Construction method of mountainous large-span deck-type open-web arch bridge

Abstract

The application discloses a kind of mountainous area large-span deck type open-web arch bridge construction methods, it is related to bridge engineering construction technical field, including the following steps: erecting support system at bridge site and preloading to support system;Establish construction whole process monitoring system, collect multi-source monitoring data in construction process, input multi-source monitoring data into pre-trained deep learning model, obtain support deformation prediction result and arch ring stress prediction result;At least a part of preloading load is maintained on support system, and the main arch ring concrete is poured in multiple segments according to the order of symmetry from arch foot to arch top, according to the segment construction order, the preloading load corresponding to the segment to be poured is removed after pouring the segment concrete, and construction control is carried out based on support deformation prediction result and arch ring stress prediction result. It aims at solving the technical problems of arch ring linear control difficulty, high construction risk, poor support stability and the like.

Description

Technical Field

[0001] This invention relates to the field of bridge engineering construction technology, and in particular to a construction method for a long-span, upper-bearing, hollow arch bridge in mountainous areas. Background Technology

[0002] Due to their advantages such as strong span, beautiful appearance, and reasonable stress distribution, the open-spandrel arch bridge is increasingly widely used in the construction of highways in mountainous areas. However, the complex terrain, inconvenient transportation, and sensitive ecological environment in mountainous areas pose enormous challenges to the construction of long-span arch bridges.

[0003] The existing construction techniques mainly have the following problems: Separating the pre-stressing of the support frame from the arch ring pouring process: In the traditional method, the arch ring is poured after the pre-stressing of the support frame is unloaded. The elastic rebound of the support frame will cause additional stress to the already poured concrete, affecting the accuracy of the arch ring's shape.

[0004] The determination of preloading load is unscientific: Traditional preloading usually directly loads to 120%, which results in large loads, high costs, and fails to effectively distinguish between inelastic and elastic deformation of the support, thus providing weak guidance for setting the precamber.

[0005] Insufficient construction control precision: The lack of a real-time monitoring and feedback control mechanism for displacement and stress throughout the entire arch construction process makes it difficult to detect and correct construction deviations in a timely manner, posing a risk of structural damage.

[0006] Overall stability risk of the support structure: For long-span arch bridges, especially side arch supports, instability is prone to occur under asymmetrical or segmented unloading conditions, and there is a lack of targeted construction control methods.

[0007] Therefore, developing a construction method that can improve alignment control accuracy, enhance construction stability, and reduce construction risks and safety costs is of great significance for promoting the construction of long-span arch bridges in mountainous areas. Summary of the Invention

[0008] The purpose of this invention is to provide a construction method for a long-span, upper-bearing, hollow-web arch bridge in mountainous areas, aiming to solve technical problems such as difficulty in controlling the arch alignment, high construction risks, and poor support stability.

[0009] To achieve the above objectives, the present invention provides the following solution: This invention provides a construction method for a long-span, open-arch bridge in mountainous areas, comprising the following steps: A support system was erected at the bridge site and the support system was pre-stressed. A construction process monitoring system is established, multi-source monitoring data is collected during the construction process, and the multi-source monitoring data is input into a pre-trained deep learning model to obtain the deformation prediction results of the support and the stress prediction results of the arch ring. Maintain at least a portion of the preload on the support system, pour the main arch ring concrete in multiple segments in a symmetrical order from the arch foot to the arch crown, and pour the concrete of the segment after removing the preload corresponding to the segment to be poured in accordance with the segment construction sequence, and carry out construction control based on the deformation prediction results of the support and the stress prediction results of the arch ring.

[0010] Preferably, the multi-source monitoring data includes support displacement data, arch stress data, ambient temperature data, and concrete strength data. The multi-source monitoring data is aggregated to form a construction big data set, and the deep learning model is trained based on the construction big data set to establish a nonlinear mapping relationship between the monitoring data and support deformation and arch stress.

[0011] Preferably, the preloading is graded preloading, which records the deformation data of the support under each load level, calculates the inelastic and elastic deformation of the support based on the deformation data, and determines the construction precamber based on the elastic deformation; the deep learning model also optimizes and corrects the construction precamber based on the deformation data collected during the graded preloading process.

[0012] Preferably, construction control based on the predicted deformation results of the support and the predicted stress results of the arch ring includes: comparing the predicted deformation results of the support and the predicted stress results of the arch ring with preset warning values; when the prediction results trigger the warning conditions, suspending construction and taking adjustment measures.

[0013] Preferably, the preset warning values ​​include displacement warning values ​​and stress warning values. The displacement warning value is when the deviation between the measured displacement and the theoretical value exceeds 15 mm or the differential settlement between adjacent measuring points exceeds 5 mm. The stress warning value is when any monitoring point experiences tensile stress exceeding 0.5 MPa or compressive stress reaching 80% of the design strength value.

[0014] Preferably, the deep learning model includes a deformation prediction sub-model and a stress prediction sub-model. The deformation prediction sub-model is used to predict the displacement change trend of the support in the subsequent construction stage based on current and historical monitoring data, and the stress prediction sub-model is used to predict the stress evolution trend of the arch ring in the subsequent construction stage based on current and historical monitoring data.

[0015] Preferably, the support system is a composite support system, including side supports and middle section supports. The side supports include strip foundations, columns, horizontal bracing, load-bearing beams, longitudinal beams, square timber, and bottom formwork. The middle section supports include strip foundations, columns, horizontal bracing, load-bearing beams, truss beams, distribution beams, modular supports, curved square steel pipes, square timber, and bottom formwork.

