Bridge structure health monitoring method based on big data and cloud computing

Abstract

The invention relates to the technical field of bridge structure health monitoring, and discloses a bridge structure health monitoring method based on big data and cloud computing, and the method comprises the following steps: S1, multi-modal data collection: collecting bridge operation data in real time through a multi-modal sensor network arranged at a key part of a bridge, the operation data comprises stress, vibration and temperature data, and the collected data is transmitted to a cloud computing platform through a wireless network; and S2, data fusion and feature extraction: carrying out modal feature extraction on the operation data, and carrying out data fusion according to a modal weight to generate a comprehensive health index. Through multi-modal data acquisition and fusion, dynamic priority distribution and trend prediction technologies, comprehensive monitoring of bridge health states and advanced identification of potential risks are realized, the monitoring accuracy and response capability are effectively improved, hierarchical visual display of a cloud platform is combined, and the system has a good application prospect. And the intuition of a monitoring result and the efficiency of management decision are obviously enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge structure health monitoring, and specifically to a bridge structure health monitoring method based on big data and cloud computing. Background Art

[0002] During the long-term use of bridges, they are affected by complex external factors, such as frequent traffic loads, environmental temperature changes, and wind impacts. The interaction of these factors may cause problems such as stress concentration, abnormal vibration, and thermal expansion and contraction inside the bridge structure. Existing bridge health monitoring technologies have made progress in the monitoring of some key components, such as monitoring methods based on single data such as stress, vibration, or temperature. However, with the complexity of bridge structural forms and usage environments, these technologies still face many challenges in application.

[0003] The monitoring method centered on single-modal data can analyze specific problems, but has obvious limitations in the overall health assessment of bridges. For example, when evaluating the operating state of a bridge only through stress data, it is difficult to capture the potential impact of temperature changes on structural stability. And relying solely on vibration signal analysis may also ignore the long-term fatigue effect caused by the slow change of bridge stress. Therefore, under the interaction of multiple factors, a single data source cannot comprehensively reflect the overall health state of the bridge, and some potential risks may be ignored.

[0004] In addition, existing health monitoring strategies are often static and lack the ability to dynamically adjust. Even when the monitoring data indicates that the health state of some areas has changed rapidly, the allocation of monitoring resources is still carried out in a fixed manner. This way may result in insufficient monitoring of high-risk areas, while data in low-risk areas is over-collected, reducing the monitoring efficiency. In some special cases, such as when a key node of a bridge is subjected to a sudden load, existing technologies may not be able to quickly identify the severity of the problem, let alone flexibly adjust the key monitoring areas.

[0005] Furthermore, another problem with existing technologies is their limited ability to predict future states. Most monitoring technologies focus on the analysis of real-time states, and often lack effective methods for judging the long-term trends of health states. If the manifestation of some risks is relatively slow in the early stage, such as the change range of health indicators is small, existing static analysis means are difficult to capture this trend in time. When the risks gradually accumulate until they reach the critical point, the best intervention time may have been missed.

[0006] The deficiency in data presentation ability is also an obvious shortcoming of the existing technologies. Traditional monitoring data are mostly presented in numerical values or simple charts, making it difficult to reflect the overall health status of the bridge and its risk distribution. Especially when faced with multi-modal data and a large number of monitoring points, it is very difficult for managers to quickly grasp the overall situation through a single data form. The lack of intuitive presentation directly leads to the lag in bridge health assessment and maintenance decision-making.

[0007] Therefore, although certain progress has been made in some aspects of bridge health monitoring in the existing technologies, there are still obvious deficiencies in dealing with the complexity of the bridge operation environment and the integration and analysis of multi-modal data. Summary of the Invention

[0008] In view of the deficiencies of the existing technologies, the present invention provides a bridge structural health monitoring method based on big data and cloud computing, which solves the problems of single data, insufficient dynamic response, limited trend prediction ability, and non-intuitive visual presentation in bridge structural health monitoring.

