An intelligent management and control method and system for communication tower maintenance

Abstract

This invention discloses an intelligent management and control method and system for communication tower maintenance, belonging to the field of communication facility maintenance and management technology. It constructs a comprehensive digital archive; achieves closed-loop management of vehicle operation trajectories and fuel consumption, calculating theoretical fuel consumption through a dynamic fuel consumption model and fitting it with actual fuel consumption for analysis; implements collaborative management of personnel work trajectories and work quality, triggering standardized work processes through geofencing, and using convolutional neural networks to identify compliance in process images; establishes material and work order associations through RFID tags, compares actual consumption with standard usage deviations, constructs a cost prediction model using a long short-term memory network, and outputs visual reports. This invention, by constructing a multi-dimensional closed-loop management and control logic, realizes the transformation of operation and maintenance resources from post-event statistics to real-time verifiable data, significantly improving cost control accuracy, work quality, and resource utilization efficiency.

Description

Technical Field

[0001] This invention belongs to the field of communication facility maintenance and management technology, specifically an intelligent control method and system for communication tower maintenance. Background Technology

[0002] As the unified construction and operation entity for communication towers, base station supporting facilities, and indoor distribution systems, China Tower manages a vast and geographically dispersed network of outdoor infrastructure. Against this backdrop, the daily inspection and maintenance of base stations, equipment rooms, and communication towers is not only a crucial link in ensuring continuous communication network signal coverage but also a core support for maintaining the efficient operation of the communication industry chain. To address the massive and dispersed asset management needs, the existing maintenance management model has gradually shifted from early fully manual inspections to basic digital management.

[0003] In existing technological implementations, the industry commonly employs dispatch systems based on mobile communication technology and the basic Global Positioning System (GPS) to assist in asset management. Specifically, GPS terminals are installed on maintenance vehicles to monitor their driving trajectories, and mobile applications (APPs) guide maintenance personnel to upload work orders and clock in on-site. This approach has, to some extent, achieved traceability of work processes, improving upon the shortcomings of traditional paper records, which are prone to loss and difficult to statistically analyze, and played a positive role in improving maintenance efficiency during a specific historical period. However, with the continuous expansion of communication maintenance business and the increasing complexity of the operation and maintenance environment, existing technological solutions have exposed a series of deep-seated inherent contradictions in practical applications, which have become increasingly prominent in the context of pursuing refined cost control.

[0004] The underlying reason lies in the core limitation of existing control measures, which stems from the isolation of management dimensions and the loose spatiotemporal correlation. Firstly, regarding vehicle operation and cost expenditure, while existing GPS monitoring can provide driving routes, this single-dimensional physical trajectory lacks strong logical verification between actual fuel consumption and fuel card swiping behavior. Due to the lack of real-time alignment between fuel level sensor data, driving trajectories, and work order tasks, the management platform struggles to identify abnormal fuel consumption during non-operational periods or in non-task areas. This data silo phenomenon leads to the risk of false reimbursements and consumption for non-project purposes, leaving vehicle operating costs consistently uncontrollable.

[0005] Secondly, the existing control logic for the core elements of maintenance operations—personnel, consumables, and tools—often exhibits a fragmented nature. In the management of low-value consumables (such as cables and connectors), existing processes primarily rely on manual data entry after the fact, lacking a closed-loop verification mechanism for requisition, consumption, and balance. Specifically, material requisition and specific maintenance actions are decoupled at the system level, resulting in an inaccurate match between actual material losses on-site and standard operating procedures, leading to significant resource waste. Meanwhile, maintenance tools, as high-value, high-frequency assets, are highly susceptible to loss or idleness in discrete maintenance scenarios. The existing static management model cannot accurately and in real-time perceive the requisition and return status of tools, increasing the company's repetitive procurement costs and management burden.

[0006] More complexly, because maintenance work often takes place in dynamic outdoor environments, the instability of network conditions and the variability of on-site working conditions make it difficult to achieve true closed-loop management of personnel's work quality and attendance efficiency. Simple location-based check-in cannot accurately reflect the depth and compliance of on-site maintenance. This weak supervision model inevitably leads to a serious imbalance between human resource input and actual output when dealing with large-scale project operations across multiple provinces and regions. Furthermore, due to the lack of an intelligent decision-making model capable of multi-dimensional correlation analysis of personnel trajectories, vehicle fuel consumption, material flow, and work quality, management struggles to uncover the underlying patterns of losses when faced with massive amounts of fragmented data.

[0007] In summary, existing communication maintenance and management solutions have revealed significant fundamental bottlenecks in addressing the urgent need for cost reduction and efficiency improvement. How to break through the single-dimensional regulatory model and construct a comprehensive governance system capable of achieving spatiotemporal correlation and closed-loop management of all elements—people, vehicles, materials, and tasks—with intelligent early warning capabilities has become a key technical challenge and an industry-wide problem urgently needing to be solved in the field of communication tower maintenance. Summary of the Invention

[0008] The purpose of this invention is to provide an intelligent management and control method and system for communication tower maintenance, so as to solve the problems of single management and control dimension, loose spatiotemporal correlation, serious resource waste and uncontrollable operation and maintenance costs in the existing communication tower maintenance process mentioned in the background art.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: An intelligent management and control method for communication tower maintenance includes the following steps: Step 1: Construct a full-element digital archive, which should include at least vehicle maintenance records, maintenance personnel records, a list of low-value consumables, a fixed asset database for maintenance tools, and a spatial geographic information database for communication towers. Step 2: Real-time collection of maintenance vehicle operation status data, spatial location data, and fuel consumption data; real-time collection of maintenance personnel location data and work process image data; real-time collection of maintenance tools and low-value consumables issuance and return status data. Step 3: Based on the collected data, establish at least one closed-loop control logic among the following: the correlation verification logic between vehicle operation trajectory and fuel consumption, the collaborative verification logic between personnel operation location and operation quality, the deviation analysis logic between material consumption and standard usage, and the matching and supervision logic between tool circulation status and work order time limit. Step 4: After fusing the multi-dimensional data, input it into the cost prediction model and output the operation and maintenance cost prediction results and visualization reports.