[0016] Preferably, the construction process monitoring system includes support displacement monitoring points located at the top of the columns and in the middle of the modular support span, arch displacement monitoring points located at the arch feet, quarter span, half span and three-quarter span sections on the upstream and downstream sides of the main arch ring, and stress sensors located at the top plate, bottom plate and web positions of the arch feet section, quarter span section and arch crown section of the main arch ring.

[0017] Preferably, after the arch ring is closed and the concrete strength reaches 90% of the design strength, the support system is dismantled symmetrically and evenly from the arch top to the arch foot.

[0018] Preferably, after the supports are removed, the arch columns are constructed using a symmetrical construction method on the main arch ring. After the construction of the arch columns and cap beams is completed, the precast prestressed beams are symmetrically hoisted and the bridge deck system is constructed.

[0019] The present invention achieves the following technical effects compared to the prior art: This invention provides a construction method for a long-span, upper-bearing, hollow-web arch bridge in mountainous areas. It enables advanced prediction of construction status, and through deep learning models analyzing multi-source monitoring data, it accurately predicts the deformation of the support structure and the stress changes in the arch ring during subsequent construction stages. This upgrades traditional passive monitoring to prediction-driven active control, significantly improving the accuracy and timeliness of construction control. It also improves the accuracy of the arch ring alignment; the simultaneous pressing and pouring process eliminates the adverse effects of support structure elastic rebound on the arch ring alignment, and the deep learning model's optimization and correction of the pre-camber further enhances the accuracy of alignment control. Furthermore, it reduces construction costs and risks; graded pre-pressing accurately distinguishes between inelastic and elastic deformation, and the deep learning model provides scientific decision-making based on big data analysis, reducing the amount of pre-pressing materials used and reliance on manual experience, thus lowering construction risks. A multi-dimensional early warning and pre-control mechanism compares the deep learning prediction results with preset early warning values, achieving a shift from post-event discovery to pre-event prediction, effectively preventing the risks of support collapse and concrete cracking. Finally, it enhances support stability; the composite support system combined with the simultaneous pressing and pouring construction method avoids excessive local stress on the support structure, ensuring the stability of the construction platform. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A construction flowchart of the construction method for a long-span, upper-bearing, hollow-arch bridge in mountainous areas provided by the present invention; Figure 2A schematic diagram of the structure of the upper-bearing hollow arch bridge in the construction method of the large-span upper-bearing hollow arch bridge in mountainous areas provided by the present invention. Figure 3 A segmented diagram of the arch ring casting in the construction method of a long-span, upper-bearing, hollow-arch bridge in mountainous areas provided by the present invention; Figure 4 This is a schematic diagram of the support system in the construction method of a long-span, upper-bearing, hollow-arch bridge in mountainous areas provided by the present invention. In the diagram: 1. Arch ring; 2. Columns on the arch; 3. Bridge deck; 4. Support system. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] The purpose of this invention is to provide a construction method for a long-span, upper-bearing, hollow-web arch bridge in mountainous areas, aiming to solve technical problems such as difficulty in controlling the arch alignment, high construction risks, and poor support stability.

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] This application provides a construction method for a long-span, upper-bearing, hollow-arch bridge in mountainous areas, such as... Figures 1-4 As shown, the process includes the following steps: erecting a support system 4 at the bridge site and preloading the support system 4; establishing a monitoring system for the entire construction process, collecting multi-source monitoring data during the construction process, inputting the multi-source monitoring data into a pre-trained deep learning model, and obtaining the deformation prediction results of the support system and the stress prediction results of the arch ring 1; maintaining at least a portion of the preload on the support system 4, pouring the main arch ring 1 concrete in multiple segments in a symmetrical order from the arch foot to the arch top, simultaneously unloading the corresponding preload of each segment when pouring each segment, and carrying out construction control based on the deformation prediction results of the support system and the stress prediction results of the arch ring 1.

[0026] In its implementation, this invention addresses core technical challenges in the construction of long-span, open-arch bridges in mountainous areas, including difficulties in controlling the arch ring 1's alignment, high construction risks, poor support stability, and the reliance on manual experience and lack of predictive capabilities in traditional construction control methods. It proposes an intelligent construction method that deeply couples deep learning prediction with a simultaneous pressure-and-cast process. This method comprises three core innovations: First, the simultaneous pressure-and-cast process involves maintaining the preload on the support during arch ring 1 concrete pouring. Following the segmental construction sequence, the preload corresponding to the segment to be poured is removed before pouring the concrete for that segment, eliminating the adverse effects of support elastic rebound on the arch ring 1's alignment. Second, a full-process construction monitoring system collects real-time multi-source monitoring data such as displacement, stress, and temperature by deploying sensors at key locations on the support and arch ring 1. Third, deep learning prediction-driven construction control inputs the collected multi-source monitoring data into a pre-trained deep learning model to obtain predicted results of support deformation and arch ring 1 stress, thereby proactively identifying and controlling construction risks.

[0027] This embodiment takes a 148m main span reinforced concrete open-spandrel arch bridge on a mountain highway as the engineering background. The following describes each step in detail with reference to the attached drawings.

[0028] By adopting the above technical solution and organically combining deep learning prediction with the simultaneous pressing and pouring process, the construction process has been upgraded to intelligent control from monitoring-response to prediction-pre-control. This effectively solves the technical problems of insufficient accuracy in the control of the arch ring 1 alignment, high construction risk and lack of forward-looking control in traditional construction methods.

