A method and system for intelligent control of a catenary warning light

Abstract

The application relates to the technical field of intelligent control of warning lights, and discloses a cable-stayed line warning light intelligent control method and system; the method comprises the following steps: collecting light brightness and three-axis vibration signals to obtain a preliminary signal sequence through noise correction; if the light brightness mutation trend or vibration track stability exceeds a threshold value, an enhanced sampling mode is activated; in the mode, high-frequency light brightness sampling and vibration short-time spectrum analysis are performed to obtain light brightness fluctuation time domain sequences and vibration frequency domain characteristics, time sequence alignment and feature fusion are performed, and an enhanced light-vibration combined feature sequence is formed; the current environment abnormality score is calculated, if the score exceeds a preset threshold value, a clustering center is extracted and a dynamic adjustment parameter set is generated, the acquisition frequency is updated in combination with environment feedback, light-vibration combined triggering judgment rules are constructed; the current sequence is compared with historical benchmarks, if the rules are met, adjustment instructions are generated, and the brightness and flickering mode of the warning light are dynamically adjusted. The application improves the environmental adaptability and early warning accuracy of the warning light.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology for warning lights, and in particular to an intelligent control method and system for a drawbar warning light. Background Technology

[0002] As a crucial supporting component for infrastructure such as power, communication, and transportation, guy wires are widely distributed in complex outdoor environments, and their safety warnings directly impact the safety of pedestrians, vehicles, and the facilities themselves. Traditional guy wire warning lights often employ fixed-brightness flashing or single-sensor triggering modes, which have significant technical drawbacks: Firstly, fixed operating modes continuously consume power, resulting in poor battery life, especially in remote areas where equipment maintenance costs are high; secondly, single sensors are susceptible to environmental interference, such as vibrations caused by gusts of wind or changes in natural light and shadow, which can easily lead to false triggering, while in actual collision or close-range risk scenarios, insufficient sampling frequency may cause response delays.

[0003] In existing technologies, while some warning lights attempt to use multiple sensors to collect signals, they lack effective cross-modal signal fusion and adaptive noise processing mechanisms, resulting in poor signal interference filtering. Furthermore, the sampling frequency is often fixed and cannot be dynamically adjusted according to environmental conditions, making it difficult to balance the requirements of low power consumption and high sensitivity. In addition, the trigger judgment rules are simplistic and do not incorporate temporal correlation and anomaly degree quantitative analysis of light intensity and vibration signals, leading to insufficient warning accuracy and an inability to accurately match complex and changing outdoor environments, severely impacting the practical application effect of maneuverable warning lights.

[0004] Therefore, there is an urgent need for an intelligent control scheme with adaptive signal processing, dynamic parameter adjustment and precise trigger control capabilities to solve the problems of high power consumption, high false trigger rate and slow response of traditional technologies. Summary of the Invention

[0005] To address the aforementioned technical issues, this application provides an intelligent control method and system for guy wire warning lights, used to identify risks such as approaching vehicles and collisions with foreign objects around outdoor guy wires, adapting to complex outdoor environments, and intelligently adjusting the warning light mode through optical-vibration collaborative sensing and adaptive control, balancing power consumption and warning effect, and improving facility safety.

[0006] In a first aspect, this application provides an intelligent control method for a guy wire warning light, the method comprising:

[0007] Step S1: Acquire light intensity data and triaxial vibration signals through a cable-stayed sensor array, and perform adaptive noise correction to obtain a preliminary signal sequence;

[0008] Step S2: Extract spatiotemporal features from the preliminary signal sequence. If the brightness change trend feature or vibration trajectory stability feature exceeds the corresponding threshold, activate the enhanced sampling mode.

[0009] Step S3: In enhanced sampling mode, perform high-frequency light intensity sampling and short-time spectrum analysis of vibration signal to obtain the light intensity fluctuation time domain sequence and vibration frequency domain feature vector, and perform cross-modal time sequence alignment and feature fusion to form a joint light-vibration feature sequence;

[0010] Step S4: Calculate the anomaly score of the current environmental state based on the optical-vibration joint feature sequence. If the score exceeds a preset threshold, extract the feature cluster centers in the optical-vibration joint feature sequence and generate a dynamic adjustment parameter set.

[0011] Step S5: Based on the dynamically adjusted parameter set and environmental feedback, update the signal acquisition frequency, determine the final acquisition frequency, and construct a joint triggering judgment rule for optical vibration with multiple constraints.

[0012] Step S6: Obtain the current optical-vibration joint feature sequence in real time and compare it with historical benchmark data. If the optical-vibration joint trigger judgment rule is met, generate a warning light adjustment command to adjust the brightness and flashing mode of the inclined wire warning light.

[0013] Secondly, this application provides an intelligent control system for a guy wire warning light, the system comprising:

[0014] The acquisition module is used to acquire brightness data and triaxial vibration signals through the cable-stayed sensor array, and perform adaptive noise correction to obtain a preliminary signal sequence.

[0015] The extraction module is used to extract spatiotemporal features from the preliminary signal sequence. If the brightness change trend feature or the vibration trajectory stability feature exceeds the corresponding threshold, the enhanced sampling mode is activated.

[0016] The fusion module is used to perform high-frequency light intensity sampling and short-time spectrum analysis of vibration signals in enhanced sampling mode, obtain the time-domain sequence of light intensity fluctuations and the frequency-domain feature vector of vibration, and perform cross-modal time-series alignment and feature fusion to form a joint optical-vibration feature sequence.

[0017] The analysis module is used to calculate the anomaly score of the current environmental state based on the optical-vibration joint feature sequence. If the score exceeds a preset threshold, the feature cluster centers in the optical-vibration joint feature sequence are extracted to generate a dynamically adjusted parameter set.

[0018] The adjustment module is used to update the signal acquisition frequency based on the dynamic adjustment parameter set and environmental feedback, determine the final acquisition frequency, and construct a joint triggering judgment rule for optical vibration with multiple condition constraints.

[0019] The control module is used to acquire the current optical-vibration joint feature sequence in real time, compare it with historical benchmark data, and if the optical-vibration joint trigger judgment rule is met, generate a warning light adjustment command to adjust the brightness and flashing mode of the inclined wire warning light.

[0020] Compared with the prior art, the beneficial effects of this application are at least as follows:

[0021] This application provides an intelligent control method and system for a guy wire warning light. It synchronously collects light intensity and triaxial vibration data through a sensor array, and dynamically suppresses interference through an adaptive noise baseline model to ensure the reliability of the original data. By combining spatiotemporal feature extraction and enhanced sampling mode activation mechanism, it achieves rapid identification and accurate capture of potential risks. With the help of short-time Fourier transform and cross-modal feature fusion, it constructs a joint feature sequence of light and vibration that includes time domain, frequency domain information and enhanced light and vibration, providing comprehensive data support for risk assessment.

[0022] Furthermore, by calculating anomaly scores through dynamic weight allocation and combining cluster analysis and nonlinear mapping to generate dynamically adjustable parameter sets adapted to different risk scenarios, adaptive optimization of the acquisition frequency is achieved. Based on a multi-condition constrained optical-vibration joint triggering rule and confidence scoring mechanism, the accuracy of risk assessment and the targeted nature of warning adjustments are ensured. This application dynamically adapts to complex outdoor environments and diverse risks, effectively balancing equipment power consumption and warning effects while ensuring monitoring sensitivity and real-time performance. It significantly improves the practicality, adaptability, and reliability of guy wire risk monitoring and warning, avoiding the drawbacks of traditional warning lights that are fixed in mode and prone to accidental triggering. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of an intelligent control method for a guy wire warning light in an embodiment of this application;

[0025] Figure 2 This is a schematic diagram of the sensor deployment for the guy wire warning light system according to an embodiment of this application;

[0026] Figure 3 This is a schematic diagram of the three types of cluster centers and their corresponding risk regions in an embodiment of this application;

[0027] Figure 4 This is a schematic diagram of the structure of an intelligent control system for a guy wire warning light according to an embodiment of this application. Detailed Implementation

[0028] This application provides an intelligent control method and system for a maneuvering warning light. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0029] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent control method for a guy wire warning light in this application includes:

[0030] Step S1: Obtain light intensity data and triaxial vibration signals through the inclined wire sensor array, and perform adaptive noise correction to obtain a preliminary signal sequence.

[0031] Step S1 includes: synchronously acquiring ambient light intensity data and triaxial vibration signals through a sensor array to form a time-stamped original signal sequence; establishing adaptive noise baseline models for the ambient light intensity data and triaxial vibration signals respectively; dynamically estimating the current ambient noise level based on the historical statistical characteristics of the signals; and performing real-time noise suppression and baseline drift compensation on the original signal sequence to generate a preliminary signal sequence after noise suppression.

