Real-time analysis methods and systems for urban infrastructure safety monitoring

Abstract

This application relates to the technical field of infrastructure data analysis, and discloses a real-time analysis method and system for urban infrastructure safety monitoring. The method first acquires infrastructure type and distribution data, matches corresponding IoT sensors from a sensor database and scientifically deploys them, and generates a digital twin model based on the IoT sensor network of multiple types of facilities. Then, based on this model, it acquires the IoT networks corresponding to first and second-type facilities with a first IoT connection, extracts sensor data for each, identifies abnormal data through a preset identification strategy, and determines and triggers corresponding abnormal alerts based on the data volume ratio. Finally, based on the two types of abnormal alerts, it links and warns other facility types related to the first IoT connection. This application achieves precise adaptation of facility monitoring, reliable identification of abnormal data, and coordinated prevention and control of related facilities, improving the real-time performance, relevance, and comprehensiveness of urban infrastructure safety monitoring.

Description

Technical Field

[0001] This application relates to the technical field of infrastructure data analysis, and in particular to a real-time analysis method and system for urban infrastructure safety monitoring. Background Technology

[0002] In current urban building and infrastructure maintenance, manual work combined with existing blueprints remains the mainstream operating mode. Staff must verify facility locations, structural parameters, and historical maintenance records on-site based on paper or electronic blueprints to conduct inspections, repairs, and troubleshooting. However, this method has significant limitations: blueprints are prone to information lag or loss due to facility modifications and wear and tear over time; manual verification is inefficient; and it is difficult to synchronize dynamic changes in facilities in real time, failing to meet the needs of efficient maintenance for large-scale, highly complex infrastructure.

[0003] As cities expand and infrastructure becomes increasingly complex, the importance of digital operations and maintenance (O&M) for urban building and infrastructure management is becoming increasingly prominent. Traditional manual maintenance models suffer from fragmented data storage and severe information silos, making full lifecycle management difficult. Digital O&M, by integrating BIM models, IoT sensing data, and historical maintenance records, can build a unified digital management platform, breaking down data silos and enabling centralized and visualized management of facility information. This significantly improves the scientific basis of maintenance decisions, reduces labor costs, and provides crucial technical support for the safe operation of infrastructure.

[0004] In recent years, extreme weather and prolonged high-load operation of equipment have led to frequent failures in urban buildings and infrastructure, creating an urgent need for real-time analysis and early warning of their operational status. Existing management models largely rely on periodic inspections, making it difficult to dynamically capture minor anomalies during facility operation, such as micro-deformation of bridge structures or sudden changes in pipeline pressure. Often, responses are reactive only after a failure occurs, easily leading to safety accidents and economic losses. Therefore, there is an urgent need to establish a technical system capable of real-time monitoring of facility status, automatic analysis of operational data, and timely risk warnings to ensure the safe operation of facilities and the normal functioning of the city. Summary of the Invention

[0005] In order to enable automatic monitoring of urban infrastructure, this application provides a real-time analysis method and system for urban infrastructure safety monitoring.

[0006] Firstly, this application provides a real-time analysis method for urban infrastructure safety monitoring, employing the following technical solution:

[0007] A real-time analysis method for urban infrastructure safety monitoring includes the following steps:

[0008] The system obtains the facility type of the infrastructure, matches the corresponding IoT sensors from the preset sensor database according to the facility type, obtains the distribution data corresponding to the infrastructure, deploys IoT sensors according to the distribution data, and obtains a digital twin model of the IoT sensor network based on multiple infrastructures.

[0009] Based on the digital twin model, a first Internet of Things (IoT) network corresponding to a first facility type and a second IoT network corresponding to a second facility type are obtained, wherein the first facility type and the second facility type have a first IoT connection;

[0010] First sensor data is obtained from the first Internet of Things network. First abnormal data is identified using a preset first identification strategy based on the first sensor data. The proportion of the first abnormal data in the first sensor data is calculated. If the proportion of the first abnormal data is greater than the preset first reference proportion, a first Internet of Things abnormality prompt is issued.

[0011] Second sensor data is obtained from the second Internet of Things network. Second abnormal data is identified using a preset second identification strategy based on the second sensor data. The proportion of the second abnormal data in the second sensor data is calculated. If the proportion of the data is greater than the preset second reference proportion, a second Internet of Things abnormality prompt is issued.

[0012] Based on the first and second IoT anomaly alerts, obtain other facility types corresponding to the first IoT connection and display warnings for these other facility types.

[0013] By adopting the above technical solutions, accurately matching IoT sensors according to facility type and scientifically deploying them in conjunction with distributed data, and intuitively replicating the associated status of multiple types of infrastructure based on digital twin models, differentiated anomaly identification strategies are adopted for different facility types, with the proportion of data volume used as the judgment criterion. This not only solves the problems of poor sensor adaptability and high false alarm rate in traditional monitoring, but also realizes the linkage early warning of related facilities based on the IoT connection between facilities, breaking the limitations of independent monitoring of a single facility. The entire process is automated, promoting the complete link from data collection to early warning and handling, effectively improving the pertinence, accuracy and comprehensiveness of urban infrastructure safety monitoring, and reducing the chain risks caused by the spread of anomalies.

[0014] Furthermore, the first IoT network includes a strain sensing network, and the first identification strategy includes the following steps:

[0015] Calculate the change range of elements in the first sensing data, and identify elements whose change range is greater than the preset first reference range as abnormal elements and other elements as normal elements. Calculate the absolute value of the difference between the average data of abnormal elements and the average data of normal elements as the first difference value.

[0016] If the first difference value is greater than the preset strain difference value, then the abnormal element is regarded as the first abnormal data.

[0017] By adopting the above technical solution, potential anomalies are initially screened by calculating the change range of data elements, and secondary verification is performed by the absolute value of the difference between the average value of the abnormal and normal elements and the preset strain difference value. This not only accurately captures the significant fluctuations in strain data, but also avoids misjudgments caused by a single abnormal data, greatly improving the accuracy of anomaly identification for strain-related infrastructure.

[0018] Furthermore, the first IoT network includes a vibration sensing network, and the first identification strategy includes the following steps:

[0019] Calculate the rate of change of elements in the first sensing data. Element with a rate of change greater than a preset first reference rate is considered an abnormal element, and other elements are considered normal elements. The absolute value of the difference between the average data of abnormal elements and the average data of normal elements is the first difference value.

[0020] If the first difference value is greater than the preset vibration difference value, then the abnormal element is regarded as the first abnormal data.

[0021] By adopting the above technical solution, potential vibration anomalies can be initially identified by calculating the rate of change of data elements. Secondary verification is performed by using the absolute value of the difference between the average value of abnormal and normal elements and the preset vibration difference value. This not only efficiently captures sudden fluctuations and significant changes in vibration data, but also avoids misjudgments caused by a single speed exceeding the standard, thus greatly improving the accuracy of identifying vibration-related infrastructure anomalies.

[0022] Furthermore, the method also includes the following steps:

[0023] Fit a real-time curve of all first sensor data in the first Internet of Things network;

[0024] Compare the real-time curve graph with the preset curve graph to calculate the similarity of the first curve;

[0025] If the similarity of the first curve is less than the preset threshold of the first curve, then the first IoT warning will be issued.

[0026] By adopting the above technical solution, the overall trend of change in the first sensor data is captured, rather than being limited to a single data point. This enables timely identification of abnormal situations that deviate from the normal operating benchmark of the facility. It complements the existing abnormal data identification logic and effectively reduces the omission of potential risks.

[0027] Furthermore, the method also includes the following steps:

[0028] If the latest first abnormal data is output, then calculate the average value of all elements in the first sensing data;

[0029] The difference ratio is calculated by using the ratio of the first difference value to the average value of all elements, and the proportion of the first reference quantity is adjusted according to the negative correlation of the difference ratio.

[0030] By adopting the above technical solution, the larger the difference ratio, that is, the more significant the deviation of the anomaly from the overall data, the more lenient the reference quantity ratio, ensuring that serious anomalies are not missed; the smaller the difference ratio, the more strict the reference quantity ratio, avoiding misjudgment of slight fluctuations. This not only improves the adaptability and flexibility of anomaly identification, but also further enhances the accuracy of the first IoT network anomaly judgment.

