A smart lamp post collision recognition method and structural health estimation method and device based on multi-dimensional time sequence features

Abstract

The present application relates to the technical field of smart city, and more particularly to a smart lamp pole collision identification method and a structure health estimation method and device based on multi-dimensional time sequence characteristics. The collision identification method comprises: collecting multi-dimensional time sequence data of the smart lamp pole; extracting feature values of the multi-dimensional time sequence data based on a preset time length sliding data window; inputting the extracted feature values into a preset state classification determination model to identify the operating conditions, including natural wind deviation condition, road resonance condition and real collision accident condition, and executing corresponding disposal logic according to the identification result. The structure health estimation method comprises: extracting posture time sequence data, fitting a curve to calculate a basic settlement rate, deducing a cumulative settlement amount, and generating structure health warning information when the cumulative settlement amount meets a preset warning condition. The technical scheme of the present application can greatly reduce the false positive rate, improve the fault repair efficiency, and realize the whole cycle health management, which is suitable for the long period operation and maintenance demand of EMC project for more than ten years.

Description

Technical Field

[0001] This invention relates to the field of smart city technology, and in particular to a smart light pole collision recognition method and structural health prediction method and device based on multi-dimensional temporal features. Background Technology

[0002] In the operation system of urban road lighting energy management (EMC) projects, the service period is generally ten years or more. The industry has put forward strict requirements for the quality of street light operation and maintenance, requiring not only that the overall lighting rate be maintained at over 98%, but also that faults in main road lights be repaired within 24 hours. Tilting and collapsing of light poles are typical safety faults of urban street lights. In order to identify such hazards in a timely manner, most mainstream single-lamp controllers in the industry now have built-in tilt sensors to monitor the posture status of the light pole and issue fault alarms.

[0003] At present, the static threshold triggering method is commonly used for light pole tilt monitoring. The technical principle is to preset a fixed tilt angle threshold, such as 10°, in the single lamp controller. When the tilt sensor detects that the real-time tilt angle of the light pole exceeds the preset threshold, the device immediately pushes the light pole tilt fault alarm information to the cloud management platform through the wireless communication network.

[0004] This solution, with its simple architecture and low implementation cost, was widely used in early street light monitoring scenarios. However, in the complex and ever-changing urban operating environment, it has gradually revealed many inherent defects. First, the static threshold triggering mechanism has poor resistance to environmental interference, resulting in a high false alarm rate and significantly increasing ineffective maintenance costs. In urban road scenarios, phenomena such as strong winds, roadside branches hitting light poles, and road surface resonance caused by heavy vehicles are very common. These external disturbances can all cause instantaneous angular shifts in the light poles, briefly exceeding the preset threshold. Traditional solutions cannot distinguish between actual structural faults and attitude fluctuations caused by environmental interference, leading to frequent false alarms and forcing maintenance personnel to repeatedly travel to the site for verification, resulting in unnecessary consumption of human and material resources and severely reducing overall maintenance efficiency. Second, existing monitoring solutions have a single data dimension and do not achieve multi-source data fusion, making it difficult to complete fault correlation analysis. In real-world scenarios, serious accidents such as vehicles colliding with light poles are usually accompanied by electrical faults such as broken underground power cables and tripped street light control boxes. Current tilt monitoring systems only collect and report pole posture data, without combining it with comprehensive analysis of street light operating electrical parameters. After receiving alarms, maintenance personnel cannot predict in advance whether there are derivative problems such as line damage or electrical faults on site, and cannot prepare appropriate repair materials such as cables and insulating tools, directly affecting the progress of fault repair. Finally, traditional monitoring methods can only identify instantaneous posture changes and lack the ability to assess the long-term health status of the pole structure. Streetlights are subject to long-term rain erosion and soil loss, making them prone to foundation loosening and slow settlement. These hidden dangers cause continuous, gradual, and minute changes in the pole's tilt angle. Existing technology only provides post-event alarms for sudden, large tilts, and cannot continuously track and analyze the long-term drift trend of the pole's tilt angle. It is difficult to identify the collapse risk caused by slow deformation in advance, and preventive maintenance intervention is impossible, posing a long-term safety hazard.