[0009] To achieve the above objectives, the present invention is realized through the following technical solutions: A bridge structural health monitoring method based on big data and cloud computing, comprising the following steps: S1. Multi-modal data acquisition: Real-time acquisition of bridge operation data through a multi-modal sensor network deployed at key parts of the bridge, the operation data including stress, vibration, and temperature data, and transmitting the acquired data to the cloud computing platform through a wireless network; S2. Data fusion and feature extraction: Extract modal features from the operation data, and perform data fusion according to modal weights to generate a comprehensive health index; S3. Dynamic priority assignment: Calculate the sensor priority based on the comprehensive health index and its change rate, and classify the risk of the monitoring area; S4. Graphic image generation and optimization: Generate graphic images from the overall to the local based on the priority results, and perform dynamic compression on low-risk data and detailed display on high-risk data; S5. Trend prediction and early warning: Predict the future health status of the bridge through a time series prediction model, and generate early warning information according to the prediction results; S6. Cloud platform data storage and visual presentation: Store the acquired operation data and the generated health index in the cloud platform, and display the real-time monitoring information of the bridge through a visual interaction interface.

[0010] Preferably, the sensors include stress sensors, vibration sensors, and temperature sensors. The stress sensors are used to monitor the stress state of the bridge, the vibration sensors are used to monitor the dynamic response of the bridge, and the temperature sensors are used to monitor the influence of the ambient temperature on the material properties of the bridge.

[0011] Preferably, in the step S2, the comprehensive health index is generated by using the following formula:

[0012] where is the comprehensive health index of the i-th sensor at time t, M is the number of modes, is the mode weight, which is dynamically calculated based on the change range of the mode data, and its formula is:

[0013] where is the variance of the j-th mode data, is the feature extraction value of the j-th mode, specifically the maximum difference of the stress mode, the root mean square value of the vibration mode, and the mean value of the temperature mode.

[0014] Preferably, in the dynamic priority allocation of the step S3, the sensor priority is calculated by the following formula:

[0015] where is the priority of the i-th sensor at time t, is the comprehensive health index, is the change rate of the comprehensive health index, and the calculation formula is:

[0016] where is the time interval, and α, β are priority weight coefficients, satisfying α + β = 1, where α > β.

[0017] Preferably, in the step S4 of generating and optimizing the graphic image, the graphic image is generated in the following manner: Generate a global graphic of the overall health score of the bridge; Generate a heat map of the health status of local key components to show the health distribution of the monitored components, where the color depth is inversely proportional to the health index value; Generate a detailed time series curve of the high-risk area to show the historical change trend of the comprehensive health index.

[0018] Preferably, in the step S6 of cloud platform data storage and visualization display, the real-time collected operation data, the calculated comprehensive health index, and the historical health trend data are stored through the cloud platform, and the following information is displayed through the visual interaction interface: The global health status of the bridge; The health status distribution of local key components; The time series change trend of the high-risk area; Prediction results of the future health status of the bridge and related early warning information.

[0019] The present invention also provides a bridge detection system, including a processor and a memory coupled to the processor. Program instructions are stored in the memory. When the program instructions are executed by the processor, the processor is caused to execute a method for bridge structural health monitoring based on big data and cloud computing as described above.

[0020] The present invention also provides a storage medium. Program instructions are stored in the storage medium. The program instructions are used to execute a method for bridge structural health monitoring based on big data and cloud computing as described above.

[0021] The present invention provides a method for bridge structural health monitoring based on big data and cloud computing. It has the following beneficial effects: 1. The present invention adopts a technical solution of multi-modal data acquisition and fusion. Through the comprehensive extraction and weighted processing of various modal data such as stress, vibration, and temperature, the technical effect of comprehensively evaluating the health status of the bridge is achieved. Compared with the monitoring methods that only rely on single-modal data in the prior art, the problems of incomplete health status evaluation and misjudgment risk caused by single data are solved.

[0022] 2. The present invention introduces a dynamic priority allocation mechanism. By combining health indicators and their change rates, the priority of each sensor is calculated in real time, achieving the technical effect of quickly identifying high-risk areas and optimizing resource allocation. Compared with the technical solutions in the prior art that cannot adjust the monitoring focus according to real-time changes, the problems of lagging response in high-risk areas and uneven resource allocation are solved.

[0023] 3. The present invention, through the trend prediction and early warning module, accurately predicts the health status based on the second-order time series model, achieving the technical effect of identifying potential risks in advance and issuing early warnings. Compared with the method of only relying on static monitoring data for passive analysis in the prior art, the deficiencies of insufficient future risk prediction ability and inability to intervene in advance are solved.

