Urban bridge sidewalk multi-mode vibration monitoring and early warning system

Abstract

The invention relates to the technical field of bridge structure health monitoring, and discloses an urban bridge sidewalk multi-mode vibration monitoring and early warning system, which comprises a vibration sensing execution unit, an MEMS (Micro Electro Mechanical System) three-axis acceleration sensor with a built-in temperature compensation sheet, and a vibration detection unit, and is characterized in that the MEMS three-axis acceleration sensor is packaged in a 304 stainless steel base with a rubber damping pad; the self-adaptive data acquisition and transmission unit is connected with a LoRa wireless node of the vibration sensing execution unit; the edge analysis execution unit is used for receiving an embedded AI chip of data of the acquisition and transmission unit and executing vibration spectrum fingerprint matching and comfort level rating operation; and the early warning output execution unit is connected with an audible and visual alarm and a 4G communication module of the edge analysis unit. Through an embedded AI chip of the edge analysis unit, a 1 / 3 octave frequency spectrum is generated in real time and is matched with a fingerprint database, progressive early warning is realized through sound-light alarm and 4G communication, digital mapping of sidewalk vibration characteristics and pedestrian perception is established for the first time, and the defect of lack of a correlation model in the prior art is overcome.

Description

Technical Field

[0001] This invention relates to the field of bridge structural health monitoring technology, specifically to a multimodal vibration monitoring and early warning system for pedestrian walkways on urban bridges. Background Technology

[0002] Bridge structural health monitoring, as a core technology area in intelligent transportation and urban infrastructure management, has consistently focused its technological development on structural safety assessment and optimization of public pedestrian experience. Current mainstream monitoring systems primarily target the main load-bearing structures of bridges, such as main beams and piers, employing sensing devices like accelerometers and strain gauges to collect vibration and stress data via wired or wireless transmission networks, and then performing modal analysis and safety ratings based on cloud servers. While this type of technology has become a mature system for long-term performance monitoring of bridge main structures, significant technological gaps remain in the area of ​​monitoring localized vibrations in pedestrian walkways.

[0003] Existing bridge monitoring systems mainly focus on the safety monitoring of the main structure, such as the main beam, but seriously neglect the key area of ​​local vibration of the sidewalk. The following significant problems exist: a correlation model between the vibration characteristics of the sidewalk and pedestrian perception has not been established, and the correspondence between vibration parameters (such as acceleration and frequency) and pedestrian comfort cannot be clearly defined, which makes it impossible to scientifically assess the operating status of the bridge sidewalk from the perspective of pedestrian experience.

[0004] The inventors of this application have discovered that the prior art has at least the following technical problems: a lack of real-time monitoring of multimodal vibration of sidewalks and quantitative correlation analysis with pedestrian comfort. The technical causes include the lack of integration of vibration-flow coupling model in data processing, which makes it impossible to analyze the correlation between pedestrian behavior and vibration modes. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multimodal vibration monitoring and early warning system for pedestrian walkways on urban bridges. This system solves the problem in existing technologies where data processing does not integrate a vibration-flow coupling model, making it impossible to analyze the correlation between pedestrian behavior and vibration modes.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multimodal vibration monitoring and early warning system for urban bridge sidewalks, comprising:

[0007] Vibration sensing actuator: MEMS triaxial accelerometer with built-in temperature compensation plate, encapsulated in a 304 stainless steel base with rubber damping pad;

[0008] Adaptive data acquisition and transmission unit: The LoRa wireless node connected to the vibration sensing and actuation unit has a sampling circuit that automatically switches between 50Hz, 200Hz or 1000Hz sampling rates according to the peak acceleration.

[0009] Edge analysis execution unit: An embedded AI chip that receives data from the acquisition and transmission unit and performs vibration spectrum fingerprint matching and comfort rating calculations;

[0010] Early warning output execution unit: connected to the audible and visual alarm and 4G communication module of the edge analysis unit.

