Safety monitoring system for cement pole structures based on digital twin
By constructing a digital twin-based safety monitoring system for cement pole structures, the problems of low efficiency and high false alarm rate in cement pole structure monitoring have been solved. This system enables real-time monitoring and graded response of cement poles, meeting the structural safety management and control requirements of power distribution lines throughout their entire life cycle.
Patent Information
- Application Number
- CN202611400057.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-10
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-09-10
AI Technical Summary
Existing structural monitoring schemes for cement poles suffer from low efficiency, high latency, and high false alarm rates. They are unable to identify the causes of early structural deterioration and damage, and cannot achieve local real-time mechanical assessment and early warning, making it difficult to meet the needs of structural safety management and control throughout the entire life cycle of power distribution lines.
A safety monitoring system for cement pole structures based on digital twins is constructed, including a cloud-based twin platform, a standardized master twin template construction module, an instantiated twin construction module, a real-time monitoring module, and an early warning analysis module. By reducing the order of the full-order finite element benchmark simulation model and performing instantiated fine-tuning, a standardized master twin template is generated to achieve real-time monitoring and graded response.
It has enabled large-scale twinning of massive cement poles, reducing engineering costs, ensuring local emergency response capabilities, improving the professional depth of anomaly diagnosis and the efficiency of operation and maintenance resources, shortening the fault location and handling cycle, and enhancing the scientific nature of operation and maintenance decisions.
Smart Images

Figure CN122904639B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power facility structural health monitoring and digital twin technology, specifically a cement pole structural safety monitoring system based on digital twins. Background Technology
[0002] Currently, cement poles, as core supporting components of power distribution lines, are widely deployed in urban and rural environments. Their structural health directly affects power supply reliability and public safety. Existing structural safety monitoring of cement poles largely relies on regular manual inspections and single-threshold sensor alarms. Manual inspections suffer from low efficiency, high latency, and difficulty in detecting hidden damage. Fixed-threshold alarms are easily affected by environmental load fluctuations, resulting in a high false alarm rate and failing to identify early structural deterioration and the causes of damage. Some monitoring solutions that introduce digital twin technology generally adopt a pole-by-pole independent modeling approach, which involves a large modeling workload and inconsistent parameter standards, making it difficult to adapt to the large-scale deployment needs of the massive number of existing poles. At the same time, due to limitations in computing power, high-precision full-order finite element models cannot be deployed to the edge side. The edge devices can only perform simple data collection and threshold comparison, making it difficult to achieve local real-time mechanical assessment and early warning. High-precision analysis in the cloud suffers from strong communication dependence and high response latency. In addition, the existing solutions lack a graded and differentiated handling mechanism for alarm events, making it difficult to balance the emergency response speed of high-risk faults with the diagnostic accuracy of early degradation. Furthermore, damage location and cause analysis rely heavily on manual experience, and model parameters cannot be iteratively optimized during service. Under long-term operation, the monitoring and diagnostic accuracy continues to decline, failing to meet the actual needs of structural safety management throughout the entire life cycle of power distribution lines. Summary of the Invention
[0003] To address the aforementioned technical problems, the present invention aims to provide a safety monitoring system for cement pole structures based on digital twins, including a cloud-based twin platform. The cloud-based twin platform is communicatively connected to a standardized master twin template construction module, a standardized master twin template library, an instantiated twin construction module, a real-time monitoring module, an early warning analysis module, and an alarm analysis module. The standardized master twin template construction module is used to construct full-order finite element benchmark simulation models corresponding to different pole types under different typical environments. It performs cross-section-level degree of freedom merging, master deformation mode screening, and low-dimensional mechanical equation reconstruction operations on the full-order finite element benchmark simulation model to generate a reduced-order finite element benchmark simulation model. Based on the full-order finite element benchmark simulation model, the reduced-order finite element benchmark simulation model, the preset full-parameter three-dimensional geometric model, and the preset physical parameter database, a standardized master twin template is constructed. The standardized master twin template library is used to store the standardized master twin templates corresponding to different pole models under different typical environments; The instantiated twin construction module matches the standardized master twin template of each cement pole according to the basic attribute parameters and deployment parameters of each cement pole, and performs instantiated fine-tuning of the full-order finite element reference simulation model and the reduced-order finite element reference simulation model in the standardized master twin template in sequence to obtain the instantiated full-order finite element reference simulation model and the instantiated reduced-order finite element reference simulation model. The real-time monitoring module is used to monitor each cement pole in real time based on the instantiated reduced-order finite element benchmark simulation model of each cement pole, obtain the measured simulation residual rate of each type of index, analyze the measured simulation residual rate of each type of index, and determine whether a transient event alarm response or a warning-level response is triggered. The early warning analysis module is used to activate the instantiated full-order finite element reference simulation model when the early warning level response is triggered, and to compare and verify the output results of the instantiated full-order finite element reference simulation model with the output results of the instantiated reduced-order finite element reference simulation model. Based on the comparison and verification results, the early warning level response of the edge gateway is cancelled or damage is traced. The alarm analysis module is used to quickly perform event authenticity verification when a transient event alarm response is triggered, and to deactivate the transient event alarm response of the edge gateway or perform damage tracing based on the verification result.
[0004] Furthermore, the process of constructing full-order finite element reference simulation models for different pole types under various typical environments, and building a standardized master twin template based on the full-order finite element reference simulation model, the reduced-order finite element reference simulation model, the preset full-parameter three-dimensional geometric model, and the preset physical parameter database, includes: Based on the national standard pole production drawings, a full-parameter three-dimensional geometric model of poles of different inherent models was constructed. Based on laboratory test data and industry standard pre-calibrated physical parameter database, the physical parameter database includes a basic material parameter library, a damage evolution parameter library and an environmental correction coefficient library corresponding to different pole inherent models and typical environments. Based on the full-parameter three-dimensional geometric model and physical parameter database, a full-order finite element benchmark simulation model corresponding to different pole types under different typical environments is constructed. The full-order finite element benchmark simulation model corresponding to different pole types under different typical environments is subjected to cross-section-level degree of freedom merging, main deformation mode screening and low-dimensional mechanical equation reconstruction operations to generate a reduced-order finite element benchmark simulation model corresponding to different pole types under different typical environments. Lightweight encapsulation and edge adaptation are performed on the reduced-order finite element baseline simulation models of different pole types under different typical environments to generate pre-reduced-order models of different pole types under different typical environments after encapsulation. By binding the pre-reduced order model corresponding to different pole inherent models under different typical environments with the full-parameter three-dimensional geometric model, physical parameter database, and full-order finite element benchmark simulation model of the corresponding pole inherent model and typical environment, a standardized master twin template corresponding to different pole inherent models under different typical environments is generated.
[0005] Furthermore, the process of merging cross-sectional degrees of freedom and screening dominant deformation modes for the full-order finite element benchmark simulation model includes: The rod of the full-order finite element reference simulation model is divided into p horizontal sections along the rod height. The rod element nodes of the full-order finite element reference simulation model are grouped according to their respective horizontal sections. Three main control variables are set. Based on the three main control variables, the degrees of freedom of all rod element nodes in each group of horizontal sections are merged into three main control degrees of freedom to generate a section-level full-order model. Simulation is performed on the full-order model at the section level to generate a full-scene deformation sample set. The following steps are then executed: Step 1: Obtain the total structural strain energy corresponding to each deformation sample in the full-scene deformation sample set, sort all deformation samples from high to low according to the total structural strain energy, and generate a sample queue; Step 2: Construct a pattern set, select the deformation sample with the highest total strain energy in the sample queue as the first dominant deformation mode, and add the first dominant deformation mode to the pattern set; Step 3: Select deformation samples that have not been added to the model set from the sample queue in descending order of total structural strain and mark them as the main deformation modes. Add the main deformation modes to the model set. Whenever a new main deformation mode is added to the model set, obtain the average error of restoring all deformation samples using the current model set. Step 4: When a new master deformation mode is added to the pattern set, obtain the decrease in the average error of restoring all deformation samples using the current pattern set. When the decrease is lower than the set threshold, stop the filtering and mark the current pattern set as the master deformation pattern set.
[0006] Furthermore, the process of reconstructing the low-dimensional mechanical equations includes: The first master-controlled deformation mode in the master-controlled deformation mode set is marked as the first independent mode. For the second master-controlled deformation mode in the master-controlled deformation mode set, the deformation components that completely overlap with the first independent mode are removed to form the second independent mode. For the third master-controlled deformation mode in the master-controlled deformation mode set, all overlapping components between the third master-controlled deformation mode and the first independent mode, as well as all overlapping components between the third master-controlled deformation mode and the second independent mode are removed to form the third independent mode. This process is repeated until all overlapping components between the last master-controlled deformation mode in the master-controlled deformation mode set and existing independent modes are removed to form the last independent mode, thus obtaining the independent deformation mode set. Generate an independent mode matrix based on all independent modes in the independent deformation mode set, obtain the structural dynamic equilibrium equations followed by the full-order finite element benchmark simulation model, construct the deformation superposition mapping relationship between the structural dynamic equilibrium equations and the independent mode matrix, construct a low-dimensional equivalent mechanical equilibrium equation based on the deformation superposition mapping relationship and the virtual work equilibrium principle, and generate a reduced-order finite element benchmark simulation model based on the low-dimensional equivalent mechanical equilibrium equation.
[0007] Furthermore, based on the basic attribute parameters and deployment parameters of each cement pole, a standardized master twin template is matched for each cement pole, and the full-order finite element benchmark simulation model in the standardized master twin template is instantiated and fine-tuned. The process of obtaining the instantiated full-order finite element benchmark simulation model includes: Obtain the basic attribute parameters and deployment parameters for each concrete pole; For each cement pole, based on the inherent model in the basic attribute parameters and the latitude, longitude, geographical environment type and soil type in the deployment parameters, a standardized master twin template is matched from the standardized master twin template library. The corresponding inherent model is consistent with the inherent model of the cement pole, and the corresponding typical environment is consistent with the latitude, longitude, geographical environment type and soil type of the cement pole. Based on the standardized master twin template, the full-parameter three-dimensional geometric model, physical parameter database, full-order finite element reference simulation model and reduced-order finite element reference simulation model of the cement pole are loaded. Based on the basic attribute parameters, the full-parameter three-dimensional geometric model and the full-order finite element reference simulation model of the cement pole are geometrically corrected to generate an instantiated three-dimensional high-precision geometric model and a full-order finite element reference simulation model with geometric correction. Based on the deployment parameters, the damage parameters corresponding to the damage evolution parameter library in the physical parameter database of the cement pole and the damage parameters and environmental correction coefficients corresponding to the environmental correction coefficient library are matched. Based on the basic parameters corresponding to the basic material parameter library, the damage parameters corresponding to the damage evolution parameter library, and the damage parameter and environmental correction coefficients corresponding to the environmental correction coefficient library in the physical parameter database, the material parameters and initial damage parameters of the full-order finite element benchmark simulation model that has completed geometric correction are corrected to obtain the instantiated full-order finite element benchmark simulation model. The instantiated full-order finite element benchmark simulation model and the instantiated 3D high-precision geometric model are packaged into a cloud twin data package, and the cloud twin data package is sent to the cloud twin platform.
