtained through historical data analysis of the correlation between the number of base stations in the neighborhood of the faulty base station and the compensation effect, ensuring that the neighborhood contains a sufficient number of base stations to meet the compensation requirements. All base stations within the neighborhood are defined as candidate base stations. All candidate base stations are iterated through; if the base station health credit value of the i'-th candidate base station is higher than the safety threshold and the risk intensity value of the grid cell where the candidate base station is located is lower than the high-risk intensity threshold Ψ, then... red Then, the i'th candidate base station is added to the candidate compensation base station set as a candidate compensation base station, where i' is the index variable of the candidate base station. The safety threshold is set based on the distribution characteristics of the base station health credit value in step S13, and a credit value that can reflect the stable operation of the base station is selected to ensure that the base station participating in the compensation has the health basis to bear the additional load; Ψ red Consistent with the high-risk intensity threshold in step S20, this avoids base stations in high-risk areas from participating in compensation, which could exacerbate the risk accumulation in that area.

[0094] Step S33: Sort the candidate compensation base stations in the candidate compensation base station set in descending order according to the base station health credit value, and allocate compensation power in sequence until the coverage hole value σ is met. Define the candidate compensation base station participating in the compensation task as the participating base station.

[0095] Step S33 sorts the candidate compensation base stations in the candidate compensation base station set in descending order according to their base station health credit values, and allocates compensation power sequentially until the coverage hole value σ is met. The candidate compensation base stations participating in the compensation are defined as participating base stations. The sorting is based on the base station health credit values ​​generated in step S13; base stations with higher health credit values ​​are allocated priority, ensuring that base stations with better health conditions take on compensation tasks first, reducing the possibility of base stations with poor health conditions failing due to overload. Starting with the first-ranked candidate compensation base station, compensation power ΔP is allocated sequentially to it. j' Where j' is the index variable of the candidate compensation base station, ΔP j' This represents the compensation power allocated to the j'-th candidate compensation base station. During the allocation process, a single-site power increase limit check is enforced: the compensation power ΔP allocated to candidate compensation base station j'. j' It must not exceed γ times its rated power Pe (i.e., ΔP) j' ≤γ×Pe). If the calculated ΔP j' If the power exceeds this upper limit, it is truncated to the upper limit value γ×Pe, and the remaining unmet power gap is assigned to the next candidate base station. This process is repeated until the total compensation power allocated to all candidate base stations is greater than or equal to σ, ensuring that coverage holes are fully filled. γ represents the maximum allowable increase ratio of the compensation power allocated to a candidate base station relative to its rated power when undertaking compensation tasks. It is usually 0.3 to 0.5. The determination of γ needs to refer to the design parameters of the base station power system and the correlation between power increase and equipment aging in historical operating data, and select a multiple that will not significantly accelerate the aging of the equipment under short-term power increases. This strategy ensures that the healthiest base station takes priority in undertaking compensation tasks without overburdening it.

[0096] In step S34, a compensation instruction is broadcast to each participating base station, and a three-way handshake confirmation is performed for each participating base station. If any handshake phase fails, the compensation task of the participating base station is cancelled and it is removed from the candidate compensation base station set. If the candidate compensation base station set is still not empty, the allocation process in step S33 is re-executed. If the candidate compensation base station set is empty, an emergency maintenance alarm is immediately triggered.

[0097] The cloud platform generates a compensation instruction, which includes the base station ID, the allocated compensation power, and the duration of the compensation task. The duration is determined based on the estimated repair time of the faulty base station, combined with the average repair time of similar faults in historical maintenance data. This instruction is broadcast to the corresponding base station in JSON format via the 4G communication module. The three-way handshake confirmation mechanism is implemented as follows: First handshake: Upon receiving the instruction, the base station returns a reception confirmation, including the integrity verification result of the instruction packet, ensuring the instruction has been correctly received and parsed. Second handshake: After completing the power adjustment, the base station returns an execution confirmation, including the actual adjusted power value, facilitating cloud platform verification of whether the adjustment meets the instruction requirements. Third handshake: After the base station's operation stabilizes, it returns a stability confirmation. The stabilization duration refers to the stabilization period after the equipment power adjustment, typically the time it takes for the equipment to operate at the new power until the fluctuations of various parameters are less than the preset fluctuation threshold. The confirmation includes current operating parameters such as output power and temperature, ensuring stable equipment operation after power adjustment. If any handshake phase fails, a rollback operation is immediately performed, canceling the compensation task for that base station, removing it from the candidate compensation base station set, and reallocating it to avoid power allocation chaos caused by communication failures or equipment malfunctions. A real-time compensation status monitoring panel is established to display information such as the current status of participating compensation base stations, the duration of compensation, and remaining power, allowing maintenance personnel to monitor the compensation progress in real time. This mechanism ensures the reliability of the compensation process and avoids power allocation chaos caused by communication failures.