[0029] In some embodiments of this application, at least a portion of the preload is maintained on the support system 4. The main arch ring 1 concrete is poured in multiple segments in a symmetrical order from the arch foot to the arch crown. According to the segment construction sequence, the preload corresponding to the segment to be poured is removed before pouring the concrete for that segment. Construction control is carried out based on the deformation prediction results of the support and the stress prediction results of the arch ring 1. The specific implementation process is as follows: In its implementation, the core technological innovation of this invention is simultaneous pressure-and-pouring. Essentially, it maintains at least a portion of the preload on the support structure throughout the concrete pouring process of the arch ring 1, ensuring the support structure remains under compressive deformation throughout the entire pouring period. This fundamentally eliminates the adverse effects of elastic rebound caused by preload unloading on the arch ring 1's alignment, a problem inherent in traditional methods. In traditional construction, the arch ring 1 concrete is poured after the support structure is preloaded and unloaded. At the moment of unloading, the support structure rebounds elastically, changing the formwork elevation. Since the concrete has not yet gained sufficient strength, this elastic rebound generates additional tensile stress in the poured concrete, potentially leading to cracking and affecting the accuracy of the arch ring 1's final bridge alignment. The simultaneous pressure-and-pouring process proposed in this invention maintains the elastic deformation of the support structure stable throughout the pouring process, preventing abrupt changes in the formwork elevation and ensuring the geometric stability of the formwork system during concrete pouring. This eliminates the adverse effect of elastic rebound from a technological perspective.

[0030] Specifically, the implementation of the simultaneous pressing and casting process includes the following key steps: The first step is the segmentation and preloading zoning planning of arch ring 1. After the graded preloading of the support frame is completed, based on the structural characteristics and construction organization design of arch ring 1, the main arch ring 1 is divided into multiple casting segments along the longitudinal direction of the bridge. Simultaneously, the preloading load is zoned according to the segments. Each casting segment corresponds to a specific preloading load zone on the support frame. After removing the preloading load corresponding to the segment to be cast, the concrete for that segment is poured. The segment division needs to comprehensively consider factors such as the stress characteristics of arch ring 1, concrete supply capacity, and support bearing capacity. Generally, it is divided symmetrically along the longitudinal direction of arch ring 1 to ensure that the segments on the left and right sides correspond one-to-one, so as to implement symmetrical casting construction. In this embodiment, the main arch ring 1 is divided into 8 segments along the bridge direction, numbered sequentially from the arch foot at one end to the arch crown and then to the arch foot at the other end as segment 1, segment 2, segment 3, segment 4, segment 5, segment 6, segment 7, and segment 8. Among them, segment 1 and segment 8, segment 2 and segment 7, segment 3 and segment 6, and segment 4 and segment 5 are symmetrical segment pairs.

[0031] The second step is to determine the pouring sequence. The pouring sequence adopts a symmetrical approach from the arch foot to the arch crown, meaning that the segments in the arch foot area are constructed first, gradually progressing towards the arch crown, and finally the arch crown closure segment is poured. This sequence is determined based on the mechanical characteristics of arch bridges: the arch foot area is the most sensitive part of arch ring 1 to stress and deformation; prioritizing the construction of the arch foot segments can establish fixed-end constraints at the arch foot as early as possible, providing stable mechanical boundary conditions for subsequent segment construction; at the same time, symmetrical progress from the arch foot to the arch crown ensures that the forces on both sides of arch ring 1 remain symmetrical, avoiding torsional deformation or lateral displacement caused by asymmetrical loading. In this embodiment, the construction sequence is: the first round constructs the first and eighth segments (arch foot segments); the second round constructs the second and seventh segments; the third round constructs the third and sixth segments; and the fourth round constructs the fourth and fifth segments (arch crown closure segment). The third step is cyclical construction with simultaneous pressing and pouring. Each round of construction follows a standardized cyclical process of prediction, unloading, and pouring monitoring, with the deep learning model playing a core role in the prediction stage of each cycle. (1) Prediction process: Before each round of construction, the current multi-source monitoring data (including support displacement, arch ring 1 stress, ambient temperature, concrete strength, etc.) are input into the deep learning model to obtain the support deformation prediction results and arch ring 1 stress prediction results for the next construction stage. The deformation prediction sub-model predicts the rebound displacement of the support after the preload of the current segment is unloaded and the displacement change trend of each monitoring point. The stress prediction sub-model predicts the stress evolution trend of the key section of arch ring 1 during unloading and pouring. The prediction results are compared with the preset warning values. If the prediction results do not trigger the warning conditions, the next unloading operation is carried out. If the prediction results trigger the warning conditions, construction is suspended, the reasons are analyzed and adjustment measures are taken. Construction can only continue after the adjustment is completed and the prediction is reconfirmed to be safe.

[0032] (2) Unloading Phase: After confirming safety through deep learning model prediction, the preload corresponding to the current segment to be poured is symmetrically unloaded. It is particularly important to emphasize that only the preload corresponding to the current segment to be poured is unloaded, while the preloads of the remaining unpoured segments are retained. The core significance of this operation is that the preload retained on the support continues to exert pressure on the support, keeping the elastic deformation of the support unchanged and the formwork elevation stable, thereby ensuring the geometric accuracy of the formwork system during concrete pouring. Taking the first round of construction as an example, the preloads corresponding to the first and eighth segments are symmetrically unloaded, while the preloads of the remaining segments are retained. At this time, most areas of the support are still in a state of preload deformation. Only the first and eighth segments experience a slight elastic rebound due to the removal of the preload. However, since concrete is poured immediately after the preload in these areas is removed, the self-weight of the newly poured concrete quickly replaces the supporting effect of the original preload. The actual stress state of the support hardly changes abruptly, and the formwork elevation changes very little. During the unloading process, the rebound displacement of the support and the stress change in the arch foot area are monitored in real time to confirm that the monitoring data are within the prediction range.