[0024] 4. The present invention adopts cloud platform unified storage and visualization display technology. Through hierarchical health status display and intuitive presentation of the three-dimensional model, the technical effect of real-time monitoring and efficient decision-making support is achieved. Compared with the solutions in the prior art that lack multi-level data integration and intuitive display, the problems of users' difficulty in quickly grasping the overall status and low utilization rate of detailed data are solved. Description of the Drawings

[0025] Figure 1 It is a schematic flow chart of the method of the present invention. Detailed Embodiments

[0026] Next, in conjunction with the accompanying drawings of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0027] Please refer to the attached Figure 1 , the embodiment of the present invention provides a bridge structure health monitoring method based on big data and cloud computing, including the following steps: S1. Multimodal data acquisition: Real-time acquisition of bridge operation data through a multimodal sensor network deployed at key parts of the bridge. The operation data includes stress, vibration, and temperature data, and the acquired data is transmitted to the cloud computing platform through a wireless network; Generally, the health state of a bridge involves multiple different physical quantities, and real-time monitoring needs to be carried out through various types of sensors respectively. As an option, in this embodiment, a multimodal sensor network is deployed to obtain multi-dimensional monitoring data such as stress, vibration, and temperature under the bridge operation state. The sensors are installed at key structural parts such as the main girder, bearings, and suspension cables to ensure that the data can comprehensively cover the key areas of the bridge. The acquired data is transmitted to the cloud computing platform for storage and processing in a wireless transmission manner.

[0028] Specifically, in this embodiment, the multimodal sensors mainly include the following categories: Stress sensor: Used to monitor the stress state of the bridge. The stress data directly reflects whether there is an obvious stress concentration phenomenon in the bridge structure. In some embodiments, the sensor is installed at the mid-span position of the main girder to record the tensile and compressive stresses borne by the bridge in real time. The data acquisition results include the instantaneous value of the stress and its fluctuation range.

[0029] Vibration sensor: Used to monitor the vibration response of the bridge, including acceleration, frequency, and amplitude information. Generally, the vibration data can characterize the dynamic performance of the bridge. For example, a change in the natural vibration frequency may indicate a decrease in the structural stiffness. In a possible implementation, the vibration sensor is installed at the key connection parts and mid-span area of the bridge to collect dynamic response data.

[0030] Temperature sensor: Used to monitor the ambient temperature and the temperature change on the surface of the bridge material. The temperature data is used to analyze the influence of the thermal expansion and contraction effect on the bridge structure. In some embodiments, the temperature sensors are mainly arranged near the suspension cables and bearings of the bridge to capture the possible influence of temperature changes on the structural displacement.

[0031] The acquired original data is formatted into the following matrix:

[0032] Among them, represents the monitoring data set of the i-th sensor at the j-th mode at time t. i represents the sensor number, ranging from 1 to N, where N is the total number of sensors. j represents the mode number, including stress, vibration, and temperature modes. is the k-th data point of the i-th sensor at time t; k represents the sequence number of the sampling moment.

[0033] As an option, the data acquisition frequency is set according to the importance of the mode. Specifically, the acquisition frequencies of stress data and vibration data are relatively high to cope with their fast-changing characteristics. For example, the sampling frequency of the stress sensor is 50Hz, and the sampling frequency of the vibration sensor is 100Hz. While the acquisition frequency of temperature data is relatively low, generally set to 1Hz, to reduce the storage and transmission burden of redundant data.

[0034] To improve the reliability of the acquired data, in some embodiments, a redundant layout method is adopted. Multiple sensors are arranged at key positions, and the acquired data is transmitted to the cloud platform after consistency verification. The advantage of redundant layout is that even if a certain sensor fails, valid data of the same position can be obtained through other sensors, ensuring the continuity and accuracy of system monitoring.

[0035] After data acquisition, the data of all sensors are uploaded to the cloud computing platform in real time through a wireless transmission module. The choice of wireless network can be flexibly configured according to the network conditions of the bridge location. Generally, low-power wide-area network (LPWAN) technologies such as LoRa and NB-IoT are used to reduce the power consumption and bandwidth occupancy of data transmission.

[0036] As a possible implementation manner, all sensors in the present invention are equipped with edge computing units, which can perform preliminary denoising processing on the acquired data. Specifically, high-frequency noise interference is eliminated through an average filtering algorithm based on a sliding window, and at the same time, simple anomaly detection is performed on the data to eliminate possible acquisition errors. For example, some abnormal data points may exceed the upper limit of the preset normal data range. In this case, the edge computing unit will mark them as invalid points and send an alarm signal.

[0037] The data wirelessly transmitted to the cloud computing platform is further uniformly formatted and then used as the input for subsequent data fusion and analysis, providing basic data support for generating comprehensive health indicators.