[0098] In step S30, the calculation of the coverage hole value σ combines the average output power and the coverage attenuation coefficient. Compared with the existing technology that only uses output power as the compensation basis, this more accurately reflects the amount of compensation required after a base station failure in different geographical locations. For example, the σ value is larger after a failure of an edge base station, ensuring that the compensation power can fully cover the hole area and avoiding signal quality degradation due to insufficient compensation. Through dual screening of base station health credit value and grid risk intensity value, base stations with poor health status or located in high-risk areas are more strictly excluded than those selected based on a single condition. For example, a base station may meet the health credit value requirement, but its grid risk intensity value may be higher than Ψ. redExcluded base stations are excluded to prevent them from exacerbating risks in their respective areas after participating in compensation. This, combined with the risk heatmap generated in step S20, achieves a comprehensive consideration of spatial risk and individual health status, improving the rationality of candidate base station selection. Sorting by health credit value and setting a single-station power limit ensures that base stations with good health status are prioritized and not overloaded with compensation tasks. This reduces the aging rate of healthy base stations compared to random allocation. For example, base stations with high health credit values ​​are prioritized and their power does not exceed the limit, preventing them from entering a stress state due to overload. Combined with the health credit value assessment in step S10, the quantitative results of the health credit value directly affect the power allocation decision, achieving linkage between health status assessment and actual scheduling. The three-way handshake confirmation mechanism ensures more reliable command execution than a single confirmation, reducing power adjustment anomalies caused by communication packet loss or equipment failure. For example, if the first handshake fails, the command can be resent in time; if the second handshake fails, the power adjustment deviation can be corrected in time; and if the third handshake fails, equipment instability can be detected and rolled back in time. Combined with real-time monitoring of the cloud platform, a closed-loop compensation control process is formed, improving the reliability of the compensation process. When the power allocation in step S30 is combined with the rotation strategy in step S40, the base stations participating in compensation enter a rotation period after the task is completed, and the decline in their health credit value is significantly reduced compared to the case without rotation. The dynamic adjustment mechanism of the candidate compensation base station set can quickly reallocate after some base stations fail to handshake, shortening the compensation delay time compared to the fixed allocation method and ensuring that coverage holes can be filled in a timely manner. If step 30 is missing, targeted power compensation cannot be achieved after a base station failure, which may lead to the long-term existence of coverage holes affecting communication quality, or blindly selecting base stations with poor health status to participate in compensation, causing fault contagion and accelerating the deterioration of the base station power system in the area. The power allocation in step S30, together with the health credit value in step S10, the risk heat map in step S20, and the rotation strategy in step S40, form a synergy, making health assessment, risk identification, power allocation, and subsequent maintenance form a complete closed loop, ensuring that the "fault contagion" phenomenon is effectively suppressed. If this step is missing, these technical features cannot be linked, making it difficult to achieve the risk control objectives of the entire solution.

[0099] Step S40: Implement an active rotation strategy for candidate compensation base stations that have completed the power compensation task;

[0100] Further, step S40 includes:

[0101] Step S41: After the compensation task is completed, write a rest status label for all participating base stations and define them as rest base stations, and remove the rest base stations from the candidate compensation base station set.

[0102] Step S42: During the off-duty period, the load on the off-duty base station is actively reduced.

