A BIM-based tower full life cycle management method, system, device and storage medium

Abstract

The application discloses a kind of based on BIM's pole tower full life cycle management method, system, equipment and storage medium, comprising: acquisition pole tower full life cycle multidimensional data, is formed standardized data system by unified standard preprocessing.And parallelly constructs state prediction intelligent model and BIM management model.According to prediction result, dynamic trigger hierarchical targeted data acquisition, supplement key information.Fusion prediction and targeted data to BIM model, use multi-objective optimization algorithm to generate decision instruction, and scheduling execution unit will virtual guide superimposed to target physical object.For each base pole tower allocation unique ID, establish the digital archives of associated full-dimensional data and record evidence.Collect feedback data to build closed-loop optimization loop, drive model iteration.Can shorten decision cycle, improve operation success rate, realize comprehensive cost saving.

Description

Technical Field

[0001] This invention relates to the technical field of digital management of power engineering, and in particular to a BIM-based method, system, equipment and storage medium for the full life cycle management of power poles and towers. Background Technology

[0002] Transmission towers, as core infrastructure of the power system, are crucial to the safety and stability of the power grid. With the expansion of the power grid coverage, the environment in which towers operate is becoming increasingly complex. Various extreme environments and long-term losses continuously threaten structural safety, and traditional manual inspection and phased data management models are no longer sufficient to meet the demands of modern power grid's refined and intelligent management. Against this backdrop, Building Information Modeling (BIM) technology, with its advantages of visualization, parametric analysis, and information integration, offers the possibility of digital management of towers throughout their entire lifecycle. The power engineering field has gradually begun related explorations, such as using RFID technology for tower resource management, combining BIM and GIS technologies to assist in site selection and construction planning, and leveraging drones and sensors to improve the efficiency of operation and maintenance data collection.

[0003] Existing pole management technologies and related full lifecycle management methods lack data integration capabilities. Data standards at different stages are inconsistent and poorly correlated, making it difficult to form complete digital identity files for poles. Furthermore, data collection lacks specificity, failing to achieve risk-driven, precise data collection, resulting in low resource utilization efficiency. Moreover, the depth of technology integration is insufficient, with BIM models and intelligent algorithms disconnected, leading to weak decision support capabilities. In addition, the management mechanism is imperfect, lacking full-process traceability and dynamic iteration capabilities, and exhibiting insufficient resilience in extreme scenarios. Therefore, there is an urgent need for a management method that can achieve full-cycle data connectivity, accurate risk prediction, and intelligent and efficient decision-making. Summary of the Invention

[0004] In view of the aforementioned existing problems, this invention is proposed. Therefore, this invention provides a BIM-based method, system, device, and storage medium for the full lifecycle management of power poles and towers, addressing the problems of inconsistent data standards and poor correlation leading to missing digital identity files, and insufficient risk-driven data collection resulting in low resource efficiency in existing power pole and tower management technologies.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, embodiments of the present invention provide a BIM-based method for full lifecycle management of poles and towers, including: collecting full-dimensional data of the pole and tower lifecycle, and preprocessing the full-dimensional data based on a unified data standard to form a standardized full-process data system; Based on standardized end-process data, an intelligent prediction model for state prediction and a BIM pole management model for management simulation are constructed in parallel. Based on the prediction results output by the intelligent prediction model, a hierarchical targeted data collection command is dynamically triggered to supplement data collection for key scenes and key parts. The prediction results are fused with targeted supplementary data and input into the building information model tower management model. Decision instructions are generated through a multi-objective optimization algorithm, and the execution unit is scheduled to overlay the virtual operation guidance generated based on the decision onto the target physical object. Assign a unique physical ID to each tower, establish a digital identity file that links data from all dimensions of its entire lifecycle, and preserve records of key operations and status changes. The feedback data generated after the decision is executed is collected, a closed-loop optimization loop is constructed, and the intelligent prediction model and the BIM pole management model are continuously iterated and optimized.

[0006] As a preferred embodiment of the BIM-based full lifecycle management method for power poles described in this invention, the method includes: collecting full-dimensional data of the power pole's lifecycle, and preprocessing the full-dimensional data based on a unified data standard to form a standardized full-process data system, including: The comprehensive data includes production phase data, installation phase data, and operation phase data. By adopting a fusion standard based on Building Information Modeling (BIM) data standards and power industry tower operation and maintenance data specifications, multi-source heterogeneous data is converted into a unified data format, and unstructured data is encapsulated using XML. The multi-source heterogeneous data is then verified and cleaned to form a standardized end-to-end data system.

[0007] As a preferred embodiment of the BIM-based pole and tower full lifecycle management method described in this invention, the method comprises: constructing, in parallel, an intelligent prediction model for state prediction and a BIM pole and tower management model for management simulation based on standardized full-process data, including: Construct a smart prediction model for pole status and a BIM-based pole management model based on a dual machine learning framework; The intelligent prediction model integrates LSTM and random forest algorithms, reduces parameter interference through Neyman orthogonality, takes operational phase data as input, and outputs tilt, stress warning, and remaining lifetime prediction based on the Weibull proportional risk model. The BIM management model is a digital twin that integrates a finite element analysis engine, an operation and maintenance knowledge rule base, and a GIS module, and automatically matches management solutions based on terrain slope and soil bearing capacity. The dual-model linkage is achieved through a parameter server architecture under the federated learning framework. The prediction model uploads the risk confidence and prediction parameters as intermediate parameters, while the management model sends out the simulation verification results and rule matching degree as feedback parameters. After encryption and aggregation, both models are updated locally, and the prediction error and simulation results are used as a joint loss function for iterative optimization.

[0008] As a preferred embodiment of the BIM-based full life cycle management method for poles and towers described in this invention, the method includes: dynamically triggering hierarchical targeted data collection instructions based on the prediction results output by the intelligent prediction model to supplement data collection for key scenarios and key parts, including: executing a three-level targeted data collection triggering mechanism, wherein the triggering level is jointly determined by the risk confidence level output by the intelligent prediction model, the predicted values ​​of key parameters, and the historical failure mode matching results. When a low-risk level is triggered, the basic sensors are activated to perform routine data collection. When a medium-risk level is triggered, additional drones will be deployed for high-definition inspection and AI image recognition, and 3D point cloud data of the designated area will be generated. When a high-risk level is triggered, the fiber optic stress sensor network deployed on the main structure of the tower is activated for real-time monitoring, and high-precision on-site detection equipment carrying infrared thermal imagers is dispatched. All data collection instructions are pushed to the corresponding terminals via 5G network slicing technology. The collected data is encrypted and directly transmitted to the BIM model database after being associated with the unique ID and timestamp of the tower on the edge side.

[0009] As a preferred embodiment of the BIM-based pole and tower full lifecycle management method described in this invention, the following steps are included: fusing the prediction results with targeted supplementary data and inputting them into the building information model pole and tower management model; generating decision instructions through a multi-objective optimization algorithm; and overlaying the virtual operations of the decision instructions onto the target physical object, including: A multi-objective optimization algorithm is adopted, with safety, economy and reliability as optimization objectives, inputting risk level and real-time data, and outputting differentiated decisions; For high-risk decisions, generate structured maintenance work orders that include BIM model sectional views, material code lists of components to be replaced, and precise process parameters; The decision-making instructions are pushed to the AR terminals of the operation and maintenance personnel through the production management system, and the virtual operation guidance is superimposed on the real tower components using 3D registration technology; During the maintenance process, key step data is transmitted back through the terminal to drive the digital twin to perform synchronous simulation and effect verification, and the verification results are fed back to the multi-objective optimization algorithm as decision quality assessment parameters.

[0010] As a preferred embodiment of the BIM-based full lifecycle management method for poles and towers described in this invention, the method includes: assigning a unique physical ID to each pole and tower, establishing a digital identity file associated with its data across all dimensions throughout its lifecycle, and storing records of key operations and status changes, including: Each tower is assigned a unique physical ID, and based on ontology semantic mapping technology, a full lifecycle digital identity file is constructed that links its design drawings, production batches, construction and acceptance records, operation and maintenance data, forecast reports and decommissioning assessments. Hash operations are performed on key status changes, major maintenance operations, and parameter correction records exceeding preset thresholds in the digital identity file. The hash values ​​are stored on the blockchain as evidence, and smart contracts ensure that the operation logs are tamper-proof and fully traceable.