[0033] (3) Pouring process: Immediately after unloading, symmetrically pour concrete for the current segment. The concrete's self-weight gradually replaces the preload, allowing for a smooth transition in the stress state of the support, avoiding the adverse effects of the support's elastic rebound on the concrete caused by unloading all components before pouring in traditional processes. During the pouring process, continuously monitor the support displacement and the stress at the bottom of arch ring 1. The deep learning model runs synchronously, comparing the real-time monitoring data with the prediction results to confirm the consistency between the actual deformation and stress and the predicted values. If the deviation between the measured data and the predicted values ​​exceeds the allowable range, promptly analyze the cause and take corresponding measures.

[0034] (4) Confirmation Phase: After the concrete pouring of the current segment is completed, the deep learning model updates its prediction based on the latest monitoring data. After confirming that the current construction status meets the safety requirements, it enters the next cycle. Subsequent cycles repeat the above prediction, unloading, pouring, and monitoring cycle, completing the construction of the second and seventh segments, the third and sixth segments, and the fourth and fifth segments of the closure segment in sequence. Throughout the entire construction process of simultaneous pressing and pouring, the deep learning model continuously provides advanced prediction capabilities, upgrading the traditional passive control mode of construction first, monitoring later, and then reacting to an active control mode of prediction first, risk assessment, construction later, and real-time verification. The unloading and pouring operations of each construction segment are carried out under the prediction and verification of the deep learning model, realizing intelligent and precise control of the entire construction process.

[0035] Compared with traditional construction methods, the simultaneous pressing and pouring process has the following significant technical advantages: First, the impact of elastic rebound is eliminated. In traditional processes, elastic rebound after the preload of the support causes changes in the formwork elevation, and concrete is poured on the rebounded formwork, resulting in the arch 1's alignment deviating from the design alignment after completion. However, the simultaneous pressing and pouring process maintains the preload, keeping the formwork elevation stable during concrete pouring, resulting in the arch 1's alignment closer to the design alignment after completion. In this embodiment, the maximum deflection of the arch 1 after completion is 33.7mm, far less than the control standard of L / 1000 (L being the main span length), i.e., 148mm, and significantly better than the bridge alignment accuracy achieved by traditional processes.

[0036] Secondly, it improves the stress state of the support structure. In traditional processes, the load on the support structure suddenly decreases after preloading and unloading, and then gradually increases again during re-pouring. The support structure undergoes repeated loading-unloading-reloading processes, resulting in an unstable stress state. In the simultaneous loading and pouring process, the unloading of the preloading load and the application of the concrete's self-weight are carried out simultaneously. The stress state of the support structure achieves a smooth transition from the preloading load to the concrete's self-weight, avoiding abrupt and repeated load changes and improving the overall stress state of the support structure. In particular, for side arch supports, the stability is significantly improved under asymmetrical or segmented unloading conditions.

[0037] Third, the preloading load requirement is reduced. Since the preloading load in the simultaneous pressing and casting process does not require complete unloading and reloading, the inelastic deformation of the support is essentially eliminated during the staged preloading phase. Theoretically, the conventional 110% preloading load can be reduced to 80%, decreasing the amount of preloading material used and construction procedures, thus lowering construction costs and risks. In this embodiment, the staged preloading data shows that inelastic deformation is essentially complete after the 60% to 80% load range. In actual construction, the preloading load can be dynamically adjusted based on the prediction results of the deep learning model, achieving optimization of the preloading load while ensuring construction safety.

[0038] In some preferred embodiments of this application, the multi-source monitoring data includes support displacement data, arch ring 1 stress data, ambient temperature data, and concrete strength data. The multi-source monitoring data is aggregated to form a construction big data set, and the deep learning model is trained based on the construction big data set to establish a nonlinear mapping relationship between the monitoring data and the support deformation and arch ring 1 stress.

[0039] In the specific implementation process, the acquisition of multi-source monitoring data and the construction of a construction big data set are the foundation for training and predicting deep learning models. Support displacement data is acquired through displacement sensors deployed at key nodes of the support, reflecting the vertical deformation and horizontal displacement of the support under various load levels and construction stages. Stress data for arch ring 1 is acquired through strain sensors embedded in key sections of arch ring 1, reflecting the stress state of arch ring 1 during construction. Ambient temperature data is acquired through temperature sensors deployed at the construction site, as temperature changes cause thermal expansion and contraction of concrete and temperature deformation of the support, significantly affecting displacement and stress monitoring data. Concrete strength data is obtained through on-site curing test blocks under the same conditions or through rebound testing, reflecting the development of the mechanical properties of arch ring 1 concrete at different ages. The above multi-source monitoring data are aligned and fused according to a unified timestamp and spatial coordinates to form a construction big data set. This dataset not only includes construction data from this project but can also integrate construction monitoring data from similar projects to increase data diversity and the model's generalization ability. The deep learning model is trained on this large construction dataset. It automatically learns the complex nonlinear mapping relationship between monitoring data and support deformation and arch stress through a multi-layer neural network. During training, historical monitoring data is used as input features, and the corresponding actual deformation and stress values ​​are used as labels. The backpropagation algorithm is employed to optimize the model parameters. After the model is trained, in actual construction, only the current multi-source monitoring data needs to be input to quickly obtain the predicted results of support deformation and arch stress, providing a scientific basis for construction control decisions.