[0103] Specifically, when the compensation task involving the base station ends, the system writes a duration of τ into its state database. r The rest status label REST-τ r At the same time, start the countdown timer T for the remaining rest time. r . τ r The determination of τ needs to be based on the duration of the compensation task, the power increase of the participating base stations during the compensation period, and the power system recovery parameters provided by the equipment manufacturer. It is derived by analyzing the time required for the base stations to recover to a stable state under different compensation intensities in historical data: the longer the compensation duration and the greater the power increase, the higher the τ value. r The longer the timeout period, the more time the base station has to recover to a healthy state; conversely, the shorter the timeout period, the better. The off-duty status label acts as a global scheduling lock, its function being to exclude the base station from the set of all candidate compensation base stations during the off-duty period, preventing it from undertaking compensation tasks again before it has fully recovered. The validity of this label is ensured through a database transaction mechanism, meaning it only becomes valid when the countdown timer T is reached. r When the status is reset to zero, the off-duty status tag will automatically become invalid, and the base station can then re-enter the candidate pool, i.e., the candidate compensation base station set.

[0104] During the off-duty period, the load on the off-duty base station is actively reduced until its operating power does not exceed η×Pe. η is the load reduction coefficient, typically ranging from 0.5 to 0.8. Its determination requires reference to the design redundancy capacity of the base station power module and historical data on the correlation between load and equipment aging. Through experimental analysis of the changing trends of equipment temperature stress amplitude and battery state of health (SOH) under different load ratios, the load ratio that keeps the equipment aging rate at a low level is selected as the value of η, ensuring that the power level corresponding to η×Pe can both maintain the operation of the base station's core services and significantly reduce equipment losses.

[0105] The off-duty status tagging mechanism, by forcibly isolating compensated base stations, prevents their health status from continuously deteriorating due to frequent participation in compensation tasks. Compared to the traditional no-off-duty strategy, this significantly slows down the rate of decline in the health credit value of participating base stations. The off-duty status tag works in conjunction with the candidate compensation base station screening in step S30. When screening candidate compensation base stations, the off-duty status tag automatically excludes base stations in their off-duty period, ensuring that the screening results only include base stations with good health and not overused, thus improving the rationality of power compensation allocation. Load reduction, by lowering the operating power of off-duty base stations, reduces the temperature stress amplitude and load utilization of the power module. This is linked to the four-dimensional key performance indicators in step S10, allowing these indicators to recover during the off-duty period. For example, the load utilization rate drops from its high level during compensation to the level corresponding to η×Pe, curbing the downward trend of the battery health status indicator (SOH) and extending the effective service life of the equipment.

[0106] Compared to existing technologies that rely solely on passive maintenance, the proactive rotation strategy intervenes early to restore the base station's state after compensation, resolving the issue of accelerated equipment aging caused by the cumulative effect of compensation tasks. Load migration during the rotation period not only protects the rotating base station but also optimizes the load distribution of neighboring base stations, preventing a single base station from operating at high load for extended periods. This results in a more balanced load distribution among base stations within the region and reduces fluctuations in the grid risk intensity value. The countdown timer T for the rotation status label... r This synergizes with the health credit score calculation in step S10 along the time dimension, when T r At the end of the process, the base station's health credit value has recovered to its pre-compensation state, sufficient to support new compensation tasks. Re-entering the candidate pool at this point ensures its ability to undertake new compensation tasks. Compared to the absence of a rotation strategy, this mechanism allows base stations to regain their candidate status earlier, reducing the idle period of base station resources and thus improving the overall utilization efficiency of base station resources in the area. Without rotation, although base stations can enter the candidate pool faster, the load pressure and temperature stress accumulated during the compensation task have not been released. The four key performance indicators—battery health status, historical mean time between failures, load utilization, and power module temperature stress—are still deteriorating. In this state, even if selected to participate in a new compensation task, the rapid decline in health credit value may trigger a stress state, even a fault alarm, ultimately leading to forced shutdown for maintenance and a longer period of passive idleness. If step S40 is missing, the base station that has completed the compensation task will continue to be under high load or frequently called upon, causing its health credit value to drop rapidly and triggering a high-risk alarm, leading to new coverage gaps and exacerbating the "fault contagion" phenomenon.