[0005] Patent application CN121007534A discloses a street light tilt monitoring system and a rapid repair construction method. This solution includes a data standardization module, a data initialization module, a model building module, and a result output module. It can standardize various data such as wind speed, foundation settlement, and vibration. By splicing state vectors, constructing causal graphs, and calculating reward and punishment indices, it identifies the causes of street light tilt, thus improving the susceptibility of single-threshold alarms to environmental interference to some extent. It also attempts to monitor long-term deformation states such as foundation settlement. However, this solution still has limitations. Its data processing and logical analysis architecture leans towards a generalized model design and does not integrate deep multi-source data fusion with street light electrical operating parameters, making it unable to achieve linked analysis of impact-related accidents and electrical faults. Furthermore, there is still room for optimization in long-term structural health quantification assessment, graded early warning, and matching of operation and maintenance strategies, making it difficult to fully meet the operation and maintenance needs of long-term, high-efficiency, and high-precision urban lighting EMC projects.

[0006] Based on the aforementioned technical problems and shortcomings, there is an urgent need to provide a smart light pole collision recognition and structural health prediction method that can reduce false alarms caused by environmental interference, achieve fusion analysis of attitude data and electrical parameters to accurately support emergency repair operations, and simultaneously complete long-term health status tracking and prediction of light pole structures, so as to fully adapt to the high-standard operation and maintenance requirements of urban road lighting EMC projects. Summary of the Invention

[0007] To address the aforementioned deficiencies or improvement needs of existing technologies, this invention provides a smart light pole collision identification and structural health prediction method based on multi-dimensional temporal features. Through the fusion analysis of attitude, electrical, and meteorological multi-dimensional temporal data, it accurately distinguishes between environmental interference and actual collision faults, reducing false alarm rates. Combined with electrical parameter linkage analysis to determine fault types, it improves emergency repair efficiency. Simultaneously, it conducts long-term monitoring and trend prediction of light pole foundation settlement, enabling early warning of structural health deterioration, fully adapting to the high-standard operation and maintenance requirements of urban road lighting EMC projects.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a smart light pole collision recognition method based on multi-dimensional temporal features, comprising the following steps: Simultaneously collect multi-dimensional time-series data of smart light poles, including attitude time series, electrical parameter series, and environmental meteorological series; Based on a sliding data window of preset duration, feature values ​​are extracted from multidimensional time-series data to obtain attitude feature values, electrical feature values, and meteorological feature values; The extracted feature values ​​are input into a preset state classification and judgment model to identify the operating conditions of the smart light pole, and the corresponding handling logic is executed according to the identification results; the operating conditions include natural wind deflection conditions, road resonance conditions, and real collision accident conditions; Furthermore, the synchronous acquisition of multi-dimensional time-series data of the smart light pole includes: The attitude time series includes the tilt angle and angular velocity of the X, Y, and Z axes, which are collected by the attitude sensor built into the smart light pole. The electrical parameter sequence includes input voltage, current, and active power, which are acquired through an electrical parameter acquisition module. The environmental meteorological sequence includes real-time wind speed and wind direction, which are obtained by a physical small anemometer installed on a street light micro base station or by a cloud platform through an external meteorological interface and then sent to the edge computing gateway. Furthermore, after all data collection is completed, time-series alignment preprocessing is performed to ensure the consistency of the time dimension of multi-source data.

[0009] Furthermore, the attitude characteristics include the peak tilt angle, deformation rate (attitude change derivative), oscillation frequency, and oscillation damping coefficient; the electrical characteristics include the current step change rate within the sliding data window; and the meteorological characteristics include the gust frequency. Performing local characteristic calculations at the edge computing gateway effectively reduces the amount of raw data transmitted uplink, improving processing efficiency.