[0038] S2. Data fusion and feature extraction: Extract modal features from the operation data, and perform data fusion according to modal weights to generate comprehensive health indicators; After completing the multi-modal data acquisition, in order to obtain the comprehensive evaluation results of the bridge health status, it is necessary to fuse and extract features from the collected multi-modal data. The goal of this step is to extract representative information from data of different modalities and generate a comprehensive health index according to the importance of the modalities, providing a data basis for subsequent priority allocation.

[0039] Generally, the fusion of multi-modal data needs to solve problems such as different data types and different importance levels. Therefore, in this step, a fusion method based on modal weight allocation is designed. Through feature extraction and weighting processing of multi-modal data, a comprehensive health score for each sensor is generated. As an option, the process of data fusion and feature extraction can combine statistical characteristics to dynamically adjust the weights of modal data, so as to more accurately reflect the health status of the bridge.

[0040] In this embodiment, data fusion first performs feature extraction on the collected stress, vibration, and temperature data, and the extraction methods are respectively set according to modal characteristics. Specifically: The data feature extraction of the stress modality is based on the maximum stress difference. Generally, the stress difference can intuitively reflect the fluctuation of the force, and its calculation formula is:

[0041] where represents the set of collected data of the i-th sensor in the j-th modality, and represent the maximum and minimum values of the data respectively.

[0042] The data feature extraction of the vibration modality is based on the root mean square value. The root mean square value of vibration is used to characterize the intensity of vibration, and its calculation formula is:

[0043] where K represents the total number of sampling points of the vibration data, is the k-th vibration data point, represents the root mean square value of the vibration signal of the i-th sensor in the j-th modality, represents the set of vibration signal data of the i-th sensor in the j-th modality.

[0044] The data feature extraction of the temperature modality adopts average value processing. As an option, the average value of temperature reflects the impact of the current environment on the overall bridge, and its calculation formula is:

[0045] where represents the average value of the temperature data of the i-th sensor in the j-th modality, Denote the sum of the temperature data of all sampling points of the \(i\)-th sensor in the \(j\)-th mode The mean value of the temperature data can provide basic information on the thermal expansion and contraction trend.

[0046] After extracting the eigenvalue of each mode, the multi-modal data is fused by weighted calculation to generate the comprehensive health index of each sensor. In one possible implementation, the calculation formula of the comprehensive health index is:

[0047] where denotes the comprehensive health index of the \(i\)-th sensor at time \(t\); \(M\) is the number of modes, including stress, vibration and temperature modes; denotes the weight of the \(j\)-th mode, which is used to measure the contribution of the mode to the health state; denotes the data eigenvalue of the \(j\)-th mode.

[0048] Generally, the mode weight is dynamically adjusted according to the volatility of the mode data. Specifically, the calculation formula of the mode weight is:

[0049] where is the variance of the data of the \(j\)-th mode. As an option, the mode data with a larger variance contributes more to the comprehensive health index, so its weight value also increases accordingly.

[0050] In one possible implementation, for a scenario with large dynamic environmental changes, the accuracy of the fusion result can be improved by increasing the real-time sampling frequency of the mode data. For example, when the bridge is under extreme climate conditions, the weight of the temperature mode will increase dynamically, which can reflect the increased impact of temperature changes on the structural health state.

[0051] In some embodiments, in order to improve the robustness of data fusion, for the outliers in the original data, a filtering algorithm based on a sliding window is used for preprocessing before the fusion process. The width of the sliding window can be dynamically set according to the sampling frequency of the mode data. For example, for vibration data, the width of the sliding window is generally 10 sampling points. The filtering algorithm has a good suppression effect on high-frequency noise and can reduce the interference of abnormal points on the calculation of the comprehensive health index.

[0052] The comprehensive health index generated in this step can not only reflect the health state of the location where the sensor is located, but also provide basic data support for the next dynamic priority allocation. In one possible implementation, these health indexes can also be recorded and stored to form historical data for subsequent trend prediction and long-term analysis.

[0053] S3. Dynamic Priority Allocation: Calculate the sensor priority based on the comprehensive health index and its change rate, and classify the risk levels of the monitoring areas. After data fusion and feature extraction are completed, the task of dynamic priority allocation is to further identify the risk levels of the monitoring areas. Through the comprehensive health index and its change rate classify the sensors. Priority allocation ensures that high-risk areas can be quickly identified and concerned, while optimizing the efficiency of data processing and resource allocation. The method of setting priorities not only depends on the health index itself but also needs to consider its change trend to more comprehensively evaluate potential risks.