[0008] Furthermore, the process of sequentially instantiating and fine-tuning the reduced-order finite element baseline simulation model to obtain the instantiated reduced-order finite element baseline simulation model includes: Obtain the physical correction scaling factor between the instantiated full-order finite element reference simulation model and the full-order finite element reference simulation model. Based on the physical correction scaling factor, correct the low-dimensional equivalent mechanical equilibrium equation followed by the reduced-order finite element reference simulation model to generate the instantiated reduced-order finite element reference simulation model. A lightweight geometric model is constructed based on an instantiated 3D high-precision geometric model. A sensor point mapping table is constructed based on the pole sensor deployment data in the deployment parameters and the lightweight geometric model. The instantiated reduced-order finite element baseline simulation model, lightweight geometric model, and sensor point mapping table are packaged into an edge twin data package, which is then sent to the edge gateway.
[0009] Furthermore, based on the instantiated reduced-order finite element benchmark simulation model of each cement pole, real-time monitoring is performed on each cement pole to obtain the measured simulation residual rate of various types of indicators. The process of analyzing the measured simulation residual rate of each type of indicator to determine whether a transient event alarm response or a warning-level response is triggered includes: The acquisition cycle of various indicators of the current cement pole sensor terminal is obtained. The measured data of various indicators of the cement pole are collected by the sensor terminal at the end timestamp of the acquisition cycle of each type of indicator and cached to the edge gateway. The judgment cycle of the edge gateway is set. The edge gateway pulls the measured data cached by the sensor terminal at the start timestamp of the judgment cycle. At the same time, the local meteorological data of the current cement pole is obtained. The measured data and local meteorological data are input into the instantiated reduced-order finite element benchmark simulation model. The simulation benchmark value of the current judgment cycle is output according to the instantiated reduced-order finite element benchmark simulation model. The simulation benchmark value is sent back to the sensor terminal. The newly received simulation benchmark value of the sensor terminal is marked as the judgment standard. The sensor terminal compares the measured data of various indicators with the judgment criteria in real time to obtain the measured simulation residual rate of each type of indicator. It presets the measured simulation residual rate warning threshold and national standard safety limit for each type of indicator. It compares each type of indicator with the national standard safety limit to obtain the cumulative time of each type of indicator exceeding the national standard safety limit. If the cumulative time corresponding to a type of indicator exceeds the preset time threshold, the type of indicator is marked as an abnormal indicator, triggering a transient event alarm response. The judgment cycle of the edge gateway is switched to a second-level cycle. The measured data of the sensor terminal, the abnormal indicators and the simulation benchmark value of the edge gateway are packaged into a transient alarm data packet at a second-level cycle and uploaded to the cloud twin platform. If the cumulative time corresponding to each type of indicator does not exceed the preset time threshold, the measured simulation residual rate of each type of indicator is compared with the corresponding measured simulation residual rate warning threshold. If the measured simulation residual rate corresponding to the type of indicator exceeds the measured simulation residual rate warning threshold within two consecutive judgment periods, a warning-level response is triggered. The measured data of the sensor terminal and the simulation benchmark value of the edge gateway are packaged into a warning data package according to the judgment period and uploaded to the cloud twin platform.
[0010] Furthermore, the process of obtaining the current data acquisition cycle of the cement pole sensor terminal includes: Obtain the measured simulation residual rate of each type of indicator in the measured data of the previous acquisition cycle, and preset the acquisition cycle mapping table. The acquisition cycle mapping table includes the acquisition cycle corresponding to different measured simulation residual rates of each type of indicator. Based on the measured simulation residual rate of each type of indicator in the measured data of the previous acquisition cycle and the acquisition cycle mapping table, obtain the acquisition cycle of the sensor terminal for acquiring each type of indicator of the cement pole at the current time.
[0011] Furthermore, the process of activating and instantiating the full-order finite element benchmark simulation model, comparing and verifying the output results of the instantiated full-order finite element benchmark simulation model with the output results of the instantiated reduced-order finite element benchmark simulation model, and then deactivating the early warning level response of the edge gateway or performing damage tracing based on the comparison and verification results includes: After receiving the early warning data packet uploaded by the edge gateway, the cloud twin platform locates the cloud twin data packet corresponding to the cement pole through the unique identity identifier of the pole pre-bound by the edge gateway, and activates the instantiated full-order finite element benchmark simulation model and the instantiated three-dimensional high-precision geometric model in the cloud twin data packet. The measured data from the sensor terminal in the early warning data packet and the local meteorological data from the cement pole are input into the instantiated full-order finite element reference simulation model. The simulation reference value is output according to the instantiated full-order finite element reference simulation model. The simulation reference value in the early warning data packet is compared with the simulation reference value output by the instantiated full-order finite element reference simulation model. The deviation of each type of feature is output, and the average deviation value of each type of feature deviation is obtained. The average deviation value is compared with the preset deviation threshold. If the average deviation value is less than or equal to the preset deviation threshold, the early warning level response of the edge gateway is cancelled. If the average deviation value is greater than the preset deviation threshold, the measured data of the sensor terminal corresponding to the unique identification of the pole in several historical acquisition cycles are extracted. The simulation benchmark value output by the instantiated full-order finite element benchmark simulation model, the deviation of each type of feature, the measured data of the sensor terminal in the early warning data package, and the measured data of several historical acquisition cycles are input into the preset damage source estimation model. According to the damage source estimation model, the damage type, damage location, and the degree of deterioration corresponding to the damage location are output. The damage location and the damage type and degree of deterioration corresponding to the damage location are marked on the instantiated three-dimensional high-precision geometric model.
[0012] Furthermore, the process of quickly performing event authenticity verification and, based on the verification results, deactivating the transient event alarm response of the edge gateway or performing damage tracing includes: After receiving the transient alarm data packet uploaded by the edge gateway, the cloud twin platform locates the cloud twin data packet corresponding to the cement pole through the unique identity identifier of the pole pre-bound by the edge gateway, activates the instantiated full-order finite element benchmark simulation model and the instantiated three-dimensional high-precision geometric model in the cloud twin data packet, and obtains the abnormal indicators in the transient alarm data packet. The sensor terminal that collects abnormal indicators is marked as the terminal to be verified. The backup sensor terminal of the terminal to be verified is obtained and deployed on the same cement pole. The difference between the abnormal indicators collected by the terminal to be verified and the abnormal indicators collected by the backup sensor terminal is obtained from the transient alarm data packet. The difference threshold is set. If the difference between the values of two consecutive second-level cycles is greater than the difference threshold, a sensor fault alarm is generated and the transient event alarm response of the edge gateway is deactivated. If there is no numerical difference greater than the difference threshold between two consecutive second-level cycles, the measured data of the sensor terminal and the local meteorological data of the cement pole in the transient alarm data packet are input into the instantiated full-order finite element reference simulation model. The simulation reference value is output according to the instantiated full-order finite element reference simulation model. The simulation reference value in the transient alarm data packet is compared with the simulation reference value output by the instantiated full-order finite element reference simulation model, and the deviation of each type of feature is output. Subsequently, the measured data of the sensor terminal corresponding to the unique identification of the pole in several historical acquisition cycles are extracted. The simulation benchmark value output by the instantiated full-order finite element benchmark simulation model, the deviation of various types of features, the measured data of the sensor terminal in the transient alarm data packet, and the measured data of several historical acquisition cycles are input into the preset damage source prediction model. According to the damage source prediction model, the damage type, damage location, and the degree of deterioration corresponding to the damage location are output. The damage location and the damage type and degree of deterioration corresponding to the damage location are marked on the instantiated three-dimensional high-precision geometric model.
[0013] Compared with the prior art, the beneficial effects of the present invention are: I. Unlike the traditional approach of independent modeling for each pole with inconsistent parameter configuration standards, this invention generates a full-order benchmark model through a unified parametric geometric template and a pre-calibrated physical parameter database. Based on this, it performs order reduction processing to form a standardized master twin template that corresponds one-to-one with the pole type and environment. All poles of the same type have the same simulation benchmark and uniform specifications, avoiding redundancy and standard deviation caused by repeated modeling. Relying on the standardized template, single pole instantiation and adaptation can be quickly completed without starting modeling and parameter debugging from scratch. This enables large-scale twin retrofitting of a massive number of existing concrete poles, significantly shortening the construction cycle of single-pole twin models and reducing the engineering cost of global deployment.
[0014] Second, this invention brings a lightweight, reduced-order model to the edge, enabling real-time structural status assessment and alarm triggering locally without relying on cloud communication links. Even in scenarios with unstable field communication, it ensures local emergency response capabilities, completely resolving the technical bottleneck of high computational requirements and inability to deploy full-order finite element models on the terminal side. A high-precision full-order baseline model is retained in the cloud for refined verification and damage analysis after alarm triggering, serving as a truth reference and accuracy safety net for edge-side assessment. This collaborative approach satisfies both the low-latency requirements of routine monitoring and sudden alarms while ensuring professional depth in anomaly diagnosis, achieving optimal hierarchical allocation of computing resources.
[0015] Third, this invention distinguishes between two response levels: early deterioration warning and high-risk transient alarm. For slowly developing structural deterioration, an early warning response is triggered, enabling planned, high-precision diagnosis and maintenance arrangements. For high-risk scenarios such as sudden impacts and structural instability, a transient alarm is triggered, activating an emergency green channel for priority handling. The two response types are matched with corresponding data upload strategies, computing power scheduling rules, and handling procedures. This avoids missed detections and delayed responses to high-risk events, reduces ineffective maintenance scheduling caused by non-urgent anomalies, significantly reduces the interference of false alarms on daily maintenance, and improves the return on investment of maintenance resources.