[0010] Furthermore, the step of inputting the extracted feature values ​​into a preset state classification and judgment model to identify the operating conditions of the smart light pole, and executing corresponding handling logic based on the identification results, includes: If the tilt angle is detected to exhibit periodic damped oscillation, the mean change in tilt angle is less than a preset first angle threshold, the change in electrical parameters meets preset conditions, and the correlation between the oscillation frequency and the gust frequency meets preset conditions, then it is determined to be the natural wind deflection condition, and a fault alarm is triggered. Furthermore, the method for determining whether the tilt angle exhibits periodic damped oscillation is as follows: if the tilt angle signal is detected to continuously cross the zero-position baseline and generate alternation, and the peak values ​​(absolute values ​​of peaks and troughs) of adjacent oscillations in the same direction show a monotonically decreasing trend, and the ratio of adjacent peak values ​​in the same direction is within the preset damping coefficient range, then it is determined to be a periodic damped oscillation.

[0011] Furthermore, the change range of the electrical parameter meets the preset conditions, specifically including: within the sliding data window, the variance and / or standard deviation of the electrical parameter is less than a preset statistical fluctuation threshold, and the maximum step change within the sliding data window is less than a preset step threshold, then it is determined that the change range of the electrical parameter meets the preset conditions; the electrical parameter includes at least one of current, voltage, or active power; preferably, the electrical parameter is current.

[0012] Furthermore, the correlation satisfies the preset conditions, including: calculating the Pearson correlation coefficient between the oscillation frequency and the gust frequency; if the correlation coefficient is greater than a preset threshold, it is determined that the preset correlation conditions are met.

[0013] Furthermore, it should be noted that the specific value of the first angle threshold can be adaptively adjusted according to the specific physical parameters of the street light (such as pole height, material, windward area, natural frequency, etc.). For example, in practical applications, for conventional municipal street lights with a height of 8-10 meters, the first angle threshold is set to 3° to 5°; for high-mast lights or flexible poles with a height of 12 meters or more, the threshold is set to 1° to 3°. Those skilled in the art can determine the optimal threshold through a limited number of conventional calibration experiments.

[0014] If the posture of the light pole experiences instantaneous high-frequency oscillations and decays to the initial reference posture within a preset first time threshold, and the rate of change of the electrical parameters is less than a preset first rate of change threshold, then it is determined to be the road resonance condition, and a fault alarm is triggered. Furthermore, the instantaneous high-frequency oscillation is defined as follows: within a set time window, the frequency change of the attitude signal exceeds a preset multiple of the inherent natural frequency of the lamp post; further, the preset multiple is ≥3 times. It should be noted that the specific values ​​of the preset first time threshold and the first rate of change threshold can be adaptively calibrated according to the actual physical and electrical parameters of the street light.

[0015] Furthermore, the first time threshold is used to limit the decay time of oscillations caused by transient impacts on the road surface. Considering the transient characteristics of road surface excitation and the structural damping of the lamppost, the first time threshold is preferably set to 1 to 30 seconds, more preferably 3 to 15 seconds. If the duration of the oscillation exceeds this threshold, the possibility of road surface resonance is ruled out.

[0016] Furthermore, the first rate of change threshold is used to characterize the steady-state maintenance capability of electrical parameters during mechanical oscillations. To avoid differences in electrical baselines between streetlights of different power ratings, the rate of change of the electrical parameters is preferably calculated using a relative rate of change or a normalized rate of change. For example, the rate of change can be defined as the ratio of the change in electrical parameters within a sliding window to the rated electrical parameters. Accordingly, the first rate of change threshold is preferably set to 0.5% to 5% of the rated value. In absolute numerical embodiments, when the electrical parameter is current, the first rate of change threshold is preferably set to 0.1 A / s to 0.5 A / s. Those skilled in the art can determine the optimal threshold through a limited number of conventional calibration experiments.

[0017] If the tilt angle undergoes a step change within the preset second time threshold and the change angle is greater than the preset angle change threshold, and the tilt angle produces an irreversible angle shift after the oscillation ends, and a sudden surge in current (characteristic of cable short circuit) or a sudden drop in current to zero (characteristic of circuit breaker tripping and power failure) occurs simultaneously, then it is determined to be the actual collision accident condition. The system immediately reports the fault to the cloud platform, generates an emergency repair work order, and marks the damage to the electrical lines in the work order. Furthermore, it should be noted that the specific values ​​of the preset second time threshold and second angle mutation threshold are adaptively calibrated based on the physical characteristics of real collision accidents.