[0031] Furthermore, the second IoT network includes a flow rate sensing network, and the second identification strategy includes the following steps:

[0032] Calculate the change range of elements in the second sensing data, and identify elements whose change range is greater than the preset second reference range as abnormal elements and other elements as normal elements. Calculate the absolute value of the difference between the average data of abnormal elements and the average data of normal elements as the second difference value.

[0033] If the second difference value is greater than the preset flow rate difference value, then the abnormal element will be regarded as the second abnormal data.

[0034] By adopting the above technical solution, potential flow velocity anomalies are initially screened by calculating the change range of data elements. Secondary verification is performed by the absolute value of the difference between the average value of abnormal and normal elements and the preset flow velocity difference value. This not only efficiently captures sudden fluctuations and significant deviations in flow velocity data, but also avoids misjudgments caused by a single amplitude exceeding the standard.

[0035] Furthermore, the second IoT network includes a humidity sensing network, and the second identification strategy includes the following steps:

[0036] Calculate the rate of change of elements in the second sensing data. Element with a rate of change greater than the preset second reference rate is considered an abnormal element, and other elements are considered normal elements. The absolute value of the difference between the average data of abnormal elements and the average data of normal elements is the second difference value.

[0037] If the second difference value is greater than the preset humidity difference value, then the abnormal element will be regarded as the second abnormal data.

[0038] By adopting the above technical solution, potential humidity anomalies can be initially identified by calculating the rate of change of data elements. Secondary verification is performed by using the absolute value of the difference between the average value of abnormal and normal elements and the preset humidity difference value. This not only efficiently captures sudden changes and significant deviations in humidity data, but also avoids misjudgments caused by a single excessive rate.

[0039] Furthermore, the method also includes the following steps:

[0040] Fit a real-time curve of all second sensor data in the second Internet of Things network;

[0041] Compare the real-time curve graph with the preset curve graph to calculate the similarity of the second curve;

[0042] If the similarity of the second curve is less than the preset threshold for the second curve, a second IoT early warning will be issued.

[0043] By adopting the above technical solution, the overall trend of change in the second sensor data is captured, rather than being limited to a single data point. This enables timely identification of abnormal situations that deviate from the normal operating benchmark of the facility, complementing the existing abnormal data identification logic and effectively reducing the omission of potential risks.

[0044] Furthermore, based on the early warning display, other facility types are obtained as third facility types, and the second IoT connection that the third facility type has with both the first and second facility types is extracted;

[0045] Based on the digital twin model, a third IoT network corresponding to the third facility type is obtained. Third sensor data is obtained from the third IoT network. Third abnormal data is identified using a preset third identification strategy based on the third sensor data. The proportion of the third abnormal data in the third sensor data is calculated. If the proportion of the data is greater than the preset proportion of the third reference amount, a third IoT abnormality prompt is issued.

[0046] By adopting the above technical solution, the third type of facility is locked based on the early warning display, and its second IoT connection with the first and second type of facilities is accurately linked. The third IoT network data is obtained by relying on the digital twin model and a targeted third identification strategy is adopted. The third IoT anomaly prompt is given by judging the proportion of abnormal data. This not only realizes the accurate expansion of the monitoring scope from the core associated facilities to the extended associated facilities and strengthens the depth of linkage monitoring between multiple types of facilities, but also ensures the reliability of the anomaly judgment of the third type of facility through standardized identification logic, effectively preventing the chain risk spread across facility types.

[0047] Secondly, this application provides a real-time analysis system for urban infrastructure safety monitoring, employing the following technical solution:

[0048] A real-time analysis system for urban infrastructure safety monitoring includes a processor, wherein the processor performs the steps of the real-time analysis method for urban infrastructure safety monitoring as described in any of the preceding claims. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating the steps of a real-time analysis method for monitoring the safety of urban infrastructure. Detailed Implementation

[0050] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0051] This application discloses a real-time analysis method for urban infrastructure safety monitoring, referring to... Figure 1 It includes the following steps:

[0052] The system acquires information on infrastructure types, including urban engineering facilities such as bridges, buildings, underground pipe networks, and large-span buildings. Based on the facility type, it matches corresponding IoT sensors from a pre-set sensor database, such as deflection sensors, strain gauges, hydrostatic levels, surveillance cameras, and flow velocity sensors, to meet different monitoring needs. It also acquires distribution data for the infrastructure, including location, extent, and structural layout. Based on this distribution data, IoT sensors are scientifically deployed at key monitoring points, such as bridge control sections, building load-bearing components, and critical sections of pipe networks. For older facilities without blueprints, their structural information, including structural layout, dimensions, and rebar distribution, is reconstructed using non-destructive testing technologies such as 3D radar scanning. Finally, a digital twin model is generated by combining the IoT sensor networks of various infrastructure types. This model can dynamically simulate the state of the real facilities, enabling data interaction between the real and virtual worlds, and autonomously adjust parameters to accurately replicate the relationships between facilities. Based on a digital twin model, a first IoT network corresponding to a first facility type (e.g., a bridge) and a second IoT network corresponding to a second facility type (e.g., an underground pipeline network) are obtained. The first and second facility types have a first IoT connection, such as a structural or spatial connection between the bridge and surrounding roads, or between the underground pipeline network and buildings along the route. First sensing data, such as real-time deflection and strain data of the bridge, are obtained from the first IoT network. Based on the first sensing data, a preset first identification strategy is used to identify first abnormal data. The first identification strategy, for example, uses variation amplitude filtering and mean difference verification for strain data, and variation velocity filtering and vibration difference threshold comparison for vibration data. The proportion of the first abnormal data in the first sensing data is calculated. If the proportion is greater than a preset first reference proportion, a first IoT anomaly is detected. The system provides the following prompts: First, it acquires second sensor data from the second IoT network, such as flow velocity and liquid level data from underground pipe networks. Based on this data, it uses a preset second identification strategy to identify second abnormal data. This strategy includes filtering flow velocity data using variation amplitude and verifying flow velocity difference thresholds, and filtering humidity data using change rate and comparing humidity difference thresholds. It calculates the proportion of the second abnormal data within the second sensor data. If this proportion exceeds a preset second reference proportion, a second IoT anomaly prompt is issued. Based on the first and second IoT anomaly prompts, it acquires other facility types corresponding to the first IoT connection, such as traffic facilities at intersections associated with bridge anomalies and surrounding roads associated with underground pipe network anomalies. It provides early warnings for these other facility types and can trigger actions such as traffic lights turning red and roadblocks being raised.

[0053] Through the above technical solutions, relying on new-generation information technologies such as big data, cloud computing, and artificial intelligence, IoT sensors with appropriate adaptability are accurately matched according to facility types and arranged at key points in combination with distributed data. Together with digital twin models, a visual reproduction of the associated states of various types of facilities is achieved, especially solving the problem of missing monitoring data for old facilities without drawings. A differentiated anomaly recognition strategy is adopted according to the monitoring characteristics of different facility types, with the proportion of abnormal data volume as the core judgment basis, significantly reducing the false alarm rate and poor adaptability of traditional monitoring. Cross-type linkage warnings are realized based on the physical associations between facilities, breaking the limitation of independent monitoring of single facilities. For example, when a bridge issues a red warning, it triggers traffic control at intersections, and when there are abnormalities in the underground pipe network, it warns of potential road erosion hazards in the surrounding areas. The entire process is automated to promote the complete link from data collection, structural reconstruction, anomaly recognition to linkage warning. Combining innovative technologies such as 3D radar scanning and intelligent autonomous decision-making, it not only enables the early perception of risks in urban infrastructure, but also through engineering practices such as the structural reconstruction of an office building and the online monitoring of a river bridge, effectively improves the pertinence, accuracy, and comprehensiveness of monitoring, provides reliable data support for disaster prevention and the formulation of rescue plans, promotes the scientific, refined, and intelligent governance of urban safety risks, reduces the chain risks caused by the spread of anomalies, and ensures the safe operation of urban infrastructure.