[0040] By adopting the above technical solution, the fusion of multi-source monitoring data and the construction of a construction big data set provide sufficient data support for the deep learning model, enabling the model to learn the complex correlation between support deformation and arch stress and multiple factors, establish accurate nonlinear mapping relationships, and significantly improve prediction accuracy and the scientific nature of construction control.

[0041] In some preferred embodiments of this application, the preloading is staged preloading, recording the deformation data of the support under each level of load, calculating the inelastic and elastic deformation of the support based on the deformation data, and determining the construction precamber based on the elastic deformation; the deep learning model also optimizes and corrects the construction precamber based on the deformation data collected during the staged preloading process.

[0042] In the specific implementation process, graded preloading is a key step in determining the construction precamber and also provides important training and validation data for the deep learning model. This invention employs a four-level loading scheme, with loading levels sequentially at 60%, 80%, 100%, and 120% of the preload value. After each loading level, the displacement of key nodes of the support is measured, and displacement-load curves are plotted. Data analysis reveals that a clear inflection point appears in the 60% to 80% load range, after which the displacement and load exhibit a essentially linear relationship, indicating that the inelastic deformation of the support has been largely completed in this range. Based on this, inelastic and elastic deformation can be accurately separated: inelastic deformation is eliminated through preloading, while elastic deformation serves as the primary basis for setting the construction precamber. In this embodiment, the total preloading load is 1.1 times the sum of the self-weight of the arch ring 1 and the construction load, i.e., 3742.07 tons. The loading is carried out in four stages: 60% load is 2041.13 tons, 80% load is 2721.50 tons, 100% load is 3401.88 tons, and 120% load is 4082.25 tons. Deformation data collected at each stage during the graded preloading process are incorporated into the construction big data set. The deep learning model learns the deformation patterns of the support under different load levels based on this data, and combines this with measured deformation data from subsequent construction stages to optimize and correct the construction pre-camber determined based on elastic deformation calculations. Specifically, the model can identify the influence of factors such as temperature effects, material creep, and node relaxation, which were not considered in the elastic deformation calculations, on the pre-camber, and output corrected suggested pre-camber values, making the template elevation setting more accurate. In this embodiment, the pre-camber of the construction bridge, which has been optimized and corrected by a deep learning model, shows a significantly better match with the final bridge alignment than the pre-camber determined solely based on elastic deformation calculations.

[0043] By adopting the above technical solution, graded preloading provides a scientific basis for setting the pre-camber during construction. The deep learning model optimizes and corrects the pre-camber based on the graded preloading data, further improving the accuracy of the arch ring 1 alignment control and realizing the pre-camber optimization from experience-based determination to data-driven optimization.

[0044] In some preferred embodiments of this application, construction control based on prediction results includes: comparing the predicted deformation results of the support and the predicted stress results of the arch ring 1 with preset warning values; and suspending construction and taking adjustment measures when the prediction results trigger the warning conditions.

[0045] In practical implementation, construction control based on deep learning prediction results is the core element of this invention for achieving proactive control. Unlike traditional methods that only compare measured data with theoretical values ​​after the fact, this invention uses a deep learning model to predict the deformation of the support and the stress of the arch ring 1 in subsequent construction stages, enabling the identification of risks and the implementation of countermeasures before abnormal conditions actually occur.

[0046] The specific control process is as follows: Before the start of each construction segment, the current multi-source monitoring data is input into the deep learning model to obtain the predicted deformation results of the support and the predicted stress results of the arch ring 1 for the next construction stage; the predicted results are compared with the preset warning values. If the predicted results are within the range of the warning values, construction can proceed normally; if the predicted results trigger the warning conditions, that is, if it is predicted that there may be abnormalities in the subsequent construction, construction is suspended in the current construction stage, the cause is analyzed and adjustment measures are taken, such as slowing down the pouring speed, adding temporary counterweights, and reinforcing the support. Construction can only continue after the adjustment is completed and the safety is confirmed by re-prediction.

[0047] In this embodiment, the predictive control process runs continuously during the construction of the eight segments of the arch ring, namely the first, second, third, fourth, fifth, sixth, seventh, and eighth segments. Step 1: Pour the first and eighth segments, and simultaneously unload the preload in the corresponding areas of the first and eighth segments; Step 2: Pour the second and seventh segments, and simultaneously unload the preload in the corresponding areas; Step 3: Pour the third and sixth segments, and simultaneously unload the preload in the corresponding areas; Step 4: Pour the fourth and fifth segments, and finally pour the arch closure segment. The entire pouring process maintains symmetry and balance, and each step strictly follows the predictive-pre-control process. The unloading and pouring operations of each segment are predicted and verified by a deep learning model to ensure that the entire construction process is under control.

[0048] By adopting the above technical solution and comparing the prediction results with the warning values ​​through deep learning, the construction risks can be identified and proactively controlled in advance. This transforms construction safety management from post-event handling to pre-event prevention, significantly improving the safety and controllability of construction.

[0049] In some preferred embodiments of this application, the preset warning values ​​include displacement warning values ​​and stress warning values. The displacement warning value is when the deviation between the measured displacement and the theoretical value exceeds 15 mm or the differential settlement between adjacent measuring points exceeds 5 mm. The stress warning value is when any monitoring point experiences tensile stress exceeding 0.5 MPa or compressive stress reaching 80% of the design strength value.