[0018] Furthermore, the second time threshold is used to define the time window of the impact. Since vehicle collisions are transient high-energy impacts of extremely short duration, in order to effectively distinguish them from conditions such as slow foundation settlement and slow bending caused by strong winds, the second time threshold is preferably set to 0.5 seconds to 5 seconds, more preferably 0.5 seconds to 1 second.

[0019] Furthermore, the angle abrupt change threshold is used to characterize the instantaneous change in tilt angle caused by the collision. Considering that minor scratches or wind deflection usually do not cause large-angle deflection, while real collisions are often accompanied by significant instantaneous large-angle displacement, the second angle abrupt change threshold is preferably set to 10° to 30°, more preferably 15° to 20°.

[0020] Furthermore, the method for determining the irreversible angle shift is as follows: after the collision oscillation ends and a preset rest recovery time (10s~30s) has elapsed, if the difference between the average value of the current tilt angle and the average value of the tilt angle of the initial reference attitude before the collision is greater than a preset permanent variation difference threshold (2°~5°), then it is determined that an irreversible angle shift has occurred.

[0021] In practical applications, when a sudden change in tilt angle greater than 10° occurs within a timeframe of 0.5 to 5 seconds, accompanied by irreversible permanent deformation and short-circuit / open-circuit characteristics of the electrical system, it can be confidently determined as a genuine collision accident. Those skilled in the art can determine the optimal threshold combination through a limited number of routine calibration experiments based on the height of the light pole, material stiffness, and vehicle speed limits on the road.

[0022] Furthermore, the state classification and determination model is a determination model based on artificial logical rules and fixed feature thresholds; or, the state classification and determination model is an intelligent learning model, which includes support vector machines, random forest classifiers or long short-term memory neural networks, and is trained on a training dataset consisting of multi-source time-series data slices containing historical wind drift disturbances, vehicle resonance and real light pole collision accidents.

[0023] Furthermore, the method is implemented through collaboration between an edge computing gateway and a cloud platform; the acquisition of multi-dimensional time-series data, feature value extraction, and identification of operating conditions are performed on the edge computing gateway side; the edge computing gateway uploads the extracted feature values ​​and operating condition judgment results to the cloud platform, and the cloud platform performs data storage, long-term attitude data extraction, foundation settlement drift curve fitting, and work order distribution.

[0024] Furthermore, the emergency repair work orders and preventive reinforcement work orders generated by the cloud platform are pushed to the handheld terminals of maintenance personnel. The work orders include auxiliary information such as fault type, electrical damage status, and light pole settlement risk level.

[0025] Furthermore, the duration of the preset sliding data window is adaptively adjusted in conjunction with urban road traffic density and / or regional meteorological conditions.

[0026] Secondly, this invention provides a method for predicting the structural health of a smart light pole, comprising the following steps: The attitude time series data of the smart light pole in a resting state over a long period is extracted. Based on the attitude time series data, the foundation settlement drift curve is fitted to calculate the foundation settlement rate. The cumulative settlement amount within a specified time period is extrapolated based on the foundation settlement rate. When the cumulative settlement amount meets the preset early warning conditions, structural health early warning information is generated.

[0027] Furthermore, the long period is >3 months; the resting state is: avoiding urban traffic rush hours and severe weather periods, and during the period with the least environmental interference (such as 2:00 to 3:00 am and the wind speed is lower than the set value), the light pole is in a natural static state without significant external interference.

[0028] Furthermore, when the cumulative settlement meets the preset early warning conditions, structural health early warning information is generated, including: when the difference between the calculated cumulative settlement and the preset safety threshold is less than the preset early warning threshold, or when the cumulative settlement reaches the preset proportion of the preset safety threshold, it is determined that the preset early warning conditions are met, and a preventive reinforcement work order is generated. The attitude time series data includes the tilt angles and angular velocities of the X, Y, and Z axes, which are collected by the attitude sensors built into the smart light pole; the settlement amount is the settlement angle.