[0032] Specifically, guy wires are important supporting components for outdoor infrastructure such as power lines, communication towers, and streetlights. Made of high-strength steel wire or composite materials, one end is fixed to the main structure of the facility, and the other end is diagonally fixed to an anchor on the ground. They are used to balance lateral forces and prevent the facility from tilting or collapsing. Their deployment in outdoor environments is characterized by variable lighting, complex interference, strict energy consumption constraints, harsh operating conditions, and dispersed distribution. For example... Figure 2The diagram shows the sensor deployment of the guy wire warning light system. The sensor array, adapted to the requirements of this scenario, includes a photosensor and a triaxial accelerometer. It synchronously collects ambient light intensity data and triaxial vibration data of the outdoor environment where the guy wire is located at a low power interval of once per second. The light intensity data records the ambient light intensity in lux, such as strong light during the day, weak light at night, and light and shadow fluctuations. The triaxial vibration data records the vibration amplitude data of the guy wire in the X, Y, and Z spatial dimensions in the unit of gravitational acceleration. Through synchronous triaxial acquisition, it can comprehensively capture vibration events of different directions and types, including low-amplitude vibrations caused by gusts and slight environmental disturbances, as well as high-amplitude and severe vibrations in various directions caused by vehicle scrapes, foreign object impacts, and strong winds. During the acquisition process, a corresponding timestamp is added to each set of data to form a raw signal sequence containing timestamps, light intensity values, and triaxial vibration acceleration values. The adaptive noise baseline model is an adaptive model that dynamically adjusts noise estimation based on the historical statistical characteristics of the signal. Its core function is to update the noise baseline in real time according to the signal change patterns under different outdoor environments in order to accurately suppress interference. The brightness data and triaxial vibration history data within the most recent preset time period from the original signal sequence are input into the adaptive noise baseline model. The implementation of the adaptive noise baseline model includes: for brightness data, the moving average and standard deviation of the data in the queue are calculated in real time, and the range of the moving average ± k times the standard deviation is used as the current noise baseline; for vibration data, the statistical peak value of the data in the queue in a static state, such as when identified by low-pass filtering, is calculated in real time, and this peak value is used as the vibration noise baseline. The coefficient k is dynamically adjusted according to the environmental stability. When the data fluctuates drastically over several consecutive periods, the value of k is increased to widen the baseline, and vice versa to tighten the baseline. For the characteristics of different outdoor scenarios, the two noise baselines are optimized by weighted averaging to generate a comprehensive noise baseline. For example, in outdoor high wind noise scenarios, in order to suppress the impact of high-frequency vibration interference caused by wind on signal judgment, the contribution of vibration data to the comprehensive noise baseline should be reduced. Therefore, the weight of the brightness baseline is set to 0.7 and the weight of the vibration baseline is set to 0.3. By weighted fusion, the vibration noise component is weakened to ensure the accuracy of the signal in low light environment. The integrated noise baseline is then applied to the corresponding sampling points in the original signal sequence. Baseline drift compensation is performed on each sampling point through subtraction, and finally, a preliminary signal sequence after noise suppression is generated.

[0033] By acquiring low-power interval data to adapt to the energy consumption constraints of outdoor environments, the device's battery life is extended. By using an adaptive noise baseline model to dynamically adapt to noise changes in complex outdoor environments, irrelevant interference such as gusts and light fluctuations is effectively filtered out, improving the accuracy and reliability of signal data and providing high-quality data support for subsequent spatiotemporal feature extraction and enhanced sampling mode judgment.

[0034] Step S2: Extract spatiotemporal features from the preliminary signal sequence. If the brightness change trend or vibration trajectory stability features exceed the corresponding threshold, activate the enhanced sampling mode.

[0035] Step S2 includes: calculating the abrupt change slope within a sliding time window for the ambient light brightness data in the preliminary signal sequence, as the abrupt change trend feature of the brightness in the time domain; calculating the sequence of changes in the direction angle of the acceleration vector between consecutive sampling points for the triaxial vibration signal in the preliminary signal sequence, and statistically analyzing the proportion of changes in the direction angle less than a preset angle threshold, as the spatial vector vibration trajectory stability feature of the vibration signal.

[0036] Specifically, after obtaining the preliminary signal sequence after noise suppression, in order to accurately identify potential risk scenarios around the cable-stayed bridge, such as vehicle approach, object collision, or continuous contact, it is necessary to extract key features reflecting environmental changes from the preliminary signal sequence. By comparing the features with preset thresholds, it is determined whether there are signs of risk, providing an accurate basis for the subsequent activation of enhanced sampling modes. Specifically, for the brightness signal in the preliminary signal sequence, a sliding time window of fixed duration, such as 2 seconds, is used to extract the abrupt change slope as the abrupt change trend feature of brightness in the time domain. The abrupt change slope is the rate of change of the brightness signal per unit time, used to quantify the severity of sudden changes in illumination. In the calculation, the brightness difference between adjacent sampling points within the window is first taken and divided by the 1-second sampling interval to obtain a single slope. Then, the maximum absolute value of all slopes is taken as the brightness abrupt change slope corresponding to that window. The larger this feature value is, the more drastic the change in illumination has occurred in the corresponding time period, such as caused by direct or strong reflection of vehicle headlights; conversely, the smaller the feature value is, the more gradual the change in illumination is, belonging to natural environmental changes or no obvious external interference. Meanwhile, for the triaxial vibration signal in the preliminary signal sequence, the proportion of the direction angle change sequence of the acceleration vector between continuous sampling points is calculated, and the proportion of the direction angle change less than a preset angle threshold is statistically used as the spatial vector trajectory stability feature of the vibration signal. The acceleration vector of the triaxial vibration signal refers to the vector A(x, y, z) composed of the vibration accelerations in the three orthogonal directions X, Y, and Z corresponding to each sampling point. Before calculation, each acceleration vector is normalized to eliminate the interference of vibration amplitude on direction judgment. Then, the cosine of the direction angle between the acceleration vectors A1 and A2 of two consecutive sampling points is calculated by vector dot product, and then the direction angle is obtained. Assuming the preset angle threshold is 15 degrees, the proportion of sampling points with a direction angle less than 15 degrees within the sliding time window is counted to the total number of sampling points in the window. This proportion is the spatial vector trajectory stability characteristic. The larger the characteristic value, the smoother the change of vibration direction and the more consistent the trajectory during the observation period. This usually corresponds to the cable being subjected to continuous and directional external forces such as continuous wind load or regular impact. Conversely, the smaller the characteristic value, the more chaotic the change of vibration direction and the lack of a stable trajectory, which is mostly caused by random disturbances such as gusts or instantaneous impacts.

[0037] Assuming a preset threshold of 100 lux / s for the brightness abrupt change slope and a preset threshold of 0.8 for the vibration spatial vector trajectory stability feature, the extracted brightness abrupt change slope and spatial vector trajectory stability feature are compared with their corresponding preset thresholds. If the brightness abrupt change slope exceeds 100 lux / s, it indicates a brief and drastic change in light intensity in the environment. This type of change is usually associated with direct vehicle headlights, strong reflections, or sudden artificial light source interference, and is a light signal anomaly event requiring close attention. If the vibration spatial vector trajectory stability feature exceeds 0.8, it indicates that the vibration of the cable-stayed cable exhibits a high degree of directional consistency during the observation period. Vibration-like patterns are usually not caused by random environmental disturbances, but are more likely to originate from external forces with clear directions, such as continuous wind loads, regular contact, or scraping by external objects. As long as either the slope of the sudden change in light intensity or the stability characteristics of the vibration spatial vector trajectory exceed its corresponding preset threshold, it is determined that there are obvious signs of abnormality in the current environment, which meets the conditions for activating the optical-vibration co-enhanced sampling mode. The optical-vibration co-enhanced sampling mode is a high-frequency sampling mode that simultaneously improves the acquisition accuracy of light intensity signals and vibration signals. Its core is to accurately capture subtle signal changes in risky scenarios by co-optimizing the sampling parameters of the two signals, which is different from the low-frequency acquisition in the low-power mode.

[0038] By clarifying the extraction logic of two types of spatiotemporal features and the core connotation of the optical-vibration collaborative enhancement sampling mode, potential risk signs around the cable-stayed line are accurately identified, which not only ensures timely response to risk scenarios, but also improves the ability to capture signal details through high-frequency collaborative sampling.

[0039] Step S3: In enhanced sampling mode, perform high-frequency light intensity sampling and short-time spectrum analysis of vibration signal to obtain the light intensity fluctuation time domain sequence and vibration frequency domain feature vector, and perform cross-modal time sequence alignment and feature fusion to form a light-vibration joint feature sequence.

[0040] Step S3 includes: in the enhanced sampling mode, increasing the ambient light brightness sampling frequency to a preset high frequency value, dividing a sampling window of fixed duration based on the adjusted sampling frequency, obtaining the light brightness fluctuation time-domain sequence within each window, performing a short-time Fourier transform on the synchronously acquired vibration signal, extracting the energy distribution characteristics of the vibration signal in each frequency band, forming a vibration frequency domain feature vector, interpolating and aligning the light brightness fluctuation time-domain sequence and the vibration frequency domain feature vector on the time axis, concatenating the aligned light brightness time-domain feature vector and the vibration frequency domain feature vector into a joint feature vector, and enhancing the joint feature vector to form an enhanced light-vibration joint feature sequence.