[0011] As a preferred embodiment of the BIM-based pole and tower lifecycle management method described in this invention, the method includes: collecting feedback data generated after the decision execution, constructing a closed-loop optimization circuit, and driving the intelligent prediction model and the BIM pole and tower management model to continuously iterate and optimize, including: Construct a closed-loop optimization loop that takes as input tower status change data, resource consumption data, and cost-benefit data after decision execution, and outputs the loss function of the prediction model and the parameter adjustment of the objective function of the management model; The difference between the actual operation and maintenance effect and the expected effect is used as a reinforcement learning reward signal to dynamically update the weight of the prediction model; at the same time, the confidence of the finite element analysis parameters and operation and maintenance rules in the BIM management model is adjusted through backpropagation by using the difference between the decision simulation verification results and the on-site execution results. After extreme scenarios occur, transfer learning techniques are used to fine-tune the key layers of the pre-trained model by using new scenario data as the source domain.

[0012] Secondly, the present invention provides a BIM-based pole and tower full life cycle management system, including: a data acquisition module, used to collect full-dimensional data of the pole and tower life cycle, and preprocess the full-dimensional data based on a unified data standard to form a standardized full-process data system; The dual-model parallel construction module is used to build, in parallel, an intelligent prediction model for state prediction and a BIM pole management model for management simulation based on standardized full-process data. The hierarchical data acquisition triggering module is used to dynamically trigger hierarchical targeted data acquisition instructions based on the prediction results output by the intelligent prediction model, so as to supplement data acquisition for key scenes and key parts. The virtual-real fusion decision module is used to fuse the prediction results with targeted supplementary data and input them into the building information model tower management model. It generates decision instructions through a multi-objective optimization algorithm and schedules the execution unit to overlay the virtual operation guidance generated based on the decision onto the target physical object. The digital identity storage module is used to assign a unique physical ID to each tower, establish a digital identity file that links data from all dimensions of its entire lifecycle, and store key operation and status change records. The iterative optimization module is used to collect feedback data generated after the decision is executed, construct a closed-loop optimization circuit, and drive the intelligent prediction model and the BIM pole management model to continuously iterate and optimize.

[0013] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the BIM-based pole and tower full life cycle management method are implemented.

[0014] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the BIM-based pole and tower full lifecycle management method.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention replaces the continuous, comprehensive data collection mode with intelligent prediction and preliminary assessment. Based on historical operational and environmental data, it predicts risk points and critical periods, dynamically triggering tiered data collection commands to drive high-cost resources such as drones to perform precise inspections on demand, reducing data collection, transmission, and storage costs from the source. Simultaneously, it constructs an integrated closed loop of prediction, decision-making, and execution: prediction results directly drive the Building Information Model to generate targeted response plans, and use augmented reality and other means to guide on-site operations, replacing the traditional process of first collecting data, then manually analyzing it, and finally formulating a general plan. This shortens the decision-making cycle, improves the first-time success rate of maintenance, and comprehensively reduces labor, time, and material costs.

[0016] This invention integrates multi-source, heterogeneous data from the entire lifecycle of power poles into standardized, interconnected digital identity archives by establishing unified data standards and processing procedures. Based on this, blockchain technology is introduced to immutably record key operations and status changes, ensuring the integrity and credibility of the archives and resolving data silos and trust issues. The resulting digital twin of the power pole provides a reliable data foundation for risk prediction, accurate decision-making, residual value assessment, and liability determination, avoiding duplicate investments and disputes caused by information distortion or difficulties in traceability.

[0017] This invention incorporates self-optimization and emergency response mechanisms, enabling the system to continuously evolve. Utilizing feedback data after each decision execution, reinforcement learning and other methods drive iterative optimization of the prediction and management models, achieving continuous improvement in prediction accuracy and cost control effectiveness with use. Simultaneously, an emergency mechanism for extreme scenarios is established to ensure system stability and rapid recovery under abnormal conditions. Furthermore, the application of BIM models is extended to the decommissioning stage for residual value assessment and green dismantling planning, achieving precise management and control throughout the entire lifecycle from construction to decommissioning and regeneration. Thus, an intelligent system with self-learning capabilities, resilience against risks, and continuous cost optimization capabilities is constructed, ultimately minimizing the total cost of the asset's entire lifecycle. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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. Wherein: Figure 1 This is a schematic diagram of the process flow of a BIM-based full life cycle management method for poles and towers according to an embodiment of the present invention; Figure 2 This is a schematic diagram showing the prediction accuracy comparison curve of a BIM-based pole and tower full life cycle management method according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the prediction relative error comparison curve of a BIM-based pole and tower full life cycle management method according to an embodiment of the present invention; Figure 4 This is a bar chart illustrating the multi-scenario confidence level of a BIM-based full lifecycle management method for poles and towers, as described in one embodiment of the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0020] Example 1, referring to Figure 1 As one embodiment of the present invention, this embodiment provides a BIM-based method for the full lifecycle management of power poles, including: S100: Collects full-dimensional data of the tower's life cycle and preprocesses the full-dimensional data based on a unified data standard to form a standardized full-process data system; S200: Based on standardized full-process data, it constructs in parallel an intelligent prediction model for state prediction and a BIM pole management model for management simulation. S300: Based on the prediction results output by the intelligent prediction model, it dynamically triggers hierarchical targeted data collection instructions to supplement data collection for key scenarios and key parts. S400: The prediction results are fused with the targeted supplementary data and input into the building information model tower management model. Decision instructions are generated through a multi-objective optimization algorithm, and the execution unit is scheduled to overlay the virtual operation guidance generated based on the decision onto the target physical object. S500: Assign a unique physical ID to each tower, establish a digital identity file that links data from all dimensions of its entire lifecycle, and preserve records of key operations and status changes. S600: Collects feedback data generated after decision execution, constructs a closed-loop optimization circuit, and drives the intelligent prediction model and BIM pole management model to continuously iterate and optimize.

[0021] It should be noted that existing pole management technologies and related full lifecycle management methods lack data integration capabilities. Data standards at different stages are inconsistent and poorly correlated, making it difficult to form complete digital identity files for poles. Furthermore, data collection lacks specificity, failing to achieve risk-driven precision collection, resulting in low resource utilization efficiency. Moreover, the depth of technology integration is insufficient, with BIM models and intelligent algorithms disconnected, leading to weak decision support capabilities. In addition, the management mechanism is imperfect, lacking full-process traceability and dynamic iteration capabilities, and exhibiting insufficient resilience in extreme scenarios. This invention adopts an intelligent prediction-first model, analyzing data to predict risks and triggering the precise, on-demand deployment of resources such as drones, reducing data collection and processing costs from the source. Simultaneously, relying on a closed loop of prediction, decision-making, and execution, the prediction results directly generate BIM solutions and use AR to guide on-site operations, skipping the traditional manual analysis stage and focusing resources on key issues, thereby significantly shortening the decision-making cycle, improving the success rate of operation and maintenance, and achieving overall cost savings.

[0022] In this embodiment of the invention, step S100 involves collecting full-dimensional data of the tower's lifecycle and preprocessing the full-dimensional data based on a unified data standard to form a standardized end-to-end data system, including: The comprehensive data includes data from the production phase, installation phase, and operation phase. Specifically, the production stage data includes the material composition, processing parameters, material specifications, and strength parameters of the tower components. The recording accuracy of the steel yield strength is no less than ±5MPa, the concrete compressive strength data is retained to two decimal places, and the production batch and quality inspection report. During the installation phase, data such as the terrain slope and soil bearing capacity of the installation area are obtained through GIS technology, with the soil bearing capacity accuracy not less than 0.01MPa, the distribution of underground pipelines, and meteorological statistics and real-time monitoring data of the area in recent years are collected simultaneously, with the sampling interval for wind speed and ice thickness not exceeding 15 minutes, as well as the geographic coordinates and foundation settlement monitoring data during the installation phase. The operational data includes the tower's tilt angle, stress and strain, ambient temperature and humidity, monitoring data on geological activity around the tower, and monitoring data on tree growth along the transmission corridor.