[0010] According to the above technical solution, the correlation verification logic between vehicle operating trajectory and fuel consumption includes: obtaining the cumulative fuel injection quantity calculated by the engine electronic control unit through the vehicle controller local area network bus, or obtaining the fuel level height through a level sensor inserted inside the fuel tank and converting it into fuel volume data using a fuel tank geometric model mapping table to obtain the actual fuel consumption. ; Based on engine displacement Instantaneous speed Instantaneous speed Calculated engine load rate Approved load capacity and road surface slope coefficient Calculate the instantaneous theoretical fuel consumption rate: in To comprehensively correct the factors, and for The theoretical fuel consumption is calculated by accumulating points over the duration of the trip. ; Will With preset threshold In comparison, when the threshold is exceeded and the vehicle is located in a non-preset operating area or non-operating period, an abnormal fuel consumption warning is triggered.

[0011] According to the above technical solution, the logic for verifying the correlation between vehicle operating trajectory and fuel consumption also includes verification of refueling compliance: Get fuel card transaction records, including the transaction time. Geographic coordinates of the gas station ; Extract the real-time GPS location coordinates of the vehicle at the time of card swipe. ;when At that time, it was determined to be an illegal card-swipe for refueling, among which This is a preset spatial matching threshold.

[0012] According to the above technical solution, the collaborative verification logic between personnel work location and work quality includes: Geofencing for task orders is generated based on the polygonal footprint of the communication tower base or the boundary of the standard work area. Once the maintenance personnel's location enters the geofence, the on-site check-in function is activated and they are guided to execute a standardized work process; Images of key processes are captured, and a convolutional neural network model is used to identify compliance of the images and output confidence scores. ; Set dynamic confidence threshold ,when The work order progress is locked in time and a rework instruction is sent. The initial value is determined by the precision and recall curves and periodically optimized based on field feedback.

[0013] Based on the above technical solution, the deviation analysis logic between material consumption and standard usage includes: Attach RFID tags or QR code tags to low-value consumables, and scan the tags during the requisition process to establish a link between the material and the work order; Calculate standard usage based on work order type and bill of materials database. Confirm the actual consumption during the receipt process. Calculate the consumption deviation rate: when Exceeding the preset allowable deviation rate When this occurs, it is determined to be an abnormal material consumption and recorded in the cost deviation report.

[0014] According to the above technical solution, the matching and supervision logic between tool workflow status and work order deadlines includes: To maintain the RFID tags attached to the tools, the RFID reader array embedded in the tool storage box is used to periodically scan the tool tags in the box to monitor the tool's storage status in real time. Record tool outbound time and related work orders; When the preset buffer time for the issued tool is detected after the work order ends. If the item is not returned to the warehouse, a reminder instruction will be sent to the maintenance personnel; if more than 2... If it is still not returned, then copy it to the administrator interface.

[0015] According to the above technical solution, the cost prediction model adopts a long short-term memory network architecture. Its input feature sequence includes internal operation and maintenance variables and external environmental factors. The internal operation and maintenance variables include historical period fuel consumption per 100 kilometers, per capita labor hours output, consumable deviation rate, tool turnover rate and vehicle maintenance frequency. The external environmental factors include weather conditions, seasonal failure frequency and terrain complexity. The model learns the nonlinear mapping relationship between time series features and operation and maintenance costs through the forget gate, input gate and output gate structure of the long short-term memory network, and outputs the predicted value of operation and maintenance expenditure for the next period.

[0016] Based on the above technical solution, Kalman filtering is used to perform multi-source fusion of vehicle positioning data, and GPS positioning output is used as the observation vector. The speed obtained based on the controller area network bus and heading angle The pose data calculated using the dead reckoning formula is used as the state prediction vector. : in, This represents the state estimate at the previous time step. Indicates the instantaneous velocity at the previous moment. Indicates the sampling time interval. This indicates the heading angle at the previous moment. The cosine value of the heading angle is represented by a Kalman filter, which dynamically adjusts the weights to correct the state estimate.

[0017] An intelligent management and control system for the maintenance of communication towers includes: The cloud management platform is used to store all digital archives and run the fuel cost intelligent verification module, operation quality monitoring module, and resource scheduling module. The vehicle-mounted intelligent terminal, installed in the maintenance vehicle, integrates a positioning module, a controller area network bus communication interface, and a liquid level sensor interface, and is used to collect vehicle operation data and perform local preprocessing. The mobile operation terminal, configured for maintenance personnel, integrates a positioning module, camera, biometric module and short-range wireless communication module, and is used for task reception, on-site operation interaction, quality data collection and material scanning and requisition. The intelligent tool storage box is equipped with an array of radio frequency identification readers to enable real-time monitoring of maintenance tool assets and automated retrieval and return of certificates.

[0018] According to the above technical solution, the cloud management platform and the vehicle-mounted intelligent terminal, mobile operation terminal and intelligent tool storage box use a lightweight message queue telemetry transmission protocol for message transmission. Data packets are serialized using a protocol buffer format. The mobile operation terminal supports an offline caching mechanism, which encrypts and stores operation data locally in communication blind spots. After the network is restored, it uses breakpoint resume technology to synchronize the data to the cloud management platform.

[0019] As a preferred option, the cloud management platform and each terminal use the lightweight MQTT protocol for message transmission, and combine it with the TLS 1.3 protocol to ensure the security of the transmission process; the data packets are serialized in Protobuf format to reduce bandwidth consumption in outdoor weak network environments; the mobile operation terminal supports an offline caching mechanism, which encrypts and stores the operation data locally in communication blind spots, and uses breakpoint resume technology to synchronize the data to the cloud management platform after the network is restored, ensuring that the delay in issuing control commands is less than 3 seconds.