[0011] By adopting the above technical solution, a triaxial accelerometer with a built-in temperature compensation plate and encapsulated in a 304 stainless steel base with rubber damping pads is used as the vibration sensing execution unit. This resolves the contradiction between non-invasive sensor installation and vibration signal fidelity, effectively attenuating interference components in specific frequency bands and suppressing thermal drift effects to obtain accurate raw electrical signals. The LoRa wireless node sampling circuit of the adaptive data acquisition and transmission unit automatically switches between 50Hz, 200Hz, or 1000Hz sampling rates based on the acceleration peak, achieving accurate acquisition of low, medium, and high frequency multimodal vibration data and avoiding redundancy in low-amplitude vibration data and loss of details in high-amplitude vibration data. The embedded AI chip in the edge analysis execution unit... The FFT processor generates 1 / 3 octave band spectrum data in real time and inputs it into a pre-trained convolutional neural network classifier to perform vibration spectrum fingerprint matching and comfort rating calculations. This establishes a correlation model between sidewalk vibration characteristics and pedestrian perception, enabling quantitative analysis of the correspondence between vibration parameters and pedestrian comfort. The audible and visual alarm and 4G communication module of the early warning output execution unit can respond progressively based on the comfort rating index output by the edge analysis unit. This solves the problems in existing technologies, such as the lack of real-time monitoring of multimodal vibration of sidewalks and quantitative correlation analysis with pedestrian comfort, the lack of integration of vibration-flow coupling model in data processing, and the inability to analyze the correlation between pedestrian behavior and vibration modes.

[0012] Preferably, the bottom surface of the stainless steel base is provided with a wedge-shaped rubber damping pad and an annular guide groove.

[0013] Preferably, the switching logic of the sampling circuit is executed as follows: when the peak acceleration is <0.3g, a 50Hz clock signal is output; when 0.3g ≤ peak value < 0.7g, a 200Hz clock signal is output; and when the peak value is ≥0.7g, a 1000Hz clock signal is output.

[0014] Preferably, the edge analysis execution unit includes an FFT operator that generates 1 / 3 octave band spectrum data in real time and inputs it into a pre-trained convolutional neural network classifier.

[0015] Preferably, the output layer of the convolutional neural network classifier is connected to a four-position execution switch:

[0016] Class I switch: Turns on the green LED, indicating comfort mode;

[0017] Level II switch: Turning on the yellow LED causes mild discomfort;

[0018] Level II switch: Turns on the red LED and triggers a 90dB buzzer, causing significant discomfort;

[0019] Level IV switch: Activates the 4G module to send risk codes to the municipal platform, indicating structural risk.

[0020] Preferably, the LoRa wireless node is equipped with an SX1276 chipset, performs data packet transmission at a range of ≥2km in a line-of-sight environment, and the bit error rate verification circuit meets the EN300328V2.2.2 standard.

[0021] Preferably, the temperature compensation sheet is an NTC thermistor array mounted on the accelerometer PCB, and its compensation circuit has an output drift of ≤±0.02g in an environment of -25℃ to +65℃.

[0022] Preferably, it also includes a solar execution unit, which consists of a monocrystalline silicon solar panel connected to a lithium titanate battery and powered by a Buck-Boost voltage regulator circuit, maintaining system operation for ≥30 days in cloudy or rainy conditions.

[0023] Preferably, a sensor data acquisition unit is deployed every 30 meters on the bridge pedestrian walkway, and each unit is connected to the edge gateway at the bridge pier via a star network using LoRa.

[0024] Preferably, the edge gateway pushes data to the monitoring platform via a 4G module, and the data includes:

[0025] Real-time acceleration time-domain waveform, compression ratio ≥50%;

[0026] Spectral fingerprint matching similarity, with a 0-1 scale value;

[0027] Comfort level code, integer from 1 to 4.