[0023] In step S100 of this embodiment of the invention, a fusion standard based on building information modeling data standards and power industry tower operation and maintenance data specifications is adopted to convert multi-source heterogeneous data into a unified data format and encapsulate unstructured data using XML; and the multi-source heterogeneous data is verified and cleaned to form a standardized full-process data system.

[0024] In one feasible approach, a fusion standard based on the ISO 19650 Building Information Modeling (BIM) data standard and the power industry tower operation and maintenance data specification is adopted to convert multi-source heterogeneous data into a unified JSON-LD format and encapsulate unstructured data using XML. MD5 verification is used to ensure data integrity, and regular expressions are used to remove abnormal characters, box plots are used to identify and correct outliers, and time series consistency checks are used to eliminate data jumps. Abnormal data with errors exceeding ±3% is removed, forming a unified data system covering the entire process of production, installation, operation and maintenance, and decommissioning.

[0025] In this embodiment of the invention, step S200 involves constructing, in parallel, an intelligent prediction model for state prediction and a BIM tower management model for management simulation based on standardized full-process data, including: Construct a smart prediction model for pole status and a BIM-based pole management model based on a dual machine learning framework; The intelligent prediction model integrates LSTM and random forest algorithms, reduces parameter interference through Neyman orthogonality, takes operational phase data as input, and outputs tilt and stress warnings as well as remaining lifetime predictions based on the Weibull proportional risk model. The BIM management model is a digital twin that integrates a finite element analysis engine, an operation and maintenance knowledge rule base, and a GIS module. It automatically matches management solutions based on terrain slope and soil bearing capacity. In one possible implementation, a pole status intelligent prediction model based on a dual machine learning framework and a BIM pole management model with LOD400 accuracy are constructed. The intelligent prediction model integrates LSTM and random forest algorithms, reduces parameter interference through Neyman orthogonality, takes historical and environmental data as input, and outputs tilt angle, stress warning, and remaining life prediction based on the Weibull proportional hazards model; wherein, in this embodiment of the invention, the tilt angle threshold is set to ±0.1° and the stress threshold is set to ±1MPa.

[0026] The BIM management model is a digital twin that integrates a finite element analysis engine, an operation and maintenance knowledge rule base, and a GIS module. It can automatically match reinforced designs or strengthening schemes based on terrain slope >25° and soil bearing capacity <0.15MPa.

[0027] In step S200 of this embodiment of the invention, dual-model linkage is achieved through a parameter server architecture under the federated learning framework. The prediction model uploads the risk confidence and prediction parameters as intermediate parameters, while the management model sends the simulation verification results and rule matching degree as feedback parameters. After encryption and aggregation, both models are updated locally, and the prediction error and simulation results are used as a joint loss function for iterative optimization to improve the prediction accuracy.

[0028] Furthermore, the BIM tower management model supports AR interaction. Maintenance personnel can scan the physical ID of the tower with their mobile devices to overlay virtual conductor sag standard lines or component maintenance standards onto the real-time image from the equipment's camera. The BIM tower management model automatically compares the deviation between the on-site conditions and the standard model through image recognition and automatically generates adjustment or maintenance work orders.

[0029] Furthermore, BIM model-based decommissioning management includes: The BIM management model calls upon structural performance degradation data, material corrosion data, and historical load spectra from the digital identity archives of the tower throughout its entire life cycle, calculates residual values ​​through a residual strength assessment algorithm, and generates a stepped dismantling scheme. The disassembled components are classified according to the extrapolation results of their corrosion rate models; For concrete foundations, the crushing and screening process is planned based on the strength grade and aggregate information in the BIM model.

[0030] In this embodiment of the invention, step S300 dynamically triggers a hierarchical targeted data collection instruction based on the prediction results output by the intelligent prediction model, so as to collect supplementary data for key scenarios and key parts. This includes: executing a three-level targeted data collection triggering mechanism, where the triggering level is jointly determined by the risk confidence level output by the intelligent prediction model, the predicted values ​​of key parameters, and the historical failure mode matching results. When a low-risk level is triggered, the basic sensors are activated to perform routine data collection. When a medium-risk level is triggered, additional drones will be deployed for high-definition inspection and AI image recognition, and 3D point cloud data of the designated area will be generated. When a high-risk level is triggered, the fiber optic stress sensor network deployed on the main structure of the tower is activated for real-time monitoring, and high-precision on-site detection equipment carrying infrared thermal imagers is dispatched. All data collection instructions are pushed to the corresponding terminals via 5G network slicing technology. The collected data is encrypted and directly transmitted to the BIM model database after being associated with the unique ID and timestamp of the tower on the edge side.

[0031] Specifically, when the risk confidence level is ≥85% and all predicted parameters do not exceed 80% of the threshold, a low-risk level is triggered, and the basic sensor is activated to perform routine data collection at a frequency of ≤24 hours / time. When the predicted ice thickness is ≥30mm, or the predicted wind speed is ≥10, or the predicted key stress exceeds 80% of the threshold, a medium-risk level is triggered. Drones with a positioning accuracy of ±2cm are dispatched to conduct high-definition inspections, and AI image recognition based on deep learning with an accuracy of ≥96% is used to generate three-dimensional point cloud data for the specified location. When the earthquake intensity monitoring value is ≥8, or any monitoring data exceeds 120% of the historical statistical extreme value, a high-risk level is triggered, and the fiber optic stress sensors deployed on the main structure of the tower are activated to conduct real-time monitoring at a network frequency of ≥1 time / minute, and high-precision on-site detection equipment carrying infrared thermal imagers is dispatched.

[0032] In this embodiment of the invention, step S400 involves fusing the prediction results with targeted supplementary data and inputting them into the building information model (BIM) tower management model. A multi-objective optimization algorithm is used to generate decision instructions, and the virtual operations of these decision instructions are superimposed onto the target physical object. This includes: A multi-objective optimization algorithm is adopted, with safety, economy and reliability as optimization objectives, inputting risk level and real-time data, and outputting differentiated decisions; For high-risk decisions, generate structured maintenance work orders that include BIM model sectional views, material code lists of components to be replaced, and precise process parameters; The decision-making instructions are pushed to the AR terminals of the operation and maintenance personnel through the production management system, and the virtual operation guidance is superimposed on the real tower components using 3D registration technology; During the maintenance process, key step data is transmitted back through the terminal to drive the digital twin to perform synchronous simulation and effect verification, and the verification results are fed back to the multi-objective optimization algorithm as decision quality assessment parameters.

[0033] Specifically, a multi-objective optimization algorithm is adopted, with safety (target ≥99.9%), economy (cost ≤110% of budget) and reliability as optimization objectives. The risk level and real-time data are input, and differentiated decisions are output. For high-risk decisions, a structured maintenance work order containing BIM model sectional views, material code list of components to be replaced, and precise process parameters is automatically generated. For example, the bolt tightening torque requirement is ±5 N·m of the design value, and the foundation verticality deviation is ≤1‰.

[0034] Instructions are pushed to the AR terminals of maintenance personnel through the production management system, and virtual operation guidance is superimposed on real tower components using 3D registration technology. Key step data in the maintenance process, such as actual torque values ​​and replacement part serial numbers, are automatically transmitted back through the terminal and drive the digital twin to perform synchronous simulation and effect verification. The verification results are fed back to the optimization algorithm as decision quality assessment parameters.