[0020] As a preferred option, the system also includes an inventory dynamic early warning module and an access control module; the inventory dynamic early warning module utilizes the Economic Order Quantity (EOQ) model, based on the historical daily average consumption rate of materials. Single purchase cost and annual holding cost per unit of material Through formula Automatically calculate the optimal order quantity and take into account the procurement lead time. and safety stock Through formula Automatic calculation of reorder point When the actual inventory is lower than The system automatically pushes procurement suggestions to the supply chain system. The access control module is based on the role-based access control (RBAC) model, which divides system access and operation permissions into four levels: provincial company administrator, municipal project manager, maintenance team leader, and front-line maintenance personnel. Each level of permission is strictly restricted by policies to ensure that no user can access sensitive data without authorization.

[0021] Compared with the prior art, the present invention has the following beneficial effects: This invention solves the fundamental problem of isolated management dimensions in existing technologies by constructing a comprehensive digital archive that maps dispersed physical assets, personnel, materials, tools, and spatial locations into a unified digital space. Based on this, it collects vehicle operation data through in-vehicle intelligent terminals, calculates theoretical fuel consumption using a dynamic fuel consumption model, and analyzes the results by fitting them to actual fuel consumption. Simultaneously, it uses spatiotemporal matching logic to verify the compliance of refueling behavior, establishing a strong causal relationship between fuel consumption and vehicle operating status. This transforms post-event statistics into real-time verifiable costs, eliminating the risks of false reimbursements and non-project-related fuel consumption. Furthermore, it collects personnel locations through mobile work terminals and matches them with geofences, activating standardized work processes and using convolutional neural networks to verify the compliance of process images. The system achieves a qualitative leap from simply having people on-site to ensuring tasks are done correctly, forming a fully automated closed loop. By binding RFID tags to materials and tools, it establishes a link between requisition and work orders, compares actual consumption with standard usage deviations, and uses tool storage boxes to monitor tool status online, achieving refined closed-loop management of material and asset flow. Through a distributed big data processing engine and long short-term memory network, it constructs a cost prediction model, integrates internal operational variables and external environmental factors, and uncovers the drivers of cost anomalies, achieving a decision-making upgrade from experience-driven to data-driven. Ultimately, it constructs a full-dimensional intelligent management and control closed loop from resource scheduling and process monitoring to quality acceptance and cost optimization, achieving significant technological progress in cost control, quality improvement, and resource utilization efficiency. Attached Figure Description

[0022] Figure 1 This is a flowchart of the intelligent control method of the present invention. Detailed Implementation

[0023] 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.

[0024] Example 1 like Figure 1 As shown, an intelligent management and control method for communication tower maintenance is presented. The method relies on a system architecture consisting of a cloud management platform, vehicle-mounted intelligent terminals, portable IoT terminals, and mobile operation terminals. The method includes the following steps: The first step is to construct a comprehensive digital archive of all aspects of communication tower maintenance. This archive is stored on non-volatile storage media within a cloud-based management platform and includes maintenance vehicle files, maintenance personnel files, a Bill of Materials (BOM) database for low-value consumables, a fixed asset database for maintenance tools, and a spatial geographic information database for communication towers. Specifically, maintenance vehicle files must include at least the vehicle's unique identification number, fuel card number, engine displacement, rated fuel consumption parameters, and approved load capacity; maintenance personnel files must include personnel identification codes, their project team affiliation, professional qualification level, and biometric data.

[0025] The second step involves achieving closed-loop management of vehicle maintenance and fuel consumption. The onboard intelligent terminal collects vehicle operating data in real time via the vehicle controller local area network (CAN) bus interface, and simultaneously acquires GPS coordinates, altitude, and timestamp information. The onboard intelligent terminal integrates a high-precision fuel level sensor monitoring module, employing an ultrasonic sensor or capacitive level gauge, with a sampling frequency of no less than 1Hz and data acquisition accuracy error controlled within 5%.

[0026] During vehicle operation, the onboard intelligent terminal will collect real-time oil level data. Instantaneous speed Engine speed and mileage The data is encapsulated into an encrypted data packet and transmitted to the cloud management platform via a mobile communication network. The cloud management platform then uses an intelligent fuel cost verification algorithm to construct a fuel consumption model. in, This is the theoretical fuel consumption. For load parameters, This is the road surface slope coefficient calculated based on a Geographic Information System (GIS). The cloud management platform will display the actual fuel consumption. Compared with theoretical fuel consumption Perform a fitting analysis. If ( The system automatically triggers a first-level warning if the vehicle is in a non-operational area or during a non-operational period, and the GPS trajectory shows that the vehicle is at a preset threshold offset (initial value is set to 10% of the theoretical fuel consumption). This warning is determined to be suspected non-project fuel consumption.

[0027] Meanwhile, the cloud management platform obtains fuel card swipe records, including the swipe time, through an interface. Fuel volume and the geographical coordinates of the gas station The system verifies the compliance of refueling behavior through spatiotemporal matching logic: the judgment condition is... This means that the vehicle's actual location must coincide with the gas station's location when the card is swiped; otherwise, it will be considered an illegal card swipe.

[0028] The third step involves collaborative management of maintenance personnel's work trajectories and quality. Mobile work terminals collect real-time location data from maintenance personnel and match it with the geofence of the task work order. The geofence is generated based on a high-precision polygon of the communication tower's base area, with its boundaries precisely defined by a set of latitude and longitude coordinates. Personnel must enter the boundary of this polygon before activating the mobile work terminal's attendance check-in function. The system optionally performs secondary verification via Bluetooth beacons or RTK positioning to ensure personnel accurately arrive at the core work area.