[0028] Working Principle: First, the vibration sensing and actuation unit accurately captures the vibration signal of the sidewalk slab. This unit employs a high-precision MEMS triaxial accelerometer with a built-in NTC thermistor array to compensate for temperature changes. This sensor is encapsulated in a 304 stainless steel base with wedge-shaped rubber damping pads and annular guide grooves. This suppresses the zero-point drift that may occur in the sensor's output in environments ranging from -25 degrees Celsius to 65 degrees Celsius. This drift is controlled within a range of no more than ±0.02 grams, thus ensuring the high fidelity of the original vibration signal.

[0029] The acquired signals are then transmitted to an adaptive data acquisition unit. This unit consists of a LoRa wireless node configured with an SX1276 chipset. The peak detection logic within the circuit analyzes the acquired acceleration peak amplitude in real time: a low sampling rate of 50 Hz is used when the peak value is less than 0.3 grams to save power; a medium sampling rate of 200 Hz is used when the peak value is between 0.3 and 0.7 grams; and a high sampling rate of 1000 Hz is used when the peak value reaches or exceeds 0.7 grams to ensure sufficient capture of the high-frequency details of the impact load. This dynamic sampling strategy provides a wireless transmission range of at least 2 kilometers in line-of-sight environments, and its bit error rate verification strictly adheres to the EN300328V2.2.2 standard. This significantly reduces invalid data traffic by 40% to 80% while effectively improving the system's accuracy in identifying abnormal vibrations.

[0030] Data transmitted wirelessly over a long distance eventually reaches the edge analytics execution unit deployed at the edge. At the core of this unit is an embedded AI chip that uses a Fast Fourier Transform (FFT) operator to process the input data in real time, generating 1 / 3 octave band spectral feature vectors with center frequencies ranging from 0.8 Hz to 80 Hz. These feature vectors are then fed into a pre-trained convolutional neural network classifier for deep analysis, performing the core tasks of "vibration spectral fingerprint" matching and modal identification. This neural network model, based on the international standard ISO 2631-1, precisely quantifies the vibration level of the sidewalk into four asymptotic comfort / risk levels.

[0031] Finally, the analysis results are transmitted to the early warning output execution unit for dynamic response. This unit strictly follows the four-level discrimination results output by the convolutional neural network to perform progressive graded early warning: when identified as Level I, indicating a comfortable state, only the green indicator light is lit; when identified as Level II, mild discomfort, the yellow indicator light flashes as a warning; when the level reaches Level III, significant discomfort, a red indicator light and a buzzer with a sound pressure level of approximately 90 decibels are simultaneously triggered to issue an audible warning; once the system determines that the highest Level IV structural risk level has been reached, the 4G communication module will be immediately activated, and an encrypted data packet containing effectively compressed real-time acceleration time-domain waveform data, spectral fingerprint matching similarity values, and corresponding key information on the comfort level will be pushed to the municipal monitoring platform to realize remote emergency alarm.

[0032] This invention provides a multimodal vibration monitoring and early warning system for pedestrian walkways on urban bridges. It has the following beneficial effects:

[0033] 1. This invention integrates an FFT operator and a pre-trained CNN classifier into an embedded AI chip in the edge analysis unit, generates a 1 / 3 octave band spectrum in real time and matches it with a fingerprint database, outputs a four-level comfort rating (Level I comfort to Level IV structural risk) based on the ISO2631-1 standard, and achieves progressive early warning through audible and visual alarms and 4G communication. It is the first to establish a digital mapping between sidewalk vibration characteristics and pedestrian perception, solving the deficiency of existing technologies in lacking a correlation model.

[0034] 2. The adaptive data acquisition and transmission unit of this invention dynamically switches the sampling rate through peak detection. The hardware-level comparator and clock generator work together to avoid frequent switching, reducing data redundancy by 40%-80% while capturing high-frequency details of impact loads. Compared with the traditional fixed sampling scheme, it improves the accuracy of abnormal vibration identification by more than 30%.