[0050] In the specific implementation process, the specific values ​​of each early warning threshold need to be determined in conjunction with the actual engineering situation and specification requirements. The displacement threshold of 15mm is determined based on the usual accuracy requirements for construction monitoring of long-span arch bridges and the structural characteristics of this project. For an arch bridge with a main span of 148m, the allowable deviation is approximately ±L / 1500, or ±98.7mm. The 15mm early warning threshold is much smaller than the allowable deviation, providing sufficient safety margin. The differential settlement threshold of 5mm reflects the control standard for deformation coordination between adjacent measuring points. When the differential settlement between two adjacent points exceeds 5mm, it may indicate uneven stress on the support or a risk of local instability. The tensile stress threshold of 0.5MPa is determined based on the design value of concrete tensile strength. Tensile stress should be avoided as much as possible during construction. The compressive stress ratio of 80% is based on the safety reserve of concrete under compressive stress. When the compressive stress reaches 80% of the design strength value, the safety margin is already relatively limited. It should be particularly noted that in the deep learning prediction mode, the above early warning thresholds are not only used for the evaluation of measured data but also for the advanced evaluation of predicted data. Deep learning models can predict the displacement and stress values ​​of each monitoring point in the next construction phase. If the predicted values ​​are close to but have not yet reached the warning threshold, preventive measures can be taken in advance to avoid the actual occurrence of risks.

[0051] The above technical solutions have been verified through engineering practice, ensuring both construction safety and efficiency. The combination of deep learning prediction and early warning thresholds upgrades the early warning mechanism from post-event alarm to pre-event prediction, significantly improving the timeliness of risk prevention.

[0052] In some preferred embodiments of this application, the deep learning model includes a deformation prediction sub-model and a stress prediction sub-model. The deformation prediction sub-model is used to predict the displacement change trend of the support in the subsequent construction stage based on current and historical monitoring data, and the stress prediction sub-model is used to predict the stress evolution trend of the arch ring 1 in the subsequent construction stage based on current and historical monitoring data.

[0053] In practical implementation, the deep learning model adopts a dual-model architecture, with deformation prediction and stress prediction handled by dedicated sub-models. This facilitates specialized training and optimization of each sub-model for different prediction tasks. The deformation prediction sub-model uses historical time-series data from the support displacement monitoring points as its primary input, while integrating auxiliary features such as ambient temperature data, current construction stage information, and preload status. It outputs the predicted displacement values ​​for each monitoring point in one or more subsequent construction stages. This sub-model can employ a time-series prediction architecture based on Long Short-Term Memory (LSTM) networks or Gated Recurrent Units (GRUs) to fully utilize the time-dependent characteristics of displacement data and capture the evolution of support deformation across different construction stages. The stress prediction sub-model uses historical time-series data from the stress monitoring point of arch ring 1 as its primary input, while integrating auxiliary features such as concrete age data, pouring progress information, and temperature data. It outputs the predicted stress values ​​for each monitoring section in subsequent construction stages. This sub-model can also employ an LSTM or GRU architecture, with added stress distribution constraints in the output layer to ensure that the prediction results conform to the mechanical equilibrium conditions of the arch bridge structure. The two sub-models share a common feature extraction network at the bottom, enabling joint learning and cross-utilization of features from multiple data sources while maintaining the specialized predictive capabilities of their respective outputs. In this embodiment, both the deformation prediction sub-model and the stress prediction sub-model are trained based on a large construction dataset. The input features of the training data include dimensions such as support displacement, arch stress, ambient temperature, concrete strength, and construction stage. The output labels are the measured displacement and stress values ​​of subsequent construction stages. The model training adopts a rolling update strategy, meaning that as construction progresses, newly collected monitoring data is continuously added to the training set, and the model is periodically retrained to adapt to changes in the construction status and maintain prediction accuracy.

[0054] By adopting the above technical solution, the division of labor and cooperation between the deformation prediction sub-model and the stress prediction sub-model realizes the professional prediction of the deformation of the support and the stress of the arch ring 1. The shared feature extraction network improves the data utilization efficiency, and the rolling update strategy ensures the continuous adaptability of the model, providing accurate and reliable prediction support for construction control.

[0055] In some preferred embodiments of this application, the support system 4 is a composite support system, including side supports and middle section supports. The side supports include strip foundations, columns, horizontal bracing, load-bearing beams, longitudinal beams, square timber, and bottom formwork. The middle section supports include strip foundations, columns, horizontal bracing, load-bearing beams, truss beams, distribution beams, modular supports, curved square steel pipes, square timber, and bottom formwork.

[0056] In the specific implementation process, the composite support system is a differentiated support scheme designed according to the characteristics of mountainous terrain and the stress characteristics of arch bridge structure. Since the main arch ring 1 of a long-span, upper-bearing, open-web arch bridge is usually of variable cross-section, with a large cross-sectional height at the arch foot and a small cross-sectional height at the arch crown, different support structure forms are required in different sections. The side supports are mainly arranged on both sides of the arch ring 1, bearing the self-weight of the larger cross-section near the arch foot and the construction load. Its structure from bottom to top is as follows: strip foundations provide a stable support surface, columns serve as the main vertical load-bearing components, horizontal bracing enhances the overall stiffness and lateral stability between columns, load-bearing beams distribute the load at the top of the columns laterally, longitudinal beams transfer the load along the longitudinal direction of the bridge, square timber serves as a load distribution layer to adjust the elevation, and the bottom formwork directly supports the concrete of the arch ring 1. The intermediate support is located in the middle area of ​​the arch ring 1. The cross section of the arch ring 1 is relatively small in this area, but the span is large. Therefore, truss beams (such as Bailey beams) are used as the main longitudinal load-bearing components. Distribution beams are laid on them in sequence to distribute the load laterally. Modular supports (such as disc-lock supports) provide the ability to adjust the elevation. Curved square steel pipes adapt to the curved shape of the arch ring 1. Square timber and bottom formwork complete the final elevation adjustment and concrete support. In this embodiment, the side support adopts a C30 concrete strip foundation, on which spiral steel pipe columns with a diameter of 630mm and a wall thickness of 10mm are installed. Horizontal bracing and diagonal bracing are set between the columns using channel steel. Three-section I40a I-beams are installed on the top of the columns as load-bearing beams. Double or triple I32a longitudinal beams, 9cm x 9cm square timber, and 1.5cm bamboo plywood are laid on top of the beams as the bottom formwork. The foundation and steel pipe columns of the middle section support are the same as those of the side support. Double I40a load-bearing beams are installed on the top of the columns, on which Bailey beams are assembled. I10 distribution beams, disc-locked supports, transverse I10 I-beams, and longitudinal 50mm x 3mm double-section arc-shaped square steel pipes are laid on the Bailey beams in sequence. Finally, 9cm x 9cm square timber and 1.5cm bamboo plywood are laid on top.