[0016] Fourth, after the system issues an alarm, it automatically identifies the damage type, locates the concentrated damage area, and quantifies the degree of structural degradation by combining the results of full-scale simulation field, historical service data, and environmental characteristics. This information is then presented intuitively on a 3D twin model, replacing the traditional method of relying on manual experience to troubleshoot. Maintenance personnel can directly formulate detection and hardening plans based on the diagnostic results, eliminating the need for blind on-site troubleshooting. This significantly shortens the fault location and handling cycle, while improving the scientific rigor and relevance of maintenance decisions and reducing ineffective on-site work. Attached Figure Description
[0017] Figure 1 This is a flowchart of a digital twin-based safety monitoring system for cement pole structures, as described in an embodiment of this application. Detailed Implementation
[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0019] like Figure 1 As shown, the safety monitoring system for cement pole structures based on digital twins includes a cloud twin platform. The cloud twin platform is connected to a standardized master twin template construction module, a standardized master twin template library, an instantiated twin construction module, a real-time monitoring module, an early warning analysis module, and an alarm analysis module. The standardized master twin template construction module is used to construct full-order finite element benchmark simulation models corresponding to different pole types under different typical environments. It performs cross-section-level degree of freedom merging, master deformation mode screening, and low-dimensional mechanical equation reconstruction operations on the full-order finite element benchmark simulation model to generate a reduced-order finite element benchmark simulation model. Based on the full-order finite element benchmark simulation model, the reduced-order finite element benchmark simulation model, the preset full-parameter three-dimensional geometric model, and the preset physical parameter database, a standardized master twin template is constructed. The standardized master twin template library is used to store the standardized master twin templates corresponding to different pole models under different typical environments; The instantiated twin construction module matches the standardized master twin template of each cement pole according to the basic attribute parameters and deployment parameters of each cement pole, and performs instantiated fine-tuning of the full-order finite element reference simulation model and the reduced-order finite element reference simulation model in the standardized master twin template in sequence to obtain the instantiated full-order finite element reference simulation model and the instantiated reduced-order finite element reference simulation model. The real-time monitoring module is used to monitor each cement pole in real time based on the instantiated reduced-order finite element benchmark simulation model of each cement pole, obtain the measured simulation residual rate of each type of index, analyze the measured simulation residual rate of each type of index, and determine whether a transient event alarm response or a warning-level response is triggered. The early warning analysis module is used to activate the instantiated full-order finite element reference simulation model when the early warning level response is triggered, and to compare and verify the output results of the instantiated full-order finite element reference simulation model with the output results of the instantiated reduced-order finite element reference simulation model. Based on the comparison and verification results, the early warning level response of the edge gateway is cancelled or damage is traced. The alarm analysis module is used to quickly perform event authenticity verification when a transient event alarm response is triggered, and to deactivate the transient event alarm response of the edge gateway or perform damage tracing based on the verification result.
[0020] In response to the standardized production and uniform model characteristics of cement poles, a twin modeling architecture of "master template library + instantiation mapping" is constructed to replace the traditional independent modeling mode of single poles, thereby reducing modeling costs from the source.
[0021] It should be further explained that, in the specific implementation process, the process of constructing full-order finite element reference simulation models corresponding to different pole types under different typical environments, and constructing a standardized master twin template based on the full-order finite element reference simulation model, the reduced-order finite element reference simulation model, the preset full-parameter three-dimensional geometric model, and the preset physical parameter database includes: Based on national standard pole production drawings, a parametric solid modeling engine is used to construct full-parameter 3D geometric models of poles of different inherent models. The full-parameter 3D geometric model has 18 core configurable geometric parameters, which fully cover the structural features of the pole: Core parameters of the pole: total pole length, root diameter, tip diameter, pole taper, wall thickness, number of segments; Internal reinforcement parameters: number of main bars, diameter of main bars, spacing of stirrups, thickness of protective layer, and prestressing tension value; Auxiliary structural parameters: crossarm installation position, guy wire hanging point position, and hardware installation point; Basic structural parameters: foundation depth, chassis dimensions, chuck dimensions, pull plate dimensions.
[0022] All parameters strictly match the national standard specification range. Through parameter-driven operation, a complete 1:1 3D model including the concrete structure of the pole, the internal steel reinforcement cage, auxiliary hardware, and the tower foundation can be automatically generated. The model mesh accuracy meets the requirements of finite element simulation, eliminating the need for secondary geometric correction. Unlike traditional simplified models that only construct the outer shell of the pole, this template incorporates the internal steel reinforcement cage and foundation constraints into a parametric system, ensuring the structural integrity of subsequent mechanical simulations.
[0023] Based on laboratory test data and a pre-calibrated physical parameter database according to industry standards, the physical parameter database includes a basic material parameter database, a damage evolution parameter database, and an environmental correction coefficient database corresponding to different pole types and typical environments. Specifically: Basic material parameter library: Preset basic mechanical parameters such as elastic modulus, compressive / tensile strength, Poisson's ratio, and density according to concrete strength grade; preset yield strength, elastic modulus, and linear expansion coefficient according to steel bar type (HRB335 / HRB400 / prestressed steel wire); preset foundation bearing capacity, internal friction angle, and constraint stiffness parameters according to soil type (silty clay, sandy soil, red soil, frozen soil). All parameters are taken from the concrete structure design code and geotechnical engineering code.
[0024] Damage evolution parameter library: pre-set parameters for concrete creep damage model, fatigue damage SN curve parameters, steel electrochemical corrosion rate parameters, and concrete crack propagation rate parameters. All damage parameters have been calibrated through concrete structure durability tests.
[0025] Environmental correction coefficient library: For 5 typical environments, the corresponding correction coefficients for damage parameters are preset. For example, the correction coefficient for steel corrosion rate in coastal salt spray area is 1.8~2.5, the correction coefficient for concrete creep coefficient in high-altitude cold area is 0.7~0.9, and the correction coefficient for fatigue damage coefficient in strong wind area is 1.3~1.6.
[0026] The concrete strength grade and steel reinforcement type are parameters corresponding to the inherent type of the pole, characterizing the mechanical properties and intrinsic damage characteristics of the component material; the soil type parameter is a parameter included in a typical environment.
[0027] Based on a full-parameter three-dimensional geometric model and a physical parameter database, a full-order finite element benchmark simulation model corresponding to different pole types under different typical environments is constructed using general-purpose finite element software. The full-order finite element benchmark simulation model of this invention is built on the existing full-parameter three-dimensional geometric model and physical parameter database, without independent geometric modeling and material parameter configuration, thus achieving homogenization and standardization of the model system.
[0028] First, the structural topology, dimensional parameters, element types, and meshing rules of the fully parametric geometric template are called to automatically complete the finite element geometric assembly of the rods, steel reinforcement cage, foundation, and ancillary structures. Then, the material mechanics parameters, concrete plastic damage constitutive model, steel corrosion evolution parameters, foundation constraint parameters, and environmental correction coefficients are loaded into the physical parameter database in batches to complete the attribute assignment of the full-order finite element benchmark simulation model. After the model is calibrated by static load in the laboratory and by on-site working conditions, a calibrated full-order finite element benchmark simulation model is generated.
[0029] The laboratory static load calibration and the field working condition calibration include: Laboratory static load calibration: Based on static load test data of the same type of pole, the core parameters such as the elastic modulus of concrete and the foundation constraint stiffness are corrected to ensure that the simulated values of pole deflection and root stress under standard static load conditions have an error of ≤3% compared with the test values. On-site working condition calibration: Select three in-operation poles of the same model, and correct the model damping ratio and wind load shape coefficient through on-site strain test and vibration test data to ensure that the simulation error of dynamic main frequency and steady-state stress under on-site working conditions is ≤5%.
[0030] Since the full-order finite element benchmark simulation model of cement poles essentially breaks down the pole's concrete, internal steel reinforcement, and foundation into thousands of tiny elements and tens of thousands of nodes, with each node containing multiple degrees of freedom, the stress and deformation state of the entire structure is described by tens of thousands of independent node displacement variables. During calculation, it is necessary to solve a system of mechanical equations of tens of thousands of orders simultaneously. Therefore, the full-order finite element benchmark simulation model cannot be run on the edge side. In order to solve the above technical problems, the full-order finite element benchmark simulation models of different pole types under different typical environments are subjected to section-level degree of freedom merging, master deformation mode screening, and low-dimensional mechanical equation reconstruction operations to generate reduced-order finite element benchmark simulation models of different pole types under different typical environments. Lightweight encapsulation and edge adaptation are performed on the reduced-order finite element baseline simulation models of different pole types under different typical environments. The lightweight encapsulation and edge adaptation include: The reduced-order finite element benchmark simulation model and the accompanying lightweight solver are packaged into independent binary format files. The total size of the pre-reduced-order template file for a single model is ≤2MB, the model file after single-bar instantiation is ≤500KB, and the runtime memory usage is ≤10MB. It can be directly deployed on low-power industrial edge gateways without relying on large finite element software. The solver is compiled in lightweight C language and supports static solution, dynamic time history solution and damage increment calculation. The time taken for a single static calculation is ≤10ms, and the time taken for a single 10-second dynamic time history calculation is ≤50ms, which fully meets the needs of real-time status assessment on the edge side.
[0031] The pre-reduced order models of different pole inherent models under different typical environments are generated and encapsulated. The pre-reduced order models of different pole inherent models under different typical environments are bound to the full-parameter three-dimensional geometric model, physical parameter database and full-order finite element benchmark simulation model of the corresponding pole inherent model and typical environment, and the standardized master twin templates of different pole inherent models under different typical environments are generated.
[0032] It should be further explained that, in the specific implementation process, the process of merging the cross-sectional degrees of freedom and screening the main deformation modes of the full-order finite element benchmark simulation model includes: The pole of the full-order finite element reference simulation model is divided into p (e.g., p=10) horizontal sections along its height (horizontal cuts). The pole element nodes (the vertices of the pole elements are the pole element nodes after the entire pole is cut into thousands of tiny pole elements along the pole height, circumferential direction, and radial direction) are grouped according to their respective horizontal sections. Since the core forces acting on the annular concrete pole during operation are bending and axial compression, this conforms to the basic assumption of plane sections in mechanics of materials: when the pole is subjected to bending or axial load, any cross-section always remains planar. The displacement of all nodes is entirely determined by three principal variables of the cross section: axial displacement, horizontal deflection, and cross section rotation. Therefore, three principal variables are set. Based on these three principal variables, the degrees of freedom of all rod element nodes in each group of horizontal cross sections are merged into three principal degrees of freedom: axial displacement, horizontal deflection, and cross section rotation. This generates a cross section-level full-order model. Through this step, tens of thousands of node degrees of freedom are directly compressed into "dozens of cross sections distributed along the rod height × 3 principal degrees of freedom". The model order is directly reduced by two orders of magnitude and fully conforms to the physical laws of structural deformation without any loss of fundamental accuracy.