[0035] 3. The LoRa wireless node of this invention is equipped with the SX1276 chipset, with a line-of-sight transmission range of ≥2km. The solar unit is powered by a monocrystalline silicon solar panel and a lithium titanate battery. The Buck-Boost circuit supports a battery life of ≥30 days in cloudy and rainy conditions. A single system has a coverage radius of 300 meters, making it suitable for mass application in urban bridge clusters. Attached Figure Description

[0036] Figure 1 This is a diagram illustrating the architecture of the multimodal vibration monitoring and early warning system for urban bridge sidewalks according to the present invention.

[0037] Figure 2 This is a flowchart of the edge analysis processing of the present invention;

[0038] Figure 3 This is a schematic diagram of the adaptive sampling switching process of the present invention;

[0039] Figure 4 This is a schematic diagram of the solar power supply system of the present invention. Detailed Implementation

[0040] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Please see the appendix Figure 1 - Appendix Figure 4 This invention provides a multimodal vibration monitoring and early warning system for pedestrian walkways on urban bridges, comprising:

[0042] Vibration sensing actuator: MEMS triaxial accelerometer with built-in temperature compensation plate, encapsulated in a 304 stainless steel base with rubber damping pad;

[0043] Adaptive data acquisition and transmission unit: The LoRa wireless node connected to the vibration sensing and actuation unit has a sampling circuit that automatically switches between 50Hz, 200Hz or 1000Hz sampling rates according to the peak acceleration.

[0044] Edge analysis execution unit: An embedded AI chip that receives data from the acquisition and transmission unit and performs vibration spectrum fingerprint matching and comfort rating calculations;

[0045] Early warning output execution unit: connected to the audible and visual alarm and 4G communication module of the edge analysis unit.

[0046] Specifically, the vibration energy of the bridge pedestrian walkway slab is attenuated by wedge-shaped rubber damping pads to reduce interference components in a specific frequency band (2-100Hz) before being transmitted to the vibration sensing and execution unit. The MEMS triaxial accelerometer within its 304 stainless steel base generates the original electrical signal under the condition of temperature compensation elements suppressing thermal drift. This electrical signal enters the signal conditioning circuit of the adaptive data acquisition and transmission unit for amplification and filtering. A peak comparator dynamically determines the acceleration amplitude range, triggering the sampling circuit to select a sampling rate of 50Hz, 200Hz, or 1000Hz according to preset rules for analog-to-digital conversion. The data packets are then encapsulated and transmitted via the LoRa physical layer protocol. After receiving the data packet, the embedded AI chip of the vibration analysis execution unit performs a Fourier transform, decomposes the generated vibration spectrum into feature vectors through a 1 / 3 octave bandpass filter bank, and inputs them into a pre-trained convolutional neural network model for spectral feature matching and modal recognition. Finally, it outputs a four-level pedestrian comfort rating index corresponding to the ISO2631-1 standard. The early warning output execution unit controls the sound and light alarm components to perform a progressive response according to the rating index: Level I triggers a solid green LED, Level II activates a flashing yellow LED, Level III activates a red LED and intermittent buzzer, and Level IV links the 4G communication module to upload the structural safety alarm data packet to the monitoring platform.

[0047] The stainless steel base has a wedge-shaped rubber damping pad and an annular flow guide groove on its bottom surface. The switching logic of the sampling circuit is executed as follows: when the acceleration peak value is <0.3g, a 50Hz clock signal is output; when 0.3g ≤ peak value <0.7g, a 200Hz clock signal is output; and when the peak value is ≥0.7g, a 1000Hz clock signal is output.