[0029] Thirdly, the present invention provides a smart light pole collision recognition device based on multi-dimensional temporal features, comprising: The data acquisition module is used to synchronously acquire multi-dimensional time-series data of the smart light pole, including attitude time series, electrical parameter series and environmental meteorological series; The feature extraction module is used to extract feature values ​​from preprocessed multidimensional time-series data based on a sliding data window of preset duration, to obtain attitude feature values, electrical feature values, and meteorological feature values. The operating condition identification module is used to input the extracted feature values ​​into a preset state classification and judgment model to identify the operating conditions of the smart light pole. The operating conditions include natural wind deflection, road resonance, and real collision accident conditions, and execute the corresponding handling logic according to the identification results.

[0030] Fourthly, the present invention provides a structural health prediction device for smart light poles, comprising: The data acquisition module is used to synchronously collect the attitude time series of the smart light pole; The data processing module is used to fit the foundation settlement drift curve based on the attitude time series data to calculate the foundation settlement rate, and to extrapolate the cumulative settlement within a specified time period based on the foundation settlement rate. The health alarm module is used to generate structural health warning information when the cumulative settlement meets the preset warning conditions.

[0031] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Significantly reduce false alarm rate: This invention integrates multi-dimensional time-series features of attitude, electrical, and meteorological data. By analyzing features such as oscillation patterns, frequency matching, and parameter stability, it accurately distinguishes between environmental interference such as natural wind deflection and road resonance and real faults, thereby reducing false alarms at the source and lowering the manpower and material costs of ineffective maintenance.

[0032] 2. Improve fault repair efficiency: By linking and analyzing light pole posture data with electrical parameters, derivative faults such as cable short circuits and switch tripping caused by collision accidents can be identified simultaneously. Maintenance personnel can prepare corresponding spare parts and tools in advance to achieve accurate dispatching and efficient emergency repair, meeting the short-term repair requirements of main roads.

[0033] 3. Achieve full-cycle health management: By fitting settlement and drift curves through long-term resting data, quantitatively analyze the long-term deterioration trend of light pole foundations, predict collapse risks in advance and generate preventive work orders, transforming post-event repairs into pre-event prevention and control, and adapting to the long-term operation and maintenance needs of EMC projects for more than 10 years.

[0034] 4. Reasonable computing architecture: The edge computing gateway is used to complete data collection, sliding window feature extraction and local preliminary judgment, and only feature data and results are uploaded to reduce network transmission pressure; the cloud platform is responsible for data aggregation, trend analysis and work order management. The collaborative architecture of edge and cloud takes into account both real-time performance and management capabilities. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a system architecture diagram of a smart light pole collision recognition device based on multi-dimensional temporal features; Figure 2 System architecture diagram for a smart light pole structural health prediction device; Figure 3This is a curve comparing the time sequence characteristics of normal wind deflection and actual impact. Detailed Implementation

[0037] The present invention will be further described below with reference to embodiments, but the scope of protection of the present invention is not limited thereto. It should be noted that the implementation steps in the embodiments can be further adjusted according to the specific experimental environment, and the implementation steps not specified are usually conditions in conventional experiments. To make the objectives, technical solutions, and advantages of the present invention clearer, the following embodiments are provided in conjunction with the actual operation and maintenance scenario of urban road lighting EMC projects: Example 1: False Alarm Interception Based on Natural Wind Deflection and Road Surface Resonance Scenario Description: In a main urban road scenario, a smart light pole is subjected to a sudden strong gust of wind (wind speed 15m / s), or a heavy dump truck passes by, causing road surface resonance. The handling steps are as follows: Figure 1 As shown.