[0041] Specifically, after activating the optical-vibration co-enhanced sampling mode, in order to accurately capture the subtle correlation changes between optical and vibration signals in risky scenarios, it is necessary to extract the core features of the two types of signals through high-frequency sampling and spectrum analysis. After time-series alignment and fusion, a unified feature sequence is formed, providing comprehensive and synchronous feature data support for subsequent environmental state anomaly scoring. Specifically, in the optical-vibration co-enhanced sampling mode, the ambient light brightness sampling frequency is first increased from the low-power mode to a preset high-frequency value, such as 10Hz. Based on this high-frequency sampling rate, the continuous time stream is further divided into sampling windows of fixed duration, such as 2 seconds, with each window containing 20 consecutive light brightness sampling points. Within each 2-second sampling window, luminance data is continuously collected using a photosensor, and first-order difference calculations are performed on these data to calculate the luminance difference between adjacent sampling points. This yields a luminance fluctuation time-domain sequence reflecting the short-term trend of illumination changes. Key features, such as the maximum fluctuation amplitude and the average fluctuation, are extracted from the luminance fluctuation time-domain sequence. With time as the horizontal axis and luminance fluctuation value as the vertical axis, this luminance fluctuation time-domain sequence clearly and intuitively presents the dynamic process of illumination increasing, decreasing, or fluctuating in a short period of time, providing high-resolution temporal change information for subsequent feature fusion and anomaly assessment.

[0042] Simultaneously, a short-time Fourier transform is performed on the triaxial vibration signals synchronously acquired by the sensor array. The short-time Fourier transform is an analytical method that converts time-domain signals into time-frequency domain signals. Specifically, after segmenting and windowing the vibration signal, the Fourier transform is used to convert each segment of the time-domain vibration signal into a frequency-domain signal. The signal energy value in each frequency band is calculated. The energy is calculated by squared the amplitude of the frequency domain signal. The maximum value, mean, and variance of the energy in each frequency band are extracted as core features to form a vibration frequency domain feature vector containing multi-dimensional frequency information. This vector can effectively distinguish between low-energy random vibrations such as wind vibrations and risky vibrations such as impacts. For example, low-energy random vibrations are concentrated in low-frequency bands with small variances, while specific high-frequency bands have obvious energy peaks with large variances. Because the analysis windows of high-frequency light intensity sampling and short-time Fourier transform of vibration signals may not be perfectly aligned in time, resulting in a mismatch in sampling time points, it is necessary to interpolate and align the feature sequences of the two types of signals on the time axis. This embodiment adopts a linear interpolation method, that is, based on the signal values ​​of adjacent known time points, the feature values ​​of missing time points are supplemented by linear fitting to ensure that each timestamp corresponds to both light intensity fluctuation features and vibration frequency domain features. After alignment, the key features in the light intensity fluctuation time domain sequence are concatenated with the features of each dimension in the vibration frequency domain feature vector to form an initial joint feature vector. Subsequently, this initial joint feature vector is enhanced and normalized. The enhancement process will be explained later. Through normalization, all feature values ​​are mapped to a unified interval of 0-1, eliminating the scale differences of different feature dimensions. Finally, the final joint feature vectors are arranged sequentially in chronological order to form a continuous light-vibration joint feature sequence. This sequence integrates high temporal resolution light intensity variation information and frequency-discriminative vibration energy distribution information, providing a structured multimodal data foundation for subsequent unified environmental state assessment and event confidence calculation.

[0043] High-frequency sampling enhances the ability to capture details of optical and vibration signals. Short-time Fourier transform is used to mine the frequency domain difference features of vibration signals. After time-series alignment and normalization fusion, the optical-vibration joint feature sequence contains time-domain, frequency-domain, and correlation information of both types of signals. This effectively solves the problems of incomplete feature extraction and signal asynchrony in traditional single-signal extraction, laying a high-quality feature foundation for subsequent accurate calculation of environmental state anomaly scores.

[0044] The enhancement process for the joint feature vector includes: establishing a physical coupling model between wind-induced vibration and light and shadow fluctuations; the physical coupling model predicts the corresponding theoretical light and shadow fluctuation curve based on the extracted vibration signal frequency domain feature vector; the theoretical light and shadow fluctuation curve is compared with the real-time acquired light brightness fluctuation time domain sequence to obtain a consistency score; the consistency score is used as a new feature dimension and incorporated into the light and vibration joint feature sequence to form an enhanced light and vibration joint feature sequence.

[0045] Specifically, the implementation process for enhancing the joint feature vector is as follows: First, a physical coupling model between wind-induced vibration and light and shadow fluctuations is established. This model is an analytical model constructed by combining the mechanical characteristics of the cable-stayed cable structure with the laws of environmental optical propagation. Its core function is to derive the corresponding theoretical laws of light and shadow changes through vibration signal characteristics. The model incorporates inherent and environmental parameters such as the cable-stayed cable's structural damping, windward area, and incident angle of illumination, clarifying the quantitative relationship between vibration displacement and changes in the light and shadow projection area, thus forming a complete feature mapping logic. In the model application stage, the extracted frequency domain feature vector of the vibration signal, containing vibration frequency, amplitude, and phase, is input into the physical coupling model. The real-time spatial attitude of the cable-stayed cable under the corresponding vibration state is first calculated using the mechanical transmission formula, which is: ,in, Let be the radial vibration displacement of the cable-stayed cable at time t. The structural damping coefficient of the stay cable is pre-calibrated based on the material and diameter of the stay cable, for example, when the steel strand material and diameter are 12mm. Let 0.02 be the value, and t be the time. For amplitude, The vibration frequency, Using the phase as a formula, the displacement of the cable-stayed line from its initial stationary position at different times can be accurately calculated, thereby determining its real-time spatial attitude.

[0046] Once the real-time spatial attitude is obtained, i.e. the vibration displacement at each moment, is obtained... Then, combining the geometric formula of light and shadow projection, the law of light and shadow fluctuation caused by attitude change was derived. The geometric formula of light and shadow projection used is: ,in, Let be the theoretical light and shadow brightness value at time t. The baseline light and shadow brightness for a stationary cable-stayed bridge was obtained through prior field measurements in vibration-free scenarios, such as midday on a sunny day. Take 8000 lux, is the light and shadow attenuation coefficient, which is related to the refractive index of the surrounding medium. In air, k is taken as 0.01. The incident angle of light is measured in real time by a light sensor, ranging from 0° to 90°. D is the vertical distance between the guy wire and the light-receiving surface, such as the ground or surrounding components, in meters (m). This distance is determined in advance based on the on-site installation scenario. This formula is derived from vibration displacement. With light and shadow brightness The quantitative correlation can be used to derive the theoretical light and shadow fluctuation curve that changes over time. This curve represents the ideal light and shadow change time sequence characteristics corresponding to the vibration of the cable-stayed bridge when there is no external interference.

[0047] Subsequently, a similarity check was performed, comparing the theoretical light and shadow fluctuation curve with the real-time acquired light intensity fluctuation time-domain sequence. The normalized cross-correlation coefficient algorithm was used to calculate the similarity between the two, with the coefficient ranging from -1 to 1. This correlation coefficient was then linearly mapped to a consistency score in the range of 0 to 1, the mapping logic being the correlation coefficient plus 1 and then divided by 2. Specifically, the closer the score is to 1, the closer the real-time light intensity fluctuation is to the ideal light and shadow change caused by vibration, and the smaller the external interference; the closer the score is to 0, the more severe the environmental interference in the real-time light intensity fluctuation, and the weaker the correlation with vibration.

[0048] Finally, feature enhancement is completed. The consistency score S mentioned above is used as a new feature dimension and directly incorporated into the original optical-vibration joint feature sequence. Combined with the time-domain features of brightness and the frequency-domain features of vibration, an enhanced optical-vibration joint feature sequence is formed. This enhanced sequence not only includes the original brightness and vibration features, but also adds optical-vibration coupling consistency features, which can effectively distinguish whether the signal anomaly is caused by the vibration of the cable or by environmental interference, significantly improving the accuracy of subsequent risk identification.

[0049] Step S4: Calculate the anomaly score of the current environmental state based on the optical-vibration joint feature sequence. If the score exceeds the preset threshold, extract the feature cluster centers in the optical-vibration joint feature sequence and generate a dynamically adjusted parameter set.

[0050] Step S4 includes: calculating the abnormal deviation of the brightness feature dimension and vibration feature dimension in the optical-vibration joint feature sequence; assigning dynamic weights to each feature dimension based on the statistical characteristics of historical signals; the feature dimensions include the brightness feature dimension and the vibration feature dimension; weighting and summing the abnormal deviation of each feature dimension with its corresponding weight to obtain the abnormal score of the current environmental state; when the abnormal score exceeds a preset threshold, performing cluster analysis on the optical-vibration joint feature sequence, extracting the feature cluster centers of each category; calculating the frequency adaptive adjustment parameter and the brightness response gain parameter respectively through a nonlinear mapping function based on the brightness and vibration distribution patterns reflected by the feature cluster centers; and combining the frequency adaptive adjustment parameter and the brightness response gain parameter into a dynamic adjustment parameter set.