[0035] In this embodiment of the invention, step S500 assigns a unique physical ID to each tower, establishes a digital identity file associated with its data across all dimensions throughout its entire lifecycle, and stores evidence of key operations and status change records, including: Each tower is assigned a unique physical ID, and based on ontology semantic mapping technology, a full lifecycle digital identity file is constructed that links its design drawings, production batches, construction and acceptance records, operation and maintenance data, forecast reports and decommissioning assessments. Hash operations are performed on key status changes, major maintenance operations, and parameter correction records exceeding preset thresholds in the digital identity file. The hash values ​​are stored on the blockchain as evidence, and smart contracts ensure that the operation logs are tamper-proof and fully traceable.

[0036] In one possible implementation, a unique physical ID is assigned to each tower, and a full lifecycle digital identity file is constructed based on ontology semantic mapping technology, linking its design drawings, production batches, construction and acceptance records (e.g., galvanized layer thickness ≥86μm), and maintenance data (e.g., average annual corrosion rate <0.03mm / a), prediction reports, and decommissioning assessments. Using blockchain technology based on a consortium blockchain, hash calculations are performed on key status changes, major maintenance operations, and parameter correction records exceeding ±3% in the digital identity file. The hash values ​​are stored on the blockchain, and smart contracts ensure that the operation logs are tamper-proof and fully traceable, enabling root cause tracing of faults within 3 minutes via mobile ID scanning.

[0037] In this embodiment of the invention, step S600 involves collecting feedback data generated after decision execution, constructing a closed-loop optimization circuit, and driving continuous iterative optimization of the intelligent prediction model and the BIM pole management model, including: Construct a closed-loop optimization loop that takes as input tower status change data, resource consumption data, and cost-benefit data after decision execution, and outputs the loss function of the prediction model and the parameter adjustment of the objective function of the management model; The difference between the actual operation and maintenance effect and the expected effect is used as a reinforcement learning reward signal to dynamically update the weight of the prediction model; at the same time, the confidence of the finite element analysis parameters and operation and maintenance rules in the BIM management model is adjusted through backpropagation by using the difference between the decision simulation verification results and the on-site execution results. After extreme scenarios occur, transfer learning techniques are used to fine-tune the key layers of the pre-trained model by using new scenario data as the source domain.

[0038] Furthermore, this includes emergency response mechanisms for extreme scenarios: Establish an emergency response plan database covering various extreme disaster scenarios; Establish a tiered early warning mechanism to automatically trigger the corresponding level of early warning when monitoring data is abnormal, prediction confidence is too low, or multiple risks occur simultaneously; The early warning response includes automatically initiating drone disaster reconnaissance, pre-scheduling emergency repair resources based on BIM-based disaster damage simulation, and switching to a disaster recovery backup decision model. When the overall confidence level of the intelligent prediction model is below 85%, or the monitoring data exceeds 150% of the historical statistical extreme value, or when the high-risk level in claim 5 and another independent risk source are triggered at the same time, a graded early warning will be initiated. Level 1 warning: Automatically activate drone swarms to conduct multi-angle disaster reconnaissance and transmit real-time images back to the command center; Level 2 Early Warning: Based on the disaster damage simulation results analyzed by the BIM model, automatically dispatch the nearest available repair teams and materials to the standby location; Level 3 warning: Forcefully suspend the regular forecasting-decision-making process and switch to a disaster recovery backup model based on case reasoning and reinforcement learning training. This model directly outputs an emergency plan with personnel safety and power grid stability as the highest priority, including power outage isolation strategies, optimal evacuation routes, and backup power activation sequences, ensuring that the end-to-end response time from triggering to generating an emergency plan is ≤15 minutes.

[0039] Furthermore, it also includes continuous iterative optimization of the model: constructing a closed-loop optimization loop that takes the tower state change data, resource consumption data and cost-benefit data after decision execution as input, and the loss function of the prediction model and the adjustment of the parameters of the objective function of the management model as output; By using time difference learning and gradient descent algorithms, the difference between the actual operation and maintenance effect and the expected effect is used as a reinforcement learning reward signal to dynamically update the weights of the prediction model. At the same time, by using the difference between the decision simulation verification results and the on-site execution results, the confidence of the finite element analysis parameters and operation and maintenance rules in the BIM management model is adjusted through backpropagation. After extreme scenarios occur, transfer learning techniques are used to fine-tune the key layers of the pre-trained model by using new scenario data as the source domain, thereby achieving a cumulative monthly improvement of 1%-2% in model prediction accuracy.

[0040] In another embodiment of the present invention, the method further includes decommissioning phase management based on a BIM model: The BIM management model calls upon structural performance degradation data, material corrosion data, and historical load spectra from the digital identity archives of the tower throughout its entire life cycle, calculates residual values ​​through a residual strength assessment algorithm, and generates a stepped dismantling scheme. For disassembled components, they are classified according to the extrapolation results of their corrosion rate model: if the steel corrosion rate is ≤30%, it is marked as a reusable component and included in the prefabricated component library; if it is >30%, a decomposition and regeneration process flow is generated. For concrete foundations, based on the strength grade and aggregate information in their BIM model, the crushing and screening process is planned to control the particle size distribution of the produced recycled aggregate to meet the GB / T25177-2010 standard, achieving an overall resource utilization rate of ≥95%.

[0041] The above is an illustrative scheme of a BIM-based pole and tower lifecycle management method according to this embodiment. It should be noted that the technical solution of this BIM-based pole and tower lifecycle management system belongs to the same concept as the technical solution of the aforementioned BIM-based pole and tower lifecycle management method. Details not described in detail in the technical solution of the BIM-based pole and tower lifecycle management system in this embodiment can be found in the description of the technical solution of the aforementioned BIM-based pole and tower lifecycle management method.

[0042] This embodiment presents a BIM-based full lifecycle management system for power poles, including: The data acquisition module is used to collect full-dimensional data throughout the life cycle of the tower and preprocess the full-dimensional data based on a unified data standard to form a standardized end-to-end data system. The dual-model parallel construction module is used to build, in parallel, an intelligent prediction model for state prediction and a BIM pole management model for management simulation based on standardized full-process data. The hierarchical data acquisition trigger module is used to dynamically trigger hierarchical targeted data acquisition commands based on the prediction results output by the intelligent prediction model, so as to supplement data acquisition for key scenarios and key parts. The virtual-real fusion decision module is used to integrate the prediction results with the targeted supplementary data and input them into the building information model tower management model. It generates decision instructions through a multi-objective optimization algorithm and schedules the execution unit to overlay the virtual operation guidance generated based on the decision onto the target physical object. The digital identity storage module is used to assign a unique physical ID to each tower, establish a digital identity file that links data from all dimensions of its entire lifecycle, and store key operation and status change records. The iterative optimization module is used to collect feedback data generated after decision execution, construct a closed-loop optimization circuit, and drive the intelligent prediction model and BIM pole management model to continuously iterate and optimize.

[0043] This embodiment also provides an electronic device applicable to BIM-based pole and tower lifecycle management methods, including: The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement the BIM-based full lifecycle management method for towers, as proposed in the above embodiments.

[0044] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the BIM-based full lifecycle management method for poles and towers as proposed in the above embodiments.

[0045] The storage medium proposed in this embodiment and the method for implementing BIM-based full life cycle management of poles and towers proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0046] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0047] Example 2 is an embodiment of the present invention. Unlike the first embodiment, this embodiment verifies the beneficial effects of the present invention through a specific implementation process.

[0048] In response to the characteristics of 220kV transmission line projects, a strategy of zoned data collection, categorized verification, and unified packaging was adopted: covering core data from the four stages of tower production to decommissioning to ensure no information gaps. During the production stage, data acquisition terminals were deployed in conjunction with the tower factory to obtain real-time information on the chemical composition of Q345 steel components, including C content of 0.16%-0.20% and Mn content of 1.00%-1.60%, hot-dip galvanizing process parameters such as zinc bath temperature of 450±5℃ and immersion time of 60-90s, as well as strength test data, with a yield strength ≥345MPa, and recording accuracy to ±5MPa. During the installation phase, tower location coordinates were collected using GIS-RTK equipment with a planar accuracy of ±2cm, terrain slope (maximum 32° in mountainous areas), soil bearing capacity (0.25MPa in plains and 0.12MPa in mountainous areas), and nearly 10 years of regional meteorological data were simultaneously accessed, with a maximum wind speed of 28m / s and a maximum ice thickness of 25mm. Sampling interval was 15 minutes, and foundation settlement monitoring data was also collected once a week. During operation, tilt sensors with an accuracy of ±0.1° and fiber optic stress sensors with an accuracy of ±1MPa are installed at the tower head, middle of the tower body, and tower legs of each tension tower. LiDAR is deployed in the power transmission corridor to monitor tree growth once a month to achieve real-time data acquisition.