[0033] Simulations were performed on the full-order cross-section model, specifically calculating the steady-state deformation patterns under all typical working conditions (wind load, icing load, temperature load, and uneven foundation settlement). This generated a full-scenario deformation sample set covering the entire lifecycle of the pole. The full-scenario deformation sample set includes: Bending deformation corresponding to different levels of wind load (0~40m / s); Bending deformation corresponding to ice loads of different thicknesses; Axial and bending deformations corresponding to uniform temperature change and radial temperature difference; Rigid body rotation and bending deformation corresponding to uneven foundation settlement of different degrees; Dynamic deformation patterns corresponding to different impact loads.
[0034] Each deformation sample corresponds to a set of cross-sectional displacement distributions along the rod height, representing a real deformation form that will occur in actual operation. The sample size of a single model plus a standardized master twin template under a typical environment is no less than 200 sets, fully covering the working conditions of the entire parameter range.
[0035] Since the contributions of different deformation modes to the overall structural response vary greatly, a few core deformation modes can cover the deformation characteristics of most working conditions, and deformation modes with extremely low contribution can be directly discarded without affecting the overall accuracy. Therefore, the following steps are performed: Step 1: Obtain the total structural strain energy corresponding to each deformation sample in the full scene deformation sample set. The total structural strain energy corresponding to the current deformation sample is the sum of the element strain energies of all rod element nodes in sequence. The element strain energy of a rod element node is 0.5 × the equivalent stiffness of the rod element node × the square of the relative deformation at both ends of the rod element node. Sort all deformation samples from high to low according to the total structural strain energy to generate a sample queue. Step 2: Construct a pattern set, select the deformation sample with the highest total strain energy in the sample queue as the first dominant deformation mode, and add the first dominant deformation mode to the pattern set; Step 3: Select deformation samples that have not been added to the model set from the sample queue in descending order of total structural strain and mark them as the main deformation modes. Add the main deformation modes to the model set. Whenever a new main deformation mode is added to the model set, obtain the average error of restoring all deformation samples using the current model set. Step 4: When a new master deformation mode is added to the pattern set, obtain the decrease in the average error of restoring all deformation samples using the current pattern set. When the decrease is lower than the set threshold, stop the filtering and mark the current pattern set as the master deformation pattern set.
[0036] For example, the full-scene deformation sample set includes 200+ sets of deformation samples under all working conditions (wind, icing, settlement, temperature, etc.). Each set of deformation samples is a complete displacement distribution of the rod cross-section, denoted as: ( =Total number of samples), the deformed sample selected from the deformed samples to combine and restore all deformed samples is used as the master deformed mode, denoted as: ( =Number of selected master control transformation modes). This indicates the k-th master control transformation mode; Core rule: Any real deformation sample can be approximated by the linear superposition of the selected master deformation modes, that is: In the formula, Approximate deformations obtained by piecing together the current pattern set; : The weight coefficient corresponding to each mode; set up Original, real-world deformed sample; Step 1: Select only the first master control transformation mode Current pattern set: { }; Use this single pattern to restore all of them one by one. Identify 1 original deformed sample; for each sample, calculate the actual deformation. splicing deformation The error is calculated by averaging the errors of all samples to obtain the current global average error. Because only the strongest deformation mode is used, the overall fitting effect is very poor and the average error is too large.
[0037] Step 2: Add the second main control transformation mode Current pattern set: { , }; Add a new pattern based on the existing patterns, without deleting the old patterns; combine the two patterns and restore all patterns one by one. One sample; recalculate the error of all samples to obtain a new global average error. ; Calculation error reduction magnitude : Because of the addition of a basic deformation form, the fitting ability is strengthened, and the error will definitely decrease.
[0038] Step 3: Continue to overlay the third main control transformation mode. Current pattern set: { , , }; Add another pattern, bringing the pattern set to three; then perform a full sample reconstruction and calculate the average error again. ; Calculate the magnitude of error reduction in this round : .
[0039] If the threshold is set to 0.5%; If a fourth master control transformation mode is added: average error 3.2%, reduction 3.8% > 0.5% → continue; If a fifth master control deformation mode is added: the average error is 2.9%, and the reduction is 0.3% < the threshold, then adding another mode will have a negligible effect on the overall fitting accuracy, which is redundant. Therefore, the addition is stopped, and the first 4 are finally selected as the official master control deformation mode set.
[0040] It should be further explained that, in the specific implementation process, the process of reconstructing the low-dimensional mechanical equations includes: The first master-controlled deformation mode in the master-controlled deformation mode set is marked as the first independent mode. For the second master-controlled deformation mode in the master-controlled deformation mode set, the deformation components that completely overlap with the first independent mode are removed, and only the unique deformation parts are retained to form the second independent mode. For the third master-controlled deformation mode in the master-controlled deformation mode set, all overlapping components between the third master-controlled deformation mode and the first independent mode, as well as all overlapping components between the third master-controlled deformation mode and the second independent mode are removed, and only the unique parts are retained to form the third independent mode. This process is repeated until all overlapping components between the last master-controlled deformation mode in the master-controlled deformation mode set and the existing independent modes are removed to form the last independent mode. Finally, a set of independent deformation modes that do not overlap and are completely independent is obtained. Generate an independent mode matrix based on all independent modes in the independent deformation mode set, obtain the structural dynamic equilibrium equations followed by the full-order finite element benchmark simulation model, construct the deformation superposition mapping relationship between the structural dynamic equilibrium equations and the independent mode matrix, construct the low-dimensional equivalent mechanical equilibrium equations followed by the reduced-order finite element benchmark simulation model based on the deformation superposition mapping relationship and the virtual work equilibrium principle, and generate the reduced-order finite element benchmark simulation model based on the low-dimensional equivalent mechanical equilibrium equations.
[0041] The structural dynamic equilibrium equations followed by the full-order finite element benchmark simulation model are: In the formula, : Full-order displacement vector, with dimension ( (This represents the total degrees of freedom of the full-order model, corresponding to the displacement variables of tens of thousands of nodes). , : These are the full-order acceleration vector and velocity vector, respectively, and are the second and first derivatives of displacement with respect to time; Full-order mass matrix, dimension This describes the mass distribution and inertial coupling relationship of each node; Full-order damping matrix, dimension Describes the energy dissipation characteristics during structural vibration; Full-order stiffness matrix, dimensions This describes the resistance characteristics of a structure to elastic deformation. : Full-order external load vector, dimension This describes the magnitude of the effect of external forces such as wind, icing, and collisions at each node.
[0042] Through the preceding steps, k independent modes with no overlapping components have been selected. The value of k is usually 60-80, which is much smaller than the full-order degrees of freedom. .
[0043] Each independent mode corresponds to a fixed rod deformation shape, which can be represented as a full-order displacement vector. ( ), dimension and full-order displacement vector Consistency is used to describe the displacement ratio of each node in this mode; all independent modes together form the independent mode matrix. , , dimension .
[0044] Based on the small deformation assumption in mechanics of materials and the principle of linear elastic superposition: any real deformation of a cement pole under normal operating conditions can be expressed as a linear superposition of k independent deformation modes. That is, by combining each mode according to a certain weight ratio, the complete deformation of the pole can be restored. The expression is: ,in: ,for Low-dimensional weight vector, each element This represents the weight of the i-th independent mode in the total deformation; The mapping relationship between velocity and acceleration is similar: , (pattern matrix) (It has a fixed form and is independent of time). and These represent the weight matrices for the corresponding velocities and accelerations, respectively.
[0045] Through this mapping, the descriptive variables of structural deformation are compressed from tens of thousands of node displacements to dozens of mode weights, completing the physical compression of the variable dimension. Subsequently, it is only necessary to derive the balance equation satisfied by these k weight variables.
[0046] Then select the first Individual deformation modes Substituting this as a virtual displacement into the virtual work equilibrium equation: , to obtain the corresponding number Scalar equilibrium equations for each mode: ,in, , It is a variational symbol (virtual shift operator). for The transpose of the equation has three terms on the left side corresponding to the virtual work done by inertial force, damping force, and elastic force under virtual displacement, respectively, while the right side represents the virtual work done by external force.
[0047] Deformation superposition mapping relationship , , Substituting the above scalar equilibrium equations and replacing the full-order displacement variables with low-dimensional weight variables, we get: ; Finally, Each independent deformation mode is analyzed by repeating the above steps sequentially. Each mode corresponds to a scalar equation, resulting in a total of [number] equations. There are 1 independent equations.
[0048] This Arrange the equations in order into a matrix form to obtain the low-dimensional equivalent mechanical equilibrium equations: The low-dimensional equivalent mechanical equilibrium equation has the same form as the full-order equilibrium equation, but the dimension is different. The order (tens of thousands) decreased to Order (tens), the solution size is reduced by more than three orders of magnitude, of which , , , These are the equivalent mass matrix, equivalent damping matrix, equivalent stiffness matrix, and equivalent load vector after order reduction, respectively.
[0049] It should be further explained that, in the specific implementation process, after the order reduction is completed, the accuracy of the reduced finite element reference simulation model is ensured through verification, including: using the verification working condition that is not involved in the mode construction, comparing the core outputs (rod top deflection, root maximum stress, vibration dominant frequency, damping ratio) of the reduced model and the full-order model. If the error of a certain working condition exceeds the threshold, the deformation mode of the working condition is processed according to the rule of eliminating overlapping components to obtain an independent mode, which is then added to the independent deformation mode set until the errors of all core working conditions meet the accuracy requirements.
[0050] It should be further explained that, in the specific implementation process, based on the basic attribute parameters and deployment parameters of each cement pole, a standardized master twin template is matched for each cement pole, and the full-order finite element reference simulation model in the standardized master twin template is instantiated and fine-tuned. The process of obtaining the instantiated full-order finite element reference simulation model includes: Based on a standardized master twin template library, this invention provides a fully automated instantiation pipeline that includes "automatic matching, layered correction, initial damage assignment, self-calibration, and batch issuance." This pipeline supports the batch parallel generation of twins for tens of thousands of existing utility poles. Compared to traditional 3D scanning reverse modeling, it significantly reduces modeling costs and the average generation time per pole. Furthermore, the generated twins directly match the actual operating state of the physical utility poles, rather than their brand-new factory condition, and can output accurate health assessment results immediately upon deployment.