[0107] Without a rotational rest strategy, base stations participating in compensation cannot effectively recover to a steady-state level after their health credit values ​​decline. In this case, the health credit value assessed in step S10 will continue to decrease. When selecting candidate base stations in step S30, the number of base stations with acceptable health credit values ​​may decrease, forcing the selection of base stations with deteriorated health conditions to continue undertaking compensation tasks. When these base stations are compensated again, their health credit values ​​will further decline, forming a cycle of "compensation - health deterioration - worse base stations participating in compensation - further health deterioration." This causes the health credit value assessment in step S10 to lose its effective reflection of the true health status of the base stations, and the allocation logic in step S30 to fail due to a lack of sufficiently healthy candidate compensation base stations. Ultimately, the decline in health credit values ​​becomes irreversible and cannot be contained. This significantly weakens the overall solution's ability to suppress "fault contagion," accelerates the degradation of the base station power system in the area, and increases operation and maintenance costs.

[0108] Step S50: Identify potential risky base stations within the target area and perform maintenance intention identification. When a maintenance intention is identified, combine the regionalized hexagonal grid risk heat map to push maintenance tasks.

[0109] Further, step S50 includes:

[0110] Step S51: Obtain the historical sequence of base station health credit values ​​for each base station within the target area, and obtain the future preset time ΔT using a pre-built base station health credit prediction model. p The sequence of base station health credit prediction values ​​for each base station within the range, if the base station health credit prediction value in the future ΔT p If the base station is below the preset maintenance threshold for a period of time, it will be marked as a potential risk base station.

[0111] The historical sequence of base station health credit values ​​refers to the continuous records of base station health credit values ​​generated in step S13 over multiple rolling time windows ΔT. Its length must meet the input requirements of the prediction model, typically covering at least one complete equipment operating cycle to include the health fluctuation characteristics of the equipment under different load and environmental conditions. The base station health credit prediction model employs machine learning models, such as LSTM or Transformer. These models have the ability to process time-series data and capture the trends and periodic characteristics of health credit values ​​over time. The model training process requires using historical health credit value data and corresponding fault records of base stations within the target area. The historical sequence is used as input, and the health credit value for a future time period is used as output. Model parameters are adjusted to ensure that the error between the predicted and actual values ​​is within a preset range. The error range is set with reference to the accuracy requirements of health credit value evaluation to ensure the reliability of the model's prediction results. The future preset time ΔT p The determination of ΔT needs to be combined with the response cycle of base station maintenance, and refer to the time required from the discovery of potential risks to the completion of maintenance in historical operation and maintenance, so that ΔT p Sufficient coverage of the entire maintenance preparation and execution process ensures adequate time to address anticipated potential risks. Maintenance thresholds are set based on a correlation analysis of historical data between base station health credit values ​​and fault occurrence, selecting values ​​where the probability of fault occurrence increases significantly below the maintenance threshold. When the predicted health credit value of a base station is in the future ΔT... p When the health credit score remains below the maintenance threshold for a given period of time, it indicates a clear trend of continuous deterioration in the base station's health status. This base station is then marked as potentially risky and an early intervention mechanism can be triggered. Using a machine learning model to predict the health credit score, compared to traditional threshold-based static early warning methods, can capture the dynamic trends of health status changes, identify risks in advance, and avoid failures caused by delayed warnings.

[0112] Step S52: Prioritize all identified potential risk base stations according to the risk intensity value of their respective grid cells to form a maintenance team list;

[0113] The risk intensity value of a grid cell comprehensively reflects the average health status and distribution density of base stations within the grid. A higher risk intensity value indicates a worse overall health status of base stations in that area and a higher risk of fault propagation. Therefore, potentially risky base stations located in that grid need to be prioritized. The maintenance priority of a single potentially risky base station depends not only on its own health status but also on the probability of risk propagation in its surrounding area. In grids with dense base stations and poor overall health, a failure in a single base station has a significantly higher probability of triggering a cascading compensation effect on surrounding base stations and accelerating regional degradation. Therefore, it needs to be prioritized to break the potential "fault contagion" chain. The sorting process requires first obtaining the risk intensity value of the grid where each potentially risky base station is located, and then sorting the potential risky base stations from highest to lowest value. For potential risky base stations within the same grid, the rate of deterioration of their health credit prediction value can be further considered, with base stations deteriorating faster ranked higher. The resulting maintenance team list must include information such as base station ID, grid coordinates, health credit prediction value, and the estimated time to reach the maintenance threshold, providing a clear basis for subsequent maintenance scheduling. Sorting based on grid risk intensity values ​​takes into account the overall regional risk better than simply sorting by the health status of individual base stations. For example, if two potentially risky base stations have similar individual health statuses, but one is located in a high-risk grid and the other in a low-risk grid, the base station in the high-risk grid will be prioritized for maintenance after sorting. This can prevent the risk in the area from accumulating further and causing cascading failures. It achieves a comprehensive consideration of individual and regional risks, allowing maintenance resources to be tilted towards areas with higher risks, thereby improving overall maintenance efficiency.