[0049] Data quality is ensured by employing a converged standard and multiple verification modes. A unified standard integrating ISO 19650 and the power industry standard DL / T 5210.6 is adopted to convert structured data such as steel composition into JSON-LD format, and unstructured data such as design drawings and test reports are encapsulated in XML to ensure data interoperability. The data cleaning process involves MD5 checksum to exclude damaged data during transmission, regular expressions to remove abnormal characters from sensors, box plots to identify and correct outliers in stress monitoring, data with errors exceeding ±3%, and time series consistency checks to eliminate jumps in foundation settlement data, ultimately forming an engineering database containing 120,000 standardized data entries.

[0050] The specific implementation is as follows: A prediction-decision dual model is constructed based on standardized data, and iterative optimization is achieved through a linkage mechanism: A dual machine learning framework combining LSTM and random forest is adopted. The input includes historical meteorological data, such as wind and icing records of the past 10 years, tower status data such as tilt and stress, and geographical environment data such as topography and soil. The Neyman orthogonality method is used to reduce interference between parameters. The model outputs tilt warning (with a tilt threshold of ±0.1° in this embodiment), stress warning (with a stress threshold of ±1MPa in this embodiment), and remaining life prediction based on the Weibull proportional hazards model.

[0051] The model training used nearly 5 years of operation and maintenance data of similar towers, including 32 fault records. The initial prediction accuracy reached 92%, which was improved to over 97% after joint optimization.

[0052] A 3D model with LOD400 accuracy is constructed, including the geometric dimensions of tower components with an error ≤1mm, material properties, connection node parameters, and other details. It integrates a finite element analysis engine, an operation and maintenance knowledge rule base containing over 200 types of fault handling solutions, and a GIS module to achieve precise matching with the geographic space. For tower locations in mountainous areas with slopes >25° and soil bearing capacity <0.15MPa, the model can automatically match reinforced tower leg design schemes, such as increasing the rib thickness to 12mm.

[0053] Data interaction is achieved through a parameter server architecture under the federated learning framework. The prediction model uploads the risk confidence level, such as a 91% risk confidence level for an ice thickness of 22mm in the next 48 hours, along with the prediction parameters, to the BIM model. The BIM model verifies the impact of the risk through finite element simulation. For example, when the ice thickness is 22mm, the stress at the tower head reaches 320MPa, which is close to the threshold. The simulation results and the matching degree with the rules are sent down as feedback parameters. After encryption and aggregation, both parties perform local updates. The prediction error and the deviation of the simulation results are used as a joint loss function for iterative optimization to ensure the accuracy of the model output. The prediction error is ≤3%, and the deviation of the simulation results is ≤5%.

[0054] Abandoning the traditional full-coverage, high-frequency data collection mode, this embodiment triggers three-level targeted data collection based on prediction results. This mechanism reduces the cost of UAV inspection by 60%, and the specific tiered implementation is as follows: The risk confidence level, predicted values ​​of key parameters, and historical failure mode matching results output by the intelligent prediction model are used to make a joint judgment. For example, if a tower location has experienced an icing failure, the matching weight is increased by 20%.

[0055] The low-risk level covers 102 poles in the plain section, and only the basic sensors are activated for routine data acquisition. For medium-risk areas, additional drones equipped with high-definition cameras are deployed for inspections. Deep learning algorithms are used to identify tower corrosion and loose bolts, and to generate three-dimensional point cloud data of the tower head. When three power poles near an industrial park detect an earthquake intensity of ≥8 or stress data exceeding 120% of historical extreme values, a fiber optic stress sensor network is activated for real-time monitoring. Detection equipment equipped with infrared thermal imagers is dispatched to conduct on-site investigations. All acquisition commands are pushed through 5G network slicing technology, and the data is encrypted and uploaded after being associated with the unique ID and timestamp of the power pole on the edge side.

[0056] By integrating prediction results with targeted data collection, the BIM model generates differentiated decisions and drives execution, achieving precise linkage between the model and the site: A multi-objective optimization algorithm is adopted, with safety (target ≥ 99.9%), economy (cost ≤ 110% of budget) and reliability as objectives, and risk level and real-time data as input.

[0057] For the 28 ice-covered towers with medium risk, the model generates a plan of de-icing first and then reinforcement, and automatically outputs a structured maintenance work order that includes BIM sectional views, component material codes, such as M16×50 bolts, code GB / T5782-2016, bolt tightening torque 80±5N·m, and foundation verticality deviation ≤1‰. For the three high-risk towers, emergency work orders for power outage isolation and emergency reinforcement are directly generated. These work orders are pushed to the maintenance personnel's AR terminals via the production management system. 3D registration technology overlays virtual operation guidelines, such as bolt reinforcement sequence and de-icing equipment placement, onto the actual tower components. During maintenance, maintenance personnel upload data such as actual torque values ​​and replacement part serial numbers in real time via their terminals, driving the BIM digital twin to update its status synchronously. The model verifies the reinforcement effect through simulation, showing that the stress has decreased to 280MPa, meeting safety requirements. The verification results are fed back to the optimization algorithm as decision-making quality assessment parameters.

[0058] Each tower is assigned a unique physical ID, using a dual identification system of QR code and RFID. The QR code is used for on-site identification, while RFID is used for full-process component tracking. A digital identity profile is constructed based on ontology semantic mapping technology to achieve full lifecycle traceability. The system includes tower design drawings, load calculation sheets, production batches (e.g., batch 2024-03-15), construction and acceptance records (including galvanized layer thickness ≥ 88μm), maintenance data, annual corrosion rate 0.025mm / a < 0.03mm / a, prediction reports, and decommissioning assessment results, providing comprehensive information.

[0059] The power industry consortium blockchain is adopted, with nodes including owners, construction, operation and maintenance units and regulatory departments. Hash calculations are performed on records of critical status changes, such as stress exceeding the threshold, major operation and maintenance operations, such as tower head reinforcement and parameter corrections exceeding ±3%, and the hash values ​​are stored on the blockchain for evidence.

[0060] When a bolt loosening fault occurred on a certain tower, the maintenance personnel scanned the ID and completed the source tracing within 3 minutes, locating the cause of the fault as the torque during the construction phase not meeting the standard, thus achieving precise identification of responsibility.

[0061] A dual mechanism of emergency response to extreme scenarios and continuous model optimization was established to ensure the system is both resilient and capable of evolution: 12 types of emergency plans were pre-set for extreme situations such as typhoons and earthquakes. During the project implementation, a certain area experienced a level 12 gale. The system monitored that the wind speed of 3 towers exceeded the historical extreme value by 150%, and the confidence level of the prediction model dropped to 82%, automatically triggering a three-level early warning: Level 1 early warning activated a drone swarm for reconnaissance and transmitted real-time images of tower swaying; Level 2 early warning, based on BIM disaster damage simulation, dispatched a repair team and materials 5 kilometers away to a backup point; Level 3 early warning switched to the disaster recovery backup model and generated a power outage isolation + temporary guy wire reinforcement plan within 12 minutes to prevent tower collapse.

[0062] A closed-loop optimization circuit is constructed, inputting state changes after decision execution, such as a 40MPa decrease in stress after reinforcement, resource consumption (a reduction of 12,000 yuan in maintenance costs per tower), and cost-benefit data. The prediction model weights are updated using time-difference learning and gradient descent algorithms. The differences between simulation and field results are used to adjust the finite element parameters of the BIM model. After extreme wind disasters, transfer learning is employed, using new scenario data as the source domain to fine-tune key layers of the model, improving prediction accuracy by 1.8%, achieving increasing accuracy with repeated use.

