A control system for continuous construction of prestressed concrete cast-in-place

Abstract

The present invention relates to the technical field of general control or regulation systems, and specifically to a prestressed concrete cast-in-place continuous construction control system. First, the linear monitoring data and construction parameter monitoring data of the first N beam segments are obtained through a data monitoring module, and then the linear prediction data of the N+1 beam segment is obtained through the GA-grey prediction model of the linear control module. Then, the stress prediction data of the N+1 beam segment is obtained using the time-decayed LSTM model of the stress control module. Finally, the adaptive learning rate is dynamically adjusted through the adaptive correction algorithm of the construction parameter control module to correct the construction parameters of the N+1 beam segment. The system realizes precise linear control, stress control and construction parameter control through the mutual cooperation of the data monitoring module, the linear control module, the stress control module and the construction parameter control module, thereby ensuring that the implementation control module can complete reliable and stable construction control.

Description

Technical Field

[0001] The present invention relates to the technical field of general control or regulation systems, and in particular to a prestressed concrete cast-in-place continuous construction control system. Background Art

[0002] Construction control is crucial in bridge construction projects, directly affecting the safety, stability and service life of the project. Effective construction control can not only ensure that the linear shape and stress of the structure meet the design requirements, but also reduce the safety hazards and subsequent maintenance costs caused by construction errors. Construction control usually includes linear control and stress control. The former focuses on the geometric shape and position of the structure to ensure the accuracy of elevation, axis and deflection; the latter focuses on the stress state of the material to ensure its safety and durability under load. The combination of the two realizes the effective control of construction parameters during the construction process, thereby ensuring the continuous construction control of cast-in-place prestressed concrete.

[0003] However, the current construction control technology still has some obvious drawbacks. First, the current construction control is mainly based on BIM (Building Information Modeling) technology. Although BIM provides the advantages of visualization and information integration, in actual applications, its real-time response capabilities to dynamic changes in construction are insufficient, and it is often unable to update and adjust the model in time to reflect the actual situation on site. Secondly, when dealing with complex structures and variable working conditions, BIM technology has limitations in data processing capabilities and accuracy, resulting in reduced reliability of linear and stress predictions. In addition, existing construction control systems often lack effective error correction mechanisms and cannot make rapid adjustments to construction parameters, which may lead to error accumulation during the construction process, ultimately affecting the safety and quality of the project. Therefore, it is necessary to introduce more advanced construction control systems to improve the accuracy and adaptability of construction control.

[0004] Therefore, a prestressed concrete cast-in-place continuous construction control system is proposed. Summary of the invention

[0005] The object of the present invention is to provide a prestressed concrete cast-in-place continuous construction control system. First, the present invention realizes efficient prediction of linearity during construction by combining genetic algorithm and gray system theory through GA-grey prediction model, and significantly improves the accuracy of linearity control. Secondly, in order to obtain more accurate stress prediction, the present invention proposes a time-decayed LSTM model. By introducing the time decay mechanism, the model can give priority to recent data when processing time series data, thereby enhancing the reliability of linearity control. Finally, based on the adaptive correction algorithm proposed by the present invention, the setting of construction parameters can meet the needs of the project through real-time adjustment and dynamic optimization of the learning rate, thereby improving the reliability of the construction control system.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] Data monitoring module, used to obtain the linear monitoring data of the first N beam sections and the construction parameter monitoring data of the first N beam sections; the linear shape includes elevation, axis and deflection, and the construction parameters include concrete elastic modulus, concrete shrinkage and creep, main beam dead load, main beam second-stage dead load, section temperature difference, concrete internal and external temperature difference and effective prestress;

[0008] A linear control module is used to establish N multidimensional state matrices according to the linear monitoring data of the first N beam segments and the linear observation times, establish a GA-grey prediction model according to the N multidimensional state matrices, obtain the linear prediction data of the N+1th beam segment according to the GA-grey prediction model, and then control the linear shape of the N+1th beam segment; the linear observation times are divided into before the hanging basket travels, after the hanging basket travels, before the block concrete pouring, after the block concrete pouring, before the prestressing and after the prestressing according to the construction process;

[0009] A stress control module is used to establish a time decay LSTM model based on the construction parameter monitoring data of the first N beam segments as a time feature sequence and the stress monitoring data of the first N beam segments as a feature result, obtain the stress prediction data of the N+1th beam segment according to the time decay LSTM model, and control the stress of the N+1th beam segment;

[0010] The construction parameter control module is used to establish an adaptive correction algorithm based on the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment, the linear pre-construction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment, and use the adaptive correction algorithm to dynamically adjust the adaptive learning rate to control the construction parameters of the N+1th beam segment.