[0054] In this embodiment, the priority is calculated by the following formula:

[0055] where represents the priority of the i-th sensor at time t; is the comprehensive health index of the i-th sensor at time t; represents the change rate of the comprehensive health index, and its calculation formula is:

[0056] where represents the time interval; α and β are the weights of the comprehensive health index and its change rate respectively, satisfying α + β = 1, and α > β to ensure that the health index has a greater weight in priority allocation.

[0057] Generally, the comprehensive health index reflects the current state of the location where the sensor is located, while the change rate reflects the change trend of its health state. As an option, the specific values of the weights α and β can be dynamically adjusted according to the operating scenarios of the bridge. For example, under large loads or special weather conditions (such as high temperature, heavy rain) on the bridge, the weight of the change rate can be appropriately increased to more sensitively capture the rapid changes in the state.

[0058] Specifically, according to the priority value , the monitoring area can be divided into high-risk areas and low-risk areas. The priority value of the high-risk area is higher than the preset threshold , and the low-risk area is lower than or equal to this threshold. As a possible implementation, the setting of the threshold can be dynamically adjusted according to the statistical distribution of historical monitoring data. For example, take the 80th percentile of all sensor priorities as the high-risk threshold.

[0059] In a possible implementation, to improve the sensitivity of priority allocation, the system also introduces a dynamic threshold adjustment mechanism. This mechanism can dynamically adjust the value. For example, when the bridge is in a harsh environment (such as strong wind or extreme temperature), the threshold may be appropriately reduced to expand the coverage of high-risk areas.

[0060] In some embodiments, to avoid the influence of single data fluctuations on the priority results, time series analysis can be combined to smooth the priority. For example, the sliding window averaging method can be used to smooth the priority and reduce the influence of instantaneous abnormal data. The width W of the sliding window can be dynamically adjusted according to the sampling frequency. For example, for a sampling frequency of 1 Hz, the sliding window width can be set to 10 seconds.

[0061] In this embodiment, the calculation result of the priority can be displayed in the form of a two-dimensional or three-dimensional heat map. As an option, high-priority areas are marked in red, and low-priority areas are marked in green, and the color depth is proportional to the magnitude of the priority value. This visualization method can help users quickly identify high-risk areas and facilitate managers to formulate targeted maintenance measures.

[0062] In another implementation, the result of priority allocation can also be used for subsequent trend analysis. For example, for a certain area, if its priority value shows a continuous upward trend, even if it has not reached the high-risk threshold, the system can trigger an early warning in advance. This trend-based early warning mechanism is of great significance in identifying potential risks.

[0063] To further improve the robustness of priority allocation, spatial correlation analysis is combined in some embodiments. For example, when the priorities of multiple adjacent sensors increase simultaneously, the entire area can be marked as a high-risk area. This analysis based on spatial correlation can more accurately identify potential weak points of the bridge.

[0064] The result of priority allocation not only provides a data basis for subsequent graphic image generation but also lays an important foundation for subsequent trend prediction and decision-making analysis. By dynamically adjusting the priority calculation parameters and thresholds, the system can more flexibly respond to the complex changes in the bridge health status and improve the accuracy and efficiency of overall monitoring.

[0065] S4. Graphic Image Generation and Optimization: Generate graphic images from the whole to the part based on the priority results, and perform dynamic compression on low-risk data and refined display on high-risk data; After completing the dynamic priority allocation, the system needs to graphically process the monitoring data to visually display the health status and risk distribution of the bridge. This step adopts a differential display strategy for high-risk and low-risk areas according to the priority results, ensuring that the information of key areas is highlighted while compressing the information of low-risk areas to optimize resource allocation. Through graphic image generation and optimization, the system can provide managers with a clear and hierarchical display of the health status.

[0066] In this embodiment, the graphic image generation is mainly divided into three levels, including the global graph of the overall health status, the heat map of the local risk distribution, and the detailed trend curve of the high-risk area.

[0067] Generally, the display of the overall health status is based on the comprehensive health score S(t) of the bridge. The calculation formula of the comprehensive health score is:

[0068] where S(t) represents the comprehensive health score of the whole bridge at time t; N represents the total number of sensors; is the comprehensive health index of the i-th sensor at time t.

[0069] As a possible implementation, the result of the comprehensive health score S(t) will be presented in the form of colors. For example: When S(t)>0.8, the overall state of the bridge is marked as green, indicating a good health status; When 0.5≤S(t)≤0.8, it is marked as yellow, indicating that there is a certain risk in the health status; When S(t)<0.5, it is marked as red, indicating that the bridge is in a high-risk state.