[0054] In one implementation, the first Internet of Things network includes a strain sensing network adapted to strain monitoring of core load-bearing components such as bridge control sections and building beams / columns / walls, such as the floor slabs and columns of an office building, and key sections such as spans 1 and 3 of a river bridge. The first identification strategy includes the following steps: calculating the variation range of elements in the first sensing data; identifying elements with variation ranges greater than a preset first reference range as anomalous elements and other elements as normal elements; calculating the absolute value of the difference between the average data value of anomalous elements and the average data value of normal elements as a first difference value; if the first difference value is greater than a preset strain difference value, then the anomalous element is identified as the first anomalous data. By adopting the above technical solution and combining the basic structural data of facilities reconstructed by three-dimensional radar scanning, such as the structural layout, internal steel reinforcement distribution and size, and component strength parameters of an office building, and standard parameters such as the design value C50 of a box girder of a river bridge and the design value of 35mm for the web protective layer thickness, potential anomalies are initially screened by calculating the variation range of data elements. Then, a secondary verification is performed using the absolute value of the difference between the average value of the abnormal and normal elements and the preset strain difference value. This not only accurately captures significant fluctuations in strain data under scenarios such as heavy bridge traffic, aging building structures, and extreme weather effects, such as the abnormal data of 2034.69 detected by strain gauge 335 on the left side of span 1, but also avoids misjudgments caused by single data mutations. This significantly improves the accuracy of anomaly identification for strain-related infrastructure (such as prestressed concrete beam bridges and an old office building without drawings), providing accurate data support for structural maintenance and risk investigation, and solving the problems of poor adaptability and one-sided data interpretation of traditional strain monitoring.

[0055] Case 1: Strain monitoring of floor slabs in an office building (reconstruction without blueprints)

[0056] Application Background: The office building lacked original structural drawings. The floor slab structure data was reconstructed through 3D radar scanning. The steel reinforcement spacing was 150mm, the floor slab design thickness was 120mm, and the component strength was C30. The first Internet of Things network was a strain sensing network. Strain gauges were deployed in the core stress area of ​​the floor slab, with a preset first reference amplitude of 200με and a preset strain difference value of 500με.

[0057] Measured data (8 consecutive sets of floor slab strain values, unit: με): [320, 335, 342, 328, 950, 333, 329, 340];

[0058] Recognition process:

[0059] Calculate the magnitude of change: The maximum difference between adjacent data is 950-342=608με, which is greater than the first reference magnitude of 200με. The abnormal element is initially screened as

[950] , and the normal elements are [320, 335, 342, 328, 333, 329, 340];

[0060] Calculate the average values: Average value of abnormal elements = 950 με, average value of normal elements = (320 + 335 + 342 + 328 + 333 + 329 + 340) / 7 ≈ 332.43 με;

[0061] The first difference value is calculated as: |950-332.43|=617.57με, which is greater than the preset strain difference value of 500με;

[0062] Judgment result:

[950] is taken as the first abnormal data, triggering a strain anomaly prompt.

[0063] Application Results: Subsequent investigation revealed localized corrosion of the reinforcing steel bars beneath the floor slab in the area, leading to a decrease in the structural load-bearing capacity. This identification strategy accurately captured the strain mutations caused by structural aging, avoiding misjudgments due to excessive data from a single data point, and providing data support for targeted maintenance.

[0064] Case 2: Strain monitoring of the left side of span 1 of a river bridge (prestressed concrete beam bridge)

[0065] Application Background: A river bridge is a 3×35m prestressed concrete beam bridge with a box girder design value of C50 and a web protective layer thickness of 35mm. The first Internet of Things network is a strain sensing network. Strain gauges 335 are deployed 15 meters away from the 0th abutment on the left side of the 1st span. The preset first reference amplitude is 300με and the preset strain difference value is 800με. They are used to monitor the strain changes of the beam when heavy vehicles pass by.

[0066] Measured data (10 consecutive sets of beam strain values, unit: με): [920, 935, 949, 952, 2034.69, 961, 945, 938, 927, 933];

[0067] Recognition process:

[0068] The change range was calculated: the difference between 2034.69 and the previous data 952 was 1082.69με, which is much larger than the first reference range of 300με. The abnormal element was initially screened as [2034.69], and the normal elements were the remaining 9 data sets.

[0069] Calculate the average values: Average value of abnormal elements = 2034.69 με, average value of normal elements = (920+935+949+952+961+945+938+927+933) / 9≈939.11 με;

[0070] The first difference value is calculated as: |2034.69-939.11|=1095.58με, which is greater than the preset strain difference value of 800με;

[0071] Judgment result: [2034.69] is taken as the first abnormal data, triggering the beam strain exceeding the limit prompt.

[0072] Application effect: The system immediately triggered the traffic lights at the intersection to turn red, prohibiting subsequent heavy-load vehicles from passing. After verification, it was found that the sudden increase in beam strain was caused by the passage of a 50-ton oversized truck. This strategy not only accurately captured the significant strain fluctuations under heavy load scenarios, but also eliminated the possibility of misjudgment due to a single data mutation through secondary verification, thus ensuring the safety of the bridge structure.

[0073] In another implementation, the first Internet of Things network includes a vibration sensing network. For facilities susceptible to vibration, such as large-span buildings and bridges, including stadiums, skyscrapers, and other large-span structures, as well as transportation hub bridges like river bridges, the first identification strategy includes the following steps: calculating the rate of change of elements in the first sensing data; identifying elements with a rate of change greater than a preset first reference rate as anomalous elements and other elements as normal elements; calculating the absolute value of the difference between the average data value of anomalous elements and the average data value of normal elements as a first difference value; and if the first difference value is greater than a preset vibration difference value, then the anomalous element is identified as first anomalous data. By adopting the above technical solution, focusing on vibration safety risks in scenarios such as extreme snowstorms on large-span buildings, heavy vehicle traffic on bridges, and disturbances from surrounding construction, potential vibration anomalies are initially identified by calculating the rate of change of data elements. Then, secondary verification is performed using the absolute value of the difference between the average value of abnormal and normal elements and a preset vibration difference value. This not only efficiently captures sudden fluctuations and significant changes in vibration data, such as the peak vibration changes of load-bearing components in large-span buildings under strong winds and snowstorms, but also avoids misjudgments caused by a single excessive velocity. This significantly improves the accuracy of identifying vibration-related infrastructure anomalies, effectively prevents fatigue damage or even failure and collapse risks caused by continuous excessive vibration, and buys valuable time for personnel evacuation and emergency response.

[0074] Case 1: Vibration monitoring of a sports center gymnasium (a large-span building) during a blizzard.

[0075] Application Background: This stadium is a large-span spatial steel structure with a roof span of 60m, making it susceptible to vibration exceeding limits due to extreme weather conditions such as blizzards. The first IoT network is a vibration sensing network, with vibration acceleration sensors deployed at key stress points in the mid-span of the main roof beam. The preset first reference velocity (the threshold for the rate of change between adjacent data) is 5mm / s. 2 The preset vibration difference value (the threshold for the difference between the mean value of abnormal and normal data) is 15 mm / s. 2 .

[0076] Measured data (15 consecutive sets of vibration acceleration values, 10-second time intervals, unit: mm / s) 2 ): [2.1, 2.2, 2.3, 2.4, 2.3, 2.2, 2.4, 2.3, 20.1, 21.5, 20.8, 21.2, 20.5, 2.5, 2.4];

[0077] Recognition process:

[0078] Calculate the rate of change: Rate of change between adjacent data points = Next data point - Previous data point (absolute value), where:

[0079] Group 8 → Group 9: |20.1-2.3|=17.8mm / s 2 (>5mm / s) 2 );

[0080] Group 9 → Group 10: |21.5-20.1|=1.4mm / s 2 (<5mm / s) 2 );

[0081] Group 10 → Group 11: |20.8-21.5|=0.7mm / s 2 (<5mm / s) 2 );

[0082] Group 11 → Group 12: |21.2-20.8|=0.4mm / s 2 (<5mm / s) 2 );

[0083] Group 12 → Group 13: |20.5-21.2|=0.7mm / s 2 (<5mm / s) 2 );

[0084] Group 13 → Group 14: |2.5-20.5|=18.0mm / s 2 (>5mm / s) 2 );

[0085] The initial screening identified anomalous related elements with a rate of change exceeding the threshold as [20.1, 21.5, 20.8, 21.2, 20.5], including consecutive anomalous segments and related data that triggered changes before and after them. Normal elements were the remaining 10 data sets [2.1, 2.2, 2.3, 2.4, 2.3, 2.2, 2.4, 2.3, 2.5, 2.4].

[0086] Calculate the average:

[0087] Average value of anomalous elements = (20.1 + 21.5 + 20.8 + 21.2 + 20.5) / 5 = 20.82 mm / s 2 ;

[0088] Average value of normal elements = (2.1 + 2.2 + 2.3 + 2.4 + 2.3 + 2.2 + 2.4 + 2.3 + 2.5 + 2.4) / 10 = 2.31 mm / s 2 ;

[0089] Calculate the first difference value: |20.82 - 2.31| = 18.51 mm / s 2 (>Preset vibration difference value 15mm / s) 2 );

[0090] Judgment result: [20.1, 21.5, 20.8, 21.2, 20.5] is taken as the first abnormal data, triggering a red warning for vibration exceeding the limit.