[0057] Using the above technical solution, the composite support system is designed differently for the stress and deformation characteristics of different sections. The side supports are mainly controlled by strength and stiffness, while the middle support is mainly controlled by crossing capacity and linear adaptability. This not only ensures the safety and stability of the construction platform, but also achieves accurate fitting of the curve of the arch ring 1.

[0058] In some preferred embodiments of this application, the construction process monitoring system includes support displacement monitoring points located at the top of the columns and in the middle of the modular support span, arch ring 1 displacement monitoring points located at the arch feet, quarter span, half span and three-quarter span sections on the upstream and downstream sides of the main arch ring 1, and stress sensors located at the top plate, bottom plate and web of the arch feet section, quarter span section and arch crown section of the main arch ring 1.

[0059] In the specific implementation process, the reasonable layout of the monitoring system is the foundation for realizing data collection and deep learning model prediction throughout the construction process. Displacement monitoring points for the supports are set at the top of the columns and the mid-span of the modular supports, as these two locations are the most sensitive areas to support deformation: the top of the columns directly reflects the vertical compression and tilting deformation of the supports, while the mid-span of the modular supports reflects the maximum deflection of the support system 4. Displacement monitoring points for the arch ring 1 are arranged longitudinally along the main arch ring 1 at four key sections: the arch foot, the quarter-span, the half-span, and the three-quarter-span, and are symmetrically arranged on both upstream and downstream sides. Stress sensors are installed at the arch foot section, the quarter-span section, and the arch crown section of the main arch ring 1, with sensors positioned at the top slab, bottom slab, and web within each section. In this embodiment, the displacement monitoring points for the arch ring 1 are automatically measured using a reflective prism in conjunction with a total station, and the stress sensors are automated data acquisition using vibrating wire strain gauges connected to a comprehensive testing instrument. All monitoring data is transmitted in real-time to the construction control center through a unified data interface. After data cleaning and preprocessing, the data is stored in the construction big data set for use by the deep learning model. The sampling frequency can be dynamically adjusted according to the construction stage, increasing to once per minute during key processes such as unloading and pouring, and decreasing to once per hour during routine maintenance.

[0060] By adopting the above technical solution, and by reasonably deploying monitoring points at key parts of the support and arch 1, comprehensive collection of multi-dimensional data throughout the construction process was achieved, providing high-quality training and prediction data for the deep learning model and supporting intelligent decision-making in construction control.

[0061] In some preferred embodiments of this application, after the arch ring 1 is closed and the concrete strength reaches 90% of the design strength, the support system 4 is symmetrically and evenly removed from the arch top to the arch foot.

[0062] In the specific implementation process, the dismantling of the main arch ring 1 is a critical transition stage in the construction process. At this time, arch ring 1 changes from a supported state to a self-supporting state, and the stress system undergoes a fundamental change. The timing of the dismantling must ensure that the concrete strength of arch ring 1 reaches more than 90% of the design strength after closure, at which point arch ring 1 has sufficient self-supporting capacity. The dismantling sequence adopts a symmetrical and uniform removal from the arch crown to the arch foot, determined based on the mechanical characteristics of arch bridges: the arch crown area is the most stress-sensitive part of arch ring 1. Removing the supports in the arch crown area first allows arch ring 1 to complete the system transformation in the most unfavorable stress area first, facilitating timely detection and handling of anomalies; at the same time, symmetrical dismantling from the arch crown to the arch foot ensures symmetrical stress on both sides of arch ring 1, avoiding torsional deformation or lateral displacement caused by asymmetrical unloading. During the dismantling process, a deep learning model continuously runs, predicting the deformation and stress changes of arch ring 1 at each stage of dismantling based on real-time monitoring data, ensuring that the dismantling process is safe and controllable. In this embodiment, the maximum compressive stress of the rear arch ring 1 after the lowering of the frame is 16.64 MPa and the maximum deflection is 21.1 mm, both of which are within the safe range, verifying the safety and rationality of the lowering scheme.

[0063] By adopting the above technical solution, selecting reasonable timing and sequence for dismantling the frame, and using a deep learning model to predict and control the dismantling process, the safety and stability of the arch ring 1 system conversion are ensured.

[0064] In some preferred embodiments of this application, after the supports are removed, the arch columns 2 are constructed on the main arch ring 1 using a symmetrical construction method. After the construction of the arch columns 2 and the cap beam is completed, the precast prestressed beams are hoisted symmetrically and the bridge deck 3 is constructed.