[0011] Furthermore, the cross-sectional temperature difference is fitted according to the temperature distribution index curve of the concrete box girder in the vertical direction of the cross-sectional area, and the temperature distribution index curve calculation formula is:

[0012] ;

[0013] in, is the temperature difference in the vertical direction of the calculation point, represents the maximum temperature difference in the vertical direction of the box girder, e represents the base of the natural logarithm, a represents the temperature parameter, and y represents the distance from the calculation point to the top surface of the box girder, and the calculation point is the installation point of the temperature sensor;

[0014] The temperature difference between the inside and outside of the concrete is calculated by the concrete hydration heat temperature. The concrete hydration heat temperature includes the maximum internal hydration temperature and the maximum surface hydration temperature. The calculation formula of the maximum internal hydration temperature is:

[0015] ;

[0016] The calculation formula for the maximum surface hydration temperature is:

[0017] ;

[0018] in, Indicates the maximum internal hydration temperature, express The adiabatic temperature rise of concrete at the age of represents the temperature drop coefficient, Indicates the concrete pouring temperature, represents the maximum surface hydration temperature, represents the average atmospheric temperature, Indicates the actual thickness of the concrete. represents the calculated thickness of concrete, It represents the difference between the maximum internal hydration temperature and the average temperature of the atmosphere.

[0019] Furthermore, establishing the GA-grey prediction model includes:

[0020] Establishing N multidimensional state matrices according to the linear monitoring data of the first N beam segments and the linear observation measurements;

[0021] According to the grey system theory, the GM (1,1) grey prediction model is established. By accumulating the state matrix of the first N beam segments and constructing the differential equation, the basic grey prediction model is obtained.

[0022] The parameters of the basic grey prediction model are optimized using a genetic algorithm, and selection, crossover and mutation operations are performed through a fitness function and an initial population, and the optimal parameter combination is obtained after multiple generations of iterative optimization;

[0023] The basic grey prediction model is modified according to the optimal parameter combination to obtain a GA-grey prediction model, and the GA-grey prediction model is used to predict the linear shape of the N+1th beam segment to obtain linear shape prediction data of the N+1th beam segment.

[0024] Furthermore, the stress monitoring data is acquired through strain sensors, the strain sensors are pre-buried in the stress control section of the box girder, and the strain sensors are distributed on the upper top plate and the lower bottom plate.

[0025] Furthermore, establishing the time decay LSTM model includes:

[0026] Collecting the construction parameter monitoring data of the first N beam segments and the stress monitoring data of the first N beam segments to form an input feature sequence and a target output sequence;

[0027] Normalize the input features and divide the data into training and validation sets;

[0028] Define the time-decay LSTM model architecture, including the input layer, LSTM layer, and output layer, and set the activation function and loss function;

[0029] Introduce a time decay factor in the LSTM layer and assign different weights based on the time step;

[0030] Inputting the construction parameter monitoring data of the first N beam segments into the trained LSTM model to obtain the stress prediction data of the N+1th beam segment;

[0031] The time decay factor calculation formula is:

[0032] ;

[0033] in, represents the time decay factor, t represents the time step, T represents the total time step, e represents the base of the natural logarithm, Indicates the correction factor.

[0034] Furthermore, the adaptive correction algorithm includes:

[0035] Step 1: Obtain the linear prediction data of the N+1th beam segment and the stress prediction data of the N+1th beam segment, define the number of iterations and convergence conditions, and start the algorithm;

[0036] Step 2: Calculate the linear error between the linear prediction data of the N+1th beam segment and the linear pre-construction data of the N+1th beam segment, and the stress error between the stress prediction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment;

[0037] Step 3: Determine whether the linear error and the stress error are within a preset threshold. If the linear error is less than the linear error threshold and the stress error is less than the stress error threshold, the adaptive correction algorithm converges and the iteration ends; otherwise, continue to step 4;

[0038] Step 4: Calculate the adaptive learning rate according to the linear error and the stress error, introduce the historical error to correct the linear prediction data of the N+1th beam segment and the stress prediction data of the N+1th beam segment, use the corrected linear prediction data and stress prediction data as new linear prediction data and linear observation data, add 1 to the number of iterations, and return to step 2;

[0039] Step 5: When the algorithm stops iterating, the construction parameters of the N+1th beam segment are corrected by using the adaptive learning rate of the last iteration.