[0051] Analyze the ledger data to obtain the basic attribute parameters (inherent model, pole height, concrete grade, reinforcement specifications, manufacturing date, commissioning date, etc.) and deployment parameters (line, latitude and longitude, geographical environment type, soil type, pole sensor deployment) for each cement pole. For each cement pole, based on the inherent model in the basic attribute parameters and the latitude, longitude, geographical environment type and soil type in the deployment parameters, a standardized master twin template is matched from the standardized master twin template library. The corresponding inherent model is consistent with the inherent model of the cement pole, and the corresponding typical environment is consistent with the latitude, longitude, geographical environment type and soil type of the cement pole. Based on the standardized master twin template, the full-parameter three-dimensional geometric model, physical parameter database, full-order finite element reference simulation model and reduced-order finite element reference simulation model of the cement pole are loaded. Based on the basic attribute parameters, the full-parameter three-dimensional geometric model and the full-order finite element reference simulation model of the cement pole are geometrically corrected to generate an instantiated three-dimensional high-precision geometric model and a full-order finite element reference simulation model with geometric correction. The process of correcting the full-parameter three-dimensional geometric model of a cement pole is as follows: based on the pole height, tip diameter, root diameter, reinforcement specifications, foundation dimensions and other parameters in the basic attribute parameters, the parametric geometric template is driven to automatically adjust the variable values to generate a 1:1 geometric model that is completely consistent with the design parameters of the pole, automatically divide the simulation mesh, and complete the instantiation of the geometric dimensions. Since the full-order finite element benchmark simulation model is itself a parametric finite element model constructed based on a fully parametric 3D geometric model (with a fixed mesh topology and node coordinates that can be scaled by parameters), when performing geometric correction on the full-order finite element benchmark simulation model: input the pole height, top diameter, root diameter, wall thickness, number / diameter of main reinforcement bars, protective layer thickness, and other geometric parameters. Through the parametric mapping engine, the node coordinates of the full-order finite element benchmark simulation model are directly scaled and their positions corrected, and the geometric shape of the concrete solid elements and reinforcement line elements is automatically updated.
[0052] Based on the deployment parameters, the damage parameters corresponding to the damage evolution parameter library in the physical parameter database of the cement pole and the damage parameters and environmental correction coefficients corresponding to the environmental correction coefficient library are matched. For example, for utility poles in coastal areas, the system automatically applies a salt spray environment correction factor, multiplying the benchmark value of steel bar corrosion rate by 2.0 to match the actual corrosion environment.
[0053] Based on the basic parameters corresponding to the basic material parameter library, the damage parameters corresponding to the damage evolution parameter library, and the damage parameter and environmental correction coefficients corresponding to the environmental correction coefficient library in the physical parameter database, the material parameter correction and initial damage parameter correction are performed on the full-order finite element baseline simulation model that has undergone geometric correction, thereby obtaining an instantiated full-order finite element baseline simulation model; specifically: Basic material parameter assignment: The matched basic parameters such as the elastic modulus, strength, and density of concrete, the yield strength and elastic modulus of steel reinforcement, and the soil bearing capacity are assigned in batches to the corresponding units (concrete unit group, steel reinforcement unit group, and foundation soil unit group).
[0054] Initial damage field mapping: Based on the matched damage parameters, the damage parameters are mapped to the element level. Creep damage and fatigue damage are uniform damage over the entire domain, and the material stiffness matrix is corrected globally according to the overall reduction factor. Reinforcing bar corrosion damage: This is a localized, concentrated damage, where stiffness reduction and corrosion expansion equivalent loads are applied only to the reinforcing bar units and surrounding concrete units in the 0.5~1.5m section at the root of the bar. The final result is an initial stiffness field with real service conditions, rather than a completely new zero-damage model.
[0055] The instantiated full-order finite element benchmark simulation model and the instantiated 3D high-precision geometric model are packaged into a cloud twin data package, and the cloud twin data package is sent to the cloud twin platform.
[0056] It should be further explained that, in the specific implementation process, the process of sequentially instantiating and fine-tuning the reduced-order finite element baseline simulation model to obtain the instantiated reduced-order finite element baseline simulation model includes: Obtain the physical correction scaling factor between the instantiated full-order finite element reference simulation model and the full-order finite element reference simulation model. Based on the physical correction scaling factor, correct the equivalent mass matrix, equivalent damping matrix, and equivalent stiffness matrix in the low-dimensional equivalent mechanical equilibrium equation followed by the reduced-order finite element reference simulation model to generate the instantiated reduced-order finite element reference simulation model. Based on the instantiation of a full-order finite element baseline model, there are two technical routes for instantiating a reduced-order finite element baseline model: One approach involves resampling snapshots of the instance-based full-order model, re-extracting independent deformation modes, and then reducing the order. However, this approach has the drawback of high computational cost, making it unsuitable for batch generation. Another approach is to reuse the pre-solved low-dimensional equivalent mechanical equilibrium equations. Since the main deformation modes of cement poles of the same type remain unchanged within the normal service parameter fluctuation range, with only stiffness, mass, damping, and load amplitudes varying, the low-dimensional equivalent mechanical equilibrium equations are reusable. Therefore, only the equivalent mass matrix, equivalent damping matrix, equivalent stiffness matrix, and equivalent load vector of the low-dimensional equivalent mechanical equilibrium equations are parametrically modified; for example: Let the parameter set of the full-order finite element benchmark simulation model be: The parameter set for instantiating the full-order finite element baseline simulation model is: ,in, The elastic modulus of the concrete for the template reference pole; The actual elastic modulus of concrete is used as an example for cement poles. The equivalent density of the concrete for the template reference pole; The equivalent density of concrete for example utility poles; The moment of inertia of the corresponding section of the template reference pole; The example pole has a corresponding section with bending moment of inertia. The area of the annular cross-section at the root of the template reference rod; This is the cross-sectional area of the example pole; The total height of the template reference pole; This is the actual total height of the example pole; Calculate the physical correction ratio for each item: , , , and ;in, This represents the ratio of elastic modulus. This refers to the density ratio; The ratio of the moment of inertia of the cross section; This represents the cross-sectional area ratio; This represents the ratio of rod length.
[0057] Simultaneously import: Basic constraint stiffness correction coefficient Global stiffness reduction scalar coefficient and local region reduced degree of freedom stiffness attenuation correction amount .
[0058] Apply layer-by-layer physical scaling to the equivalent mass matrix, equivalent damping matrix, and equivalent stiffness matrix in the equivalent mechanical equilibrium equation: Equivalent mass matrix correction: For slender rods primarily subjected to bending, mass is proportional to volume. ; This is the corrected equivalent mass matrix; The equivalent stiffness matrix shows that the basic corrected bending stiffness is proportional to the elastic modulus and the moment of inertia of the section, while the deflection is related to the cube of the rod length. , This is the intermediate reduced-order stiffness matrix after scaling by geometric and material parameters; Introducing basic boundary constraint correction: Basic constraints only affect low-order bending modes, and only correct the diagonal elements of the first few modal orders corresponding to the matrix. In the formula, This represents a local element overlay correction, without altering higher-order modal terms. The additional stiffness correction matrix resulting from the basic constraints, and the basic constraint stiffness correction coefficients. As input parameters; Introducing initial damage stiffness degradation correction: In the formula, The reduced-order equivalent stiffness matrix of a single concrete pole example after correction of geometric parameters, material parameters, foundation boundary constraints, global creep / fatigue damage reduction, and local steel reinforcement corrosion damage. To correct the sparse matrix of local stiffness of steel reinforcement corrosion, with the steel reinforcement corrosion damage degree As input; The scalar reduction factor for global stiffness is derived from the concrete creep damage degree. Fatigue damage degree under alternating load Determined jointly by the inputs; Synchronous correction of damping matrix: Damping is coupled with stiffness and mass, and is synchronously scaled according to the stiffness correction ratio. , This is the corrected equivalent damping matrix.
[0059] A lightweight geometric model is constructed based on an instantiated 3D high-precision geometric model. Specifically, key geometric parameters such as pole height, root diameter, tip diameter, pole taper, and foundation dimensions are extracted. The outer contour surface of the pole is generated through parametric reconstruction. Redundant geometric features such as internal steel reinforcement and minor auxiliary components are removed. The surface mesh is then thinned to obtain a lightweight geometric model usable on the edge side. Simultaneously, the positions of each horizontal section along the pole height are marked as state alignment references. Based on the pole sensor deployment data in the deployment parameters and the lightweight geometric model, a sensor point mapping table is constructed. Specifically, the pole taper and height parameters of the lightweight geometric model are read; the horizontal section number (p-segment) to which the sensor belongs is determined based on the sensor installation height; the 3D coordinates of the sensor on the outer surface of the pole wall are calculated based on the circumferential angle; and the sensor point mapping table is generated and stored in the edge twin data package. The instantiated reduced-order finite element baseline simulation model, lightweight geometric model, and sensor point mapping table are packaged into an edge twin data package (single rod volume ≤ 15MB), and the edge twin data package is sent to the edge gateway.
[0060] It should be further explained that, in the specific implementation process, each cement pole is monitored in real time based on an instantiated reduced-order finite element benchmark simulation model, and the measured simulation residual rates of various types of indicators are obtained. The process of analyzing the measured simulation residual rates of various types of indicators to determine whether a transient event alarm response or a warning-level response is triggered includes: The system acquires the acquisition cycle of various indicators of the current cement pole sensor terminal. It collects measured data (tilt angle, strain, temperature) of various indicators of the cement pole at the end timestamp of each acquisition cycle and caches it to the edge gateway. The edge gateway sets its analysis cycle (10 minutes). At the start timestamp of the analysis cycle, the edge gateway retrieves the cached measured data (tilt angle, strain, temperature) from the sensor terminal. Simultaneously, it acquires local meteorological data (real-time wind speed, ambient temperature data) for the current cement pole. For example, an integrated micro-meteorological terminal can be installed on a cement pole in a certain area, including a wind speed sensor and an air temperature sensor, with optional humidity and air pressure sensors. The micro-meteorological terminal has an independent low-power control unit. It communicates with the edge gateway via wired RS485 or wireless LoRa, inputting measured data and local meteorological data into an instantiated reduced-order finite element benchmark simulation model. Based on the instantiated reduced-order finite element benchmark simulation model, it outputs the simulation benchmark value for the current judgment period, including core mechanical indicators such as the stress of key sections of the pole, the deflection at the top of the pole, the inclination angle of the pole, and the first-order vibration frequency. The time for a single simulation solution is ≤5ms, with almost no additional power consumption. The simulation benchmark value is then transmitted back to the sensor terminal, and the newly received simulation benchmark value by the sensor terminal is marked as the judgment standard. By comparing the measured data of various indicators with the judgment criteria in real time through sensor terminals, the measured simulation residual rate of each type of indicator is obtained. ,in, This represents the measured simulation residual rate. This represents the measured data. The criteria for judgment are defined by setting the measured simulation residual rate warning threshold and national standard safety limit for each type of indicator. Each type of indicator is compared with the national standard safety limit to obtain the cumulative time for each type of indicator to exceed the national standard safety limit. If the cumulative time for any type of indicator exceeds the preset time threshold (10ms), the type of indicator is marked as an abnormal indicator, triggering a transient event alarm response. Specifically: 1. On-site audio-visual alarm triggering: The edge gateway directly triggers the audio-visual alarm module of the on-site terminal through the IO port, activating the high-frequency flashing light and high-decibel warning sound to warn irrelevant personnel on site to stay away from the dangerous power pole and avoid secondary accidents; Real-time push of operation and maintenance information: The gateway pre-stores alarm templates and the address book of operation and maintenance personnel. After being triggered, the alarm information is immediately pushed to the operation and maintenance platform and the mobile phone of the operation and maintenance person in charge of the area via 4G / power private network. The content includes: unique pole ID, precise latitude and longitude location, and abnormal indicator type. The end-to-end delivery delay of alarm information is ≤100ms. After a transient event alarm response is triggered, the alarm link is activated immediately. The alarm link adopts a hardware bypass direct design, and the total alarm link latency is ≤100ms. The alarm link is independent of the twin simulation calculation link and adopts hardware pass-through logic. It does not need to wait for simulation calculation, data packaging and other processes. From the capture of abnormal signal to the alarm output, it only goes through two links: hardware interrupt response and instruction issuance, which physically guarantees an alarm latency of ≤100ms.