[0065] In the specific implementation process, the construction of the arch support column 2 was carried out symmetrically. The construction sequence was either symmetrical from the arch foot to the arch crown or an optimized sequence of arch crown first and then arch foot. Regardless of the sequence, symmetrical construction on both sides of the support column was ensured to avoid asymmetrical loads on the arch ring 1. In this embodiment, the arch support column 2 was constructed symmetrically from the arch foot to the arch crown, and the stress changes of the arch ring 1 were continuously monitored during the construction process. During the superstructure construction stage, 13m precast prestressed T-beams were symmetrically hoisted to ensure balanced loads on both sides of the arch ring 1. Subsequently, the construction of wet joints, bridge deck paving, and guardrails and other ancillary facilities was carried out to complete the construction of the entire bridge. After the bridge was completed, the alignment of the arch ring 1 was finally re-measured. The overall alignment was smooth, and the deviations of all monitoring points were within the design allowable range. The final maximum deflection of the arch ring 1 was 33.7mm, which was much less than the control standard of L / 1000, i.e., 148mm, verifying the effectiveness and advancement of the invention.

[0066] By adopting the above technical solution, through symmetrical construction and standardized subsequent procedures, the safety of the arch ring 1 during the construction of the arch column 2 and the superstructure, as well as the overall quality of the completed bridge, were ensured.

[0067] In summary, this invention, through the deep integration of deep learning models and the simultaneous pressing and pouring process, constructs a new construction control paradigm of data acquisition, intelligent prediction, and active control. It systematically solves the technical problems of difficult control of the arch ring 1 alignment, high construction risk, and lack of forward-looking control in the construction of long-span upper-bearing hollow arch bridges in mountainous areas. It realizes intelligent and precise control of the construction process, resulting in a smooth bridge alignment and a safe and reliable structure, and has good engineering application value.

[0068] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A construction method for a long-span, upper-bearing, hollow-arch bridge in mountainous areas, characterized in that: Includes the following steps: A support system was erected at the bridge site and the support system was pre-stressed. A construction process monitoring system is established, multi-source monitoring data is collected during the construction process, and the multi-source monitoring data is input into a pre-trained deep learning model to obtain the deformation prediction results of the support and the stress prediction results of the arch ring. Maintain at least a portion of the preload on the support system, pour the main arch ring concrete in multiple segments in a symmetrical order from the arch foot to the arch crown, and pour the concrete of the segment after removing the preload corresponding to the segment to be poured in accordance with the segment construction sequence, and carry out construction control based on the deformation prediction results of the support and the stress prediction results of the arch ring.

2. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: The multi-source monitoring data includes support displacement data, arch stress data, ambient temperature data, and concrete strength data. The multi-source monitoring data is aggregated to form a construction big data set. The deep learning model is trained based on the construction monitoring data of similar projects and the construction big data set to establish a nonlinear mapping relationship between the monitoring data and support deformation and arch stress.

3. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: The preloading is a graded preloading, which records the deformation data of the support under each load level, calculates the inelastic and elastic deformation of the support based on the deformation data, and determines the construction precamber based on the elastic deformation; the deep learning model also optimizes and corrects the construction precamber based on the deformation data collected during the graded preloading process.

4. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: Construction control based on the predicted deformation of the support and the predicted stress of the arch ring includes: comparing the predicted deformation of the support and the predicted stress of the arch ring with preset warning values; when the prediction results trigger the warning conditions, suspending construction and taking adjustment measures.

5. The construction method for a long-span, upper-bearing, hollow-arch bridge in mountainous areas according to claim 4, characterized in that: The preset warning values ​​include displacement warning values ​​and stress warning values. The displacement warning value is when the deviation between the measured displacement and the theoretical value exceeds 15 mm or the differential settlement between adjacent measuring points exceeds 5 mm. The stress warning value is when any monitoring point experiences tensile stress exceeding 0.5 MPa or compressive stress reaching 80% of the design strength value.

6. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: The deep learning model includes a deformation prediction sub-model and a stress prediction sub-model. The deformation prediction sub-model is used to predict the displacement change trend of the support in the subsequent construction stage based on current and historical monitoring data. The stress prediction sub-model is used to predict the stress evolution trend of the arch ring in the subsequent construction stage based on current and historical monitoring data.

7. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: The support system is a composite support system, including side supports and middle section supports. The side supports include strip foundations, columns, horizontal bracing, load-bearing beams, longitudinal beams, square timber, and bottom formwork. The middle section supports include strip foundations, columns, horizontal bracing, load-bearing beams, truss beams, distribution beams, modular supports, curved square steel pipes, square timber, and bottom formwork.

8. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: The construction process monitoring system includes support displacement monitoring points located at the top of the columns and in the middle of the modular support spans, arch displacement monitoring points located at the arch feet, quarter span, half span, and three-quarter span sections on the upstream and downstream sides of the main arch ring, as well as stress sensors located at the top plate, bottom plate, and web positions of the arch feet section, quarter span section, and arch crown section of the main arch ring.

9. The construction method for a long-span, open-arch bridge in mountainous areas according to claim 1, characterized in that: After the arch ring is closed and the concrete strength reaches 90% of the design strength, the support system is dismantled symmetrically and evenly from the top of the arch to the bottom of the arch.

10. The construction method for a long-span, upper-bearing, hollow-arch bridge in mountainous areas according to claim 1, characterized in that: After the scaffolding is removed, the arch columns are constructed using a symmetrical construction method on the main arch ring. After the construction of the arch columns and cap beams is completed, the precast prestressed beams are hoisted symmetrically and the bridge deck system is constructed.

Images