[0029] During the operation, mobile work terminals guide personnel to execute standardized operating procedures (SOPs) and require the use of high-definition cameras to capture image or video data at key stages (such as cable connection, equipment debugging, and tool return). A cloud-based management platform deploys a deep learning-based computer vision recognition model to extract features and compare compliance with uploaded work photos. The model uses a convolutional neural network (CNN) architecture to classify and identify aspects such as connector tightness, cable straightness, and label placement in maintenance scenarios. Instead of using fixed thresholds, the system sets dynamic confidence thresholds. , The initial value is determined by the precision and recall curves on the validation set and is periodically optimized based on on-site feedback. If the identification result is lower than the current threshold, the system immediately issues a rework instruction, forming a closed loop of task issuance, on-site verification, work implementation, intelligent acceptance, and completion archiving.

[0030] The fourth step is to implement refined flow control of low-value consumables and maintenance tools. Each low-value consumable (such as feeder, connector, tape) and each maintenance tool (such as fusion splicer, theodolite, handheld tester) is bound with a unique radio frequency identification (RFID) electronic tag or QR code tag.

[0031] During the requisition process, maintenance personnel scan material / tool ​​tags using mobile work terminals, and the system automatically associates this with the current maintenance work order number. The cloud management platform establishes a material consumption prediction model, automatically calculating the standard usage of consumables required based on the work order type and equipment specifications. During the post-operation return process, personnel need to scan the remaining materials upon returning them to the warehouse and upload photos of the remaining materials. The system then calculates the actual consumption. Calculate the consumption deviation rate ,like Exceeding the preset allowable deviation rate (The value is usually 5%-8%). The system will record this material consumption anomaly in the project cost deviation report and restrict the person from using excessive amounts in the future.

[0032] For maintenance tools, the system adopts an IoT-based online inventory mode. The maintenance tool storage box is equipped with an intelligent control module integrating an RFID reader / writer, with a read / write frequency of 902MHz-928MHz, supporting the ISO / IEC 18000-6C protocol. This module uses multi-antenna time-division multiplexing technology to eliminate signal blind spots caused by metal partitions inside the box, achieving comprehensive scanning of the tools inside. When a tool's status changes from "used" to "not returned" and exceeds the work order deadline, the system pushes a supervisory instruction to maintenance personnel and managers via mobile devices to prevent tools from being lost or left idle for extended periods.

[0033] The fifth step involves conducting multi-dimensional data fusion analysis and intelligent decision-making. The cloud management platform stores massive amounts of operational data based on the Hadoop Distributed File System (HDFS) and utilizes the Spark computing engine to execute multi-dimensional data processing tasks. The platform constructs a cost prediction model, using a Long Short-Term Memory (LSTM) network to serialize and model historical operational cost data. Input features include internal operational variables (fuel consumption, working hours, consumable deviation rate, etc.) and external environmental factors (weather, season, terrain, etc.), outputting predicted cost expenditures for future periods.

[0034] The system generates multi-dimensional visualization reports, including vehicle fuel efficiency distribution maps, personnel output intensity curves, consumable deviation rate heatmaps, and tool turnover rate statistics. Managers can identify high-cost and low-efficiency anomalies based on these reports.

[0035] An intelligent management and control system for the maintenance of communication towers, the system comprising: Cloud Management Platform: Serving as the central processing layer of the system, this platform consists of a high-performance server cluster, deploying database servers, application servers, and a big data processing engine. The database servers employ an architecture combining relational and NoSQL databases, storing comprehensive digital archives and various management rules. The application servers run intelligent fuel cost verification modules, operation quality monitoring modules, and resource scheduling modules.

[0036] Vehicle-mounted intelligent terminal: Installed on maintenance vehicles, it integrates a central processing unit (CPU), a GPS / BeiDou dual-mode positioning module, a mobile communication module (supporting 4G / 5G transmission), a CAN bus communication interface, and a high-precision liquid level sensor interface. The central processing unit has an embedded edge computing engine, which has basic local recognition capabilities for abnormal behavior.

[0037] Mobile work terminal: A ruggedized smartphone or tablet equipped for maintenance personnel, with a dedicated maintenance management app installed. The terminal utilizes built-in GPS, camera, fingerprint / facial recognition module, and near field communication (NFC) / QR code scanning function to achieve on-site operation interaction and feedback.

[0038] Intelligent tool storage box: It is equipped with an RFID reader array, an embedded main control board and a wireless transmission unit to realize real-time on-site monitoring of tool assets and automated retrieval and return of certificates.

[0039] Furthermore, the vehicle-mounted intelligent terminal adopts multi-source heterogeneous data fusion technology, which uses Kalman filtering algorithm to fuse GPS positioning data and CAN bus speed data, eliminating positioning drift caused by tunnels and tall buildings, ensuring absolute accuracy of vehicle trajectory and mileage calculation, with trajectory deviation of less than 2 meters.

[0040] Furthermore, the cloud management platform and each terminal use the lightweight MQTT protocol for message transmission, combined with TLS 1.3 encryption to ensure the security of the transmission process. Data packets are serialized using Protobuf format to reduce bandwidth consumption in outdoor weak network environments and ensure that the delay in issuing control commands is less than 3 seconds.

[0041] Furthermore, the intelligent verification algorithm for fuel cost authenticity also includes an anomaly pattern recognition submodule based on machine learning. This submodule learns from historical illegal refueling data using the random forest algorithm, extracting 24 feature dimensions, including refueling frequency, fuel volume fluctuations, and deviation from the driving route, to build an illegal refueling classifier with a classification accuracy of no less than 95%.

[0042] Furthermore, the mobile operation terminal's operation quality control module supports an offline caching mechanism. In areas with no communication base station signal, the terminal can locally store operation data and photos, and upon detecting network recovery, use breakpoint resume technology to synchronize the data to the cloud, ensuring data continuity and integrity.