[0061] 2. Upon alarm triggering, the data acquisition and transmission strategy is simultaneously upgraded to provide high-density data support for fault tracing and damage diagnosis: The edge gateway sends backup terminal activation and frequency increase commands to all sensor terminals, activating the backup sensor terminals of all sensor terminals. All sensor terminals immediately exit the low-power mode and all sensors operate at full load: tilt and strain sensors switch to continuous sampling mode of 1 time / second to continuously collect structural response data throughout the process; vibration sensors switch to continuous high-frequency sampling mode (default 100Hz) to fully record the vibration time history curve of the impact process. The entire switching process takes less than 1 second, achieving a seamless transition from low frequency to high frequency. At the same time, it automatically retransmits the cached data from the 30 seconds prior to the alarm trigger, ensuring the integrity of the event process.
[0062] The edge gateway's assessment cycle was switched to a second-level cycle (1s), changing from a 10-minute cycle to real-time, second-by-second analysis: the reduced-order finite element model is solved once per second, synchronously tracking the real-time evolution of stress, deflection, and damage, and quickly determining whether the structure will experience instability, collapse, or other secondary risks; the trend of the measured-simulation residual is calculated in real time, and if the residual continues to expand, the alarm level is automatically upgraded, and the risk assessment results are pushed to the cloud.
[0063] The measured data, abnormal indicators, and simulation benchmark values of the edge gateway from the sensor terminal are packaged into transient alarm data packets at second-level intervals and uploaded to the cloud twin platform. If the cumulative time corresponding to each type of indicator does not exceed the preset time threshold, the measured simulation residual rate of each type of indicator is compared with the corresponding measured simulation residual rate warning threshold. If the measured simulation residual rate corresponding to the type of indicator exceeds the measured simulation residual rate warning threshold within two consecutive judgment periods, a warning-level response is triggered. The measured data of the sensor terminal and the simulation benchmark value of the edge gateway are packaged into a warning data package according to the judgment period and uploaded to the cloud twin platform.
[0064] It should be further explained that, in the specific implementation process, the process of obtaining the current data acquisition cycle of the cement pole sensor terminal includes: Obtain the measured simulation residual rate of each type of indicator in the measured data of the previous acquisition cycle, and preset the acquisition cycle mapping table. The acquisition cycle mapping table includes the acquisition cycle corresponding to different measured simulation residual rates of each type of indicator. Based on the measured simulation residual rate of each type of indicator in the measured data of the previous acquisition cycle and the acquisition cycle mapping table, obtain the acquisition cycle of the sensor terminal for acquiring each type of indicator of the cement pole at the current time.
[0065] It should be further explained that, in the specific implementation process, independent mapping tables were designed for the three core indicators of cement poles, with different indicators matched to their respective data collection periods; an example is shown below: Deformation type (tilt sensor) mapping table: Residual rate range <2%, corresponding to a sampling period of 15 minutes / time; residual rate range 2%~5%, corresponding to a sampling period of 5 minutes / time; residual rate range 5%~10%, corresponding to a sampling period of 1 minute / time; residual rate range ≥10%, corresponding to a sampling period of 1 second / time; Stress-related (strain sensor) mapping table: Residual rate range <2%, corresponding to an acquisition cycle of 30 minutes / time; residual rate range 2%~5%, corresponding to an acquisition cycle of 10 minutes / time; residual rate range 5%~10%, corresponding to an acquisition cycle of 1 minute / time; residual rate range ≥10%, corresponding to an acquisition cycle of 1 second / time; Dynamic (vibration sensor) mapping table: Residual rate range <2%, corresponding to a sampling period of 60 minutes / time; residual rate range 2%~5%, corresponding to a sampling period of 15 minutes / time; residual rate range 5%~10%, corresponding to a sampling period of 5 minutes / time; residual rate range ≥10%, corresponding to a sampling period of continuous 100Hz sampling. When the residual rate falls within the interval boundary, a higher frequency acquisition period is selected, which is more conservative and safer.
[0066] It should be further explained that, in the specific implementation process, activating the instantiated full-order finite element benchmark simulation model, comparing and verifying the output results of the instantiated full-order finite element benchmark simulation model with the output results of the instantiated reduced-order finite element benchmark simulation model, and then, based on the comparison and verification results, deactivating the edge gateway's warning-level response or performing damage tracing includes: After receiving the early warning data packet uploaded by the edge gateway, the cloud twin platform locates the cloud twin data packet corresponding to the cement pole through the unique identity identifier of the pole pre-bound by the edge gateway, and activates the instantiated full-order finite element benchmark simulation model and the instantiated three-dimensional high-precision geometric model in the cloud twin data packet. The measured data from the sensor terminals in the early warning data packet and the local meteorological data of the cement pole are input into an instantiated full-order finite element reference simulation model. The simulation reference value is output based on the instantiated full-order finite element reference simulation model. The simulation reference value in the early warning data packet is compared with the simulation reference value output by the instantiated full-order finite element reference simulation model, and the deviation of each type of feature is output. For example, for the feature of pole top deflection in the simulation reference value, the deviation calculation formula is as follows: ,in, This indicates the deviation of the deflection at the top of the rod. The deflection at the top of the rod is the simulated baseline value in the early warning data packet. Take the rod top deflection in the simulation reference value output by the instantiated full-order finite element reference simulation model; obtain the average deviation value of each type of feature deviation (sum and average the deviations of each type of feature deviation to obtain the average deviation value), compare the average deviation value with the preset deviation threshold (5%, the 5% threshold is set according to the accuracy benchmark of the full-order model field calibration error ≤ 5% in the instruction manual, exceeding this range indicates that the instantiation correction or parameter update of the reduced-order model is inaccurate). If the average deviation value is less than or equal to the preset deviation threshold, the warning level response of the edge gateway is released, and the instantiation twin construction module is re-executed to generate a new instantiation reduced-order finite element reference simulation model and send it back to the edge gateway. If the average deviation value is greater than the preset deviation threshold, the measured data of the sensor terminal corresponding to the unique identification of the pole in several historical acquisition cycles (acquisition cycles within the most recent six weeks) are extracted. The simulation benchmark value output by the instantiated full-order finite element benchmark simulation model, the deviation of each type of feature, the measured data of the sensor terminal in the early warning data package, and the measured data of several historical acquisition cycles are input into the preset damage source estimation model. According to the damage source estimation model, the damage type, damage location, and the degree of deterioration corresponding to the damage location are output. The damage location and the damage type and degree of deterioration corresponding to the damage location are marked on the instantiated three-dimensional high-precision geometric model.
[0067] It should be further explained that the preset damage source estimation model is a multi-input physical-driven damage classification and regression model. The input includes two types of core data: 1. Simulation baseline values output by the instantiated full-order finite element baseline simulation model (simulation baseline values such as global stress, deflection, section rotation angle, and dynamic main frequency); 2. Multi-dimensional measured time-series data from field sensors (tilt angle, strain, vibration, temperature and humidity, wind speed); the output is the spatial coordinates of the rod damage (horizontal section + circumferential region) + corresponding damage type (creep / corrosion / fatigue / foundation settlement) + quantified deterioration degree (0~1 normalized damage degree). The model is built entirely based on the instantiated full-order finite element baseline simulation model, the damage evolution parameter library in the physical parameter database, and the standardized working condition sample library. There is no pure black-box fitting, and each set of input and output has a basis in structural mechanics and concrete durability physics. The process of constructing a standardized operating condition sample library includes: The samples are derived from the full-order finite element simulation results of the standardized master twin template library, specifically: traversing all pole models and 5 typical environments in the standardized master twin template library; For each combination (model + environment), batch modify the damage parameters: Global damage: Gradient settings for multiple creep damage degrees. Fatigue damage degree (0, 0.2, 0.4, 0.6, 0.8, 1.0); Localized corrosion: Gradient corrosion damage is applied only to the root section from 0.5 to 1.5 m. The remaining pole segments showed no corrosion; foundation damage: different foundation settlement amounts and foundation constraint reduction coefficients were set at different gradients; For each combination of damage parameters, run an instantiated full-order finite element baseline simulation model and output a complete output set, including: Global field characteristics: deflection sequence along p horizontal sections, maximum tensile / compressive stress sequence of the sections, rotation angle of each section, and natural frequencies of each order; Load environment characteristics: corresponding average wind speed, ambient temperature, and soil parameters; Manually labeled: damage type, cross section number to which the damage belongs, circumferential damage region, and normalized damage degree.
[0068] All samples are shuffled and divided into three parts in a 7:2:1 ratio: 70% training set, 20% validation set, and 10% test set. Each sample is stored in a uniform format: input feature vector + standardized damage label.
[0069] After the standardized working condition sample library is built, the sample data is input into the damage source tracing prediction model for training. The damage source tracing prediction model pre-extracts features from the sample data, including: Feature extraction of output data from instantiated full-order finite element benchmark simulation model (first input branch): Deformation characteristics: mean deflection of each section, deflection gradient along the height of the rod, deflection difference between the top and root of the rod, and rotation gradient of each section; Stress characteristics: maximum tensile / compressive stress at each section, stress ratio at the root to in the middle of the rod, and index of stress abrupt change locations; Dynamic characteristics: first / second order dominant frequency offset, damping ratio change rate; Load environment characteristics: average wind speed, average annual ambient temperature, and soil constraint stiffness under the corresponding working conditions.