[0070] In some embodiments, the local risk distribution is displayed in the form of a heat map, which is generated based on the spatial distribution of the sensor priorities Specifically, the high-risk area is displayed in red, and the low-risk area is displayed in green. The depth of the color is proportional to the priority value. To enhance the accuracy of the heat map, this embodiment uses an interpolation algorithm to spatially expand the priority value to generate a continuous color transition effect. As an option, the bilinear interpolation method can be used to smooth the priority values of adjacent sensors, thereby reducing the visual error caused by data discreteness.

[0071] In a possible implementation, the detailed trend of the high-risk area is displayed through a time series curve, focusing on analyzing the change process of the comprehensive health index For example, for the monitoring data of vibration modes, the following change curve can be generated:

[0072] Among them, is the root mean square value of vibration at time t, is the vibration acceleration value at the k-th sampling point at time t; K is the total number of sampling points of vibration data.

[0073] This change curve can intuitively display the fluctuation of the vibration signal, which helps to identify possible structural damages of the bridge.

[0074] In this embodiment, in order to further optimize the display effect of the graphic image, the system will dynamically compress the information in the low-risk area. For example, for the data of low-priority sensors, only statistical features such as their mean value and maximum value are retained, and they are displayed in a lighter color in the heat map, so as to reduce the information occupancy in the low-risk area.

[0075] In some embodiments, the generation of the graphic image will also combine with the three-dimensional structure model of the bridge to enhance the intuitiveness of visualization. For example, by mapping the health status information onto the three-dimensional model of the bridge, the risk area can be more accurately located. This implementation method is applicable to the health monitoring of large bridges, which helps the manager quickly judge the location and nature of high-risk components.

[0076] As an extended function, the system can also generate an automatic report based on the graphical results. For example, by extracting the change trend data of the high-risk area, a quantitative analysis report is generated, which includes the specific location of the high-risk area, the change rate of the comprehensive health index, and the possible damage types.

[0077] Through the above optimization measures, the graphic image generated in this step can comprehensively reflect the health status of the bridge and provide decision-making support for the manager. The hierarchical display method of the graphic image not only ensures the detail of the information in the high-risk area, but also avoids the redundancy of the information in the low-risk area, improving the overall usability and display efficiency of the system.

[0078] S5. Trend Prediction and Early Warning: Predict the future health status of the bridge through a time series prediction model, and generate early warning information according to the prediction result; After completing the dynamic priority allocation and graphic image generation, the next step is to conduct trend prediction based on the historical data of health indicators and priorities, and generate real-time early warning information. This step uses a time series analysis model to predict the future health status of the bridge to ensure that potential risks can be identified in advance. Trend prediction and early warning are crucial parts of the bridge health monitoring system, providing a basis for subsequent maintenance decisions.

[0079] In this embodiment, the trend prediction adopts a second-order time series prediction model. The comprehensive health indicator The prediction formula is as follows:

[0080] Among them, represents the comprehensive health index predicted at the time ; represents the comprehensive health index at time t; is the change rate of the comprehensive health index, and its calculation formula is:

[0081] is the second-order change rate of the comprehensive health index, and the calculation formula is:

[0082] Δt is the time interval.

[0083] Generally, the second-order time series prediction model can capture the linear and non-linear change trends of the comprehensive health index. As an option, the time interval Δt of this model can be dynamically set according to the sampling frequency of the sensor. For example, when the sampling frequency is 10 Hz, the time interval Δt can be set to 0.1 second, so as to improve the accuracy of the prediction result.

[0084] Specifically, the result of the trend prediction is compared with the preset health threshold to generate a warning message. The health threshold includes two levels: yellow warning and red warning: When the predicted value is lower than the yellow warning threshold , the system generates a yellow warning, indicating that there are potential risks in the health status; When the predicted value is lower than the red warning threshold , the system generates a red warning, indicating that there are serious risks in this area and maintenance measures need to be taken immediately.

[0085] As a possible implementation method, the health thresholds and can be set based on the statistical analysis of historical monitoring data. For example, the yellow warning threshold can be set to the 20th percentile of the health index distribution, and the red warning threshold can be set to the 10th percentile.

[0086] In some embodiments, in order to improve the accuracy of the trend prediction, the system preprocesses the input data. For example, the time series of the health index is smoothed by the moving window average method to reduce the interference of instantaneous fluctuations on the prediction result. The width of the moving window can be dynamically adjusted according to the sampling frequency of the sensor, for example, set to 50 sampling points.