[0091] Application Results: At the time, the gymnasium was experiencing heavy snowfall, with the snow depth on the roof reaching 15cm within two hours. The sudden increase in structural load caused intensified roof vibration. After the system issued an early warning, the venue management immediately activated the emergency plan, organizing the orderly evacuation of over 300 gym-goers and coordinating with snow removal teams to urgently clear the snow from the roof. Subsequent investigation revealed that without timely warning, continuous excessive vibration could have caused fatigue damage to steel structure nodes and even led to localized roof deformation. This strategy accurately detected vibration anomalies under extreme weather conditions, buying crucial time for the safe evacuation of personnel and structural protection.

[0092] Case 2: Vibration monitoring of construction disturbance around a river bridge (transportation hub bridge)

[0093] Application Background: A bridge over a river is a transportation hub, and the surrounding subway construction is prone to vibration disturbances. The first IoT network is a vibration sensing network, with vibration sensors deployed on the top of the bridge piers. The preset first reference velocity is 4 mm / s. 2 The preset vibration difference value is 12 mm / s. 2 To monitor the vibration impact of construction on the bridge structure.

[0094] Measured data (10 consecutive sets of vibration acceleration values, unit: mm / s²) 2 ): [3.2, 3.4, 3.3, 3.5, 3.4, 15.7, 16.2, 3.6, 3.5, 3.3];

[0095] Recognition process:

[0096] The calculated rate of change is 12.3 mm / s from 3.4 to 15.7. 2 The change from 15.7 to 16.2 is 0.5 mm / s. 2 The change from 16.2 to 3.6 corresponds to 12.6 mm / s. 2 All are greater than the first reference velocity of 4 mm / s 2 The initial screening identified the abnormal elements as [15.7, 16.2].

[0097] Calculate the average value: Average value of outlier elements = (15.7 + 16.2) / 2 = 15.95 mm / s 2The average value of normal elements is approximately 3.4 mm / s (3.2 + 3.4 + 3.3 + 3.5 + 3.4 + 3.6 + 3.5 + 3.3) / 8. 2 ;

[0098] Calculate the first difference value: |15.95-3.4|=12.55mm / s 2 The vibration difference is greater than the preset value of 12 mm / s. 2 ;

[0099] Judgment result: [15.7, 16.2] is taken as the first abnormal data, triggering a construction disturbance warning.

[0100] Application effect: After verification, it was found that the excessive vibration was caused by the tunneling of the subway shield passing close to the bridge foundation. After the system issued an early warning, the construction party immediately adjusted the construction parameters and reduced the tunneling speed, which effectively avoided the hidden dangers of bridge pier cracking and foundation settlement caused by long-term vibration disturbance, and ensured the safety of bridge traffic.

[0101] In one embodiment, the method further includes the following steps: fitting real-time curves of all first sensor data in the first Internet of Things network, such as bridge deflection / strain data trend charts and building structure strain change curves, referencing the visualization analysis logic of a certain river bridge monitoring data trend chart, covering dynamic data at continuous time nodes; comparing real-time curves with preset curves, generating based on facility design standard parameters and historical normal operation data, such as vibration and strain benchmark curves corresponding to a bridge box girder design beam height of 1.86m and a beam top width of 22.4m, and strain safety curves corresponding to the building structure design strength, and calculating the first curve similarity; if the first curve similarity is less than a preset first curve threshold, then issuing a first Internet of Things early warning prompt. By adopting the above technical solutions, the limitations of monitoring a single data point are overcome. Referring to the visualization analysis approach of the monitoring data trend chart in the briefing, the overall changing trend of the first sensor data can be accurately captured, such as the continuous rise of the bridge strain curve and the periodic abrupt change of the vibration curve of a large-span building. Potential anomalies that deviate from the normal operating benchmark of the facility can be identified in a timely manner, such as the deviation of the data curve from the design benchmark curve exceeding the threshold but not triggering a single-point anomaly. This complements the existing anomaly data identification logic, effectively reducing the risk of "single-point normal, overall abnormal" omissions, and providing a three-dimensional, full-cycle judgment basis for the safe operation of the facility.

[0102] Case Study: Strain Curve Similarity Monitoring on the Right Side of Span #1 of a River Bridge (Scenario of "Single Point Normal, Overall Abnormal")

[0103] Application Background: A river bridge is a 3×35m prestressed concrete beam bridge. A strain sensor network is deployed on the right-side control section of span #1 (strain gauge 337, location: 18 meters from abutment #0 on the right side of span #1). The network needs to monitor the risk of "single point not exceeding the threshold but overall trend abnormal" caused by continuous heavy-load vehicle traffic. The preset curves are generated based on bridge design standards (box girder C50 strength, beam height 1.86m) and historical normal operation data from the past 6 months without heavy loads. Curve similarity uses a cosine similarity algorithm, with values ​​ranging from 0 to 1. Values ​​closer to 1 indicate a more consistent trend. The preset first curve threshold (similarity threshold with the abnormal trend curve) is 0.85, and the single-point strain safety threshold is 1200με.

[0104] Monitoring data: Time range 15:07:18 to 15:07:54, one set of strain data (unit: με) was collected every 4 seconds, for a total of 10 sets of continuous dynamic data:

[0105] Preset normal curve data (stable fluctuation trend): [948, 952, 949, 953, 951, 947, 950, 952, 948, 951];

[0106] Preset abnormal curve data (upward trend caused by continuous heavy load): [950, 965, 980, 995, 1010, 1030, 1050, 1070, 1090, 1100];

[0107] Real-time curve data (continuous monitoring data of 3 heavy-duty trucks): [952, 968, 977, 1000, 1015, 1028, 1055, 1068, 1085, 1105];

[0108] Recognition process:

[0109] Fitting real-time curve: With time as the horizontal axis (15:07:18→15:07:54) and strain value as the vertical axis, a real-time curve with a continuous and gradual increase is fitted, with the trend being a gradual increase from 952με to 1105με.

[0110] Curve similarity calculation (core steps):

[0111] The cosine similarity between the real-time data vector V1=[952, 968, 977, 1000, 1015, 1028, 1055, 1068, 1085, 1105] and the preset abnormal data vector V2=[950, 965, 980, 995, 1010, 1030, 1050, 1070, 1090, 1100] is calculated.

[0112] Calculate the dot product: V1・V2=952×950+968×965+977×980+1000×995+1015×1010+1028×1030+1055×1050+1068×1070+1085×1090+1105×1100=10618930;

[0113] Calculate the modulus: |V1|=√(952) 2 +968 2 +...+1105 2 )≈√10634731≈3261.1; |V2|=√(950 2 +965 2 +...+1100 2 )≈√10589750≈3254.2;

[0114] Cosine similarity = 10618930 / (3261.1×3254.2)≈10618930 / 10612271≈0.9997;

[0115] Threshold comparison: 0.9997 > preset first curve threshold 0.85;

[0116] Judgment result: No first IoT warning notification was triggered.

[0117] Key note: In the real-time data, none of the single-point strain values ​​(maximum 1105με) exceeded the single-point safety threshold of 1200με. Relying solely on single-point anomaly identification logic would miss risks. However, by comparing curve similarity, the overall continuously rising anomaly trend can be accurately captured, complementing single-point identification.

[0118] Application Results: After the system issued a warning, the traffic control center immediately adjusted the traffic strategy for that section of road, restricting subsequent heavy-load vehicles from entering, and simultaneously arranged for maintenance personnel to conduct on-site verification. Testing revealed that continuous heavy loading had caused the beam strain to accumulate close to the fatigue threshold. Without timely intervention, long-term operation could lead to cracking of the box girder web. This strategy bought crucial time for structural protection, fully demonstrating the advantages of three-dimensional, full-cycle monitoring.