[0040] Furthermore, the linear error includes an elevation error, an axis error and a deflection error, and the linear error threshold includes an elevation error threshold, an axis error threshold and a deflection error threshold.

[0041] Furthermore, the construction parameters of the N+1th beam section are corrected by a parameter correction formula, and the parameter correction formula is:

[0042] ;

[0043] in, represents the modified construction parameters of the N+1th beam segment, represents the pre-construction parameters of the N+1th beam segment, represents the adaptive learning rate of the N+1th beam segment.

[0044] Furthermore, controlling the linear shape of the N+1th beam segment specifically comprises: modifying the linear shape pre-construction data in the pre-construction plan of the N+1th beam segment into the linear shape prediction data of the N+1th beam segment; controlling the stress of the N+1th beam segment specifically comprises: modifying the stress pre-construction data in the pre-construction plan of the N+1th beam segment into the linear shape prediction data of the N+1th beam segment; controlling the construction parameters of the N+1th beam segment specifically comprises: modifying the pre-construction parameters of the N+1th beam segment into the modified construction parameters of the N+1th beam segment.

[0045] Furthermore, the construction control system also includes:

[0046] The implementation control module is used to implement the engineering of the N+1th beam segment according to the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment and the revised construction parameters of the N+1th beam segment.

[0047] Furthermore, the construction control system also includes:

[0048] The implementation control module is used to implement the engineering of the N+1th beam segment according to the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment and the revised construction parameters of the N+1th beam segment.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] 1. The present invention proposes a GA-grey prediction model, which combines the adaptive optimization capability of genetic algorithms with the high efficiency of grey system theory to improve the prediction accuracy and reliability of structural states during construction. Through multidimensional analysis of historical observation data, the model can effectively handle uncertainty and small sample problems, accurately capture complex nonlinear relationships, and more accurately achieve linear prediction, providing a reliable reference solution for linear control during construction.

[0051] 2. The present invention proposes a time-decay LSTM model, which introduces a time-decay mechanism to enable the model to more effectively capture the importance of recent data when processing time series data and gradually reduce the influence on historical data. The model uses a decay factor to weight the input features, so that the model pays more attention to the latest information related to the current construction status when predicting, thereby improving the accuracy of stress prediction.

[0052] 3. The present invention proposes an adaptive correction algorithm, which dynamically adjusts the correction strategy by iterating the error between the prediction result and the pre-construction plan to optimize the prediction performance of the model. The algorithm first defines the error threshold and adjusts the learning rate according to the current error situation, making corrections when the error is large and keeping stable when the error is small. This flexible adjustment mechanism enables the algorithm to adaptively control the setting of construction parameters, thereby realizing construction control more accurately and reliably. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A structural diagram of a prestressed concrete cast-in-place continuous construction control system provided by the present invention;

[0054] Figure 2 A schematic diagram of a flow chart of an adaptive correction algorithm provided by the present invention;

[0055] Figure 3 A schematic flow chart of a prestressed concrete cast-in-place continuous construction control system provided by the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] Embodiment 1

[0058] A prestressed concrete cast-in-place continuous construction control system, such as Figure 1 As shown, including:

[0059] Data monitoring module, used to obtain the linear monitoring data of the first N beam sections and the construction parameter monitoring data of the first N beam sections; the linear shape includes elevation, axis and deflection, and the construction parameters include concrete elastic modulus, concrete shrinkage and creep, main beam dead load, main beam second-stage dead load, section temperature difference, concrete internal and external temperature difference and effective prestress;

[0060] A linear control module is used to establish N multidimensional state matrices according to the linear monitoring data of the first N beam segments and the linear observation times, establish a GA-grey prediction model according to the N multidimensional state matrices, obtain the linear prediction data of the N+1th beam segment according to the GA-grey prediction model, and then control the linear shape of the N+1th beam segment; the linear observation times are divided into before the hanging basket travels, after the hanging basket travels, before the block concrete pouring, after the block concrete pouring, before the prestressing and after the prestressing according to the construction process;

[0061] A stress control module is used to establish a time decay LSTM model based on the construction parameter monitoring data of the first N beam segments as a time feature sequence and the stress monitoring data of the first N beam segments as a feature result, obtain the stress prediction data of the N+1th beam segment according to the time decay LSTM model, and control the stress of the N+1th beam segment;

[0062] The construction parameter control module is used to establish an adaptive correction algorithm based on the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment, the linear pre-construction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment, and use the adaptive correction algorithm to dynamically adjust the adaptive learning rate to control the construction parameters of the N+1th beam segment.