[0043] Furthermore, the intelligent management module for low-value consumables includes a dynamic inventory early warning function. Based on real-time consumption rates and procurement lead times, the system automatically calculates the optimal inventory level using an Economic Order Quantity (EOQ) model. When the actual inventory falls below the set safety stock threshold, the cloud platform automatically generates procurement recommendations and pushes them to the supply chain management system.

[0044] Furthermore, the system also includes a hierarchical access control module. Based on the organizational structure of the maintenance projects, four levels of access are set: provincial company administrator, municipal project manager, maintenance team leader, and frontline maintenance personnel. Each level of access strictly restricts its access to and operation scope of sensitive data (such as specific salary costs and detailed fuel consumption of vehicles throughout the province), and ensures the compliance of system operations through a role-based access control (RBAC) model.

[0045] Example 2 In one specific implementation of this invention, by transforming the maintenance elements of the physical world into dynamic data flows within a digital twin system, a highly integrated, real-time feedback, and self-learning closed-loop control architecture is constructed. The implementation of this invention relies on a multi-level hardware matrix consisting of a cloud management platform, vehicle-mounted intelligent terminals, portable IoT terminals, and mobile operation terminals. High-frequency data exchange between each hardware level is achieved through encryption protocols, ensuring the immediacy of control command issuance and the authenticity of on-site data collection.

[0046] In the initial phase of this invention, the first step is to construct a comprehensive digital archive of all aspects of communication tower maintenance. This step forms the foundation for all subsequent intelligent decision-making. The digital archive is deployed in a non-volatile storage cluster on a cloud management platform. Its storage architecture employs a hybrid model combining distributed and relational databases to balance efficient querying of structured data with large-scale storage of unstructured multimedia data. Specifically, the vehicle maintenance file not only records the vehicle's unique identification code but also deeply links it to key engineering parameters such as fuel card number, engine displacement, rated fuel consumption curve, and rated load capacity. These parameters are not statically displayed but serve as fundamental input variables for the fuel consumption model. Simultaneously, the maintenance personnel file achieves refined modeling of personnel attributes through personnel identification coding, encompassing the administrative affiliation of the project team, skill dimensions of professional qualification levels, and security dimensions of biometric data. Furthermore, the low-value consumables BOM and maintenance tool fixed asset databases achieve atomized management of materials, with each item assigned a unique digital characteristic. The communication tower spatial geographic information database maps tens of thousands of towers distributed in discrete geographical locations to a unified spatiotemporal coordinate system through a high-precision three-dimensional coordinate system.

[0047] In achieving closed-loop management of vehicle maintenance and fuel consumption, this invention utilizes an onboard intelligent terminal installed on the maintenance vehicle to capture all data in real time. The onboard intelligent terminal is deeply coupled with the vehicle's internal network via the vehicle controller LAN bus interface, allowing real-time reading of the engine's operating status. Specifically, in terms of hardware implementation, the onboard intelligent terminal integrates a high-performance central processing unit (CPU). This processor uses a built-in edge computing engine to parse the raw CAN bus messages, extracting key parameters including instantaneous vehicle speed, engine speed, throttle opening, and cumulative mileage. Simultaneously, the terminal refreshes the vehicle's dynamic spatial displacement data at a frequency of at least 1Hz using a built-in GPS and BeiDou dual-mode positioning module. For fuel level monitoring, this invention employs a non-contact capacitive level sensor. This sensor inserts a probe into the fuel tank, utilizing the difference in dielectric constant between oil and air to obtain the fuel level in real time. The sensor integrates a temperature compensation algorithm to eliminate volume measurement deviations caused by fuel thermal expansion and contraction, ensuring sampling accuracy errors are controlled within 5%.

[0048] When the vehicle is in operation, the onboard intelligent terminal collects real-time fuel level data, instantaneous vehicle speed, engine speed, and mileage, serializes the data in Protobuf format, encapsulates it into encrypted data packets, and transmits it to the cloud management platform via a 5G or 4G mobile communication network. Upon receiving the data, the cloud management platform immediately calls a fuel consumption authenticity verification algorithm for fitting analysis. The core of the algorithm lies in constructing a dynamic theoretical fuel consumption model. This model derives the theoretical fuel consumption value for the current trip by multiplying the engine displacement, instantaneous speed, and load rate calculated based on GIS. The cloud management platform compares the actual fuel consumption with the theoretical fuel consumption to determine if the deviation exceeds a preset threshold. As a key identification mechanism of this invention, if the system detects abnormal fuel consumption deviation and the GPS trajectory shows that the vehicle is not currently in the preset work path or work area, the system will automatically trigger a first-level warning, determining that the fuel consumption is suspected to be non-project fuel consumption. Furthermore, the cloud management platform obtains real-time fuel card swipe records, including swipe time, fuel volume, and gas station location, through settlement interfaces with third-party petrochemical companies. The system executes spatiotemporal matching logic to determine if the distance between the vehicle's GPS coordinates and the gas station location at the moment of swipe is less than 100 meters, thereby eliminating the risk of fuel card theft from the source.

[0049] In the collaborative management of maintenance personnel's work trajectories and quality, this invention achieves rigid constraints on personnel's on-site behavior through a mobile work terminal. The mobile work terminal activates a high-frequency positioning mode at the task initiation stage and compares it in real-time with the work order geofence issued by the cloud. The geofence is a dynamic area generated by the precise polygonal boundary of the target communication tower's base. Only when the mobile work terminal detects that a personnel's location has entered the geofence can its built-in attendance check-in module be unlocked. This spatial location-based logical control ensures the authenticity of personnel attendance. During the work implementation stage, the terminal guides personnel to operate according to a preset standardized workflow. Each key process, such as tightening cable connectors, sealing outdoor cabinets, and installing new equipment, requires personnel to capture images using a high-definition camera. After these image data are uploaded to the cloud, they are automatically reviewed by a convolutional neural network model deployed in the cloud. The CNN model has undergone reinforcement learning from hundreds of thousands of historical standard work photos, enabling it to accurately identify whether the tightness of connectors meets standards, whether cables are laid straight, and whether signage adheres to ergonomic requirements. The confidence level of the model output is compared with a dynamic threshold, which is periodically recalculated and optimized on the validation set based on field feedback. Any work image falling below the current threshold will trigger the system to immediately issue a rework instruction, thereby eliminating potential quality problems on the construction site.