[0070] Sensor measured data feature extraction (second input branch): Static characteristics: mean dip angle within the period, mean strain, and mean ambient temperature; Trend characteristics: residual growth rate and annual cumulative increase in strain over the past 6 weeks; Impact characteristics: peak vibration value, proportion of high-frequency vibration (used to distinguish between external impact and fatigue damage). Residual characteristics of simulation and actual measurement: deviations between actual and simulation for each index (various types of characteristic deviations), and the gradient of deviation distribution along the rod height.
[0071] Finally, all features are normalized and mapped to the [0,1] interval to unify the dimensions and eliminate numerical fluctuations caused by differences in pole height and model size.
[0072] The damage source prediction model adopts a dual-branch fusion multi-task physical model. The two branches process full-order simulation features and sensor measured features respectively. The fusion layer outputs a shared feature vector, and the back-end completes three tasks in parallel: Task 1: Damage area classification, outputting the horizontal section number and circumferential position of the damage (classification task); Task 2: Damage type classification, outputting creep / corrosion / fatigue / foundation settlement / composite damage (classification task); Task 3: Deterioration degree regression, outputting the single-type damage degree between 0 and 1 (regression task).
[0073] Damage source tracing and prediction models include: Dual-branch coding layer: Simulation field branch encoder: multi-layer fully connected layer, inputting cross-sectional mechanical features extracted from full-order simulation, and extracting the spatial distribution law of stress deformation in the whole domain; Tested temporal branch encoder: Lightweight temporal network, input sensor temporal measured features, extract long-term evolution trend of structure; Feature fusion layer: The two encoded outputs are spliced and merged, and physical constraint regularization terms are added: the output range is limited based on structural mechanics rules. For example, corrosion damage can only appear in the 0.5~1.5m section at the root. If the model output exceeds this range, a loss penalty is automatically applied to ensure that the output conforms to physical common sense and to eliminate unreasonable prediction results.
[0074] Multi-task output head: Spatial classification head: Outputs the damage probability distribution of p horizontal cross sections, and the cross section with the highest probability is the estimated damage location; Damage type classification head: 4 basic damage types + complex damage types, totaling 5 classification channels; Damage regression head: Outputs the degree of degradation of a single type of damage, with a value range constraint of 0~1.
[0075] Loss function design: Total loss = Spatial classification loss + Damage type classification loss + Damage degree regression loss. The spatial classification loss and damage type classification loss adopt the cross-entropy loss function; the damage degree regression loss adopts the MSE loss function. After training, the accuracy of each output index is verified using a test set: 1. Damage section location error: not exceeding 1 horizontal segment; 2. Damage type identification accuracy ≥ 95%; 3. Damage degree prediction relative error ≤ 5% (matching the accuracy threshold of ≤ 5% in full-order simulation calibration); if the accuracy is not met, the corresponding working condition samples are expanded and training is iterated again.
[0076] After the damage source prediction model is trained, an offline verification process is set up to ensure that the model is compatible with all pole models and environments: 1. Randomly select any type and environment of utility poles from the database, and randomly set a set of damage parameters as the true values; 2. Run the full-order model to obtain the simulation baseline values; 3. Input the simulation baseline and measured data into the damage source prediction model; 4. Compare the model output (damage location, damage type, and degree of degradation) with the preset true value. If the error exceeds the standard, supplement the corresponding working condition samples and retrain. After completing offline verification covering all models and environments, the model was finalized.
[0077] It should be further explained that, in the specific implementation process, the process of quickly performing event authenticity verification and, based on the verification results, deactivating the transient event alarm response of the edge gateway or performing damage tracing includes: After receiving the transient alarm data packet uploaded by the edge gateway, the cloud twin platform locates the cloud twin data packet corresponding to the cement pole through the unique identity identifier of the pole pre-bound by the edge gateway, activates the instantiated full-order finite element benchmark simulation model and the instantiated three-dimensional high-precision geometric model in the cloud twin data packet, and obtains the abnormal indicators in the transient alarm data packet. The sensor terminal that collects abnormal indicators is marked as the terminal to be verified. The backup sensor terminal of the terminal to be verified is obtained and deployed on the same cement pole. The difference between the abnormal indicators collected by the terminal to be verified and the abnormal indicators collected by the backup sensor terminal is obtained from the transient alarm data packet. The difference threshold is set. If the difference between the values of two consecutive second-level cycles is greater than the difference threshold, a sensor fault alarm is generated and the transient event alarm response of the edge gateway is deactivated. If there is no numerical difference greater than the difference threshold between two consecutive second-level cycles, the measured data of the sensor terminal and the local meteorological data of the cement pole in the transient alarm data packet are input into the instantiated full-order finite element reference simulation model. The simulation reference value is output according to the instantiated full-order finite element reference simulation model. The simulation reference value in the transient alarm data packet is compared with the simulation reference value output by the instantiated full-order finite element reference simulation model, and the deviation of each type of feature is output. Subsequently, the measured data of the sensor terminal corresponding to the unique identification of the pole in several historical acquisition cycles are extracted. The simulation benchmark value output by the instantiated full-order finite element benchmark simulation model, the deviation of various types of features, the measured data of the sensor terminal in the transient alarm data packet, and the measured data of several historical acquisition cycles are input into the preset damage source prediction model. According to the damage source prediction model, the damage type, damage location, and the degree of deterioration corresponding to the damage location are output. The damage location and the damage type and degree of deterioration corresponding to the damage location are marked on the instantiated three-dimensional high-precision geometric model.
[0078] The above embodiments are only used to illustrate the technical methods 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 methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.
Claims
1. A safety monitoring system for cement pole structures based on digital twins, characterized in that, It includes a cloud-based twin platform, which has communication connections to a standardized master twin template construction module, a standardized master twin template library, an instantiated twin construction module, a real-time monitoring module, an early warning analysis module, and an alarm analysis module. The standardized master twin template construction module is used to construct full-order finite element benchmark simulation models corresponding to different pole types under different typical environments. It performs cross-section-level degree of freedom merging, master deformation mode screening, and low-dimensional mechanical equation reconstruction operations on the full-order finite element benchmark simulation model to generate a reduced-order finite element benchmark simulation model. Based on the full-order finite element benchmark simulation model, the reduced-order finite element benchmark simulation model, the preset full-parameter three-dimensional geometric model, and the preset physical parameter database, a standardized master twin template is constructed. The standardized master twin template library is used to store the standardized master twin templates corresponding to different pole models under different typical environments; The instantiated twin construction module matches the standardized master twin template of each cement pole according to the basic attribute parameters and deployment parameters of each cement pole, and performs instantiated fine-tuning of the full-order finite element reference simulation model and the reduced-order finite element reference simulation model in the standardized master twin template in sequence to obtain the instantiated full-order finite element reference simulation model and the instantiated reduced-order finite element reference simulation model. The real-time monitoring module is used to monitor each cement pole in real time based on the instantiated reduced-order finite element benchmark simulation model of each cement pole, obtain the measured simulation residual rate of each type of index, analyze the measured simulation residual rate of each type of index, and determine whether a transient event alarm response or a warning-level response is triggered. The early warning analysis module is used to activate and instantiate a full-order finite element benchmark simulation model when an early warning level response is triggered. It then compares and verifies the output of the instantiated full-order finite element benchmark simulation model with the output of the instantiated reduced-order finite element benchmark simulation model. Based on the comparison and verification results, it either cancels the early warning level response of the edge gateway or performs damage tracing, including: After receiving the early warning data packet uploaded by the edge gateway, the cloud twin platform locates the cloud twin data packet corresponding to the cement pole through the unique identity identifier of the pole pre-bound by the edge gateway, and activates the instantiated full-order finite element benchmark simulation model and the instantiated three-dimensional high-precision geometric model in the cloud twin data packet. The measured data from the sensor terminal in the early warning data packet and the local meteorological data from the cement pole are input into the instantiated full-order finite element reference simulation model. The simulation reference value is output according to the instantiated full-order finite element reference simulation model. The simulation reference value in the early warning data packet is compared with the simulation reference value output by the instantiated full-order finite element reference simulation model. The deviation of each type of feature is output, and the average deviation value of each type of feature deviation is obtained. The average deviation value is compared with the preset deviation threshold. If the average deviation value is less than or equal to the preset deviation threshold, the early warning level response of the edge gateway is cancelled. If the average deviation value is greater than the preset deviation threshold, the measured data of the sensor terminal corresponding to the unique identification of the pole in several historical acquisition cycles are extracted. The simulation benchmark value output by the instantiated full-order finite element benchmark simulation model, the deviation of each type of feature, the measured data of the sensor terminal in the early warning data package and the measured data of several historical acquisition cycles are input into the preset damage source estimation model. According to the damage source estimation model, the damage type, damage location and the degree of deterioration corresponding to the damage location are output. The damage location and the damage type and degree of deterioration corresponding to the damage location are marked on the instantiated three-dimensional high-precision geometric model. The alarm analysis module is used to quickly perform event authenticity verification when a transient event alarm response is triggered, and to deactivate the transient event alarm response of the edge gateway or perform damage tracing based on the verification result.
2. The safety monitoring system for cement pole structures based on digital twins according to claim 1, characterized in that, The process of constructing full-order finite element baseline simulation models for different pole types under various typical environments, and building a standardized master twin template based on the full-order finite element baseline simulation model, the reduced-order finite element baseline simulation model, the preset full-parameter three-dimensional geometric model, and the preset physical parameter database, includes: Based on the national standard pole production drawings, a full-parameter three-dimensional geometric model of poles of different inherent models was constructed. Based on laboratory test data and industry standard pre-calibrated physical parameter database, the physical parameter database includes a basic material parameter library, a damage evolution parameter library and an environmental correction coefficient library corresponding to different pole inherent models and typical environments. Based on the full-parameter three-dimensional geometric model and physical parameter database, a full-order finite element benchmark simulation model corresponding to different pole types under different typical environments is constructed. The full-order finite element benchmark simulation model corresponding to different pole types under different typical environments is subjected to cross-section-level degree of freedom merging, main deformation mode screening and low-dimensional mechanical equation reconstruction operations to generate a reduced-order finite element benchmark simulation model corresponding to different pole types under different typical environments. Lightweight encapsulation and edge adaptation are performed on the reduced-order finite element baseline simulation models of different pole types under different typical environments to generate pre-reduced-order models of different pole types under different typical environments after encapsulation. By binding the pre-reduced order model corresponding to different pole inherent models under different typical environments with the full-parameter three-dimensional geometric model, physical parameter database, and full-order finite element benchmark simulation model of the corresponding pole inherent model and typical environment, a standardized master twin template corresponding to different pole inherent models under different typical environments is generated.