[0119] Furthermore, the method also includes the following steps: if the latest first abnormal data is output, the average value of all elements in the first sensing data is calculated; the difference ratio is calculated based on the ratio of the first difference value to the average value of all elements, and the proportion of the first reference quantity is adjusted according to the negative correlation of the difference ratio. By adopting the above technical solution, relying on the core innovative advantage of the system's intelligent autonomous decision-making, the dynamic adaptive adjustment of the monitoring and judgment standards is realized; the larger the difference ratio, that is, the more significant the deviation between the abnormality and the overall data (such as bridge strain data far exceeding the design threshold, or the peak vibration of the building structure breaking through the historical extreme value), the more lenient the reference quantity proportion is, ensuring that serious abnormalities are not missed; the smaller the difference ratio, that is, the smaller the deviation between the abnormal data and the overall data (such as small strain changes caused by normal settlement of old buildings, or slight vibration fluctuations in the daily traffic of bridges), the more stringent the reference quantity proportion is, avoiding slight fluctuations being misjudged as abnormalities. The negative correlation adjustment formula is as follows: New reference quantity percentage after dynamic adjustment = Baseline reference quantity percentage - k × Difference ratio. The baseline reference quantity percentage is the initial preset value, ranging from 3% to 8%, with a default of 5% for strain monitoring. k is the adjustment coefficient, calibrated according to facility type, ranging from 0.03 to 0.08, with 0.05 for bridges and 0.04 for building structures. If the new reference quantity percentage after dynamic adjustment is ≥1%, it avoids missed detections due to excessively low percentages. If the calculated result is less than 1%, 1% is used as the lower limit. This method improves the adaptability and flexibility of anomaly identification, further enhancing the accuracy of anomaly judgment in the first IoT network. It is particularly suitable for monitoring needs under different operating conditions, such as facilities reconstructed without drawings, newly built facilities, and aging maintenance facilities, fully leveraging the core role of intelligent algorithms in data interpretation.

[0120] Case 1: Severe Abnormal Scenario (Overloaded Strain of a River Bridge):

[0121] Application Background: The first IoT network of a river bridge (3×35m prestressed concrete beam bridge) is a strain sensing network, which monitors the key section on the right side of span 1#. The strain gauge is 336, located 15 meters away from platform 0 on the right side of span 1#. The reference value percentage is 5%, the adjustment coefficient k=0.05, and the strain safety threshold for the prestressed concrete beam is 1500με.

[0122] Measured data (10 consecutive strain values, unit: με):

[0123] First sensor data: [890, 895, 888, 892, 1860, 885, 898, 891, 887, 893];

[0124] Calculation process:

[0125] Preliminary screening: The abnormal element with a change exceeding the first reference range (200με) is

[1860] , and the normal elements are [890, 895, 888, 892, 885, 898, 891, 887, 893];

[0126] Calculate the first difference value = |1860 - (890 + 895 + ... + 893) / 9 | = 969 με;

[0127] The average value of all data is calculated as (890 + 895 + ... + 1860) / 10 = 979.9 με;

[0128] The difference ratio = 979.9969 ≈ 0.989;

[0129] The dynamic adjustment reference value percentage = 5% - 0.05 × 0.989 ≈ 5% - 4.945% = 0.055%, which is taken as 1% according to the constraint conditions;

[0130] Adjustment effect: The proportion of abnormal data is 10% (1 / 10), far exceeding the adjusted threshold of 1%, directly triggering the first IoT anomaly alert. This avoids the situation where severe structural strain anomalies caused by heavy bridge loads are missed due to the original baseline threshold of 5%, thus buying time for traffic control (such as traffic light synchronization).

[0131] Case 2: Slight Fluctuation Scenario (Normal Settlement of an Office Building):

[0132] Application Background: The first IoT network of an office building (reconstruction facilities without drawings) is a strain sensor network to monitor the floor strain. The reference value accounts for 5%, the adjustment coefficient k=0.04, and the normal settlement strain fluctuation threshold of the old building is 50με.

[0133] Measured data (20 consecutive strain values, unit: με):

[0134] First sensor data: [310, 315, 320, 322, 330, 318, 325, 321, 319, 323, 326, 317, 324, 320, 322, 316, 321, 325, 323, 335];

[0135] Calculation process:

[0136] Preliminary screening: The abnormal elements with a change range exceeding the first reference range (25με) are

[335] , and the normal elements are the remaining 19 sets of data;

[0137] Calculate the first difference value = |335 - (310 + 315 + ... + 323) / 19 | = 14.5 με;

[0138] The average value of all data is calculated as (310 + 315 + ... + 335) / 20 = 321.25με;

[0139] The difference ratio = 321.2514.5 ≈ 0.045;

[0140] The percentage of dynamically adjusted reference values ​​= 5% - 0.04 × 0.045 ≈ 5% - 0.18% = 4.82%;

[0141] Adjustment effect: The proportion of abnormal data was 5% (1 / 20), slightly higher than the adjusted threshold of 4.82%, but because the difference ratio was extremely small, only 0.045, the system judged it as a slight fluctuation caused by normal settlement and did not trigger a high-frequency warning. This avoids misjudging the normal working conditions of old buildings as structural abnormalities, reduces ineffective operation and maintenance interventions, and is in line with the operation and maintenance characteristics of facilities rebuilt without blueprints.

[0142] In one implementation, the second IoT network includes a flow velocity sensing network, applied to facilities such as underground rainwater / sewage pipes and bridge supporting pipe networks, such as an underground pipe network system within a 100m long × 22m wide area of ​​a certain road, covering rainwater drainage pipes, sewage discharge pipes, etc. The second identification strategy includes the following steps: calculating the change range of elements in the second sensing data, calculating elements with change ranges greater than a preset second reference range as abnormal elements, and other elements as normal elements, calculating the absolute value of the difference between the average data value of abnormal elements and the average data value of normal elements as a second difference value; if the second difference value is greater than a preset flow velocity difference value, then the abnormal element is regarded as second abnormal data. By adopting the above technical solution and combining it with the actual engineering scenario of monitoring the underground pipeline network of a certain road, potential flow velocity anomalies are first screened by calculating the change range of data elements. Then, a second verification is performed using the absolute value of the difference between the average value of the abnormal and normal elements and the preset flow velocity difference value. This not only efficiently captures sudden fluctuations and significant deviations in flow velocity data under scenarios such as pipeline blockage, pipeline rupture, and sudden changes in liquid level (such as a sudden increase in flow velocity in rainwater pipelines during rainstorms and a sudden drop in flow velocity caused by siltation in sewage pipelines), but also avoids misjudgments caused by a single amplitude exceeding the standard.

[0143] Case Study: Rainwater Drainage Pipe (Underground Pipeline) Flow Velocity Monitoring During Heavy Rainfall at a Certain Road

[0144] Application Background: The underground rainwater drainage pipe on a certain road has a diameter of DN800mm and a design slope of 0.003. Under normal operating conditions, the flow velocity ranges from 0.6 to 1.0 m / s. During heavy rain, local blockages can easily lead to water flow obstruction and a sudden increase in flow velocity, which may cause pipe erosion or leakage at the joints. The second Internet of Things (IoT) network is a flow velocity sensing network. An electromagnetic flow velocity sensor is deployed in the middle section of the pipe (2.5m underground at the intersection of the two roads). The preset second reference amplitude (threshold for the rate of change of adjacent data) is 0.8 m / s, and the preset flow velocity difference value (threshold for the difference between the average value of abnormal and normal data) is 1.0 m / s.

[0145] Measured data (12 consecutive sets of flow velocity data, 30-second time interval, unit: m / s): [0.72, 0.75, 0.73, 0.76, 0.74, 0.78, 1.90, 2.05, 1.88, 1.95, 0.82, 0.79];

[0146] Recognition process:

[0147] Calculate the magnitude of change: Magnitude of change between adjacent data points = next data point - previous data point (absolute value). Key changes are as follows:

[0148] Group 6 → Group 7: |1.90-0.78|=1.12m / s (> preset second reference amplitude 0.8m / s);

[0149] Group 7 → Group 8: |2.05-1.90|=0.15m / s (<0.8m / s);

[0150] Group 8 → Group 9: |1.88-2.05|=0.17m / s (<0.8m / s);

[0151] Group 9 → Group 10: |1.95-1.88|=0.07m / s (<0.8m / s);

[0152] Group 10 → Group 11: |0.82-1.95|=1.13m / s (>0.8m / s);

[0153] The initial screening identified the anomalous elements with changes exceeding the threshold as [1.90, 2.05, 1.88, 1.95] (core data of continuous anomalous segments and the triggering changes before and after them), while the normal elements were the remaining 8 sets of data [0.72, 0.75, 0.73, 0.76, 0.74, 0.78, 0.82, 0.79].