[0050] To address the refined flow management of low-value consumables and maintenance tools, this invention introduces a full lifecycle tracking mechanism based on RFID radio frequency identification technology. Every piece of maintenance material entering the warehouse, whether it's a feeder cable several meters long or a tiny connector, is affixed with an electronic tag with a unique code. During the requisition phase, maintenance personnel bind the material to the work order using the NFC module of their mobile work terminal or an external barcode scanner. The cloud management platform automatically retrieves the standard usage model from the BOM database based on the type of work order task. For example, for a typical base station expansion maintenance, the system automatically calculates the theoretical quantities of waterproof tape, cable ties, and RFID adapters required. During the return process after the work is completed, the system mandates that personnel scan the remaining materials and return them to the warehouse, using image recognition technology to assist in confirming the remaining quantity. The system calculates the deviation rate between the actual consumption and the standard consumption. Once the deviation rate exceeds the allowable loss rate limit of 5% to 8%, the system automatically generates a material consumption anomaly report, which is used as a key indicator for the maintenance team's cost assessment.

[0051] In terms of tool management, this invention employs an intelligent tool storage box as the on-site monitoring platform. The intelligent tool storage box is equipped with an RFID reader array, operating at a frequency strictly adhering to the national standard of 902MHz-928MHz and supporting the ISO / IEC 18000-6C communication protocol. Through multi-antenna time-division multiplexing technology, the reader array achieves comprehensive scanning of the tools stored within the box, covering the entire box space and enabling batch inventory of dozens of tools within one second. When a core asset, such as a high-precision welding machine or a spectrometer, is detected to have changed from "used" to "not returned" and has exceeded the work order deadline, the system automatically initiates a multi-level supervision process. This process pushes instructions to the maintenance personnel's mobile terminals and the management's monitoring backend, achieving closed-loop supervision from tool issuance to return.

[0052] To achieve multi-dimensional data fusion analysis and intelligent decision-making, the cloud management platform utilizes the Hadoop Distributed File System to store massive amounts of historical operational data and executes complex association rule mining tasks through the Spark computing engine. The system's cost prediction model employs a Long Short-Term Memory (LSTM) network architecture. By deeply training on sequential data such as fuel consumption, labor costs, material losses, and tool depreciation from multiple past operational cycles, combined with external environmental factors such as weather, season, and terrain, it can accurately predict operational expenditure trends over a future period. Simultaneously, the system generates various visual reports, such as personnel man-hour output intensity curves and consumable deviation rate heatmaps, providing managers with intuitive operational insights, enabling them to quickly identify inefficient project nodes and implement targeted interventions.

[0053] Example 3 As a concrete manifestation of the superiority of the technical solution of this invention, the following complete engineering embodiment illustrates its operation process in a practical application scenario. In a communication tower maintenance project in a certain province, 100 representative maintenance teams were selected as implementation targets.

[0054] In this embodiment, the system first equips all 20 engineering vehicles involved in maintenance with onboard intelligent terminals integrating CAN bus reading functionality and capacitive liquid level sensors. During a maintenance task for tower number "Tower-BJ-001," the cloud platform issues a work order to the mobile terminal of maintenance personnel numbered "Staff-05." When this personnel retrieves tools from the intelligent tool storage box, the RFID reader array records the tag ID of "Welding Machine-001" and updates the status to "Out of Stock," linking it to the work order number.

[0055] During driving, the onboard terminal uploads data at a frequency of 1Hz. When the vehicle travels through a 2km stretch of mountain road with an average gradient of 5%, the cloud-based fuel consumption model calculates a theoretical fuel consumption of 0.45 liters based on engine speed, gradient, and load. However, the actual ECU data shows a consumption of 0.46 liters, a deviation rate of 2.2%, which is less than the preset threshold of 10%, and the system determines the fuel consumption is normal. When the vehicle arrives near a gas station for refueling, the system retrieves the fuel card swipe record and GPS trajectory, determining that the distance between the card swipe location and the vehicle's location is less than 20 meters, thus confirming the refueling behavior is compliant.

[0056] After staff member "Staff-05" reached the polygonal boundary of the "Tower-BJ-001" tower base, the mobile work terminal connected to the tower base beacon via Bluetooth. The RSSI value showed a distance of less than 0.3 meters, successfully triggering the on-site check-in. Subsequently, the terminal guided the staff member to complete the cable connector tightening operation and take a photo to upload. At this time, the confidence score of the CNN model output was 0.85, lower than the current dynamic threshold of 0.92. The system automatically locked the work order and pushed a rework instruction: "The connector is not tightened to the standard color mark position. Please tighten it again and take a photo." After the staff member performed the rework and took another photo, the model confidence score increased to 0.96, and the work order automatically entered the "pending acceptance" state.

[0057] After the task was completed, the personnel returned the tools. The RFID reader read the "Welding Machine-001" tag again, and the status returned to "In Stock". The system recorded the tool usage time for this task and turned off the supervision timer. In the material return process, the system compared the BOM standard usage with the actual consumption and found that the waterproof tape consumption deviation rate was 6%, which was lower than the allowable 8%, and no warning was triggered.

[0058] All data (fuel consumption, trajectory, images, materials, tool status) is stored on the Hadoop platform. At the end of the month, the LSTM cost prediction model, combined with external factors such as the month's weather (rainy) and failure frequency (high), predicts that the maintenance cost in the area will increase by 5.2% in the following month, and generates a report to prompt managers to replenish the inventory of waterproof consumables in advance. The EOQ model automatically calculates the optimal replenishment batch based on the predicted consumption rate and pushes procurement recommendations.