[0114] Step S53: Obtain the real-time location of the engineer, and based on the maintenance team list and the engineer's real-time location, identify the maintenance intent and push maintenance tasks.

[0115] Further, step S53 includes:

[0116] Step S531: Obtain the real-time location of the engineer; based on the maintenance team list and the real-time location of the engineer, perform maintenance intent identification.

[0117] Further, step S531 includes:

[0118] Step S5311: Determine the target base station from the maintenance team list, set the grid cell where the target base station is located as the target grid, and set the geometric center of the target grid as the target point;

[0119] Step S5312: Within a preset time window, periodically collect the location coordinates of engineers sorted by time.

[0120] Step S5313: For every two consecutive position coordinate points within the time window, calculate the motion vector representing the engineer's actual movement direction; and for each motion vector, calculate the target vector corresponding to the motion vector, taking its starting position coordinate point as the starting point and the target point as the ending point.

[0121] Step S5314: Calculate the angle between each motion vector and its corresponding target vector; count the number of instances where the angle is less than 90 degrees within the time window, and record them as the effective movement count; divide the effective movement count by the total number of motion vectors calculated within the time window to obtain the convergence ratio.

[0122] Step S5315: When the proximity ratio is greater than a preset intention threshold, it is determined that the engineer has the intention to maintain the target base station.

[0123] The target base station is selected as the highest priority base station in the maintenance team list, i.e., the first-ranked potential risk base station, ensuring that the base station most urgently needing maintenance is prioritized for inclusion in the intent recognition range. The length of the time window needs to be based on the typical speed of the engineer's movement and the average distance between base stations, ensuring that sufficient positional changes can be captured within the window to determine the movement trend. For example, it can be set as the average time required for the engineer to move from one grid to an adjacent grid. Position coordinates are acquired through the GPS module of the engineer's mobile terminal. The acquisition cycle needs to balance real-time performance and power consumption, typically set to an interval that reflects continuous movement trajectories, avoiding distortion in movement direction judgment due to excessively long cycles. The motion vector is calculated as the difference between the coordinates of the next position and the coordinates of the previous position, including direction and distance information, reflecting the direction and magnitude of the engineer's movement between the two acquisition points. The target vector is calculated as the difference between the coordinates of the target point and the coordinates of the starting point of the motion vector, also including direction and distance information, reflecting the direction from the current position towards the center of the grid where the target base station is located. The included angle is calculated using the vector dot product formula, obtained by dividing the dot product of two vectors by the inverse cosine of their product. An angle less than 90 degrees indicates that the engineer's movement direction is consistent with the direction pointing to the target point, meaning they are approaching the target. The proximity ratio reflects the probability that the engineer will move towards the target base station; a higher proximity ratio indicates a clearer intention to go to the target base station. The intention threshold is determined based on the distribution of proximity ratios when engineers actually go to the target base station in historical data. A value is selected where the probability of the engineer actually going to the target base station reaches a preset level when the proximity ratio is higher than the intention threshold, ensuring the accuracy of intention recognition. When the proximity ratio is greater than the preset intention threshold, it is determined that the engineer has the intention to maintain the target base station.