[0051] Specifically, after obtaining the enhanced optical-vibration joint feature sequence, to quantify the risk level of the environment surrounding the cable-stayed bridge, it is necessary to analyze the abnormal correlation between brightness, vibration characteristics, and optical-vibration coupling consistency characteristics based on this sequence. Anomaly scores are calculated through dynamic weighted summation, and then feature cluster centers are extracted based on the score results to generate dynamic parameters, providing a basis for subsequent sampling frequency optimization. Specifically, the calculation of the anomaly score first separates the brightness-related feature dimension, vibration-related feature dimension, and consistency score dimension from the enhanced optical-vibration joint feature sequence. For the brightness feature dimension, its anomaly deviation is calculated. This deviation reflects the degree of difference between the current brightness feature value and the statistical benchmark under historical normal conditions. The specific calculation method is as follows: take the absolute value of the difference between the current brightness feature value, such as the average fluctuation amplitude, and the brightness benchmark value obtained from long-term historical data, and then divide it by the standard deviation within the historical normal fluctuation range to obtain a standardized deviation measure. For the vibration feature dimension, a similar method is used to calculate its anomaly deviation, that is, based on the statistical benchmark and standard deviation of historical vibration characteristics, such as the average energy of a specific frequency band, the deviation level of the current vibration characteristic is assessed. After obtaining the anomalous deviations of the brightness and vibration feature dimensions, appropriate weights need to be assigned to each dimension for weighted summation. The weight allocation is not fixed but dynamically determined based on the statistical characteristics of historical signals. The core principle is to allocate weights according to the identification contribution of each feature dimension in historical real-world risk events. For example, analysis of historical data may reveal that vibration features exhibit more stable and significant anomalies in most real-world impact events, thus allowing for a higher weighting of vibration features. Conversely, in scenarios with strong light interference or nighttime vehicle headlights, brightness features may be more indicative, in which case the weighting of brightness features can be appropriately increased. The weighting is not fixed, but dynamically determined by combining historical signal statistical characteristics and the newly added consistency score. The core logic is as follows: First, basic weights are allocated based on the identification contribution of each feature dimension in historical real risk events. For example, analysis of historical data reveals that vibration features exhibit more stable and significant anomalies in most real impact events, thus assigning a higher basic weight to vibration features; brightness features are more indicative in certain strong light interference or nighttime vehicle headlight scenarios, so the basic weight of brightness features is appropriately increased. Second, the consistency score is used as a basis for fine-tuning the weights. When the consistency score is high, closer to 1, it indicates strong photo-vibration coupling, and signal anomalies are more likely to be caused by cable vibration. In this case, the weight ratio of brightness and vibration features is maintained or moderately increased to enhance the reliability of the collaborative evaluation of the two types of features. When the consistency score is low, closer to 0, it indicates weak photo-vibration coupling, and signal anomalies are more likely to originate from environmental interference. In this case, the weight of brightness features is reduced and the weight of vibration features is increased to avoid mis-scoring caused by environmental interference. In addition, the weighting can be further fine-tuned based on the current environmental background, such as appropriately increasing the weight ratio of vibration features in low-light environments.

[0052] After weighting, the abnormal deviation of the brightness feature dimension is multiplied by its corresponding dynamic weight to obtain a weighted brightness anomaly score; similarly, a weighted anomaly score for the vibration feature is obtained. The sum of these two weighted scores yields a comprehensive anomaly score for the current environmental state. A higher score indicates a more significant deviation of the environmental features from the normal state, and a higher potential risk. A preset scoring threshold, such as 0.7, is set. If the calculated comprehensive anomaly score exceeds this threshold, the current environmental state is determined to be abnormal, with a high probability of risk, thus triggering subsequent feature clustering center extraction and dynamic parameter generation processes. Through this dynamic weighted scoring mechanism based on multidimensional feature deviation analysis and consistency scoring, the abnormal information of both light and vibration signals can be accurately integrated, effectively distinguishing the sources of signal anomalies, accurately quantifying and classifying environmental risks, and providing a scientific decision-making basis for subsequent precise response and adaptive parameter adjustment.

[0053] When the calculated anomaly score of the current environmental state exceeds a preset threshold, it indicates a high probability of anomaly or risk state around the cable-stayed bridge. To further distinguish the risk type and generate matching system control parameters, cluster analysis of the optical-vibration joint feature sequence within the triggering period is required. When the anomaly score exceeds the preset threshold, cluster analysis is performed on the optical-vibration joint feature sequence. This embodiment uses the K-means clustering algorithm, an unsupervised clustering algorithm. Its core is to complete the category division and cluster center extraction through iterative optimization. Combined with the actual outdoor application scenario of cable-stayed bridges, the feature points are iteratively divided into a preset number of clusters, such as 3, until the cluster centers of each category tend to stabilize. After clustering, the extracted cluster centers of each category represent typical optical-vibration feature combinations under different risk modes, such as... Figure 3As shown in the figure, three types of cluster centers and their corresponding risk areas are illustrated. For example, the first type of cluster center may correspond to the characteristics of "significant changes in brightness and moderate vibration energy", which is common in scenarios where vehicle headlight interference is accompanied by minor scratches. The cluster center is located at (0.52, 0.52). The second type of cluster center may correspond to the characteristics of "gradual changes in brightness and persistently high vibration energy", which is often associated with persistent strong winds or regular contact. The cluster center is located near (0.82, 0.82). The third type of cluster center may correspond to the characteristics of "both brightness and vibration exhibit low-energy random fluctuations", indicating that the environment is in a relatively stable and risk-free state. The cluster center is located near (0.18, 0.18). Based on the light intensity and vibration distribution patterns reflected by these cluster centers, frequency adaptive adjustment parameters and brightness response gain parameters are calculated using a preset nonlinear mapping function. The frequency adaptive adjustment parameter is a key variable for dynamically controlling the signal acquisition interval of the sensor array. Its value directly determines the sampling frequency of the system in the next monitoring cycle. This parameter aims to adjust the data acquisition density according to the identified risk level: shortening the acquisition interval to improve temporal resolution and capture transient details when the risk is high, and extending the acquisition interval to reduce system power consumption when the risk is low. The brightness response gain parameter is a compensation coefficient used to dynamically adjust the system's sensitivity to changes in light intensity signals. Its value directly affects the degree of amplification or suppression of light intensity features in subsequent signal processing stages. This parameter aims to adaptively compensate for signal strength under different ambient lighting conditions, especially in low-light scenarios: enhancing the signal to improve detection reliability in insufficient light, and maintaining normal processing in sufficient light to avoid signal overload.

[0054] Specifically, for the frequency adaptive adjustment parameters, the input to its nonlinear mapping function is mainly the vibration energy-related characteristic values ​​of the cluster centers. The function is designed as an S-curve, for example: ,in, The frequency domain energy characteristic value of vibration. The coefficient for adjusting the steepness of the curve is obtained by curve fitting the relationship between vibration energy and the required sampling frequency in historical risk data, and is assumed to be 2.5. The intermediate threshold for vibration frequency domain energy is set as the 95th percentile value of the statistical distribution of vibration energy under risk-free conditions in long-term monitoring. This function calculates when the characteristic value of vibration frequency domain energy exceeds this intermediate threshold. When vibration is significant, the output frequency adjustment parameter increases rapidly, driving the system to use a higher sampling frequency to capture details; conversely, when vibration is weak, a lower frequency parameter is output to maintain low-power operation. The brightness response gain parameter is calculated using an exponential nonlinear mapping function, with the brightness fluctuation characteristic value of the cluster centers as input, and the function expression is: ,in The brightness fluctuation characteristic value of the cluster center. The reference gain coefficient is set to 1.0. The gain adjustment coefficient is coupled to the average ambient light intensity as a compensation factor. Designed in an exponential decay form, this function aims to output a higher gain coefficient when the eigenvalue is low in low-light environments, amplifying signal sensitivity and compensating for signal weakening caused by insufficient illumination. Conversely, it outputs a normal gain when illumination is sufficient to avoid over-response. Finally, the frequency adaptive adjustment parameter calculated for each feature cluster center is combined with the brightness response gain parameter to form a set of dynamically adjusted parameters that match different risk modes. This parameter set serves as the core basis for subsequent signal acquisition strategies and adaptive control of warning response intensity, enabling dynamic adjustment of monitoring frequency and response sensitivity based on the real-time identified risk category. This achieves an optimal balance between power consumption and performance while ensuring accurate risk identification.

[0055] By accurately dividing and extracting the feature cluster centers of different risk scenarios for cable-stayed bridges using K-means clustering, and relying on a nonlinear mapping function to achieve efficient conversion of feature patterns into adjustment parameters, the generated dynamic adjustment parameter set can perfectly match the real-time changes in risk characteristics and ambient lighting. This effectively solves the technical problems of traditional fixed and rigid acquisition parameters that cannot adapt to diverse risk scenarios and environmental differences. While ensuring the accuracy and sensitivity of cable-stayed bridge risk identification, it achieves dynamic adaptation of signal acquisition frequency and response gain, avoiding high power consumption and eliminating risk omissions, and greatly improving the practicality and adaptability of the solution in complex outdoor scenarios.

[0056] Step S5: Based on the dynamically adjusted parameter set and environmental feedback, update the signal acquisition frequency, determine the final acquisition frequency, and construct a joint triggering judgment rule for optical vibration with multiple constraints.