[0038] 1. Data Acquisition and Feature Extraction: Regarding wind drift characteristics: Within a 5-second sliding data window, the edge computing gateway identified that the tilt angle signal continuously crossed the zero-position baseline, exhibiting alternation. The absolute values ​​of adjacent peaks showed a monotonically decreasing trend, and the attenuation ratio remained within the preset damping coefficient range (determined as periodic damped oscillation). Simultaneously, the long-term average change in tilt angle was 0° (less than the first angle threshold of 5°), the current standard deviation was less than the statistical fluctuation threshold, and there were no abrupt changes. The Pearson correlation coefficient between the tilt angle oscillation frequency and the gust frequency obtained from the meteorological interface was calculated to be 0.85 (greater than the preset threshold). Current changes are as follows... Figure 3 As shown in (a).

[0039] Regarding the resonance characteristics: within the same time window, the attitude signal frequency abruptly reaches 3.5 times (≥3 times) the natural frequency of the lamp post, and rapidly decays to the initial reference attitude within 10 seconds (less than the first time threshold of 15 seconds); during this period, the relative change rate of the current is 1.2% (less than the first change rate threshold of 5%).

[0040] 2. Condition Determination and Handling: Based on the above characteristics, the condition classification and determination model determines with high confidence that the current condition is "a superposition of natural wind deflection and road surface resonance". The edge computing gateway actively intercepts the tilt alarm caused by this attitude fluctuation locally and does not report the fault to the cloud, effectively avoiding invalid dispatch.

[0041] Example 2: Linkage Recognition Scenario for Real Vehicle Collision Accidents Scene description: In the early hours of the morning, an out-of-control vehicle crashed into a smart light pole at high speed, causing the light pole to tilt severely and the underground power cable to be accidentally torn off.

[0042] 1. Data Acquisition and Feature Extraction: Attitude Abrupt Change and Irreversible Misalignment: Within 0.8 seconds of the sliding data window (within the second time threshold range of 0.5 to 5 seconds), a step change in the tilt angle of the lamppost was detected, instantly reaching 18° (greater than the second angle abrupt change threshold of 15°). After a brief oscillation and a 10-second pause, the current average tilt angle is 16°, which is 16° different from the baseline attitude (0°) before the collision (greater than the permanent variation difference threshold of 2°), indicating an irreversible angle misalignment.

[0043] Electrical characteristic linkage: Synchronously acquired electrical parameters show that at the instant of a sudden attitude change, the input current surges from the normal 1.2A to the short-circuit peak, then drops sharply to 0A within 0.1 seconds (triggering the circuit breaker to trip and cut off power). Current changes are as follows... Figure 3 As shown in (b).

[0044] 2. Condition Assessment and Handling: The condition classification model integrates "extremely short-time, large-angle step change," "irreversible permanent deformation," and "current short-circuit / open-circuit characteristics" to determine it as a "real collision accident condition." The edge computing gateway reports the result to the cloud, and the platform automatically generates an "urgent repair work order" and pushes it to the handheld terminal of the maintenance personnel. The work order clearly indicates "a real impact occurred, accompanied by cable damage and power outage," guiding the maintenance personnel to carry insulated tools and spare cables for precise repair.

[0045] Example 3: Structural Health Prediction and Preventive Maintenance of Smart Light Pole Scenario Description: Due to long-term erosion by rainwater, the foundation of a light pole along a river is slowly loosening. The handling steps are as follows: Figure 2 As shown.

[0046] 1. Data Acquisition: The cloud platform is set to a "quiet state" from 2:00 AM to 3:00 AM daily, when there is no traffic on the road and the wind speed is below the set value. Under this state, the platform extracts the average Z-axis tilt angle (settlement angle) data collected by the posture sensor built into the light pole every day for 4 consecutive months (long period > 3 months).

[0047] 2. Data Processing (Settlement Curve Fitting and Extrapolation): Based on the extracted long-period attitude time series data, the cloud platform fits the foundation settlement drift curve, calculating the foundation settlement rate of the light pole to be 0.5° per month, with the current cumulative settlement angle reaching 3°. Based on this, the system extrapolates that the light pole will experience approximately 2° of cumulative settlement within the next 4 months, bringing the total cumulative settlement to 5°.