[0063] In another embodiment of the present invention, optimizations were made for complex terrain: groundwater level monitoring data was added when collecting data at tower sites in mountainous areas to avoid settlement caused by foundation soaking; for towers near industrial parks, additional data on industrial corrosive gas concentrations, such as SO2 content, were collected to provide data support for corrosion rate prediction. In the standardization process, for unstructured UAV inspection images, deep learning-based image segmentation technology was used to extract defect features and convert them into structured defect type, location, and size data, further improving data reusability. Through this optimization, the accuracy rate of foundation fault early warning for towers in mountainous areas was increased to 98%, and the corrosion prediction error for towers in industrial parks was controlled within ±2%.

[0064] In another embodiment of the present invention, two enhancements were made to improve the model's adaptability to extreme weather conditions: First, an attention mechanism was introduced into the intelligent prediction model, assigning higher weights to key meteorological parameters such as strong winds and icing, thereby increasing the prediction confidence in extreme scenarios by 5%-8%; second, a material degradation model was added to the BIM model, dynamically updating component strength parameters based on corrosion rate data, such as correcting the yield strength of Q345 steel to 335MPa after 5 years of use, reducing the deviation between finite element simulation results and actual conditions from 5% to 2.5%. When the two models are linked, an encrypted transmission + local caching strategy is adopted. In mountainous areas with weak 5G signals, edge nodes can cache model parameters to ensure uninterrupted linkage.

[0065] In another embodiment of the present invention, a virtual-real integrated operation and maintenance scenario is realized: maintenance personnel scan the physical ID of the tower using a tablet equipped with AR functionality. A virtual conductor sag standard line is overlaid on the real-time image from the device's camera. Based on current weather conditions, the allowable sag deviation is calculated to be ±0.2m, and component maintenance standards, such as the weld inspection range of the tower legs, are also specified. When the conductor sag of a certain tower exceeds the standard, the AR system automatically compares the difference between the on-site sag and the standard line (exceeding the standard by 0.3m), generates an adjustment work order, and marks the tightening position. This improves the work efficiency of maintenance personnel by 35% and avoids the errors of traditional reliance on drawings.

[0066] In another embodiment of the present invention, the trigger threshold is optimized based on engineering practice: for tension towers, the stress warning threshold is adjusted from ±1MPa to ±0.8MPa to trigger data collection earlier; for straight-line towers in plains sections, the collection frequency for low-risk levels is increased from 24 hours / time to 48 hours / time, further reducing costs. For high-risk data collection, a drone + robot collaborative mode is introduced. The drone performs aerial panoramic monitoring, while the ground robot carries an ultrasonic flaw detector to perform close-range inspection of the tower leg welds, achieving comprehensive data collection from both the air and the ground, increasing the defect detection rate to 99%.

[0067] In another embodiment of the present invention, a closed-loop management system of work order-execution-acceptance-archiving is constructed: During decision generation, a cost calculation module is added to compare the prices of different maintenance solutions, such as a component replacement cost of 8000 yuan versus a reinforcement cost of 3000 yuan, prioritizing the more economical solution; during execution, GPS positioning ensures that maintenance personnel arrive on-site, and key procedures, such as tower head reinforcement, require video recording and uploading for acceptance; after acceptance, the work order is automatically archived to the tower's digital identity file, forming an immutable execution record. This optimization further reduces maintenance costs by 8%, achieving a 100% work order execution rate.

[0068] In another embodiment of this invention, to improve traceability efficiency, a hybrid model of on-chain evidence storage and off-chain storage is adopted: key data hash values, operation logs, and other core information are stored on the blockchain to ensure trustworthiness; large-capacity data such as design drawings and high-definition images are stored in a distributed file system and accessed through on-chain hash values. For decommissioned towers, additional information such as dismantling personnel, time, and recycling companies is recorded. The steel recycling rate of a certain batch of decommissioned towers reached 95%, and the traceability can accurately track the whereabouts of each piece of steel, meeting environmental protection requirements.

[0069] In another embodiment of the present invention, a grid-coordinated emergency response function is added to address the unique characteristics of the power grid: when a Level 3 warning is triggered, the system automatically synchronizes the warning information to the regional power grid dispatch center, coordinating adjustments to the power supply mode to prevent large-scale power outages caused by single-tower failures. Simultaneously, a rapid assessment model for extreme weather events is pre-set. After a disaster, tower deformation assessment is completed within one hour by comparing UAV point cloud data with the BIM model, providing a basis for prioritizing repairs. During the 2024 typhoon disaster, this mechanism reduced fault recovery time by 50%.

[0070] In another embodiment of the present invention, a quantitative optimization index system is established: prediction error, decision execution success rate, and cost reduction rate are used as core indicators, and the model is evaluated monthly. Specifically, the prediction error is ≤3%, the decision execution success rate is ≥98%, and the cost reduction rate is ≥5% / month. Through a time-difference learning algorithm, the difference between the actual operation and maintenance effect and the expected result is converted into a reward signal, such as a +10 point reward for a 0.5% reduction in error, dynamically adjusting the weights of each layer of the prediction model. The finite element parameters of the BIM model are optimized through a backpropagation algorithm, reducing the deviation between the simulation results and the measured data from the initial 8% to below 2%, achieving continuous improvement in model performance.

[0071] In another embodiment of the present invention, a refined management of the entire process of residual value assessment, dismantling planning, and resource recycling is realized: the BIM model calls the full life cycle data of a certain decommissioned tower, which has been in service for 25 years and has a steel corrosion rate of 28%. The residual value is calculated by the residual strength assessment algorithm, which is about RMB 12,000 per tower. A step-by-step dismantling plan is generated, which is to first dismantle the conductor, then the tower head, and finally the tower leg. After dismantling, the steel with a corrosion rate of ≤30% is marked as a reusable component and included in the prefabricated component library for rural power grid renovation projects. The concrete foundation is planned with a crushing and screening process according to the C30 strength grade in the BIM model. The particle size distribution of the produced recycled aggregate meets the GB / T25177-2010 standard and is used for road base paving. The overall resource utilization rate reaches 96%, which is 21% higher than the traditional method.

[0072] Example 3, referring to Figures 2-4 This is one embodiment of the present invention. This embodiment verifies the beneficial effects of the present invention through comparative experiments.

[0073] In this embodiment, 150 220kV transmission towers with similar service conditions in the same area were uniformly selected as the experimental sample. These were evenly divided into three groups of 50 towers each. The sample covered three typical operating conditions: plains, mountains, and industrial parks, and included four high-frequency risk scenarios: strong winds, icing, geological subsidence, and component corrosion. The experimental hardware was uniformly equipped with a 5G industrial transmission terminal, a ±2cm precision UAV inspection system, a fiber optic stress sensing network, and an AR maintenance terminal. The software was uniformly configured with servers of equivalent computing power, a finite element simulation platform, and a data statistical analysis system. The experiment consisted of 12 complete model iteration cycles. Each cycle integrated monthly maintenance logs, real-time environmental monitoring data, and on-site operation feedback data. The prediction accuracy, relative prediction error, scenario confidence, resource consumption, and maintenance efficiency of the three schemes were recorded round by round. All data were averaged from multiple samplings to eliminate random errors and ensure the authenticity and validity of the experimental data.

[0074] Control Group 1: This group adopted a traditional quarterly manual inspection and post-fault handling model, lacking standardized data collection specifications, intelligent prediction models, and model iteration and optimization mechanisms. Risk assessment, fault analysis, and operation and maintenance decisions relied entirely on the professional experience of operation and maintenance personnel. Data records were fragmented and lacked a unified format, making data accumulation and performance optimization impossible. This represents the traditional mainstream tower management model for power grids.

[0075] Control Group 2: Employs traditional static BIM modeling technology, only achieving 3D visualization of towers and basic information input. It lacks machine learning algorithms, a tiered targeted data collection mechanism, blockchain data storage, and a closed-loop iterative optimization system. Risk identification relies on manual analysis, and model parameters remain fixed, making dynamic optimization based on on-site conditions impossible. This represents the current mainstream lightweight BIM application solution in the power industry.