[0057] In S5, the signal acquisition frequency is updated in conjunction with environmental feedback, including: based on the frequency adaptive adjustment parameters in the dynamic adjustment parameter set, and combined with the current ambient light intensity distribution characteristics, the initial acquisition interval is calculated through a nonlinear frequency optimization function; if the initial acquisition interval is lower than the preset protection threshold, the compensation coefficient of the brightness response gain parameter is dynamically adjusted based on the difference between the current ambient brightness and historical benchmark data to form a compensated acquisition interval; the compensated acquisition interval is then fused and optimized again with the frequency adaptive adjustment parameters until the acquisition interval is not lower than the preset protection threshold, and the reciprocal of the finally optimized acquisition interval is used as the final acquisition frequency in the current scene.

[0058] Specifically, after generating the dynamically adjusted parameter set, to ensure the signal acquisition frequency accurately adapts to the real-time risk characteristics and environmental conditions around the cable-stayed cable, multiple rounds of optimization and correction of the acquisition frequency are required, taking into account environmental feedback. Simultaneously, triggering rules with multiple constraints are established to ensure both efficient capture of risk signals and low-power operation of the equipment. Specifically, updating the signal acquisition frequency based on environmental feedback involves obtaining a compliant final acquisition frequency through multiple rounds of optimization calculations. First, based on the frequency adaptive adjustment parameters in the dynamically adjusted parameter set, and combined with the current ambient light intensity distribution characteristics, a preliminary acquisition interval is calculated using a nonlinear frequency optimization function. This nonlinear frequency optimization function is a transformation function constructed based on the nonlinear correlation between the characteristic distribution and the acquisition interval, specifically: ,in, This is the basic data acquisition interval. It is a frequency adaptive adjustment parameter. It is a correction factor obtained by looking up a table based on the current average ambient light intensity. The lower the light intensity, the better. The smaller the value, the less likely it is to over-boost the frequency in low light conditions. For the initial acquisition interval, this function uses a frequency adaptive adjustment parameter as the core variable, coupled with a correction factor corresponding to the current ambient light intensity distribution characteristics, to achieve a non-linear conversion from parameter values ​​to the acquisition interval. A larger frequency adaptive adjustment parameter indicates a higher level of risk, resulting in a smaller initial acquisition interval and a higher corresponding signal acquisition frequency, meeting the high-frequency acquisition requirements in high-risk scenarios. After obtaining the initial acquisition interval, it is compared with a preset protection threshold. The preset protection threshold is the minimum acquisition interval threshold set to avoid missed detection of risk signals due to excessively low acquisition frequencies, for example, 0.5 seconds. If the initial acquisition interval is lower than the preset protection threshold, it indicates that the currently calculated acquisition interval is too small, which can easily cause a surge in device power consumption and exceed actual monitoring requirements. In this case, based on the difference between the current ambient light level and the historical benchmark, the compensation coefficient of the brightness response gain parameter needs to be dynamically adjusted. The compensation coefficient is a dynamic adjustment factor used to correct the acquisition interval. The larger the difference between the ambient light level and the historical benchmark, the more significant the deviation of the current lighting environment from the normal state, and the larger the corresponding compensation coefficient. By incorporating the compensation coefficient into the brightness response gain parameter, the initial acquisition interval is amplified and corrected to form a compensated acquisition interval, ensuring that the acquisition interval meets the basic threshold requirements.

[0059] After the initial compensation, the compensated acquisition interval and the frequency adaptive adjustment parameter are fused and optimized again. The rationality of the compensated acquisition interval is verified based on the frequency adaptive adjustment parameter. If the fused and optimized acquisition interval is still lower than the preset protection threshold, the compensation coefficient of the brightness response gain parameter is iteratively adjusted and corrected until the acquisition interval is not lower than the preset protection threshold. The iteration is then terminated and the final optimized acquisition interval is determined. The acquisition frequency and acquisition interval are reciprocals of each other. Taking the reciprocal of the final optimized acquisition interval yields the final acquisition frequency for the current scenario. This frequency can simultaneously match the actual needs of risk level and environmental conditions. Based on the determined final acquisition frequency, a joint triggering judgment rule with multiple constraints is further constructed. This rule uses the final acquisition frequency as the basic sampling benchmark and couples multiple constraints such as the brightness change slope threshold, the vibration space vector trajectory stability characteristic threshold, and the anomaly scoring threshold. It clarifies the sampling strategy switching logic when the optical vibration signal meets different characteristic conditions, realizing flexible switching between low-power low-frequency sampling and high-frequency risk sampling. At the same time, it defines the judgment criteria for joint triggering and single signal triggering of optical vibration signals, ensuring that the triggering rule can adapt to diverse risk and environmental scenarios of cable-stayed outdoor installations.

[0060] Through nonlinear optimization and multi-round correction, precise adaptation of signal acquisition frequency is achieved, which not only meets the acquisition density requirements of different risk levels, but also avoids the power consumption problem caused by excessively high frequency. At the same time, the constructed multi-condition constrained optical vibration joint trigger judgment rule effectively solves the technical problems of insufficient sensitivity and excessive power consumption caused by the fixed sampling strategy and single trigger logic of traditional methods. While ensuring the accuracy and real-time performance of cable-stayed line risk monitoring, it also takes into account the low power consumption operation requirements of the equipment, greatly improving the adaptability and practicality of the solution in complex outdoor scenarios.

[0061] In step S5, a joint triggering judgment rule for optical vibration with multiple constraints is constructed, including: dynamically segmenting historical benchmark data based on the final acquisition frequency, calculating the cumulative deviation statistics of brightness fluctuations in each segment, generating a cumulative deviation threshold for brightness fluctuations, extracting the intensity envelope features of the triaxial vibration signal based on the final acquisition frequency, and determining the lower limit of vibration intensity duration by combining the energy distribution characteristics of the vibration signal; analyzing the correlation between brightness signal and vibration signal in temporal changes, calculating the cross-covariance sequence between the two signals over time, generating a dynamic cross-modal signal consistency weight based on the statistical characteristics of the cross-covariance sequence, and combining the cumulative deviation threshold for brightness fluctuations, the lower limit of vibration intensity duration, and the cross-modal signal consistency weight into a composite condition to form the joint triggering judgment rule for optical vibration.

[0062] Specifically, after determining the final acquisition frequency, a joint triggering judgment rule with multiple constraints needs to be constructed by combining this frequency with the characteristics of the optical vibration signal. This clarifies the triggering criteria for system sampling strategy switching and risk response, ensuring a balance between the accuracy of risk identification and low-power operation. Specifically, constructing a joint triggering judgment rule with multiple constraints requires generating three core condition parameters sequentially and then combining them. First, the historical reference signal is dynamically segmented based on the final acquisition frequency. Dynamic segmentation determines the segment length according to the sampling period of the final acquisition frequency. For example, when the final acquisition frequency is 10Hz, the segment length is set to 1 second, containing 10 sampling points. The cumulative deviation statistic is calculated for the brightness signal within each segment. The cumulative deviation statistic is the sum of the absolute values ​​of the deviations between each brightness sample value within the segment and the historical reference mean for that segment. For example, if the brightness sample values ​​in a segment are 52, 55, and 48 lux, and the reference mean is 50 lux, the cumulative deviation statistic is 9 lux.

[0063] Next, the intensity envelope features of the vibration signal are extracted based on the final acquisition frequency. The intensity envelope features are the contour features of the vibration signal amplitude changing over time, which can intuitively reflect the continuous change trend of vibration intensity in the time domain. The Hilbert transform algorithm, an unsupervised signal analysis algorithm, is used for extraction. An analytical transform is performed on the original vibration time-domain signal to obtain an analytical signal. Then, a modulo operation is performed on the analytical signal to generate an intensity envelope feature sequence with the same time-domain length as the original signal. For example, performing a Hilbert transform on a vibration time-domain signal with a peak value of 1.2g and a frequency of 5Hz yields a smooth and continuous intensity envelope curve. This curve clearly shows the complete change process of vibration intensity gradually increasing from the initial 0g to the peak value of 1.2g and then falling back to the baseline value, accurately reflecting the dynamic change law of vibration intensity. Since the triaxial vibration signals acquired in this application are obtained through triaxial vibration sensors, reflecting the vibration state of the cable-stayed cable in three spatial dimensions—X (horizontal), Y (horizontal), and Z (vertical)—it is necessary to perform the aforementioned Hilbert transform on each of the X, Y, and Z triaxial vibration time-domain signals based on the final acquisition frequency to obtain the intensity envelope feature sequence for each axis. To avoid missing spatial vibration features from single-axis signals, amplitude fusion processing is required on the triaxial intensity envelope feature sequence. The maximum values ​​of the X, Y, and Z triaxial envelope amplitudes at each sampling time are taken to generate a combined-axis intensity envelope feature sequence. This sequence integrates triaxial vibration information and can comprehensively and accurately reflect the overall vibration intensity state of the cable-stayed cable. For example, if the final acquisition frequency is 10Hz, after Hilbert transformation, the X-axis envelope sequence is obtained as: [0.2g, 0.5g, 0.9g, 1.1g], the Y-axis envelope sequence is: [0.3g, 0.4g, 0.8g, 1.0g], and the Z-axis envelope sequence is: [0.2g, 0.6g, 0.9g, 1.2g]. The fused coaxial intensity envelope feature sequence is [0.3g, 0.6g, 0.9g, 1.2g], which clearly shows the trend of the overall vibration intensity gradually increasing.