[0059] Comparative Example 1 employs a traditional maintenance and control model, which involves manually recording vehicle mileage, manually filling out paper refueling slips, and manually sending back photos for quality inspection. Material management also utilizes simple manual entry and exit registration. The same external environment (equal number of towers, equal number of maintenance personnel and vehicles) is used for the same operating cycle as Example 1.

[0060] By summarizing and analyzing the data from Example 1 and Comparative Example 1 over a three-month operating period, the comparison results shown in Table 1 below were obtained: Table 1: Performance Comparison of Intelligent Control System and Traditional Control Mode As can be clearly observed from the data in Table 1, the intelligent management and control method and system provided by this invention demonstrate significant technical advantages in controlling abnormal fuel consumption, improving operational quality, and reducing material losses. In particular, the accuracy rate in determining non-project fuel consumption has increased dramatically from 35% in the traditional manual mode to 96.5%, directly proving the effectiveness of the closed-loop management and control mechanism based on fuel consumption models and spatiotemporal matching logic. Simultaneously, the overall maintenance cost of a single tower has been reduced by 30%, bringing considerable economic benefits to communication tower operation and maintenance companies.

[0061] In a further implementation of this invention, the data fusion technology of the vehicle-mounted intelligent terminal plays a crucial role. Since communication towers are often located in mountainous areas with complex terrain or densely populated urban core areas, GPS signals are highly susceptible to drift due to multipath effects. This invention introduces a Kalman filter algorithm into the edge computing module of the vehicle-mounted terminal. This algorithm uses the GPS positioning output as the observation vector and the dead reckoning system output based on CAN bus speed and heading angle as the state prediction vector. By establishing a state-space model and adjusting the gain coefficient in real time, the continuity of the vehicle trajectory can be maintained even in signal blind spots such as tunnels and high-rise buildings. Actual measurements show that the trajectory deviation remains consistently within 2 meters.

[0062] At the system communication layer, to cope with the extremely unstable network conditions in outdoor environments, this invention employs the lightweight MQTT protocol between the cloud and each terminal. This protocol, based on a publish / subscribe model, has significantly lower header overhead compared to the traditional HTTP protocol. Combined with the TLS 1.3 secure transport layer protocol, the system ensures data transmission security while keeping the delay in issuing control commands within 3 seconds. The mobile work terminal further supports an offline caching mechanism. When maintenance personnel enter completely signal-free basements or remote mountain tower sites for work, all operational data and images are temporarily stored in a local encrypted database. Once the terminal detects that the network connection has been restored, the system will initiate a breakpoint resume program, using asynchronous I / O streams to synchronize data to the cloud, ensuring that the complete link of maintenance data is not interrupted due to physical environment limitations.

[0063] To address the intelligent evolution of fuel cost verification algorithms, this invention integrates an anomaly pattern recognition submodule based on the random forest algorithm into a cloud platform. This module extracts 24 feature dimensions from massive historical data, including refueling time dispersion, linear correlation between refueling amount and mileage growth, and abnormal deviation of vehicle trajectory at night. By constructing multiple decision trees and performing integrated voting, the system can accurately identify various hidden illegal refueling behaviors, such as multiple small-amount cash withdrawals and refueling by unauthorized vehicles. This machine learning-based discrimination model overcomes the limitations of traditional fixed threshold judgments, giving the control system strong adversarial and evolutionary capabilities.

[0064] Regarding the implementation of the EOQ model in materials management, the cloud-based management platform automatically calculates the optimal inventory level for each secondary warehouse by analyzing historical consumption rates and logistics procurement lead times in different regions. For example, during the rainy summer months, the system automatically increases the safety stock threshold for vulnerable materials such as surge arresters and waterproof sleeves. When the actual inventory reaches the warning line, the system no longer relies on manual reporting and instead directly pushes procurement instructions to the supply chain system, achieving automated connection from operational needs to material supply.

[0065] Finally, the system also features in-depth design in terms of access control and security. A four-level access control system based on the RBAC model ensures that all operations, from provincial company administrators to frontline maintenance personnel, are within authorized limits. All core operation logs, including modifications to vehicle parameters and manual reconciliation of material consumption deviations, are recorded in an immutable audit trail file. In terms of physical hardware design, both the vehicle-mounted intelligent terminal and the intelligent tool storage box adopt a rugged design. Their shell materials possess excellent electromagnetic interference (EMI) resistance, capable of withstanding the effects of high-power radio frequency radiation around communication towers. Their protection level reaches IP67, ensuring long-term reliable operation in harsh outdoor environments.

[0066] In summary, this invention, by constructing a comprehensive digital archive and collaboratively utilizing vehicle-mounted sensing, mobile internet, computer vision, and big data analytics technologies, has achieved a multi-dimensional intelligent management and control system encompassing resource scheduling, process monitoring, quality acceptance, and cost optimization. The implementation of this system not only addresses the core pain points of information asymmetry and severe resource loss in traditional maintenance models but also enhances the professionalism and scientific rigor of communication infrastructure operation and maintenance through a data-driven approach.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0068] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent management and control method for the maintenance of communication towers, characterized in that: Includes the following steps: Step 1: Construct a full-element digital archive, which should include at least vehicle maintenance records, maintenance personnel records, a list of low-value consumables, a fixed asset database for maintenance tools, and a spatial geographic information database for communication towers. Step 2: Real-time collection of maintenance vehicle operation status data, spatial location data, and fuel consumption data; real-time collection of maintenance personnel location data and work process image data; real-time collection of maintenance tools and low-value consumables issuance and return status data. Step 3: Based on the collected data, establish at least one closed-loop control logic among the following: the correlation verification logic between vehicle operation trajectory and fuel consumption, the collaborative verification logic between personnel operation location and operation quality, the deviation analysis logic between material consumption and standard usage, and the matching and supervision logic between tool circulation status and work order time limit. Step 4: After fusing the multi-dimensional data, input it into the cost prediction model and output the operation and maintenance cost prediction results and visualization reports.