[0048] Specifically, the wedge-shaped rubber damping pad at the bottom of the 304 stainless steel base (preferably made of polyurethane material with a hardness of 60±5 Shore A) is installed against the sidewalk surface. Its wedge-shaped structure design can guide the vibration wave to transmit along a cone angle of 30°-45°, effectively attenuating low-frequency environmental disturbances below 10Hz and high-frequency mechanical noise above 100Hz. The annular guide groove (2-3mm wide and 1.5mm deep) coaxially opened on the bottom surface of the base can form a drainage channel in rain and snow conditions, preventing water accumulation from affecting sensor coupling. The conditioned vibration signal is input to the sampling circuit processing module, where the peak detection unit captures the absolute value of the triaxial acceleration composite vector. When the detected peak value is continuously below 0.3g, the control unit outputs a 50Hz clock signal to drive the analog-to-digital converter for low-power sampling. If the acceleration peak value is continuously in the range of 0.3g to 0.7g for more than 200ms, the sampling rate is automatically increased to 200Hz to capture mid-frequency vibration characteristics. This dynamic adjustment strategy is implemented through hardware collaboration between the comparator circuit and the programmable clock generator, avoiding oscillations caused by frequent switching of the sampling rate.

[0049] The edge analysis execution unit includes an FFT operator that generates 1 / 3 octave band spectrum data in real time and inputs it into a pre-trained convolutional neural network classifier; the output layer of the convolutional neural network classifier is connected to a four-level execution switch.

[0050] Class I switch: Turns on the green LED, indicating comfort mode;

[0051] Level II switch: Turning on the yellow LED causes mild discomfort;

[0052] Level II switch: Turns on the red LED and triggers a 90dB buzzer, causing significant discomfort;

[0053] Level IV switch: Activates the 4G module to send risk codes to the municipal platform, indicating structural risk.

[0054] Specifically, the FFT processor processes the input signal through a Hanning window to generate a time-frequency spectrum, and extracts the power spectral density vector through a 1 / 3 octave bandpass filter bank (preferably with a center frequency in the range of 0.8-80Hz). This vector is input into a pre-trained convolutional neural network classifier, which contains 3 convolutional kernels and 2 pooling layers, and outputs the probability distribution of four comfort states through a Softmax function. When the output layer probability threshold exceeds the set confidence level, the corresponding level switch is triggered: Level I switch responds to a comfort state command with a peak probability ≥ 0.85, and then... The green LED provides a continuous low-power indication; the Level II switch activates the yellow LED to perform 1Hz intermittent flashing based on mild discomfort signals in the (0.6-0.85) range; the Level III switch triggers a synchronized warning (sound pressure level 85-95dB@vertical distance 1m) by linking a red LED and a buzzer for significant discomfort in the (0.4-0.6) range; and the Level IV switch automatically activates the 4G communication module for structural risk states with a probability value <0.4, sending encrypted risk parameter data packets (including spectral characteristic values ​​and spatial positioning information) to the municipal monitoring platform.

[0055] The LoRa wireless node is equipped with an SX1276 chipset, enabling data packet transmission over a distance of ≥2km in line-of-sight environments. The bit error rate verification circuit meets the EN300328V2.2.2 standard. The temperature compensation chip is an NTC thermistor array mounted on the accelerometer PCB, and its compensation circuit has an output drift of ≤±0.02g in an environment ranging from -25℃ to +65℃. It also includes a solar-powered unit, which consists of a monocrystalline silicon solar panel connected to a lithium titanate battery and powered by a Buck-Boost voltage regulator circuit, maintaining system operation for ≥30 days in cloudy or rainy conditions.

[0056] Specifically, LoRa wireless nodes equipped with the SX1276 chipset establish communication links in line-of-sight transmission environments (with an effective transmission distance of no less than 2km in typical application scenarios). Their physical layer data packets are processed by a CRC-16 check circuit to comply with the frame error rate requirements of the EN300328V2.2.2 standard. The NTC thermistor array mounted on the MEMS accelerometer PCB board monitors the temperature gradient in real time. Through a negative feedback compensation circuit, the sensor zero-point drift is suppressed in the ambient temperature range of -25℃ to +65℃, ensuring that the deviation of the output acceleration measurement value is controlled within ±0.02g. The monocrystalline silicon solar panel charges the lithium titanate battery pack through a maximum power point tracking circuit. The 3.3V system voltage converted by the Buck-Boost voltage regulation topology sustains the device for no less than 30 days under continuous cloudy and rainy conditions (standard irradiance ≤30kLux).