[0064] After extracting the coaxial strength envelope feature sequence, the lower limit of vibration intensity duration is determined by combining the energy distribution characteristics of the vibration signal. The energy distribution characteristics of the vibration signal are core features obtained from the statistical analysis of historical monitoring data of the entire cable-stayed line scenario. Specifically, they include the temporal distribution pattern and energy proportion of the vibration intensity envelope under different risk level scenarios. Specifically, the effective amplitude threshold is the critical value that distinguishes between risk-free micro-amplitude vibration and risk-level vibration. It has been calibrated to 0.8g based on a large amount of historical data. Only when the amplitude of the coaxial strength envelope feature sequence exceeds this threshold will it be judged as a valid risk vibration feature. The statistical distribution range of vibration duration under the threshold records the distribution range of vibration duration exceeding the 0.8g threshold in historical risk scenarios, clarifying the duration characteristics of real risk vibration. The benchmark for effective energy accumulation duration under different sampling frequencies clarifies the minimum vibration accumulation duration required to stably identify risk vibration under different sampling densities, avoiding misjudgment due to differences in sampling density.

[0065] When determining the lower limit of the sampling time, the core principle is to anchor it to the benchmark of "the minimum number of effective sampling points for historical risk." Historical data shows that for stable identification of risky vibrations on a cable-stayed bridge, at least three consecutive envelope sampling points exceeding the 0.8g threshold must be captured. Vibrations exceeding the threshold with fewer than this number of points are mostly environmental interference and do not constitute a real risk. Therefore, these three sampling points represent the minimum number of effective sampling points for historical risk. Combined with the final sampling frequency, this is achieved using the formula: Calculations are performed to ensure that the lower limit of the duration matches the sampling density. If the final sampling frequency is 10Hz, the lower limit of the vibration intensity duration is calculated by substituting into the formula as 3÷10=0.3 seconds. This means that when the amplitude of three consecutive sampling points in the coaxial strength envelope feature sequence exceeds 0.8g, if one more point exceeds the threshold, the condition is met, and the vibration intensity duration is determined to have reached the lower limit requirement.

[0066] Then, the correlation between the light intensity signal and the vibration signal in terms of temporal changes is analyzed, and the cross-covariance sequence between the two signals over time is calculated. For example, taking the light intensity time series [50, 52, 55, 53] and the vibration amplitude series [0.3, 0.5, 0.6, 0.4], the covariance values ​​under different delays are calculated using a sliding window, resulting in the cross-covariance sequence [0.25, 0.18, 0.05]. The cross-covariance sequence quantifies the degree of linear correlation between two signals under different time delays. Each value in the sequence corresponds to the covariance value of the two signals under a specific delay. The sign of the covariance value indicates the direction of correlation; a positive value indicates positive correlation, and a negative value indicates negative correlation. The absolute value directly represents the strength of the correlation; the larger the absolute value, the more synchronized the changing trends of the two signals are under the corresponding delay. The maximum value in the sequence, i.e., the peak value, is the core representation of the optimal correlation between the two signals, corresponding to the time delay state of optimal synchronicity between the two signals. Synchronization is determined based on the peak value of the cross-covariance sequence. The criteria are derived from statistical analysis of historical monitoring data of the entire cable-stayed bridge scenario. Verification with extensive historical data shows that when the peak value of the cross-covariance sequence is ≥0.2, the changes in light intensity and vibration signals exhibit a significant linear correlation, indicating good synchronization. When the peak value of the cross-covariance sequence is <0.1, the linear correlation between the two signals is extremely weak, indicating poor synchronization. This is because in the cable-stayed bridge monitoring scenario, real risks such as impacts and strong wind-induced vibrations cause changes in light intensity and vibration with a natural causal relationship, resulting in a significantly higher peak value for the corresponding cross-covariance. Conversely, environmental interference such as random light and shadow changes and signal changes caused by unrelated vibrations have no causal relationship, resulting in a lower peak value for the corresponding cross-covariance. Based on the above synchronization determination results, dynamic cross-modal signal consistency weights are generated: If the peak value of the cross-covariance sequence is ≥0.2, the synchronization is good, indicating a high probability that the optical vibration anomaly is caused by the risk of the cable-stayed bridge itself. In this case, the weights are equally distributed as 0.5 for brightness and 0.5 for vibration, achieving a balanced fusion determination of the two types of signals; if the peak value of the cross-covariance sequence is <0.1, the synchronization is poor, indicating that the brightness anomaly is most likely caused by environmental interference. In this case, the weights are tilted towards the vibration signal, which has stronger anti-interference capabilities, and adjusted to 0.3 for brightness and 0.7 for vibration, weakening the impact of environmental interference signals on the determination results. Impact: If the peak value of the cross-covariance sequence is between 0.1 and 0.2, indicating moderate synchronicity, it suggests that the optical vibration anomaly may be caused by weak risk or interference combined with slight vibration. In this case, a gradient weight allocation strategy is adopted, dynamically allocating weights according to the logic of weight = base proportion + (peak value - 0.1) × weight adjustment coefficient. The base proportion for brightness is 0.4, the base proportion for vibration is 0.6, and the weight adjustment coefficient is 1. After calibration with historical data, it is ensured that the weights change linearly with synchronicity, ultimately achieving a gradient transition of brightness weight from 0.4 to 0.5 and vibration weight from 0.6 to 0.5. Continuing with the previous example, the peak value of the cross-covariance sequence is 0.25, meeting the synchronicity good judgment criterion of peak value ≥ 0.2. Therefore, a brightness weight of 0.5 and a vibration weight of 0.5 are allocated.

[0067] Finally, the cumulative deviation threshold of brightness fluctuation, the lower limit of vibration intensity duration, and the cross-modal signal consistency weight are combined to form a joint trigger judgment rule. First, the brightness fluctuation deviation and vibration intensity are multiplied by their respective cross-modal signal consistency weights to obtain a weighted deviation value. When the sum of the weighted deviation values ​​exceeds the preset cumulative deviation threshold of brightness fluctuation, and the duration of vibration intensity exceeding the corresponding threshold reaches the preset lower limit of vibration intensity duration, it indicates that abnormal risk characteristics meeting the judgment criteria have appeared around the cable-stayed line, and the anomaly of the optical vibration signal has a clear risk orientation. At this time, the joint trigger condition is determined to be met, and the system immediately switches to high-frequency sampling mode and activates the risk warning response mechanism. If the condition is not met, the current signal anomaly is determined to be environmental interference or risk-free fluctuation, and the system maintains a low-power conventional sampling state.

[0068] By dynamically adapting the final acquisition frequency to generate precise trigger parameters and combining the correlation of optical vibration signals to construct composite trigger rules, the problem of fixed and rigid rules, which are prone to false triggers and missed triggers, is effectively solved. This improves the accuracy of risk identification, enables flexible switching between high and low frequency sampling, balances monitoring sensitivity and low power consumption requirements, and enhances the adaptability and practical value of the solution in complex outdoor scenarios.

[0069] Step S6: Obtain the current optical-vibration joint feature sequence in real time and compare it with historical benchmark data. If the optical-vibration joint trigger judgment rule is met, generate a warning light adjustment command to adjust the brightness and flashing mode of the inclined wire warning light.

[0070] Step S6 includes: acquiring the current optical-vibration joint feature sequence in real time, calculating the current cumulative deviation of brightness fluctuation and the current duration of vibration intensity respectively, and weighting and fusing the first score of the current cumulative deviation of brightness fluctuation and the second score of the current duration of vibration intensity based on the cross-modal signal consistency weight to obtain the current event confidence score; if the current cumulative deviation of brightness fluctuation exceeds a preset deviation threshold, the current duration of vibration intensity exceeds a preset duration threshold, and the current event confidence score exceeds a preset score threshold, then it is determined that the optical-vibration joint triggering judgment rule meets the multi-condition constraints; based on the judgment result that meets the optical-vibration joint triggering judgment rule, generating a warning light adjustment command that matches the current environmental state, including brightness level parameters and flashing frequency parameters, and controlling the inclined wire warning light to switch to the corresponding working mode.

[0071] Specifically, by collecting and analyzing optical vibration signals in real time, verifying triggering rules, and outputting control commands, the brightness and flashing mode of the guy wire warning lights can be adaptively adjusted to accurately match the on-site risk level and improve the effectiveness and relevance of outdoor warnings. Specifically, the current optical vibration joint feature sequence is first acquired in real time. This sequence contains time-domain characteristic data of light intensity and time-domain characteristic data of vibration during the current monitoring period, fully reflecting the real-time changes in the light environment and vibration signal anomalies around the guy wire.