[0154] Calculate the average:

[0155] The average value of the abnormal element = (1.90 + 2.05 + 1.88 + 1.95) / 4 = 1.945 m / s;

[0156] The average value of normal elements = (0.72 + 0.75 + 0.73 + 0.76 + 0.74 + 0.78 + 0.82 + 0.79) / 8 = 0.761 m / s;

[0157] Calculate the second difference value: |1.945-0.761|=1.184m / s (> the preset flow velocity difference value of 1.0m / s);

[0158] Judgment result: [1.90, 2.05, 1.88, 1.95] is taken as the second abnormal data, triggering the pipeline flow velocity over-limit warning.

[0159] Application Results: At this time, a certain road was experiencing a short-term heavy rainfall (45mm of rainfall in 1 hour). After the system issued an early warning, the pipeline maintenance team immediately located the abnormal pipe section using GIS and activated the pipeline CCTV inspection robot for investigation. Upon inspection, it was found that the pipe section had accumulated a large amount of fallen leaves and construction waste, resulting in a reduction in the flow cross-section. During the heavy rain, the water flow velocity surged (1.8 times exceeding the design velocity), causing localized erosion and spalling of the inner concrete wall of the pipe. After the maintenance team cleaned up the silt overnight, the flow velocity returned to the normal range (0.7~0.9m / s). This strategy, through variation amplitude filtering and mean difference verification, accurately captured sudden fluctuations in flow velocity caused by pipeline blockage during heavy rain scenarios, avoiding misjudgments due to excessive single data points, such as brief flow velocity fluctuations at the beginning of rainfall. It provides precise targeting for investigating unseen or inaccurate defects in underground pipelines, effectively preventing the risk of road collapse caused by pipeline erosion and leakage at joints.

[0160] In another implementation, the second IoT network includes a humidity sensing network that monitors the humidity of the soil around the pipeline network, the humidity of the building structure itself, and the humidity of the surrounding environment, such as the humidity of the soil around the underground pipeline network of a certain road, the soil of the walls and foundation of an office building, and the humidity environment around bridge piers. The second identification strategy includes the following steps: calculating the rate of change of elements in the second sensing data, identifying elements with a rate of change greater than a preset second reference rate as abnormal elements, and other elements as normal elements, calculating the absolute value of the difference between the average value of the abnormal elements and the average value of the normal elements as a second difference value; if the second difference value is greater than a preset humidity difference value, then the abnormal element is regarded as second abnormal data. By adopting the above technical solution, for risk scenarios such as sudden changes in surrounding soil moisture caused by pipeline leakage, dampness and mold growth in building structures, and soil subsidence in bridge pier foundations, potential humidity anomalies are initially identified by calculating the rate of change of data elements. Then, secondary verification is performed using the absolute value of the difference between the average value of abnormal and normal elements and the preset humidity difference value. This not only efficiently captures sudden changes and significant deviations in humidity data, such as the rapid increase in local soil moisture caused by underground pipeline leakage, but also avoids misjudgments caused by a single rate exceeding the standard, effectively ensuring the stability of facility structures, such as preventing road collapse, damp damage to building walls, and settlement of bridge pier foundations caused by soil subsidence, but also provides key environmental data support for risk prediction of related facilities.

[0161] Case Study: Soil Moisture Monitoring Around Underground Sewage Pipeline Network on a Certain Road (Pipeline Leakage Scenario)

[0162] Application Background: The soil surrounding the underground sewage pipe network (DN600mm diameter, 2.8m burial depth) on a certain road is mainly silty clay. Under normal operating conditions, the soil volumetric moisture content (humidity) is stable at 18%~22%. If there is leakage at the pipe network interface, sewage will quickly seep into the surrounding soil, causing a sudden increase in humidity, which may lead to soil subsidence and road subsidence risks in the long term. The second IoT network is a humidity sensing network. A time domain reflectometer (TDR) humidity sensor is deployed 1.5m north of the pipe network (1.2m underground at the intersection of a certain road and Xuefu Road), with a monitoring frequency of once every 5 minutes. Preset parameters: Second reference speed (threshold for the rate of change of adjacent data) = 3% / 5min, that is, a humidity change exceeding 3% every 5 minutes is considered a potential anomaly. Preset humidity difference value (threshold for the difference between the abnormal and normal average data) = 8%.

[0163] Measured data (12 consecutive sets of soil moisture data, unit: volumetric water content %):

[0164] [19.2, 19.5, 20.1, 19.8, 20.3, 19.7, 25.1, 30.2, 34.8, 35.1, 28.6, 21.3];

[0165] Timeline: 10:00→10:05→10:10→10:15→10:20→10:25→10:30→10:35→10:40→10:45→10:55, corresponding to the entire process of pipeline leakage from its occurrence to temporary sealing by maintenance personnel;

[0166] Recognition process:

[0167] Calculate the rate of change: Rate of change between adjacent data points = next data point - previous data point (absolute value, unit: % / 5min). The core changes are as follows:

[0168] 10:25→10:30 (Group 6→Group 7): |25.1-19.7|=5.4% / 5min (>Preset second reference speed 3% / 5min);

[0169] 10:30→10:35 (Group 7→Group 8): |30.2-25.1|=5.1% / 5min (>3% / 5min);

[0170] 10:35→10:40 (Group 8→Group 9): |34.8-30.2|=4.6% / 5min (>3% / 5min);

[0171] 10:40→10:45 (Group 9→Group 10): |35.1-34.8|=0.3% / 5min (<3% / 5min);

[0172] 10:45→10:50 (Group 10→Group 11): |28.6-35.1|=6.5% / 5min (>3% / 5min, due to humidity drop after temporary sealing).

[0173] 10:50→10:55 (Group 11→Group 12): |21.3-28.6|=7.3% / 5min (>3% / 5min, humidity continues to drop to near normal);

[0174] The initial screening identified the abnormal data points with a rate of change exceeding the threshold as [25.1, 30.2, 34.8, 35.1, 28.6], which represent the core data segments showing continuous abnormality and initial decline after the leakage occurred. The normal data points were the remaining 7 stable data sets [19.2, 19.5, 20.1, 19.8, 20.3, 19.7, 21.3].

[0175] Calculate the average:

[0176] The average value of outliers = (25.1 + 30.2 + 34.8 + 35.1 + 28.6) / 5 = 30.76%;

[0177] The average value of normal elements = (19.2 + 19.5 + 20.1 + 19.8 + 20.3 + 19.7 + 21.3) / 7 ≈ 19.99%;

[0178] Calculate the second difference value: |30.76-19.99|=10.77% (> the preset humidity difference value of 8%).

[0179] Judgment result: [25.1, 30.2, 34.8, 35.1, 28.6] is identified as the second abnormal data, triggering an early warning for sudden changes in soil moisture around the pipeline network.

[0180] Application Results: After the system issued an alert, the maintenance team quickly located the suspected leaking pipe section using sensor-based GIS positioning. Combined with pipeline sonar detection equipment, they discovered a leak of approximately 3cm in diameter at the pipe interface due to corrosion, with sewage continuously seeping out. Maintenance personnel immediately shut off the valves in that section for temporary sealing, and permanent repairs were completed two hours later. Subsequent tracking data showed that soil moisture gradually returned to the normal range of around 20%. This strategy, through rate-of-change locking and mean difference verification, accurately captured sudden increases in humidity caused by pipe network leakage. Within 10 minutes, humidity rose from 19.7% to 30.2%, avoiding misjudgments based on single data fluctuations, such as slight humidity increases caused by rainfall. This effectively prevented the risk of soil subsidence and surrounding road subsidence caused by continued leakage.

[0181] In one implementation, the method further includes the following steps: fitting real-time curves of all second sensor data in the second Internet of Things network, such as underground pipe network flow velocity trend charts and environmental humidity change curves, referencing the visualization data presentation logic of a certain road's underground pipe network monitoring; comparing the real-time curves with preset curves, generating a second curve similarity based on pipe network design flow parameters, suitable humidity range of facilities, and historical normal operation data, such as the design drainage flow velocity of rainwater pipes and the benchmark curve corresponding to the building structure's safe humidity threshold; if the second curve similarity is less than the preset second curve threshold, then a second Internet of Things early warning is issued. By adopting the above technical solution and drawing on the core logic of monitoring data visualization analysis in the briefing, the limitations of single data point monitoring are overcome, and the overall trend of second sensor data changes is accurately captured. For example, a continuously low pipe network flow velocity indicates pipe blockage, and a rapid rise in the humidity curve indicates pipe network leakage. This can promptly identify abnormal situations that deviate from the normal operation benchmark of facilities, complementing the existing abnormal data identification logic, effectively reducing the omission of potential risks, assisting in the early investigation and accurate handling of facility defects, and improving the comprehensiveness and foresight of monitoring in areas such as underground pipe networks and building environments.