[0063] Furthermore, the cross-sectional temperature difference is fitted according to the temperature distribution index curve of the concrete box girder in the vertical direction of the cross-sectional area, and the temperature distribution index curve calculation formula is:

[0064] ;

[0065] in, is the temperature difference in the vertical direction of the calculation point, represents the maximum temperature difference in the vertical direction of the box girder, e represents the base of the natural logarithm, a represents the temperature parameter, and y represents the distance from the calculation point to the top surface of the box girder, and the calculation point is the installation point of the temperature sensor;

[0066] The temperature difference between the inside and outside of the concrete is calculated by the concrete hydration heat temperature. The concrete hydration heat temperature includes the maximum internal hydration temperature and the maximum surface hydration temperature. The calculation formula of the maximum internal hydration temperature is:

[0067] ;

[0068] The calculation formula for the maximum surface hydration temperature is:

[0069] ;

[0070] in, Indicates the maximum internal hydration temperature, express The adiabatic temperature rise of concrete at the age of represents the temperature drop coefficient, Indicates the concrete pouring temperature, represents the maximum surface hydration temperature, represents the average atmospheric temperature, Indicates the actual thickness of the concrete. represents the calculated thickness of concrete, It represents the difference between the maximum internal hydration temperature and the average temperature of the atmosphere.

[0071] The cross-sectional temperature difference will lead to nonlinear temperature distribution of the concrete box girder in the vertical direction, thus affecting the stress control of the box girder. The hydration heat temperature will cause the concrete to form an internal and external temperature difference, causing the concrete to strain. By obtaining the cross-sectional temperature difference and hydration heat temperature, the influence of weather and temperature factors on stress can be reduced.

[0072] Furthermore, establishing the GA-grey prediction model includes:

[0073] Establishing N multidimensional state matrices according to the linear monitoring data of the first N beam segments and the linear observation measurements;

[0074] According to the grey system theory, the GM (1,1) grey prediction model is established. By accumulating the state matrix of the first N beam segments and constructing the differential equation, the basic grey prediction model is obtained.

[0075] The parameters of the basic grey prediction model are optimized using a genetic algorithm, and selection, crossover and mutation operations are performed through a fitness function and an initial population, and the optimal parameter combination is obtained after multiple generations of iterative optimization;

[0076] The basic grey prediction model is modified according to the optimal parameter combination to obtain a GA-grey prediction model, and the GA-grey prediction model is used to predict the linear shape of the N+1th beam segment to obtain linear shape prediction data of the N+1th beam segment.

[0077] Specifically, the multidimensional state matrix ,in represents the observed value obtained at the jth observation of the i-th structure; by calculating the average value of all observed values ​​for each structure , converting the multidimensional state matrix into a one-dimensional time series , then the state matrix of the first N beam segments is , for the matrix Perform accumulation and generation to obtain the accumulated row vector , using the accumulated sequence Establishing the first order differential equation of grey model , a and b are calculated by the least square method, and then the basic grey prediction model is obtained; the genetic algorithm is used to optimize the parameters a and b in the basic grey prediction model, and a fitness function is selected ,in Represents the linear prediction data of the basic grey model of the Nth beam segment, Represents the linear observation data of the basic grey model of the Nth beam segment; randomly generates several initial combinations of (a, b) as the population and performs selection, crossover and mutation operations, repeats the selection, crossover and mutation process, and obtains the optimal parameter combination after multiple generations of evolution ; The optimal parameters obtained and Applied to the basic grey prediction model, the modified GA-grey prediction model is obtained, and the linear prediction data of the N+1th beam segment is restored.

[0078] The GA-grey prediction model uses the linear observation data and measurement information of the first N beam segments to construct a multidimensional state matrix, and optimizes the parameters of the grey model through genetic algorithms to achieve accurate prediction of the linear data of the N+1 beam segment. The advantages of this method are its efficiency and adaptability. It can fully explore the potential relationship of data in small sample conditions, reduce prediction errors, and flexibly process multidimensional data, making linear control during construction more reliable.