2. The intelligent management and control method for communication tower maintenance according to claim 1, characterized in that: The logic for verifying the correlation between vehicle trajectory and fuel consumption includes: The actual fuel consumption can be obtained by acquiring the cumulative fuel injection quantity calculated by the engine electronic control unit via the vehicle controller area network bus, or by acquiring the fuel level height via a fuel level sensor inserted inside the fuel tank and converting it into fuel volume data using a fuel tank geometric model mapping table. ; Based on engine displacement Instantaneous speed Instantaneous speed Calculated engine load rate Approved load capacity and road surface slope coefficient Calculate the instantaneous theoretical fuel consumption rate: in To comprehensively correct the factors, and for The theoretical fuel consumption is calculated by accumulating points over the duration of the trip. ;Will With preset threshold In comparison, when the threshold is exceeded and the vehicle is located in a non-preset operating area or non-operating period, an abnormal fuel consumption warning is triggered.

3. The intelligent management and control method for communication tower maintenance according to claim 1, characterized in that: The logic for verifying the correlation between vehicle trajectory and fuel consumption also includes verification of refueling compliance: Get fuel card transaction records, including the transaction time. Geographic coordinates of the gas station ; Extract the real-time GPS location coordinates of the vehicle at the time of card swipe. ;when At that time, it was determined to be an illegal card swipe for refueling, among which This is a preset spatial matching threshold.

4. The intelligent management and control method for communication tower maintenance according to claim 1, characterized in that: The collaborative verification logic for personnel work location and work quality includes: Geofencing for task orders is generated based on the polygonal footprint of the communication tower base or the boundary of the standard work area. Once the maintenance personnel's location enters the geofence, the on-site check-in function is activated and they are guided to execute a standardized work process; Images of key processes are captured, and a convolutional neural network model is used to identify compliance of the images and output confidence scores. ; Set dynamic confidence threshold ,when The initial value of locking the work order progress and pushing the rework instruction is determined by the precision and recall curves and periodically optimized based on on-site feedback.

5. The intelligent management and control method for communication tower maintenance according to claim 1, characterized in that: The deviation analysis logic between material consumption and standard usage includes: Attach RFID tags or QR code tags to low-value consumables, and scan the tags during the requisition process to establish a link between the material and the work order; Calculate standard usage based on work order type and bill of materials database. Confirm the actual consumption during the receipt process. Calculate the consumption deviation rate: when Exceeding the preset allowable deviation rate When this occurs, it is determined to be an abnormal material consumption and recorded in the cost deviation report.

6. The intelligent management and control method for communication tower maintenance according to claim 1, characterized in that: The matching and supervision logic between tool workflow status and work order deadlines includes: To maintain the RFID tags attached to the tools, the RFID reader array embedded in the tool storage box is used to periodically scan the tool tags in the box to monitor the tool's storage status in real time. Record tool outbound time and related work orders; When the preset buffer time for the issued tool is detected after the work order ends. If the item is not returned to the warehouse, a reminder instruction will be sent to the maintenance personnel; if more than 2... If it is still not returned, then copy it to the administrator interface.

7. The intelligent management and control method for communication tower maintenance according to claim 1, characterized in that: The cost prediction model adopts a long short-term memory network architecture. Its input feature sequence includes internal operation and maintenance variables and external environmental factors. The internal operation and maintenance variables include historical period fuel consumption per 100 kilometers, per capita labor hours output, consumable deviation rate, tool turnover rate and vehicle maintenance frequency. The external environmental factors include weather conditions, seasonal failure frequency and terrain complexity. The model learns the nonlinear mapping relationship between time series features and operation and maintenance costs through the forget gate, input gate and output gate structure of the long short-term memory network, and outputs the predicted value of operation and maintenance expenditure for the next period.

8. The intelligent control system for communication tower maintenance according to claim 1, characterized in that: Kalman filtering is used to fuse vehicle positioning data from multiple sources, with GPS positioning output used as the observation vector. The speed obtained based on the controller area network bus and heading angle The pose data calculated using the dead reckoning formula is used as the state prediction vector. : in, This represents the state estimate at the previous time step. Indicates the instantaneous velocity at the previous moment. Indicates the sampling time interval. This indicates the heading angle at the previous moment. The cosine value of the heading angle is represented by a Kalman filter, which dynamically adjusts the weights to correct the state estimate.

9. An intelligent management and control system for the maintenance of communication towers, used to implement the method described in any one of claims 1 to 8, characterized in that: include: The cloud management platform is used to store all digital archives and run the fuel cost intelligent verification module, operation quality monitoring module, and resource scheduling module. The vehicle-mounted intelligent terminal, installed in the maintenance vehicle, integrates a positioning module, a controller area network bus communication interface, and a liquid level sensor interface, and is used to collect vehicle operation data and perform local preprocessing. The mobile operation terminal, configured for maintenance personnel, integrates a positioning module, camera, biometric module and short-range wireless communication module, and is used for task reception, on-site operation interaction, quality data collection and material scanning and requisition. The intelligent tool storage box is equipped with an array of radio frequency identification readers to enable real-time monitoring of maintenance tool assets and automated retrieval and return of certificates.

10. The intelligent control system for communication tower maintenance according to claim 9, characterized in that: The cloud management platform communicates with the vehicle-mounted intelligent terminal, mobile operation terminal, and intelligent tool storage box using a lightweight message queue telemetry transmission protocol. Data packets are serialized using a protocol buffer format. The mobile operation terminal supports an offline caching mechanism, which encrypts and stores operation data locally in communication dead zones. After the network is restored, it is synchronized to the cloud management platform using breakpoint resume technology.

Images