[0057] A sensor data acquisition unit is deployed every 30 meters along the bridge's pedestrian walkway. Each unit is connected to the edge gateway at the bridge pier via a LoRa star network. The edge gateway pushes data to the monitoring platform via a 4G module. The data includes:

[0058] Real-time acceleration time-domain waveform, compression ratio ≥50%;

[0059] Spectral fingerprint matching similarity, with a 0-1 scale value;

[0060] Comfort level code, integer from 1 to 4.

[0061] Specifically, sensor data acquisition units are distributed along the sidewalk at intervals of approximately 30 meters. Each unit constructs a star-shaped topology network with the edge gateway at the bridge pier via a LoRa radio frequency module equipped with an SX1276 chip. The edge gateway pushes pre-processed data packets to the monitoring platform via a 4G communication module. These packets include: an acceleration time-domain waveform processed by a lossy compression algorithm (typical compression ratio not less than 50%); a similarity coefficient (scale value between 0 and 1) matched based on a pre-established spectrum template library; and a comfort level code (a total of 4 levels of discrete integer values) generated by comparing vibration intensity with a preset threshold.

[0062] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-modal vibration monitoring and early warning system for urban bridge sidewalks, characterized in that, Comprise: Vibration sensing execution unit: MEMS three-axis acceleration sensor with built-in temperature compensation sheet, packaged in 304 stainless steel base with rubber damping pad; Adaptive data acquisition unit: LoRa wireless node connected to vibration sensing execution unit, its sampling circuit automatically switches 50Hz, 200Hz or 1000Hz sampling rate according to acceleration peak value; Edge analysis execution unit: embedded AI chip receiving data from data acquisition unit, performing vibration frequency spectrum fingerprint matching and comfort level rating operation; Early warning output execution unit: sound and light alarm and 4G communication module connected to edge analysis unit.

2. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, The bottom surface of the stainless steel base is provided with a wedge-shaped rubber damping pad and an annular flow guide groove.

3. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, The switching logic of the sampling circuit performs: when the acceleration peak value is <0.3g, output 50Hz clock signal, 0.3g≤peak value<0.7g, output 200Hz clock signal, peak value≥0.7g, output 1000Hz clock signal.

4. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, The edge analysis execution unit contains an FFT operator that generates 1 / 3 octave spectrum data in real time and inputs a pre-trained convolutional neural network classifier.

5. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 4, wherein, The output layer of the convolutional neural network classifier is connected to four-grade execution switches: I-level switch: turn on green LED, comfortable state; Ⅱ-level switch: turn on yellow LED, mild discomfort; Ⅱ-level switch: turn on red LED and trigger 90dB buzzer, significant discomfort; Ⅳ-level switch: activate 4G module to send risk code to municipal platform, structural risk.

6. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, The LoRa wireless node is configured with SX1276 chip set, which performs ≥2km data packet transmission in line-of-sight environment, and the error code rate checking circuit meets the EN300328V2.2.2 standard.

7. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 6, wherein, The temperature compensation sheet is an NTC thermistor array attached to the acceleration sensor PCB, and its compensation circuit outputs a drift of ≤±0.02g in an environment of -25℃ to +65℃.

8. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, It also includes a solar execution unit connected to a lithium titanate battery by a single-crystal silicon solar panel, powered by a Buck-Boost voltage stabilizing circuit, which maintains system operation for ≥30 days in rainy mode.

9. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, Deploy one sensing and acquisition unit every 30 meters on the bridge sidewalk, and each unit is connected to the edge gateway at the pier through LoRa star-shaped networking.

10. The urban bridge sidewalk multi-modal vibration monitoring and warning system of claim 1, wherein, The edge gateway performs data pushing to the supervision platform through the 4G module, and the data includes: Real-time acceleration time domain waveform, compression rate ≥50%; Spectrum fingerprint matching similarity, 0-1 scale value; Comfort level code, 1-4 integer.

Images