[0076] The present invention adopts a dual-model linkage architecture, integrating the LSTM+random forest prediction algorithm with the LOD400 high-precision BIM digital twin model, and is equipped with four core mechanisms: three-level targeted data acquisition, AR virtual-real fusion decision-making, blockchain full-process evidence storage, and reinforcement learning closed-loop iteration. It can realize autonomous data accumulation, autonomous model optimization, accurate risk prediction, and intelligent operation and maintenance decision-making.

[0077] Reference Figure 2 This embodiment statistically analyzes the risk identification and prediction accuracy of three schemes across 1-12 iteration cycles. The precise data for each group in each iteration are as follows: Control Group 1: No iterative optimization capability. The prediction accuracy remained stable in the range of 75.0%-78.0% throughout the 12 iteration cycles, without any fluctuation or improvement. Specific data for each cycle: The accuracy for cycles 1-12 was 76.2%, 75.8%, 77.1%, 76.5%, 77.8%, 75.3%, 76.9%, 77.4%, 75.6%, 76.8%, 77.2%, and 76.4%, respectively, with an average accuracy of 76.6% throughout the entire process.

[0078] Control Group 2: Model parameters were fixed, with no iterative optimization effect. Slight performance degradation occurred with data accumulation. The accuracy remained at 85.0%-87.0% for 12 rounds of iteration. Specific data for each round were as follows: Round 1 86.8%, Round 2 86.5%, Round 3 86.2%, Round 4 85.9%, Round 5 86.1%, Round 6 85.7%, Round 7 85.5%, Round 8 85.3%, Round 9 85.1%, Round 10 85.2%, Round 11 85.0%, Round 12 85.1%, with an overall average accuracy of 85.7%.

[0079] The proposed solution relies on a closed-loop iterative mechanism of reinforcement learning, resulting in a continuous and steady improvement in accuracy and significant performance gains through iteration. Specific data for each round are as follows: Round 1: 92.1%; Round 2: 92.8%; Round 3: 93.5%; Round 4: 94.2%; Round 5: 94.9%; Round 6: 95.6%; Round 7: 96.1%; Round 8: 96.5%; Round 9: 96.9%; Round 10: 97.2%; Round 11: 97.5%; Round 12: 97.8%. The overall average accuracy is 95.4%. After the final iteration, it represents a 10.8% improvement compared to control group 2 and a 19.8% improvement compared to control group 1, demonstrating the significant advantages of iterative optimization.

[0080] Reference Figure 3 The relative error of prediction is the core indicator of the model's prediction accuracy. The lower the error value and the smaller the fluctuation, the higher the match between the model and the actual working conditions on site. The accuracy error data for each group from round 1 to round 12 are as follows: Control Group 1: Human subjective judgment showed significant bias and drastic error fluctuations, with the relative error remaining between 18.0% and 22.0% throughout the process. Data for each round were as follows: Round 1: 20.5%, Round 2: 19.2%, Round 3: 21.3%, Round 4: 18.9%, Round 5: 20.8%, Round 6: 21.9%, Round 7: 19.5%, Round 8: 20.2%, Round 9: 21.5%, Round 10: 18.5%, Round 11: 20.9%, Round 12: 21.2%, with an average error of 20.4% throughout the process.

[0081] Control Group 2: No dynamic parameter optimization capability, error cannot converge and remains high. Data for each round: Round 1 10.2%, Round 2 10.1%, Round 3 9.9%, Round 4 9.8%, Round 5 9.8%, Round 6 9.7%, Round 7 9.6%, Round 8 9.6%, Round 9 9.5%, Round 10 9.5%, Round 11 9.5%, Round 12 9.5%, with an average error of 9.7% throughout, showing no significant optimization effect.

[0082] The proposed solution employs a dual-model linkage error correction and backpropagation parameter optimization, resulting in continuous error convergence and improved accuracy. The error rates for each round are as follows: Round 1: 4.8%; Round 2: 4.2%; Round 3: 3.8%; Round 4: 3.5%; Round 5: 3.2%; Round 6: 2.9%; Round 7: 2.6%; Round 8: 2.4%; Round 9: 2.3%; Round 10: 2.2%; Round 11: 2.15%; Round 12: 2.1%. In the later stages of iteration, the error is stably controlled within 3%, ultimately reducing the error by 7.4% compared to control group 2 and by 18.3% compared to control group 1, achieving a qualitative improvement in prediction accuracy.

[0083] Reference Figure 4 Four high-frequency and high-risk scenarios for the power grid were selected: strong wind scenario, icing scenario, geological subsidence scenario, and component corrosion scenario. Three sets of solutions were specifically tested. The accuracy confidence data for each scenario and the differences between sets are as follows: Control Group 1: Due to limitations in human experience and incomplete data collection, the confidence levels for each scenario were generally low. Specifically, the confidence levels for strong wind scenarios were 65.0%, icing scenarios were 62.0%, geological subsidence scenarios were 60.0%, and component corrosion scenarios were 68.0%. The average confidence level for the four scenarios was 63.8%, indicating extremely poor reliability in risk prediction and an inability to support operational and maintenance decisions under complex conditions.

[0084] Control group 2: The static model has no scene adaptation and optimization capabilities, and the confidence level is significantly reduced in extreme scenarios. Specific data: 81.0% for strong wind scenario, 79.0% for icing scenario, 77.0% for geological subsidence scenario, and 83.0% for component corrosion scenario. The average confidence level of the four scenarios is 80.0%, which is an improvement over the traditional solution, but the adaptability to extremely complex scenarios is insufficient.

[0085] The solution of this invention significantly improves the confidence level of each scenario through attention mechanism weight optimization, multi-parameter fusion analysis, and scenario-based model fine-tuning. Specific data shows: 96.0% confidence level for strong wind scenarios, 95.0% for icing scenarios, 94.0% for geological subsidence scenarios, and 97.0% for component corrosion scenarios, with an average confidence level of 95.5% across the four scenarios. Compared to conventional BIM solutions, this represents a maximum improvement of 17.0% and an average improvement of 15.5% per scenario; compared to traditional manual solutions, it represents a maximum improvement of 34.0% and an average improvement of 31.7% per scenario. The stability and accuracy of identifying extreme risk scenarios far surpass existing technologies.

[0086] Beyond the core model performance, a comprehensive quantitative statistical analysis was conducted on the resource consumption, operation and maintenance efficiency, data management, and emergency response capabilities of the three solutions in engineering implementation. Regarding resource consumption, the invention's three-level targeted data collection mechanism replaces full-scale high-frequency data collection, resulting in a 42.0% reduction in inspection manpower costs, a 60.0% reduction in drone equipment wear and tear, and a 55.0% reduction in data storage and transmission bandwidth overhead compared to control groups 1 and 2, significantly optimizing resource utilization. In terms of operation and maintenance efficiency, the invention's AR virtual-real fusion intelligent decision-making replaces manual analysis and judgment, shortening the operation and maintenance decision response cycle by 70.0% and enabling one-time fault handling. The success rate of this invention reached 98.5%, while the success rate of control group 1 was only 82.3% and that of control group 2 was only 90.1%. In terms of data management, this invention uses a unified data standard and blockchain notarization to achieve 100% data traceability and immutability. The data traceability rate of control group 1 was less than 40% and that of control group 2 was only 65%, and both had problems with data tampering and fragmentation. In terms of emergency adaptation, this invention can output emergency solutions for extreme scenarios within 15 minutes. The average response time of control group 1 was 68 minutes and that of control group 2 was 42 minutes. The resilience to risks under extreme conditions is significantly better than existing technologies.