[0079] Furthermore, the monitoring stress is obtained through a strain sensor, the strain sensor is pre-buried in the stress control section of the box girder, and the strain sensor is distributed on the upper top plate and the lower bottom plate.

[0080] By embedding strain sensors in the stress control section of the box girder and distributing them on the upper top plate and the lower bottom plate, stress changes can be monitored in real time, thereby providing effective data support for stress control during the construction process.

[0081] Furthermore, establishing the time decay LSTM model includes:

[0082] Collecting the construction parameter monitoring data and stress monitoring data of the first N beam segments to form an input feature sequence and a target output sequence;

[0083] Normalize the input features and divide the data into training and validation sets;

[0084] Define the time-decay LSTM model architecture, including the input layer, LSTM layer, and output layer, and set the activation function and loss function;

[0085] Introduce a time decay factor in the LSTM layer and assign different weights based on the time step;

[0086] Inputting the construction parameter monitoring data of the first N beam segments into the trained LSTM model to obtain the stress prediction data of the N+1th beam segment;

[0087] The time decay factor calculation formula is:

[0088] ;

[0089] in, represents the time decay factor, t represents the time step, T represents the total time step, e represents the base of the natural logarithm, Indicates the correction factor.

[0090] Specifically, the construction parameter monitoring data and stress monitoring data of the first N beam sections are collected to form an input feature sequence and a target output sequence; the input data is normalized to be between 0 and 1, and the training set and the validation set are divided into a ratio of 7:3; the time-decayed LSTM model architecture is defined, including the input layer, LSTM layer, and output layer, where the activation function is ReLU and the loss function is the mean square error; the time decay factor is introduced in the LSTM layer, and the calculation formula of the forget gate is , the calculation formula for the information that needs to be updated in the input gate is , the calculation formula for updating the candidate cell state in the input gate is , the cell state update calculation formula is: , the data update calculation formula in the output gate is , the output calculation formula of the LSTM model in the output gate is in, () is the Sigmoid activation function, is the time decay factor, is the hidden state at the previous moment, is the input at the current moment, Update the weights for the forget gate, Update the bias for the forget gate, Update the weights for the information, is the information update bias, Update weights for cells, Bias for cell renewal, Update the weights for the information, is the information update bias, is a hyperbolic tangent function; the construction parameter data of the first N beam segments are input into the trained time-decayed LSTM model to predict the stress data of the N+1th beam segment, and the stress prediction data of the N+1th beam segment is obtained.

[0091] The time-decayed LSTM model is used for training and prediction, so as to accurately predict the stress state of the N+1th beam segment. This method effectively considers the dynamic change characteristics of the time series, improves the accuracy and reliability of the prediction, and provides important support for stress monitoring and control during the construction process.

[0092] Furthermore, the adaptive correction algorithm is as follows: Figure 2 As shown, including:

[0093] Step 1: Obtain the linear prediction data of the N+1th beam segment and the stress prediction data of the N+1th beam segment, define the number of iterations and convergence conditions, and start the algorithm;

[0094] Step 2: Calculate the linear error between the linear prediction data of the N+1th beam segment and the linear pre-construction data of the N+1th beam segment, and the stress error between the stress prediction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment;

[0095] Step 3: Determine whether the linear error and the stress error are within a preset threshold. If the linear error is less than the linear error threshold and the stress error is less than the stress error threshold, the adaptive correction algorithm converges and the iteration ends; otherwise, continue to step 4;

[0096] Step 4: Calculate the adaptive learning rate according to the linear error and the stress error, introduce the historical error to correct the linear prediction data of the N+1th beam segment and the stress prediction data of the N+1th beam segment, use the corrected linear prediction data and stress prediction data as new linear prediction data and linear observation data, add 1 to the number of iterations, and return to step 2;

[0097] Step 5: When the algorithm stops iterating, the construction parameters of the N+1th beam segment are corrected by using the adaptive learning rate of the last iteration.