[0072] Based on the aforementioned cross-modal signal consistency weights, feature quantization calculations are performed on the current optical-resonance joint feature sequence. Firstly, based on the historical average brightness benchmark, the cumulative deviation of the current brightness fluctuation is calculated. Then, through: The formula is normalized to obtain the first score corresponding to the cumulative deviation of the current brightness fluctuation. The first score ranges from 0 to 1. The larger the value, the greater the degree to which the current brightness deviates from the risk-free baseline state, and the more significant the risk characteristics of the light environment anomaly. for In the full-scene monitoring of the cable-stayed bridge, the extreme range difference of the cumulative deviation of historical brightness fluctuation is determined by long-term historical monitoring data. It represents the maximum fluctuation amplitude of the cumulative deviation of brightness fluctuation under real risk and environmental interference scenarios. For example, it is statistically set to 20 units. For instance, if the preset threshold for cumulative deviation of brightness fluctuation is 20 units, and the current cumulative deviation of brightness fluctuation is calculated to be 30 units, then the first score is calculated to be 0.5.

[0073] On the other hand, the intensity envelope features of the current vibration signal are extracted, the duration of vibration intensity exceeding the safety threshold is statistically analyzed, and then... The formula is normalized to obtain the second score corresponding to the duration of the current vibration intensity. The second score also ranges from 0 to 1. The larger the value, the more obvious the continuous characteristics of the current vibration anomaly and the stronger the risk indication of the vibration signal. The maximum deviation range of vibration duration is obtained from the historical monitoring data of the inclined cable. It is the maximum difference between the duration of vibration intensity exceeding the safety threshold in the actual risk scenario and the preset lower limit of vibration intensity duration. It represents the effective fluctuation range of vibration duration deviation. For example, it is statistically set to 8.33 seconds. For example, if the preset lower limit of vibration intensity duration is 5 seconds and the current duration of vibration intensity exceeding the safety threshold is statistically 10 seconds, then the second score is calculated to be 0.6. The first score and the second score are then multiplied by the corresponding cross-modal signal consistency weight to complete the weighted fusion calculation and obtain the confidence score of the current event.

[0074] The first and second scores are then multiplied by their corresponding cross-modal signal consistency weights to complete a weighted fusion calculation, yielding the current event confidence score. This score comprehensively characterizes the risk orientation and credibility of the current optical vibration signal anomaly. Subsequently, a multi-threshold verification mechanism is initiated to simultaneously verify three judgment conditions. If the current cumulative deviation of brightness fluctuation exceeds a preset cumulative deviation threshold, the current vibration intensity duration exceeds a preset lower limit for vibration intensity duration, and the current event confidence score exceeds a preset confidence threshold, it means that the current brightness fluctuation is not random environmental interference, but a risk-related fluctuation strongly correlated with the cable-stayed bridge vibration. Furthermore, the vibration intensity has reached the duration requirement for risk assessment, and the risk credibility after optical vibration signal fusion also meets the preset standard. The combination of these three factors accurately confirms the existence of a real risk signal around the cable-stayed bridge. In cases involving collisions or strong winds, subsequent risk response actions must be triggered. If any condition is not met, it means the current signal anomaly does not constitute a real risk: if only the cumulative deviation of brightness fluctuations fails to meet the standard, it indicates that the change in the light environment is small and does not have risk-indicating characteristics; if only the duration of vibration intensity fails to meet the standard, it indicates that the vibration is an instantaneous disturbance, such as a small particle impact, and does not constitute a persistent risk; if only the event confidence score fails to meet the standard, it indicates weak synchronization between brightness and vibration signals, and the brightness fluctuations are likely independent environmental interference, such as random illumination by vehicle lights, while the vibration may be an unrelated micro-vibration, all of which do not meet the collaborative judgment requirements for real risk. If the light and vibration joint trigger judgment rule is not met, i.e., any judgment condition is not met, it is confirmed that there is no real risk, and the low-power conventional mode adjustment command continues to ensure energy saving and avoid false alarms.

[0075] If the light and vibration joint triggering judgment rule is met, that is, if all three judgment conditions are met simultaneously, confirming the existence of a real risk signal, then an adjustment instruction is generated according to the three-dimensional mapping table of "risk level - ambient light intensity - warning parameters" to achieve precise matching between risk and warning effect. The specific process is as follows: First, a three-dimensional mapping table is constructed. This mapping table is pre-calibrated based on historical monitoring data and warning effect verification data of the entire scene of the cable-stayed line. The core logic is that the warning parameters are positively correlated with the risk level and adaptively matched with the ambient light intensity to ensure that the warning lights have clear visibility in different scenarios. Among them, the risk level is divided into three levels according to the current event confidence score mentioned above: confidence score of 0.7~0.8 is low risk level, 0.8~0.9 is medium risk level, and ≥0.9 is high risk level; the ambient light intensity is determined by the average value of the real-time collected light intensity signal, divided into strong light environment ≥500 lux, such as sunny noon, medium light environment 50~500 lux, such as cloudy day, early morning) and weak light environment <50 lux, such as night and tunnel perimeter. Next, based on the extracted risk level and ambient light brightness category, the corresponding brightness level parameters are matched from the mapping table. For example, if the risk level is high and the environment is in low light, the brightness level parameter is 5 and the flicker frequency parameter is 5Hz; if the risk level is medium and the environment is in medium light, the brightness level parameter is 3 and the flicker frequency parameter is 3Hz; if the risk level is low and the environment is in strong light, the brightness level parameter is 2 and the flicker frequency parameter is 0.5Hz.

[0076] Finally, the warning light operating mode is switched: based on the judgment result, a corresponding adjustment command is sent to the warning light control unit. The control unit has a built-in drive circuit that parses the parameters and converts them into current adjustment signals and pulse control signals to complete the operating mode switch. Simultaneously, the warning light's operating status feedback signal is collected in real time to verify its consistency with the adjustment command. If a deviation exists, a calibration command is reissued to ensure accurate adjustment: when the rules are met, risk linkage warnings are accurately implemented; when the rules are not met, a stable low-power conventional indicator state is maintained. Through multi-dimensional feature verification and weighted confidence assessment, the accuracy of risk trigger judgment is ensured. At the same time, relying on hierarchical control commands, the adaptive switching of the warning light's operating mode is achieved, effectively solving the technical problem that traditional warning lights have fixed brightness and flashing modes, making them unable to adapt to real-time risk levels and environmental conditions. This improves the intuitiveness and recognizability of the guy wire risk warning, and can reasonably adjust the warning power consumption according to the risk level, balancing outdoor warning effects and equipment energy-saving requirements, significantly enhancing the solution's on-site practicality and adaptability.

[0077] In summary, this application provides an intelligent control method and system for guy wire warning lights. Its core lies in synchronously acquiring ambient light intensity and vibration signals through an array of optical and vibration sensors. After adaptive noise baseline correction, spatiotemporal feature extraction, and dynamic threshold judgment, it accurately identifies potential risk events around the guy wire. When a sudden change in illumination or vibration trajectory consistency exceeds a preset threshold is detected, the system automatically activates an optical-vibration co-enhanced sampling mode. Multimodal joint features are extracted through high-frequency sampling and short-time spectrum analysis, and adaptive control parameters matching the environmental state are generated based on dynamic confidence assessment and cluster analysis. Furthermore, the signal acquisition frequency is optimized in conjunction with environmental feedback, and a joint triggering judgment rule with multiple constraints is constructed, ultimately achieving dynamic adjustment of the warning light brightness and flashing mode according to the risk level. This method effectively solves the problems of slow response, high false trigger rate, and high energy consumption of traditional warning lights in complex outdoor environments, significantly improving the real-time performance, accuracy, and energy efficiency of guy wire safety monitoring, and providing a reliable and adaptive technical solution for the intelligent protection of tall outdoor infrastructure.

[0078] The above describes an intelligent control method for a guy wire warning light in an embodiment of this application. The following describes an intelligent control system for a guy wire warning light in an embodiment of this application. Please refer to [link / reference]. Figure 4 One embodiment of the intelligent control system for a guy wire warning light in this application includes:

[0079] The acquisition module is used to acquire brightness data and triaxial vibration signals through the cable-stayed sensor array, and perform adaptive noise correction to obtain a preliminary signal sequence.

[0080] The extraction module is used to extract spatiotemporal features from the preliminary signal sequence. If the brightness change trend feature or the vibration trajectory stability feature exceeds the corresponding threshold, the enhanced sampling mode is activated.

[0081] The fusion module is used to perform high-frequency light intensity sampling and short-time spectrum analysis of vibration signals in enhanced sampling mode, obtain the time-domain sequence of light intensity fluctuations and the frequency-domain feature vector of vibration, and perform cross-modal time-series alignment and feature fusion to form a joint light-vibration feature sequence.

[0082] The analysis module is used to calculate the anomaly score of the current environmental state based on the optical-vibration joint feature sequence. If the score exceeds the preset threshold, the feature cluster centers in the optical-vibration joint feature sequence are extracted to generate a dynamically adjusted parameter set.

[0083] The adjustment module is used to update the signal acquisition frequency based on the dynamic adjustment parameter set and environmental feedback, determine the final acquisition frequency, and construct a joint triggering judgment rule for optical vibration with multiple constraints.

[0084] The control module is used to acquire the current optical-vibration joint feature sequence in real time, compare it with historical benchmark data, and if the optical-vibration joint trigger judgment rule is met, it generates a warning light adjustment command to adjust the brightness and flashing mode of the inclined wire warning light.