[0048] 3. Health Alert: The system's preset safety threshold is 5°. Since the projected total cumulative settlement (5°) has reached 100% of the preset safety threshold (meeting the warning condition for reaching the preset proportion), the system automatically generates a "Preventive Reinforcement Work Order." The work order includes auxiliary information such as "Foundation Settlement Risk Level: High" and is pushed to the maintenance terminal. The maintenance team performs foundation grouting reinforcement before the light pole actually collapses, achieving preventative control.

[0049] Example 4: Smart Light Pole Status Monitoring and Prediction Device This embodiment provides a smart light pole status monitoring and prediction device. The device includes a collision recognition subsystem and a structural health prediction subsystem in its logical architecture. In actual physical deployment, it can be implemented through the collaboration of an edge computing gateway, a local controller and a cloud platform.

[0050] 1. Collision Recognition Subsystem Data acquisition module: Through the attitude sensor built into the light pole, the electrical parameter acquisition module and the external meteorological interface, it synchronously acquires multi-dimensional time-series data such as the tilt angle and angular velocity of the light pole on the X / Y / Z axes, input current / voltage / power, and real-time wind speed / direction.

[0051] Feature extraction module: Deployed on the edge computing gateway side, it is used to filter, reduce noise and extract features from the collected multi-dimensional time series data based on a sliding data window of preset duration, so as to obtain attitude feature values, electrical feature values ​​and meteorological feature values.

[0052] Operating condition identification module: It has a built-in state classification and judgment model (such as intelligent learning models based on manual logic rules or SVM / LSTM). The extracted feature values ​​are input into the model to identify the operating conditions of natural wind deflection, road resonance or real collision accident, and execute the corresponding handling logic such as local interception alarm or reporting the fault to the cloud.

[0053] 2. Structural Health Prediction Subsystem Data acquisition module: Used to continuously collect the attitude time series data of the smart light pole under a preset resting state and upload it to the cloud platform.

[0054] Data processing module: Deployed on a cloud platform, it is used to fit the foundation settlement drift curve based on long-term attitude time series data to calculate the foundation settlement rate, and to extrapolate the cumulative settlement over a specified period of time based on the rate.

[0055] Health Alarm Module: When the simulated cumulative settlement meets the preset early warning conditions (such as reaching a preset proportion of the safety threshold), a preventive reinforcement work order is generated and pushed to the handheld terminal of the operation and maintenance personnel.

[0056] The above description is merely a preferred embodiment of the present invention and does not limit the scope of the patent. Any equivalent structural or procedural modifications made based on the content of this specification, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of this invention.

Claims

1. A smart light pole collision recognition method based on multi-dimensional temporal features, characterized in that, The collision recognition method includes the following steps: Simultaneously collect multi-dimensional time-series data of smart light poles, including attitude time series, electrical parameter series, and environmental meteorological series; Based on a sliding data window of preset duration, feature values ​​are extracted from multidimensional time-series data to obtain attitude feature values, electrical feature values, and meteorological feature values; The extracted feature values ​​are input into a preset state classification and judgment model to identify the operating conditions of the smart light pole. The operating conditions include natural wind deflection, road resonance, and real collision accident conditions. The corresponding handling logic is executed according to the identification results.

2. The method according to claim 1, characterized in that, The synchronous acquisition of multi-dimensional time-series data of smart light poles includes: The attitude time series includes the tilt angle and angular velocity of the X, Y, and Z axes, which are collected by the attitude sensor built into the smart light pole. The electrical parameter sequence includes input voltage, current, and active power, which are acquired through an electrical parameter acquisition module. The environmental meteorological sequence includes real-time wind speed and direction, which are obtained through physical meteorological sensors or cloud-based meteorological interfaces.

3. The method according to claim 1 or 2, characterized in that, The attitude characteristic values ​​include the peak tilt angle, deformation rate, oscillation frequency, and oscillation damping coefficient; the electrical characteristic values ​​include the current step change rate within the sliding data window; and the meteorological characteristic values ​​include the gust frequency.