3. The safety monitoring system for cement pole structures based on digital twins according to claim 2, characterized in that, The process of merging cross-section-level degrees of freedom and screening dominant deformation modes for a full-order finite element baseline simulation model includes: The rod of the full-order finite element reference simulation model is divided into p horizontal sections along the rod height. The rod element nodes of the full-order finite element reference simulation model are grouped according to their respective horizontal sections. Three main control variables are set. Based on the three main control variables, the degrees of freedom of all rod element nodes in each group of horizontal sections are merged into three main control degrees of freedom to generate a section-level full-order model. Simulation is performed on the full-order model at the section level to generate a full-scene deformation sample set. The following steps are then executed: Step 1: Obtain the total structural strain energy corresponding to each deformation sample in the full-scene deformation sample set, sort all deformation samples from high to low according to the total structural strain energy, and generate a sample queue; Step 2: Construct a pattern set, select the deformation sample with the highest total strain energy in the sample queue as the first dominant deformation mode, and add the first dominant deformation mode to the pattern set; Step 3: Select deformation samples that have not been added to the model set from the sample queue in descending order of total structural strain and mark them as the main deformation modes. Add the main deformation modes to the model set. Whenever a new main deformation mode is added to the model set, obtain the average error of restoring all deformation samples using the current model set. Step 4: When a new master deformation mode is added to the pattern set, obtain the decrease in the average error of restoring all deformation samples using the current pattern set. When the decrease is lower than the set threshold, stop the filtering and mark the current pattern set as the master deformation pattern set.
4. The cement pole structure safety monitoring system based on digital twin as described in claim 3, characterized in that, The process of reconstructing low-dimensional mechanical equations includes: The first master-controlled deformation mode in the master-controlled deformation mode set is marked as the first independent mode. For the second master-controlled deformation mode in the master-controlled deformation mode set, the deformation components that completely overlap with the first independent mode are removed to form the second independent mode. For the third master-controlled deformation mode in the master-controlled deformation mode set, all overlapping components between the third master-controlled deformation mode and the first independent mode, as well as all overlapping components between the third master-controlled deformation mode and the second independent mode are removed to form the third independent mode. This process is repeated until all overlapping components between the last master-controlled deformation mode in the master-controlled deformation mode set and existing independent modes are removed to form the last independent mode, thus obtaining the independent deformation mode set. Generate an independent mode matrix based on all independent modes in the independent deformation mode set, obtain the structural dynamic equilibrium equations followed by the full-order finite element benchmark simulation model, construct the deformation superposition mapping relationship between the structural dynamic equilibrium equations and the independent mode matrix, construct a low-dimensional equivalent mechanical equilibrium equation based on the deformation superposition mapping relationship and the virtual work equilibrium principle, and generate a reduced-order finite element benchmark simulation model based on the low-dimensional equivalent mechanical equilibrium equation.
5. The safety monitoring system for cement pole structures based on digital twins according to claim 4, characterized in that, Based on the basic attribute parameters and deployment parameters of each cement pole, a standardized master twin template is matched for each cement pole. The process of obtaining the instantiated full-order finite element benchmark simulation model in the standardized master twin template involves instantiating and fine-tuning the model. Obtain the basic attribute parameters and deployment parameters for each concrete pole; For each cement pole, based on the inherent model in the basic attribute parameters and the latitude, longitude, geographical environment type and soil type in the deployment parameters, a standardized master twin template is matched from the standardized master twin template library. The corresponding inherent model is consistent with the inherent model of the cement pole, and the corresponding typical environment is consistent with the latitude, longitude, geographical environment type and soil type of the cement pole. Based on the standardized master twin template, the full-parameter three-dimensional geometric model, physical parameter database, full-order finite element reference simulation model and reduced-order finite element reference simulation model of the cement pole are loaded. Based on the basic attribute parameters, the full-parameter three-dimensional geometric model and the full-order finite element reference simulation model of the cement pole are geometrically corrected to generate an instantiated three-dimensional high-precision geometric model and a full-order finite element reference simulation model with geometric correction. Based on the deployment parameters, the damage parameters corresponding to the damage evolution parameter library in the physical parameter database of the cement pole and the damage parameters and environmental correction coefficients corresponding to the environmental correction coefficient library are matched. Based on the basic parameters corresponding to the basic material parameter library, the damage parameters corresponding to the damage evolution parameter library, and the damage parameter and environmental correction coefficients corresponding to the environmental correction coefficient library in the physical parameter database, the material parameters and initial damage parameters of the full-order finite element benchmark simulation model that has completed geometric correction are corrected to obtain the instantiated full-order finite element benchmark simulation model. The instantiated full-order finite element benchmark simulation model and the instantiated 3D high-precision geometric model are packaged into a cloud twin data package, and the cloud twin data package is sent to the cloud twin platform.
6. The safety monitoring system for cement pole structures based on digital twins according to claim 5, characterized in that, The process of obtaining the instantiated reduced-order finite element baseline simulation model by sequentially instantiating and fine-tuning the model includes: Obtain the physical correction scaling factor between the instantiated full-order finite element reference simulation model and the full-order finite element reference simulation model. Based on the physical correction scaling factor, correct the low-dimensional equivalent mechanical equilibrium equation followed by the reduced-order finite element reference simulation model to generate the instantiated reduced-order finite element reference simulation model. A lightweight geometric model is constructed based on an instantiated 3D high-precision geometric model. A sensor point mapping table is constructed based on the pole sensor deployment data in the deployment parameters and the lightweight geometric model. The instantiated reduced-order finite element baseline simulation model, lightweight geometric model, and sensor point mapping table are packaged into an edge twin data package, which is then sent to the edge gateway.
7. The cement pole structure safety monitoring system based on digital twin as described in claim 6, characterized in that, Based on an instantiated reduced-order finite element baseline simulation model of each cement pole, real-time monitoring is performed on each cement pole to obtain the measured simulation residual rate of various types of indicators. The process of analyzing the measured simulation residual rate of each type of indicator to determine whether a transient event alarm response or a warning-level response is triggered includes: The acquisition cycle of various indicators of the current cement pole sensor terminal is obtained. The measured data of various indicators of the cement pole are collected by the sensor terminal at the end timestamp of the acquisition cycle of each type of indicator and cached to the edge gateway. The judgment cycle of the edge gateway is set. The edge gateway pulls the measured data cached by the sensor terminal at the start timestamp of the judgment cycle. At the same time, the local meteorological data of the current cement pole is obtained. The measured data and local meteorological data are input into the instantiated reduced-order finite element benchmark simulation model. The simulation benchmark value of the current judgment cycle is output according to the instantiated reduced-order finite element benchmark simulation model. The simulation benchmark value is sent back to the sensor terminal. The newly received simulation benchmark value of the sensor terminal is marked as the judgment standard. The sensor terminal compares the measured data of various indicators with the judgment criteria in real time to obtain the measured simulation residual rate of each type of indicator. It presets the measured simulation residual rate warning threshold and national standard safety limit for each type of indicator. It compares each type of indicator with the national standard safety limit to obtain the cumulative time of each type of indicator exceeding the national standard safety limit. If the cumulative time corresponding to a type of indicator exceeds the preset time threshold, the type of indicator is marked as an abnormal indicator, triggering a transient event alarm response. The judgment cycle of the edge gateway is switched to a second-level cycle. The measured data of the sensor terminal, the abnormal indicators and the simulation benchmark value of the edge gateway are packaged into a transient alarm data packet at a second-level cycle and uploaded to the cloud twin platform. If the cumulative time corresponding to each type of indicator does not exceed the preset time threshold, the measured simulation residual rate of each type of indicator is compared with the corresponding measured simulation residual rate warning threshold. If the measured simulation residual rate corresponding to the type of indicator exceeds the measured simulation residual rate warning threshold within two consecutive judgment periods, a warning-level response is triggered. The measured data of the sensor terminal and the simulation benchmark value of the edge gateway are packaged into a warning data package according to the judgment period and uploaded to the cloud twin platform.
8. The safety monitoring system for cement pole structures based on digital twins according to claim 7, characterized in that, The process of obtaining the current data acquisition cycle of the cement pole sensor terminal includes: Obtain the measured simulation residual rate of each type of indicator in the measured data of the previous acquisition cycle, and preset the acquisition cycle mapping table. The acquisition cycle mapping table includes the acquisition cycle corresponding to different measured simulation residual rates of each type of indicator. Based on the measured simulation residual rate of each type of indicator in the measured data of the previous acquisition cycle and the acquisition cycle mapping table, obtain the acquisition cycle of the sensor terminal for acquiring each type of indicator of the cement pole at the current time.
9. The safety monitoring system for cement pole structures based on digital twins according to claim 8, characterized in that, The process of quickly performing event authenticity verification and, based on the verification results, deactivating transient event alarm responses on the edge gateway or performing damage tracing includes: After receiving the transient alarm data packet uploaded by the edge gateway, the cloud twin platform locates the cloud twin data packet corresponding to the cement pole through the unique identity identifier of the pole pre-bound by the edge gateway, activates the instantiated full-order finite element benchmark simulation model and the instantiated three-dimensional high-precision geometric model in the cloud twin data packet, and obtains the abnormal indicators in the transient alarm data packet. The sensor terminal that collects abnormal indicators is marked as the terminal to be verified. The backup sensor terminal of the terminal to be verified is obtained and deployed on the same cement pole. The difference between the abnormal indicators collected by the terminal to be verified and the abnormal indicators collected by the backup sensor terminal is obtained from the transient alarm data packet. The difference threshold is set. If the difference between the values of two consecutive second-level cycles is greater than the difference threshold, a sensor fault alarm is generated and the transient event alarm response of the edge gateway is deactivated. If there is no numerical difference greater than the difference threshold between two consecutive second-level cycles, the measured data of the sensor terminal and the local meteorological data of the cement pole in the transient alarm data packet are input into the instantiated full-order finite element reference simulation model. The simulation reference value is output according to the instantiated full-order finite element reference simulation model. The simulation reference value in the transient alarm data packet is compared with the simulation reference value output by the instantiated full-order finite element reference simulation model, and the deviation of each type of feature is output. Subsequently, the measured data of the sensor terminal corresponding to the unique identification of the pole in several historical acquisition cycles are extracted. The simulation benchmark value output by the instantiated full-order finite element benchmark simulation model, the deviation of various types of features, the measured data of the sensor terminal in the transient alarm data packet, and the measured data of several historical acquisition cycles are input into the preset damage source prediction model. According to the damage source prediction model, the damage type, damage location, and the degree of deterioration corresponding to the damage location are output. The damage location and the damage type and degree of deterioration corresponding to the damage location are marked on the instantiated three-dimensional high-precision geometric model.
Citation Information
Patent Citations
Transmission tower vulnerability analysis method and system based on seismic simulation
CN121168116A
Rapid calculation method for POD order reduction of power transmission tower based on static and dynamic load separation modeling
CN122595723A