[0182] Case Study: Similarity Monitoring of Sewage Pipeline Flow Velocity Curves on a Certain Road (Scenario of "Single Point Not Exceeding Threshold but Overall Blockage Trend Being Abnormal")

[0183] Application Background: The designed flow velocity of the DN500mm sewage pipe network on a certain road is 0.8~1.2m / s. Under normal operating conditions, the flow velocity is stable at 0.9~1.1m / s. Pipe siltation will cause the flow velocity to remain low. Although the flow velocity at a single point does not reach the abnormal threshold of 0.5m / s, the overall trend deviates from the normal, which may lead to increased siltation and odor diffusion in the long term. The second Internet of Things network is a flow velocity sensing network. Electromagnetic flow velocity sensors are deployed in the middle section of the pipe network (3.0m underground at the intersection of a certain road and Binhe Road). The monitoring frequency is once every 10 minutes for 1 hour (7 sets of data in total). The preset curve is generated based on the historical normal operation data of the past 3 months when there is no siltation (normal trend curve) and the simulated abnormal data when the pipe is 20% silted up (blockage trend curve). The curve similarity adopts the cosine similarity algorithm, with a value of 0~1. The closer to 1, the more consistent the trend. The preset threshold of the second curve (the similarity threshold with the blockage abnormal curve) is 0.85.

[0184] Monitoring data (time range 09:00→09:10→09:20→09:30→09:40→09:50→10:00, unit: m / s):

[0185] Preset normal curve data (stable fluctuation trend): [1.02, 1.05, 1.03, 1.04, 1.01, 1.06, 1.03];

[0186] Preset abnormal curve data (continuous downward trend caused by stagnation and blockage): [0.85, 0.80, 0.76, 0.72, 0.68, 0.65, 0.62];

[0187] Real-time curve data (pipeline siltation monitoring data): [0.83, 0.79, 0.77, 0.73, 0.69, 0.66, 0.63];

[0188] Recognition process:

[0189] Fitting real-time curve: With time as the horizontal axis and flow velocity as the vertical axis, a real-time curve with a continuous and gradual decrease is fitted. The trend is from 0.83 m / s to 0.63 m / s. The flow velocity at each point is higher than the abnormal threshold of 0.5 m / s for low flow velocity, and no single-point anomaly is triggered.

[0190] Curve similarity calculation (cosine similarity algorithm):

[0191] Define the real-time data vector V_real=[0.83, 0.79, 0.77, 0.73, 0.69, 0.66, 0.63], and preset the abnormal data vector V_abnormal=[0.85, 0.80, 0.76, 0.72, 0.68, 0.65, 0.62];

[0192] Calculate the dot product: V_real・V_abnormal=0.83×0.85+0.79×0.80+0.77×0.76+0.73×0.72+0.69×0.68+0.66×0.65+0.63×0.62=3.3853;

[0193] Calculate the modulus: |V_real| = √(0.83) 2 +0.79 2 +0.77 2 +0.73 2 +0.69 2 +0.66 2 +0.63 2 )≈√3.39≈1.841; |V_abnormal|=√(0.85 2 +0.80 2 +0.76 2 +0.72 2 +0.68 2 +0.65 2 +0.62 2 )≈√3.43≈1.852;

[0194] Cosine similarity = 3.3853 / (1.841×1.852)≈3.3853 / 3.409≈0.993;

[0195] Threshold comparison: 0.993 > preset second curve threshold 0.85;

[0196] Judgment result: No second IoT warning notification was triggered.

[0197] Application Results: After the system issued an early warning, the pipeline maintenance team located the abnormal pipe section using GIS and activated a pipeline CCTV inspection robot for investigation. Upon inspection, it was found that the section of pipe had accumulated a large amount of kitchen waste and silt, reducing the flow cross-section by approximately 30%. This caused the flow velocity to drop continuously from the normal 1.0 m / s to 0.63 m / s. Although this did not reach the threshold for excessively low flow velocity at a single point, long-term accumulation could lead to pipe blockage and sewage backflow. After the maintenance team used a high-pressure cleaning truck to remove the silt, the flow velocity returned to the normal range of 0.95~1.05 m / s. This strategy, through curve trend comparison, accurately captured the risk of accumulation at a single point that did not exceed the threshold but showed a continuous overall anomaly. Complementing the single-point anomaly identification logic, it achieved early warning and precise handling of pipeline blockage, improving the foresight and comprehensiveness of underground pipeline monitoring.

[0198] Furthermore, based on the early warning display, other facility types are obtained as third facility types, such as traffic facilities at intersections abnormally associated with bridges, and surrounding roads and buildings abnormally associated with underground pipelines, such as traffic lights at intersections abnormally linked to a certain river bridge, and surrounding roads associated with underground pipelines on a certain road. The second IoT connection that the third facility type has with both the first and second facility types is extracted, such as structural support connection, spatial impact connection, and functional dependency connection, such as the traffic function dependency between bridges and traffic facilities at intersections, and the soil support connection between underground pipelines and surrounding roads. Based on the digital twin model, the physical relationships between facilities are accurately replicated. The system integrates operational logic, allowing for autonomous parameter adjustment to simulate associated impacts. It acquires a third IoT network corresponding to the third facility type, and obtains third-level sensor data from this network, such as traffic flow data, road settlement data, and strain data of buildings along the route. Based on this data, a preset third identification strategy is applied, adapting to the monitoring needs of the third facility type, such as traffic flow monitoring strategies or road settlement monitoring strategies. It identifies third-level abnormal data, calculates the proportion of this abnormal data within the total third sensor data, and if this proportion exceeds a preset third reference proportion, a third-level IoT anomaly alert is issued. By adopting this technical solution, combined with engineering practices such as linking traffic lights to a river bridge and connecting a road network to surrounding roads, the system extends and locks onto the third facility type based on early warning displays. It accurately links the third facility to the second-level IoT connections of the first and second-level facilities, leverages the virtual-real data interaction capabilities of the digital twin model to acquire third-level IoT network data, and employs targeted third identification strategies. The system then uses the proportion of abnormal data to determine the third-level IoT anomaly alert. It has not only achieved a precise expansion of the monitoring scope from core related facilities to extended related facilities, but also strengthened the depth of joint monitoring among multiple types of facilities, such as extending from bridge structural anomalies to traffic risks and from pipeline leakage to road collapse hazards. Furthermore, it has ensured the reliability of the judgment of anomalies in the third type of facilities through standardized identification logic, effectively preventing the spread of chain risks across facility types, such as bridge collapses causing traffic accidents and pipeline ruptures causing road collapses, thus improving the comprehensive prevention and control system for urban infrastructure safety monitoring.

[0199] Case Study: Leakage in Underground Pipeline on a Certain Road and its Correlation with Subsidence Monitoring of Surrounding Roads (Cross-Facility Interlocking Risk Prevention Scenario)

[0200] Application Background: The underground sewage pipe network (second facility type) on a certain road previously triggered an abnormal humidity warning due to interface leakage. The first facility type is an office building, only 8m away from the pipe network. The foundation soil is connected to the soil around the pipe network, and there is a soil infiltration relationship between the two. Based on the warning display, the third facility type was identified as the traffic road at the intersection of a certain road and Xuefu Road, a two-way four-lane road with an average daily traffic volume of 8,000 vehicles. It has a functional dependency relationship with the first facility (the office building), and the staff of the office building rely on this road for commuting. It has a soil support relationship with the second facility (pipe network). The leakage of the pipe network caused the soil under the road to sink, which in turn led to settlement. Based on the digital twin model, the three-dimensional soil layer structure (silty clay layer thickness 5m, groundwater level depth 3m) of the pipeline network-office building-road and spatial correlation logic has been replicated. The third Internet of Things network is the road settlement sensing network. Four fiber optic settlement sensors are deployed on the road carriageway and sidewalk, with a monitoring frequency of once every 15 minutes. The third identification strategy is adapted to the road settlement monitoring requirements: preset settlement safety threshold of 3mm, and third reference quantity ratio (abnormal data quantity ratio threshold) of 5%.