[0085] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0086] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligent control of a guy wire warning light, characterized in that, The method includes: Step S1: Acquire light intensity data and triaxial vibration signals through a cable-stayed sensor array, and perform adaptive noise correction to obtain a preliminary signal sequence; Step S2: Extract spatiotemporal features from the preliminary signal sequence. If the brightness change trend feature or vibration trajectory stability feature exceeds the corresponding threshold, activate the enhanced sampling mode. Step S3: In enhanced sampling mode, perform high-frequency light intensity sampling and short-time spectrum analysis of vibration signal to obtain the light intensity fluctuation time domain sequence and vibration frequency domain feature vector, and perform cross-modal time sequence alignment and feature fusion to form a joint light-vibration feature sequence; Step S4: Calculate the anomaly score of the current environmental state based on the optical-vibration joint feature sequence. If the score exceeds a preset threshold, extract the feature cluster centers in the optical-vibration joint feature sequence and generate a dynamic adjustment parameter set. Step S5: Based on the dynamically adjusted parameter set and environmental feedback, update the signal acquisition frequency, determine the final acquisition frequency, and construct a joint triggering judgment rule for optical vibration with multiple constraints. Step S6: Obtain the current optical-vibration joint feature sequence in real time and compare it with historical benchmark data. If the optical-vibration joint trigger judgment rule is met, generate a warning light adjustment command to adjust the brightness and flashing mode of the inclined wire warning light.

2. The intelligent control method for a guy wire warning light according to claim 1, characterized in that, Step S1 includes: The sensor array synchronously collects ambient light intensity data and triaxial vibration signals to form a time-stamped raw signal sequence. An adaptive noise baseline model is established for the ambient light intensity data and triaxial vibration signals respectively. The current ambient noise level is dynamically estimated based on the historical statistical characteristics of the signals. Real-time noise suppression and baseline drift compensation are performed on the raw signal sequence to generate a preliminary signal sequence after noise suppression.

3. The intelligent control method for a guy wire warning light according to claim 1, characterized in that, Step S2 includes: For the ambient light intensity data in the preliminary signal sequence, the abrupt change slope within the sliding time window is calculated as the abrupt change trend feature of the light intensity in the time domain; for the triaxial vibration signal in the preliminary signal sequence, the sequence of changes in the direction angle of the acceleration vector between consecutive sampling points is calculated, and the proportion of the direction angle changes being less than a preset angle threshold is statistically analyzed as the spatial vector vibration trajectory stability feature of the vibration signal.

4. The intelligent control method for a guy wire warning light according to claim 1, characterized in that, Step S3 includes: In the enhanced sampling mode, the ambient light brightness sampling frequency is increased to a preset high frequency value, and sampling windows of fixed duration are divided based on the adjusted sampling frequency. The light brightness fluctuation time-domain sequence within each window is obtained. The synchronously acquired vibration signal is subjected to short-time Fourier transform to extract the energy distribution characteristics of the vibration signal in each frequency band, forming a vibration frequency domain feature vector. The light brightness fluctuation time-domain sequence and the vibration frequency domain feature vector are interpolated and aligned on the time axis. The aligned light brightness time-domain feature and the vibration frequency domain feature vector are concatenated into a joint feature vector, and the joint feature vector is enhanced to form an enhanced light-vibration joint feature sequence.

5. The intelligent control method for a guy wire warning light according to claim 4, characterized in that, The enhancement process for the joint feature vector includes: A physical coupling model between wind-induced vibration and light and shadow fluctuation is established. The physical coupling model predicts the corresponding theoretical light and shadow fluctuation curve based on the extracted vibration signal frequency domain feature vector. The theoretical light and shadow fluctuation curve is compared with the real-time acquired light brightness fluctuation time domain sequence to obtain a consistency score. The consistency score is used as a new feature dimension and incorporated into the light and vibration joint feature sequence to form an enhanced light and vibration joint feature sequence.

6. The intelligent control method for a guy wire warning light according to claim 1, characterized in that, Step S4 includes: The abnormal deviation of the light intensity feature dimension and the vibration feature dimension in the optical-vibration joint feature sequence is calculated. Dynamic weights are assigned to each feature dimension according to the statistical characteristics of historical signals. The feature dimensions include the light intensity feature dimension and the vibration feature dimension. The abnormal deviation of each feature dimension is weighted and summed with the corresponding weights to obtain the abnormal score of the current environmental state. When the abnormal score exceeds a preset threshold, cluster analysis is performed on the optical-vibration joint feature sequence to extract the feature cluster centers of each category. Based on the brightness and vibration distribution patterns reflected by the feature cluster centers, frequency adaptive adjustment parameters and brightness response gain parameters are calculated respectively through a nonlinear mapping function. The frequency adaptive adjustment parameters and brightness response gain parameters are combined into a dynamic adjustment parameter set.

7. The intelligent control method for a guy wire warning light according to claim 1, characterized in that, S5 updates the signal acquisition frequency based on environmental feedback, including: Based on the frequency adaptive adjustment parameters in the dynamic adjustment parameter set, and combined with the current ambient light brightness distribution characteristics, the initial acquisition interval is calculated through a nonlinear frequency optimization function; if the initial acquisition interval is lower than the preset protection threshold, the compensation coefficient of the brightness response gain parameter is dynamically adjusted based on the difference between the current ambient brightness and historical benchmark data to form a compensated acquisition interval. The compensated acquisition interval is then fused and optimized again with the frequency adaptive adjustment parameter until the acquisition interval is not lower than the preset protection threshold. The reciprocal of the finally optimized acquisition interval is then used as the final acquisition frequency for the current scenario.

8. The intelligent control method for a guy wire warning light according to claim 7, characterized in that, Step S5 involves constructing a joint triggering rule for optical vibration with multiple constraints, including: Based on the final acquisition frequency, the historical benchmark data is dynamically segmented, the cumulative deviation statistics of brightness fluctuations in each segment are calculated, a cumulative deviation threshold of brightness fluctuations is generated, the intensity envelope features of the triaxial vibration signal are extracted according to the final acquisition frequency, and the lower limit of vibration intensity duration is determined by combining the energy distribution characteristics of the vibration signal. The correlation between the light intensity signal and the vibration signal in terms of temporal variation is analyzed, and the cross-covariance sequence between the two signals over time is calculated. Based on the statistical characteristics of the cross-covariance sequence, a dynamic cross-modal signal consistency weight is generated. The cumulative deviation threshold of light intensity fluctuation, the lower limit of vibration intensity duration, and the cross-modal signal consistency weight are combined as composite conditions to form a joint triggering judgment rule for light and vibration.

9. The intelligent control method for a guy wire warning light according to claim 1, characterized in that, Step S6 includes: The current optical vibration joint feature sequence is acquired in real time, and the current cumulative deviation of brightness fluctuation and the current duration of vibration intensity are calculated respectively. The first score of the current cumulative deviation of brightness fluctuation and the second score of the current duration of vibration intensity are weighted and fused based on the cross-modal signal consistency weight to obtain the confidence score of the current event. If the cumulative deviation of the current brightness fluctuation exceeds a preset deviation threshold, the duration of the current vibration intensity exceeds a preset duration threshold, and the confidence score of the current event exceeds a preset score threshold, then it is determined that the light vibration joint triggering judgment rule with multiple conditions is satisfied. Based on the judgment result that satisfies the light-vibration joint trigger judgment rule, a warning light adjustment command matching the current environmental state is generated, including brightness level parameters and flashing frequency parameters, to control the inclined wire warning light to switch to the corresponding working mode.

10. A smart control system for a guy wire warning light, used to implement the smart control method for a guy wire warning light as described in any one of claims 1-9, characterized in that, The system includes: The acquisition module is used to acquire brightness data and triaxial vibration signals through the cable-stayed sensor array, and perform adaptive noise correction to obtain a preliminary signal sequence. The extraction module is used to extract spatiotemporal features from the preliminary signal sequence. If the brightness change trend feature or the vibration trajectory stability feature exceeds the corresponding threshold, the enhanced sampling mode is activated. The fusion module is used to perform high-frequency light intensity sampling and short-time spectrum analysis of vibration signals in enhanced sampling mode, obtain the time-domain sequence of light intensity fluctuations and the frequency-domain feature vector of vibration, and perform cross-modal time-series alignment and feature fusion to form a joint light-vibration feature sequence. The analysis module is used to calculate the anomaly score of the current environmental state based on the optical-vibration joint feature sequence. If the score exceeds a preset threshold, the feature cluster centers in the optical-vibration joint feature sequence are extracted to generate a dynamically adjusted parameter set. The adjustment module is used to update the signal acquisition frequency based on the dynamic adjustment parameter set and environmental feedback, determine the final acquisition frequency, and construct a joint triggering judgment rule for optical vibration with multiple condition constraints. The control module is used to acquire the current optical-vibration joint feature sequence in real time, compare it with historical benchmark data, and if the optical-vibration joint trigger judgment rule is met, generate a warning light adjustment command to adjust the brightness and flashing mode of the inclined wire warning light.

Images