[0124] For example, please refer to Figure 3As shown, with the intent threshold set to 70%, within a preset time window, the position coordinates of five engineers (t1, t2, t3, t4, t5) are periodically collected in chronological order. The motion vector representing the actual movement direction of the engineer between coordinates t1 and t2 is t1→t2. The vector pointing from t1 to the target point is the target vector corresponding to motion vector t1→t2. The angle between motion vector t1→t2 and the corresponding target vector is 20°, less than 90°, indicating effective movement. The motion vector between t2 and t3 is t2→t3. The vector pointing from t2 to the target point is the target vector corresponding to motion vector t2→t3. The angle between motion vector t2→t3 and the corresponding target vector is... 40°, less than 90°, is considered a valid movement; the movement vector between t3 and t4 is t3→t4, and the vector pointing from t3 to the target point is the target vector corresponding to the movement vector t3→t4. The angle between the movement vector t3→t4 and the corresponding target vector is 10°, less than 90°, which is considered a valid movement; the movement vector between t4 and t5 is t4→t5, and the vector pointing from t4 to the target point is the target vector corresponding to the movement vector t4→t5. The angle between the movement vector t4→t5 and the corresponding target vector is 10°, less than 90°, which is considered a valid movement; the number of valid movements is 4, and the proximity ratio is 100%, which is greater than the intent threshold. Therefore, it is determined that the engineer has the intent to maintain the target base station.

[0125] Step S532: If a maintenance intention is identified, a task information package is pushed in conjunction with the regionalized hexagonal grid risk heat map. The task information package includes maintenance task information and emergency risk task information of the target base station. If no maintenance intention is identified, the task information package is not pushed to the engineer.

[0126] Maintenance task information includes the target base station's ID, location, predicted health credit value, and the expected type of maintenance operation. Sudden risk task information refers to the situation where, in the regionalized hexagonal grid risk heatmap, a non-target grid changes from green to red or from yellow to red within a preset time Δt from the current moment. If the distance between the engineer's real-time location and the geometric center of the non-target grid is less than a preset distance threshold, base station maintenance information within that non-target grid is pushed to the engineer. The setting of Δt needs to consider the response requirements for changes in risk level to ensure timely capture of sudden risks; changes in risk level are represented by color changes in the regionalized hexagonal grid risk heatmap. Δt can be set to 15 minutes and can be dynamically adjusted according to the intensity of regional network load fluctuations, shortening to 5 minutes during peak periods. The distance threshold setting needs to consider the range reachable by the engineer within the preset time, determined based on the engineer's average movement speed and the urgency of the sudden risk, ensuring that the pushed sudden tasks are within the engineer's reach. If the engineer's maintenance intention is not identified, planned tasks are not proactively pushed to avoid information interference, ensure user experience, and ensure that the engineer only receives proactive instructions when truly needed.

[0127] Compared to traditional task dispatching methods, the maintenance intent recognition mechanism reduces unnecessary information pushes and prevents engineers from being disturbed by irrelevant tasks. For example, if an engineer is handling other tasks and their intent to go to a target base station is not recognized, that task will not be pushed, improving the engineer's work efficiency. The maintenance task push, in conjunction with the maintenance team list, ensures that the highest priority tasks are pushed promptly when engineers have the intent to go there, guaranteeing that high-risk base stations receive priority maintenance. Combined with the risk heatmap in step S20, task pushes can take into account both pre-set maintenance queues and sudden regional risks. For example, if the risk level in a certain area suddenly rises, if an engineer is nearby, the task can be pushed promptly to prevent the risk from spreading. The mechanism for pushing sudden risk task information, together with the power compensation allocation in step S30, forms a collaborative emergency response mechanism. When a sudden risk occurs in a certain area, nearby engineers are promptly notified to intervene, reducing reliance on compensation for base stations in that area and preventing health deterioration caused by compensation, forming a dual guarantee of "predictive maintenance + emergency response." Without step S50, maintenance task pushes will lack focus, potentially causing engineers to receive a large amount of irrelevant information and reducing work efficiency. At the same time, the inability to identify the engineer's maintenance intentions in a timely manner will delay the execution of high-priority tasks, turning potential risks into actual failures.

[0128] Example 2

[0129] This embodiment, based on Embodiment 1, provides an intelligent operation and maintenance 5G power management system, such as... Figure 4 As shown, it includes:

[0130] Health credit construction module: used to acquire and normalize the four-dimensional key performance indicators of multiple base stations in the target area, and construct the quantified health credit value of all base stations;

[0131] Heatmap generation module: Generates a regionalized hexagonal grid risk heatmap based on the geographic coordinates and health credit values ​​of all base stations;

[0132] Compensation module: When a coverage hole is detected due to a base station going offline due to a fault, candidate compensation base stations are selected in the neighborhood of the faulty base station. Based on the base station health credit value of the candidate compensation base stations, credit-priority compensation allocation is performed.