[0087] In another possible implementation, the system can generate a time series curve for the predicted comprehensive health indicators. This curve can intuitively display the changing trend of the health status over a period of time in the future. For example, for the vibration mode data in high-risk areas, a curve of the predicted root mean square value can be plotted to help determine whether there is a trend of continuous deterioration in the structure.

[0088] In this embodiment, in order to enhance the sensitivity of the early warning function, the system also combines priority trend analysis. If the priority of a certain sensor continuously rises at multiple consecutive moments, even if its comprehensive health indicator has not reached the early warning threshold, the system will generate an early warning. This early warning mechanism based on the changing trend of priority can effectively identify early abnormalities in the health status.

[0089] In some embodiments, the prediction results and early warning information can be displayed in real time through a visualization interface. For example, the yellow early warning area is highlighted in yellow, and the red early warning area is highlighted in red. At the same time, the specific predicted values and risk levels are displayed on the interface. For high-risk areas, the system can automatically generate a detailed health status report, and the report content includes the predicted health indicator values, change rates, and possible risk types.

[0090] Through trend prediction and early warning, the system can timely identify potential changes in the health status of the bridge and provide clear decision-making basis for managers. This function not only improves the efficiency of risk assessment but also significantly enhances the sensitivity and response ability of the system to changes in the health status.

[0091] S6. Cloud platform data storage and visualization display: Store the collected operation data and generated health indicators in the cloud platform, and display the real-time monitoring information of the bridge through a visualization interaction interface After completing trend prediction and early warning, the system needs to store the collected real-time data, health status analysis results, and prediction data in the cloud platform, and present the real-time health status and historical change trends of the bridge through a visualization display module. The use of the cloud platform not only provides unified storage of data but also supports cross-regional remote monitoring and subsequent data analysis. Through the visualization interface, users can quickly grasp the overall operation status of the bridge and the risk distribution of key components.

[0092] In this embodiment, the data storage of the cloud platform adopts a distributed structure to ensure data stability and scalability. Generally, all the original data collected by sensors and the processed comprehensive health indicators , priority , and trend prediction results will be synchronously stored in the cloud platform. To improve the data query efficiency, these data are indexed according to the time dimension and space dimension.

[0093] As an option, the stored data includes the following categories: Multi-modal raw data collected in real time, such as stress, vibration, and temperature data; Comprehensive health indicators of each sensor , and its calculation formula is:

[0094] Wherein, is the modal weight, is the modal eigenvalue; Dynamic priority , and its calculation formula is:

[0095] Wherein, represents the change rate of the health indicator; Trend prediction result , and its calculation formula is:

[0096] Statistical results of historical data analysis, such as the change range of each modal data and the distribution of health indicators.

[0097] In some embodiments, in order to reduce the storage burden, the system will downsample the raw data in the low-priority area and only retain the eigenvalue and the data at key time points. For example, for the vibration data in the low-priority area, only its maximum value, mean value, and change range are stored, and the specific sampling values at each moment do not need to be stored.

[0098] Based on the data storage, this embodiment provides a real-time visualization display function. Specifically, the visualization interface adopts a multi-level design and includes the following main contents: Global health status display: The overall health score S(t) is displayed through the three-dimensional model of the bridge, where:

[0099] The color identification of S(t) corresponds to the health status. For example, green indicates health, yellow indicates warning, and red indicates high risk.

[0100] Local risk distribution display: Display the risk heat map of the high-priority area, and generate a color gradient map based on the spatial distribution of the priority . The high-priority area is highlighted in red, and the low-priority area is displayed in green.

[0101] Trend curve analysis: For the high-risk area, provide the health indicator and the priority The time series curve visually shows the changing trend of the state.

[0102] As a possible implementation, the visualization interface also supports the interactive operation between the user and the data. For example, the user can click on a high-risk area in the heat map to view the detailed data of that area, including the location of the sensor, the historical changes of the health indicators, and the prediction results, etc.

[0103] In some embodiments, to facilitate the user's quick decision-making, the system will automatically generate an analysis report on the health status of the bridge. The report content includes the following information: The overall health score S(t) and the changing trend; The specific location and priority of the high-risk area of the change; The prediction results of the health indicators within a certain period in the future ; Possible damage types and recommended repair measures.

[0104] In a possible implementation, the cloud platform also supports the remote access function, and the user can view the real-time health status and historical analysis results of the bridge through a PC or a mobile device. This remote access function is especially suitable for the management of bridges across regions.