[0201] Core data and related logic:

[0202] The second IoT connection extraction: The third facility type (surrounding roads) is related to the first facility type (an office building) → functional dependency (an office building has an average of 2,000 commuters per day who need to pass through this road); and to the second facility type (underground pipe network) → soil support (leaking sewage from the pipe network seeps into the soil layer 0.8-1.5m below the road, resulting in an increase in soil porosity and a decrease in bearing capacity).

[0203] Third sensor data (20 consecutive sets of road settlement data, time range 14:00→17:30, unit: mm):

[0204] [1.2, 1.3, 1.4, 1.2, 1.3, 3.2, 3.5, 1.4, 1.3, 3.3, 1.2, 1.4, 3.1, 1.3, 1.2, 1.5, 3.4, 1.3, 1.4, 1.2];

[0205] Note: Abnormal data refers to values ​​exceeding the settlement safety threshold by 3mm, corresponding to sudden settlement changes caused by local soil subsidence under the continuous influence of pipeline leakage;

[0206] Preset parameters: Road settlement safety threshold 3mm, third reference quantity ratio 5%, that is, if more than 1 abnormal data in 20 data sets, an early warning should be triggered.

[0207] Recognition process:

[0208] Second IoT connection extraction: Based on the facility association map of the digital twin model, automatically identify the dual association between the third facility (surrounding roads) and the first and second type facilities; functional dependence + soil support, clearly define the monitoring priority as roadway settlement;

[0209] The third IoT network data acquisition: real-time data from four settlement sensors are retrieved from the digital twin model and integrated into 20 consecutive sets of road settlement time series data;

[0210] The third abnormal data identification: Abnormal data exceeding the safety threshold of 3mm were filtered into 5 groups: [3.2, 3.5, 3.3, 3.1, 3.4].

[0211] The percentage of abnormal data is calculated as follows: 5 ÷ 20 × 100% = 25%, which is greater than the preset third reference percentage of 5%.

[0212] Judgment result: Triggered third-party IoT anomaly alert: Localized road subsidence at the intersection of a certain road and Xuefu Road is suspected to be caused by underground pipeline leakage leading to soil collapse, requiring emergency treatment.

[0213] Application Results: After the system issued an early warning, the operations and maintenance team used a digital twin model to simulate the settlement trend. The prediction showed that the maximum settlement within 24 hours could reach 5mm, exceeding the road structure's tolerance threshold. Traffic control measures were immediately implemented, including closing one lane, setting up warning signs, and simultaneously carrying out pipeline leak repair and soil reinforcement beneath the road, such as using grouting reinforcement technology. After 36 hours of treatment, the pipeline leak was completely sealed, the soil bearing capacity returned to normal levels, and the road settlement fell back to a safe range of 1.2-1.5mm. No road collapse or traffic accidents occurred. This case, relying on the associative replication capabilities of the digital twin model, expanded the monitoring scope from pipeline leaks to road settlement, avoiding the chain reaction of risks from "pipeline leak → soil subsidence → road collapse → impact on commuting to an office building," fully validating the practicality of the comprehensive prevention and control system.

[0214] This application also discloses a real-time analysis system for urban infrastructure safety monitoring, including a processor, wherein the processor executes the steps of the real-time analysis method for urban infrastructure safety monitoring as described in any of the above embodiments.

[0215] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A real-time analysis method for urban infrastructure safety monitoring, characterized in that, Includes the following steps: The system obtains the facility type of the infrastructure, matches the corresponding IoT sensors from the preset sensor database according to the facility type, obtains the distribution data corresponding to the infrastructure, deploys IoT sensors according to the distribution data, and obtains a digital twin model of the IoT sensor network based on multiple infrastructures. Based on the digital twin model, a first Internet of Things (IoT) network corresponding to a first facility type and a second IoT network corresponding to a second facility type are obtained, wherein the first facility type and the second facility type have a first IoT connection; First sensor data is obtained from the first Internet of Things network. First abnormal data is identified using a preset first identification strategy based on the first sensor data. The proportion of the first abnormal data in the first sensor data is calculated. If the proportion of the first abnormal data is greater than the preset first reference proportion, a first Internet of Things abnormality prompt is issued. Second sensor data is obtained from the second Internet of Things network. Second abnormal data is identified using a preset second identification strategy based on the second sensor data. The proportion of the second abnormal data in the second sensor data is calculated. If the proportion of the data is greater than the preset second reference proportion, a second Internet of Things abnormality prompt is issued. Based on the first and second IoT anomaly alerts, obtain other facility types corresponding to the first IoT connection and display early warnings for these other facility types. If the latest first abnormal data is output, then calculate the average value of all elements in the first sensing data; The difference ratio is calculated by using the ratio of the first difference value to the average value of all elements, and the proportion of the first reference quantity is adjusted according to the negative correlation of the difference ratio.

2. The real-time analysis method for urban infrastructure safety monitoring according to claim 1, characterized in that, The first Internet of Things (IoT) network includes a strain sensing network, and the first identification strategy includes the following steps: Calculate the change range of elements in the first sensing data, and identify elements whose change range is greater than the preset first reference range as abnormal elements and other elements as normal elements. Calculate the absolute value of the difference between the average data of abnormal elements and the average data of normal elements as the first difference value. If the first difference value is greater than the preset strain difference value, then the abnormal element is regarded as the first abnormal data.

3. The real-time analysis method for urban infrastructure safety monitoring according to claim 1, characterized in that, The first Internet of Things (IoT) network includes a vibration sensing network, and the first identification strategy includes the following steps: Calculate the rate of change of elements in the first sensing data. Element with a rate of change greater than a preset first reference rate is considered an abnormal element, and other elements are considered normal elements. The absolute value of the difference between the average data of abnormal elements and the average data of normal elements is the first difference value. If the first difference value is greater than the preset vibration difference value, then the abnormal element is regarded as the first abnormal data.

4. The real-time analysis method for urban infrastructure safety monitoring according to claim 2 or 3, characterized in that, The method also includes the following steps: Fit a real-time curve of all first sensor data in the first Internet of Things network; Compare the real-time curve graph with the preset curve graph to calculate the similarity of the first curve; If the similarity of the first curve is less than the preset threshold of the first curve, then the first IoT warning will be issued.

5. The real-time analysis method for urban infrastructure safety monitoring according to claim 1, characterized in that, The second IoT network includes a flow rate sensing network, and the second identification strategy includes the following steps: Calculate the change range of elements in the second sensing data, and identify elements whose change range is greater than the preset second reference range as abnormal elements and other elements as normal elements. Calculate the absolute value of the difference between the average data of abnormal elements and the average data of normal elements as the second difference value. If the second difference value is greater than the preset flow rate difference value, then the abnormal element will be regarded as the second abnormal data.

6. The real-time analysis method for urban infrastructure safety monitoring according to claim 1, characterized in that, The second IoT network includes a humidity sensing network, and the second identification strategy includes the following steps: Calculate the rate of change of elements in the second sensing data. Element with a rate of change greater than the preset second reference rate is considered an abnormal element, and other elements are considered normal elements. The absolute value of the difference between the average data of abnormal elements and the average data of normal elements is the second difference value. If the second difference value is greater than the preset humidity difference value, then the abnormal element will be regarded as the second abnormal data.

7. The real-time analysis method for urban infrastructure safety monitoring according to claim 5 or 6, characterized in that, The method also includes the following steps: Fit a real-time curve of all second sensor data in the second Internet of Things network; Compare the real-time curve graph with the preset curve graph to calculate the similarity of the second curve; If the similarity of the second curve is less than the preset threshold for the second curve, a second IoT early warning will be issued.

8. The real-time analysis method for urban infrastructure safety monitoring according to claim 1, characterized in that, Based on the early warning display, other facility types are obtained as the third facility type, and the second IoT connection that the third facility type has with both the first and second facility types is extracted; Based on the digital twin model, a third IoT network corresponding to the third facility type is obtained. Third sensor data is obtained from the third IoT network. Third abnormal data is identified using a preset third identification strategy based on the third sensor data. The proportion of the third abnormal data in the third sensor data is calculated. If the proportion of the data is greater than the preset proportion of the third reference amount, a third IoT abnormality prompt is issued.

9. A real-time analysis system for monitoring the safety of urban infrastructure, characterized in that, The system includes a processor in which the steps of the real-time analysis method for urban infrastructure safety monitoring as described in any one of claims 1-8 are performed.

Images