[0098] Specifically, before the algorithm is started, the linear prediction data of the N+1th beam segment is provided by the linear control module, and the stress prediction data is provided by the stress control module; the linear error ,in Indicates linear pre-construction data, Indicates linear prediction data, stress error ,in Represents stress pre-construction data, Represents stress prediction data; determines whether the linear error and stress error are within the preset threshold. If the linear error is less than the linear error threshold and the stress error is less than the stress error threshold, the algorithm converges and the iteration ends, otherwise proceed to the next step; designs an adaptive learning rate ,in Represents the learning rate adjustment factor, and the new linear prediction data is updated as ,in represents the structural correction factor, represents the structural history error influence factor, Indicates the linear error of the Nth beam segment, the new stress prediction data, and the linear prediction data of the N+1th beam segment ,in is the stress correction factor, represents the stress history error influencing factor, Represents the stress error of the Nth beam segment; when the algorithm stops iterating, the construction parameters of the N+1th beam segment will be corrected using the adaptive learning rate of the last iteration.

[0099] By systematically acquiring the linear shape and stress prediction data of the N+1th beam segment and comparing it with the construction data, the prediction error can be effectively identified and corrected. By dynamically adjusting the adaptive learning rate and introducing historical errors, the algorithm can continuously optimize the learning rate to ensure continuous iteration before the convergence conditions are met. This method enhances the model's ability to adapt to dynamic changes in construction and provides accurate data support for the construction control of subsequent beam segments.

[0100] Furthermore, the linear error includes an elevation error, an axis error and a deflection error, and the linear error threshold includes an elevation error threshold, an axis error threshold and a deflection error threshold.

[0101] The set convergence conditions ensure comprehensive monitoring of the construction accuracy of the beam segment by subdividing the linear error into elevation error, axis error and deflection error and setting corresponding thresholds for each error.

[0102] Furthermore, the construction parameters of the N+1th beam section are corrected by a parameter correction formula, and the parameter correction formula is:

[0103] ;

[0104] in, represents the modified construction parameters of the N+1th beam segment, represents the pre-construction parameters of the N+1th beam segment, represents the adaptive learning rate of the N+1th beam segment.

[0105] By using adaptive learning rate and parameter correction formula, more accurate construction parameters can be obtained. By correcting the construction parameters, the control of the construction environment, construction technology and construction feed materials can be effectively achieved.

[0106] Furthermore, controlling the linear shape of the N+1th beam segment specifically comprises: modifying the linear shape pre-construction data in the pre-construction plan of the N+1th beam segment into the linear shape prediction data of the N+1th beam segment; controlling the stress of the N+1th beam segment specifically comprises: modifying the stress pre-construction data in the pre-construction plan of the N+1th beam segment into the linear shape prediction data of the N+1th beam segment; controlling the construction parameters of the N+1th beam segment specifically comprises: modifying the pre-construction parameters of the N+1th beam segment into the modified construction parameters of the N+1th beam segment.

[0107] Controlling the alignment, stress and construction parameter correction of the N+1 beam segment can effectively improve the accuracy and safety of construction control. By correcting the alignment pre-construction data to the alignment prediction data, the accuracy of the structural shape and the beauty of the cast-in-place construction are ensured. At the same time, correcting the stress pre-construction data makes the stress state more consistent with the actual working conditions and improves the structural safety. The correction of the pre-construction parameters optimizes the construction plan, thereby ensuring that the alignment and stress can be implemented according to the corresponding prediction values ​​during the construction process.

[0108] Furthermore, the construction control system further includes:

[0109] The implementation control module is used to implement the engineering of the N+1th beam segment according to the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment and the revised construction parameters of the N+1th beam segment.

[0110] The implementation control module can effectively integrate the linear prediction data, stress prediction data and revised construction parameters of the N+1th beam segment to ensure good engineering quality during the construction process and ensure the beauty and life of the cast-in-place construction.

[0111] Through the close integration of data monitoring module, linear control module, stress control module and construction parameter control module, comprehensive monitoring and control of the continuous construction process of prestressed concrete cast-in-place is achieved. The data monitoring module ensures accurate linearity and construction parameters, providing a basis for subsequent predictions; the linear control module uses the GA-grey prediction model to improve the prediction accuracy and ensure the beautiful structure during the construction process; the stress prediction module uses the time-decayed LSTM model to make the stress prediction closer to the actual working conditions; the construction parameter control module dynamically adjusts the adaptive learning rate to continuously optimize the construction parameters, ensuring that the construction parameters can be reasonably applied to subsequent construction, thereby achieving accurate and reliable construction control.