[0133] Rotation module: Used to implement an active rotation strategy for candidate compensation base stations that have completed power compensation tasks;

[0134] Maintenance task push module: used to identify potential risky base stations within the target area, identify maintenance intentions, and push maintenance tasks when a maintenance intention is identified, combined with a regionalized hexagonal grid risk heat map.

[0135] In the maintenance task push module, the method for recognizing maintenance intent includes:

[0136] Step S5311: Determine the target base station from the maintenance team list, set the grid cell where the target base station is located as the target grid, and set the geometric center of the target grid as the target point;

[0137] Step S5312: Within a preset time window, periodically collect the location coordinates of engineers sorted by time.

[0138] Step S5313: For every two consecutive position coordinate points within the time window, calculate the motion vector representing the engineer's actual movement direction; and for each motion vector, calculate the target vector corresponding to the motion vector, taking its starting position coordinate point as the starting point and the target point as the ending point.

[0139] Step S5314: Calculate the angle between each motion vector and its corresponding target vector; count the number of instances where the angle is less than 90 degrees within the time window, and record them as the effective movement count; divide the effective movement count by the total number of motion vectors calculated within the time window to obtain the convergence ratio.

[0140] Step S5315: When the proximity ratio is greater than a preset intention threshold, it is determined that the engineer has the intention to maintain the target base station.

[0141] The methods and systems of this application may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the method is for illustrative purposes only, and the steps of the method of this application are not limited to the order specifically described above, unless otherwise specifically stated.

[0142] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0143] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart operation and maintenance method for 5G power management, characterized in that, The method includes: Obtain and normalize the four-dimensional key performance indicators of multiple base stations within the target area, and construct a quantified base station health credit value for all base stations; Based on the geographic coordinates and health credit values ​​of all base stations, a regionalized hexagonal grid risk heat map is generated. When a base station is detected to have a coverage hole due to a fault, candidate compensation base stations are selected in the neighborhood of the faulty base station. Based on the base station health credit value of the candidate compensation base stations, credit-priority compensation allocation is performed. Specifically, this includes: obtaining the average output power of the most recent rolling window before the base station went offline, and calculating the coverage hole value based on the average output power; sorting the candidate compensation base stations in the candidate compensation base station set in descending order according to their base station health credit value, and allocating compensation power in sequence. During the allocation process, a single-station power increase limit check is performed: if the compensation power allocated to the candidate compensation base station exceeds γ times the rated power Pe of the candidate compensation base station, the compensation power is truncated to the upper limit value γ×Pe, and the remaining unmet power gap is assigned to the next candidate base station in the sorting, until the sum of the compensation power allocated to all candidate base stations is greater than or equal to the coverage hole value, where γ is the maximum allowable increase ratio coefficient. For candidate compensation base stations that have completed the power compensation task, an active rotation strategy is implemented; Identify potential risky base stations within the target area, identify maintenance intentions, and when a maintenance intention is identified, push maintenance tasks based on a regionalized hexagonal grid risk heat map.

2. The intelligent operation and maintenance 5G power management method according to claim 1, characterized in that, The method for constructing the base station health credit value quantified for all base stations includes: determining the state of the key performance indicator sequence and calculating the integral increase or decrease; based on the calculated integral increase or decrease, accumulating it within a rolling time window ΔT to generate the final base station health credit value.

3. The intelligent operation and maintenance 5G power management method according to claim 2, characterized in that, The method for determining the state of the key performance indicator sequence and calculating the integral increase or decrease includes: Set the steady-state threshold α for the k-th dimension key performance indicator. k With stress threshold β k Where k is the index variable for key performance indicators, k=1,2,3,4; β k >α k ; If the sequence value of the k-th dimension key performance indicator is less than or equal to the corresponding steady-state threshold α k If the k-th key performance indicator is in a steady state, then the contribution of the k-th key performance indicator to health credit is positively accumulated by +ω; ω is the basic unit of integration. If the sequence value of the k-th key performance indicator is greater than or equal to the corresponding stress threshold β k If the key performance indicator of dimension k is in a state of stress, the contribution of the key performance indicator of dimension k to health credit is reduced by -ω / 2. If the sequence value of the k-th key performance indicator is greater than α k And less than β k If the condition is not met, it is considered a transitional state, the integral is 0, and credit neutrality is maintained.