4. The method according to claim 1 or 2, characterized in that, The process of identifying the operating status of the smart light pole and executing corresponding processing logic based on the identification results includes: If the tilt angle is detected to exhibit periodic damped oscillation, the mean change in tilt angle is less than a preset first angle threshold, the change in electrical parameters meets preset conditions, and the correlation between the attitude oscillation frequency and the gust frequency meets preset conditions, then it is determined to be the natural wind deflection condition, and a fault alarm is triggered. If the smart light pole exhibits instantaneous high-frequency oscillations in its posture, and decays to its initial reference posture within a preset first time threshold, while the rate of change of the electrical parameters is less than a preset first rate of change threshold, then it is determined to be the road resonance condition, and a fault alarm is triggered. If the tilt angle undergoes a step change within the preset second time threshold and the change angle is greater than the preset angle change threshold, and the tilt angle undergoes an irreversible angle shift after the oscillation ends, and a sudden surge in current or a sudden drop in current to zero occurs simultaneously, then it is determined to be the actual collision accident condition, and a fault repair work order is generated and reported.

5. The method according to claim 1 or 2, characterized in that, The state classification and determination model is a determination model based on artificial logic rules and fixed feature thresholds; or, the state classification and determination model is an intelligent learning model, which includes support vector machines, random forest classifiers or long short-term memory neural networks, and is trained on a training dataset consisting of multi-source time-series data slices containing historical wind drift disturbances, vehicle resonance and real collision accidents.

6. The method according to claim 1 or 2, characterized in that, The method is implemented through collaboration between an edge computing gateway and a cloud platform; the acquisition of multi-dimensional time-series data, feature value extraction, and identification of operating conditions are performed on the edge computing gateway side. The edge computing gateway uploads the extracted feature values ​​and working condition judgment results to the cloud platform. The cloud platform performs data storage, attitude data extraction within a preset long period, basic settlement drift curve fitting, and work order distribution.

7. The method according to claim 1 or 2, characterized in that, The preset duration of the sliding data window is adaptively adjusted based on urban road traffic density and / or regional meteorological conditions.

8. A method for predicting the structural health of a smart light pole, characterized in that, Including the following steps: The attitude time series data of the smart light pole under a preset resting state within a preset long period are extracted. Based on the attitude time series data, the foundation settlement drift curve is fitted to calculate the foundation settlement rate. The cumulative settlement amount within a specified time in the future is inferred based on the foundation settlement rate. When the cumulative settlement amount meets the preset early warning conditions, structural health early warning information is generated.

9. The method according to claim 8, characterized in that, When the cumulative settlement meets the preset early warning conditions, structural health early warning information is generated, including: when the difference between the calculated cumulative settlement and the preset safety threshold is less than the preset early warning threshold, or when the cumulative settlement reaches the preset proportion of the preset safety threshold, it is determined that the preset early warning conditions are met, and a preventive reinforcement work order is generated.

10. The method according to claim 8, characterized in that, The attitude time series data includes the tilt angles and angular velocities of the X, Y, and Z axes, which are collected by the attitude sensors built into the smart light pole; the settlement amount is the settlement angle.

11. A smart light pole collision recognition device based on multi-dimensional temporal features, characterized in that, include: The data acquisition module is used to synchronously acquire multi-dimensional time-series data of the smart light pole, including attitude time series, electrical parameter series and environmental meteorological series; The feature extraction and processing module is used to extract feature values ​​from multidimensional time-series data based on a sliding data window of preset duration, and obtain attitude feature values, electrical feature values ​​and meteorological feature values. The operating condition identification module is used to input the extracted feature values ​​into a preset state classification and judgment model to identify the operating conditions of the smart light pole. The operating conditions include natural wind deflection, road resonance, and real collision accident conditions, and execute the corresponding handling logic according to the identification results.

12. A structural health prediction device for a smart light pole, characterized in that, include: The data acquisition module is used to synchronously collect the attitude time series of the smart light pole; The data processing module is used to fit the foundation settlement drift curve based on the attitude time series data to calculate the foundation settlement rate, and to extrapolate the cumulative settlement within a specified time period based on the foundation settlement rate. The health alarm module is used to generate structural health warning information when the cumulative settlement meets the preset warning conditions.

Images