[0105] Through the data storage and visualization display of the cloud platform, the system realizes the unified management and efficient utilization of the bridge monitoring data. The multi-level design of the visualization interface not only ensures the detailed display of high-risk information but also avoids the redundant occupation of low-risk data, providing strong support for the scientific management of the bridge.

[0106] The present invention also provides a bridge detection system, including a processor and a memory connected to the processor. Program instructions are stored in the memory, and when the program instructions are executed by the processor, the processor executes the above-mentioned method for bridge structural health monitoring based on big data and cloud computing.

[0107] The present invention also provides a storage medium, in which program instructions are stored for executing the above-mentioned method for bridge structural health monitoring based on big data and cloud computing.

[0108] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirits of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A bridge structure health monitoring method based on big data and cloud computing, characterized in that: The following steps are involved: S1. Multimodal data acquisition: The multimodal sensor network deployed at key parts of the bridge collects bridge operation data in real time, including stress, vibration and temperature data, and transmits the collected data to the cloud computing platform through a wireless network; S2, data fusion and feature extraction: extracting modal features from the operating data, and fusing the data according to modal weights to generate comprehensive health indicators; S3, Dynamic priority allocation: Calculate sensor priority based on comprehensive health indicators and their change rates, and classify the monitoring areas by risk; S4. Graphics and image generation and optimization: Generate graphics and images from the whole to the part in layers based on the priority results, dynamically compress low-risk data and display high-risk data in detail; S5. Trend prediction and early warning: predict the future health status of the bridge through the time series prediction model, and generate early warning information based on the prediction results; S6. Cloud platform data storage and visual display: The collected operation data and generated health indicators are stored in the cloud platform, and the real-time monitoring information of the bridge is displayed through a visual interactive interface.

2. A bridge structure health monitoring method based on big data and cloud computing according to claim 1, characterized in that: The sensors include a stress sensor, a vibration sensor and a temperature sensor. The stress sensor is used to monitor the stress state of the bridge, the vibration sensor is used to monitor the dynamic response of the bridge, and the temperature sensor is used to monitor the influence of ambient temperature on the performance of bridge materials.

3. The bridge structure health monitoring method based on big data and cloud computing according to claim 1 is characterized in that: The following formula is used to generate the comprehensive health index in step S2: Among them, F i (t) is the comprehensive health index of the i-th sensor at time t, M is the number of modes, and w j is the modal weight, which is dynamically calculated based on the variation range of the modal data. The formula is: Among them, Var(D j ) is the variance of the j-th mode data, is the feature extraction value of the jth mode, specifically the maximum difference of the stress mode, the root mean square value of the vibration mode, and the mean value of the temperature mode.

4. The bridge structure health monitoring method based on big data and cloud computing according to claim 1 is characterized in that: The sensor priority in the dynamic priority allocation in step S3 is calculated by the following formula: Among them, P i (t) is the priority of the i-th sensor at time t, F i (t) is a comprehensive health indicator, is the rate of change of the comprehensive health index, and the calculation formula is: Wherein, Δt is the time interval, α, β are priority weight coefficients, satisfying α+β=1, where α>β.

5. The bridge structure health monitoring method based on big data and cloud computing according to claim 1 is characterized in that: In the step S4 of generating and optimizing the graphic image, the graphic image is generated in the following manner: Generate a global graph of the bridge's overall health score; Generate a heat map of the health status of local key components to show the health distribution of monitored components, where the color depth is inversely proportional to the health index value; Generate detailed time series curves of high-risk areas to show the historical trends of comprehensive health indicators.

6. The bridge structure health monitoring method based on big data and cloud computing according to claim 1 is characterized in that: In the cloud platform data storage and visualization in step S6, the cloud platform stores the real-time collected operation data, the calculated comprehensive health indicators and the historical health trend data, and displays the following information through a visualization interactive interface: The global health status of the bridge; Health status distribution of local key components; Time series trends of high-risk areas; Prediction results of the future health status of the bridge and related early warning information.

7. A bridge detection system, characterized in that: It includes a processor and a memory connected to the processor, wherein program instructions are stored in the memory, and when the program instructions are executed by the processor, the processor executes a bridge structure health monitoring method based on big data and cloud computing as described in any one of claims 1 to 6.

8. A storage medium, characterized in that: The storage medium stores program instructions, and the program instructions are used to execute a bridge structure health monitoring method based on big data and cloud computing as described in any one of claims 1 to 6.

Images