4. The intelligent operation and maintenance 5G power management method according to claim 3, characterized in that, The method for generating a regionalized hexagonal grid risk heatmap includes: Construct a grid system consisting of multiple regular hexagonal grid cells that divides the entire target area; The base station health credit value of each base station is projected onto the regular hexagonal grid cell in which it is located, and the average health of the grid is calculated. Calculate the grid base station density, combine the grid average health and the grid base station density to construct a risk intensity function, and obtain the risk intensity value of the grid cell; Based on the risk intensity value of the grid cells, a regionalized hexagonal grid risk heat map is generated.

5. The intelligent operation and maintenance 5G power management method according to claim 4, characterized in that, In the risk intensity function, the risk intensity value is the dependent variable, and the grid average health and grid base station density are the independent variables. The risk intensity value is negatively correlated with the grid average health and positively correlated with the grid base station density.

6. The intelligent operation and maintenance 5G power management method according to claim 5, characterized in that, The method for selecting candidate compensation base stations within the neighborhood of a faulty base station includes: defining all base stations within the neighborhood as candidate base stations; traversing all candidate base stations; and if the base station health credit value of the i'th candidate base station is higher than the safety threshold and the risk intensity value of the grid cell where the candidate base station is located is lower than the high risk intensity threshold Ψ. red If i' is added as a candidate compensation base station to the candidate compensation base station set, then i' is the index variable of the candidate base station.

7. The intelligent operation and maintenance 5G power management method according to claim 6, characterized in that, The proactive rotation strategy is as follows: candidate compensation base stations participating in the compensation task are defined as participating base stations. After the compensation task is completed, rotation status tags are written for all participating base stations and they are defined as rotation base stations. The rotation base stations are removed from the candidate compensation base station set. During the rotation period, the load of the rotation base stations is proactively reduced.

8. The intelligent operation and maintenance 5G power management method according to claim 7, characterized in that, The method for identifying potential risky base stations within the target area is as follows: Obtain the historical sequence of base station health credit values ​​for each base station within the target area, and use a pre-built base station health credit prediction model to obtain the future preset time ∆T. p The sequence of base station health credit prediction values ​​for each base station within the range, if the base station health credit prediction value in the future ∆T p If the base station is below the preset maintenance threshold for a period of time, it will be marked as a potentially risky base station.

9. The intelligent operation and maintenance 5G power management method according to claim 8, characterized in that, The method for identifying maintenance intent is as follows: Identify the target base station from the maintenance team list, set the grid cell containing the target base station as the target grid, and set the geometric center of the target grid as the target point; Within a preset time window, the location coordinates of engineers sorted by time are periodically collected; For every two consecutive position coordinate points within the time window, a motion vector representing the engineer's actual movement direction is calculated; and for each motion vector, with its starting position coordinate point as the starting point and the target point as the ending point, the target vector corresponding to the motion vector is calculated. Calculate the angle between each motion vector and its corresponding target vector; count the number of instances where the angle is less than 90 degrees within the time window, and record them as the effective number of moves; divide the effective number of moves by the total number of motion vectors calculated within the time window to obtain the convergence ratio. When the proximity ratio is greater than a preset intent threshold, it is determined that the engineer has the intent to maintain the target base station.

10. An intelligent operation and maintenance 5G power management system, used to implement the intelligent operation and maintenance 5G power management method according to any one of claims 1-9, characterized in that, The system includes: Health credit construction module: used to acquire and normalize the four-dimensional key performance indicators of multiple base stations in the target area, and construct the quantified health credit value of all base stations; Heatmap generation module: Generates a regionalized hexagonal grid risk heatmap based on the geographic coordinates and health credit values ​​of all base stations; Compensation module: When a coverage hole is detected due to a base station going offline due to a fault, candidate compensation base stations are selected in the neighborhood of the faulty base station. Based on the base station health credit value of the candidate compensation base stations, credit-priority compensation allocation is performed. Rotation module: Used to implement an active rotation strategy for candidate compensation base stations that have completed power compensation tasks; Maintenance task push module: used to identify potential risky base stations within the target area, identify maintenance intentions, and push maintenance tasks when a maintenance intention is identified, combined with a regionalized hexagonal grid risk heat map.

Images