[0087] This invention addresses the technical bottlenecks of traditional manual methods—low accuracy, lack of optimization capabilities, and severe resource waste—through an integrated innovative design that combines dual-model linkage prediction, hierarchical targeted precise data collection, closed-loop iterative optimization, and end-to-end blockchain-based evidence storage. These bottlenecks are addressed by incorporating a dual-model linkage prediction system, hierarchical targeted precise data collection, closed-loop iterative optimization, and end-to-end blockchain-based evidence storage. All quantitative experimental data demonstrate that this invention exhibits significant advantages in prediction accuracy, error control, scenario confidence, operational efficiency, resource utilization, and risk prevention capabilities, fully meeting the demands of intelligent, refined, and highly reliable routine management of power grid towers throughout their entire lifecycle.

[0088] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A BIM-based method for the full lifecycle management of power poles, characterized in that, include: Collect full-dimensional data of the tower's life cycle, and preprocess the full-dimensional data based on a unified data standard to form a standardized full-process data system; Based on standardized end-process data, an intelligent prediction model for state prediction and a BIM pole management model for management simulation are constructed in parallel. Based on the prediction results output by the intelligent prediction model, a hierarchical targeted data collection command is dynamically triggered to supplement data collection for key scenes and key parts. The prediction results are fused with targeted supplementary data and input into the building information model tower management model. Decision instructions are generated through a multi-objective optimization algorithm, and the execution unit is scheduled to overlay the virtual operation guidance generated based on the decision onto the target physical object. Assign a unique physical ID to each tower, establish a digital identity file that links data from all dimensions of its entire lifecycle, and preserve records of key operations and status changes. The feedback data generated after the decision is executed is collected, a closed-loop optimization loop is constructed, and the intelligent prediction model and the BIM pole management model are continuously iterated and optimized.

2. The BIM-based full lifecycle management method for poles and towers as described in claim 1, characterized in that, Collect comprehensive data throughout the tower's lifecycle and preprocess this data based on a unified data standard to form a standardized end-to-end data system, including: The comprehensive data includes production phase data, installation phase data, and operation phase data. By adopting a fusion standard based on Building Information Modeling (BIM) data standards and power industry tower operation and maintenance data specifications, multi-source heterogeneous data is converted into a unified data format, and unstructured data is encapsulated using XML. The multi-source heterogeneous data is then verified and cleaned to form a standardized end-to-end data system.

3. The BIM-based full lifecycle management method for poles and towers as described in claim 2, characterized in that, Based on standardized end-process data, an intelligent prediction model for state prediction and a BIM-based pole management model for management simulation are constructed in parallel, including: Construct a smart prediction model for pole status and a BIM-based pole management model based on a dual machine learning framework; The intelligent prediction model integrates LSTM and random forest algorithms, reduces parameter interference through Neyman orthogonality, takes operational phase data as input, and outputs tilt, stress warning, and remaining lifetime prediction based on the Weibull proportional risk model. The BIM management model is a digital twin that integrates a finite element analysis engine, an operation and maintenance knowledge rule base, and a GIS module, and automatically matches management solutions based on terrain slope and soil bearing capacity. The dual-model linkage is achieved through a parameter server architecture under the federated learning framework. The prediction model uploads the risk confidence and prediction parameters as intermediate parameters, while the management model sends out the simulation verification results and rule matching degree as feedback parameters. After encryption and aggregation, both models are updated locally, and the prediction error and simulation results are used as a joint loss function for iterative optimization.

4. The BIM-based full lifecycle management method for poles and towers as described in claim 3, characterized in that, Based on the prediction results output by the intelligent prediction model, a hierarchical targeted data collection command is dynamically triggered to supplement data collection for key scenarios and key parts, including: executing a three-level targeted data collection triggering mechanism, where the triggering level is jointly determined by the risk confidence level output by the intelligent prediction model, the predicted values ​​of key parameters, and the historical failure mode matching results. When a low-risk level is triggered, the basic sensors are activated to perform routine data collection. When a medium-risk level is triggered, additional drones will be deployed for high-definition inspection and AI image recognition, and 3D point cloud data of the designated area will be generated. When a high-risk level is triggered, the fiber optic stress sensor network deployed on the main structure of the tower is activated for real-time monitoring, and high-precision on-site detection equipment carrying infrared thermal imagers is dispatched. All data collection instructions are pushed to the corresponding terminals via 5G network slicing technology. The collected data is encrypted and directly transmitted to the BIM model database after being associated with the unique ID and timestamp of the tower on the edge side.

5. The BIM-based full lifecycle management method for poles and towers as described in claim 4, characterized in that, The predicted results are fused with targeted supplementary data and input into the building information model (BIM) tower management model. A multi-objective optimization algorithm is used to generate decision instructions, and the virtual operations of these decision instructions are superimposed onto the target physical object, including: A multi-objective optimization algorithm is adopted, with safety, economy and reliability as optimization objectives, inputting risk level and real-time data, and outputting differentiated decisions; For high-risk decisions, generate structured maintenance work orders that include BIM model sectional views, material code lists of components to be replaced, and precise process parameters; The decision-making instructions are pushed to the AR terminals of the operation and maintenance personnel through the production management system, and the virtual operation guidance is superimposed on the real tower components using 3D registration technology; During the maintenance process, key step data is transmitted back through the terminal to drive the digital twin to perform synchronous simulation and effect verification, and the verification results are fed back to the multi-objective optimization algorithm as decision quality assessment parameters.

6. The BIM-based full lifecycle management method for poles and towers as described in claim 5, characterized in that, Each tower is assigned a unique physical ID, and a digital identity file is established linking it to data across all dimensions of its entire lifecycle. Records of key operations and status changes are also documented, including: Each tower is assigned a unique physical ID, and based on ontology semantic mapping technology, a full lifecycle digital identity file is constructed that links its design drawings, production batches, construction and acceptance records, operation and maintenance data, forecast reports and decommissioning assessments. Hash operations are performed on key status changes, major maintenance operations, and parameter correction records exceeding preset thresholds in the digital identity file. The hash values ​​are stored on the blockchain as evidence, and smart contracts ensure that the operation logs are tamper-proof and fully traceable.

7. The BIM-based full lifecycle management method for poles and towers as described in claim 6, characterized in that, Collect feedback data generated after the decision is executed, construct a closed-loop optimization circuit, and drive the intelligent prediction model and BIM pole management model to continuously iterate and optimize, including: Construct a closed-loop optimization loop that takes as input tower status change data, resource consumption data, and cost-benefit data after decision execution, and outputs the loss function of the prediction model and the parameter adjustment of the objective function of the management model; The difference between the actual operation and maintenance effect and the expected effect is used as a reinforcement learning reward signal to dynamically update the weight of the prediction model; at the same time, the confidence of the finite element analysis parameters and operation and maintenance rules in the BIM management model is adjusted through backpropagation by using the difference between the decision simulation verification results and the on-site execution results. After extreme scenarios occur, transfer learning techniques are used to fine-tune the key layers of the pre-trained model by using new scenario data as the source domain.

8. A BIM-based pole and tower lifecycle management system, applied to the method described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect full-dimensional data of the tower's life cycle and preprocess the full-dimensional data based on a unified data standard to form a standardized full-process data system. The dual-model parallel construction module is used to build, in parallel, an intelligent prediction model for state prediction and a BIM pole management model for management simulation based on standardized full-process data. The hierarchical data acquisition triggering module is used to dynamically trigger hierarchical targeted data acquisition instructions based on the prediction results output by the intelligent prediction model, so as to supplement data acquisition for key scenes and key parts. The virtual-real fusion decision module is used to fuse the prediction results with targeted supplementary data and input them into the building information model tower management model. It generates decision instructions through a multi-objective optimization algorithm and schedules the execution unit to overlay the virtual operation guidance generated based on the decision onto the target physical object. The digital identity storage module is used to assign a unique physical ID to each tower, establish a digital identity file that links data from all dimensions of its entire lifecycle, and store key operation and status change records. The iterative optimization module is used to collect feedback data generated after the decision is executed, construct a closed-loop optimization circuit, and drive the intelligent prediction model and the BIM pole management model to continuously iterate and optimize.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the BIM-based pole and tower full life cycle management method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the BIM-based pole and tower lifecycle management method according to any one of claims 1 to 7.

Images