[0112] Based on a prestressed concrete cast-in-place continuous construction control system provided in this embodiment, its process is as follows: Figure 3 As shown, including:

[0113] Obtain the linear monitoring data of the first N beam sections and the construction parameter monitoring data of the first N beam sections;

[0114] Establishing N multidimensional state matrices according to the linear monitoring data and linear observation times of the first N beam segments, establishing a GA-grey prediction model according to the N multidimensional state matrices, obtaining linear prediction data of the N+1th beam segment according to the GA-grey prediction model, and then controlling the linear shape of the N+1th beam segment;

[0115] Based on the construction parameter monitoring data of the first N beam segments as a time feature sequence and the stress monitoring data of the first N beam segments as a feature result, a time decay LSTM model is established, and the stress prediction data of the N+1th beam segment is obtained according to the time decay LSTM model to control the stress of the N+1th beam segment;

[0116] An adaptive correction algorithm is established based on the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment, the linear pre-construction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment. The adaptive correction algorithm is used to dynamically adjust the adaptive learning rate to control the construction parameters of the N+1th beam segment.

[0117] The engineering implementation of the N+1 beam section is carried out according to the linear prediction data of the N+1 beam section, the stress prediction data of the N+1 beam section and the revised construction parameters of the N+1 beam section.

[0118] Embodiment 2

[0119] A certain bridge is composed of a prestressed concrete continuous beam bridge. During the construction of the bridge, a prestressed concrete cast-in-place continuous construction control system proposed in the present invention is adopted. In order to verify the construction control effect of the system, the construction process of each beam section is compared with the result predicted by BIM technology.

[0120] As shown in Table 1, the linear errors of some beam segments based on the linear prediction results of the GA-grey prediction model and the monitoring results are as follows: the elevation error, axis error and deflection error of the first beam segment are 4mm, 1mm and 3mm respectively, the elevation error, axis error and deflection error of the second beam segment are 0mm, 0mm and 2mm respectively, and the elevation error, axis error and deflection error of the third beam segment are 2mm, 1mm and 2mm respectively. As can be seen from Table 1, the linear prediction results based on the GA-grey prediction model are better than those based on BIM. The main reason is that the GA-grey prediction model utilizes the global search capability of the genetic algorithm and the simple and effective performance of the grey model, so that the model can provide more accurate linear prediction results when facing a complex construction environment, thereby ensuring the reliability of linear control.

[0121] Table 1. Partial comparison data of linear prediction results and linear monitoring results of GA-grey prediction model, BIM technology

[0122]

[0123] As shown in Table 2, the stress errors of the top plate of the first three beam sections between the stress prediction results based on time decay LSTM and the monitoring results are 0.46MPa, 0.17MPa and 0.09MPa respectively. In comparison, the stress errors of the top plate of the first three beam sections between the stress prediction results based on BIM and the monitoring results are 0.65MPa, 1.29MPa and 1.72MPa respectively. The results show that the stress prediction results based on time decay LSTM are also better than those based on BIM, and as the working conditions progress, the time decay LSTM is closer to the monitoring results. The reason for this phenomenon is that the time decay LSTM can effectively capture the dynamic changes and trends in the time series data, and by giving higher weights to the most recent observation data, it can more accurately reflect the stress state of the beam section, thus providing a guarantee for the accuracy of the system's stress control.

[0124] Table 2. Some comparative data of stress prediction results and stress monitoring results of GA-grey prediction model and BIM technology

[0125]

[0126] The stress parameters before and after the correction by the adaptive learning rate are shown in Table 3. The third beam segment is predicted according to the construction before and after the correction, and the linear error and stress error before and after the correction are calculated. The results are shown in Table 4. As can be seen from the table, the error after correction is smaller than the error before correction, which shows that the adaptive learning rate can effectively utilize the linear error and stress error, further correct the learning rate through continuous iteration, and guide the construction parameter control of subsequent construction, thereby reducing the overall error caused by the construction process.

[0127] Table 3. Data of the construction parameters of the third beam section before and after correction

[0128]

[0129] Table 4. Comparison of linear error and stress error before and after the correction of construction parameters of the third beam section

[0130]

[0131] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A prestressed concrete cast-in-place continuous construction control system, characterized in that: include: Data monitoring module, used to obtain the linear monitoring data of the first N beam sections and the construction parameter monitoring data of the first N beam sections; the linear shape includes elevation, axis and deflection, and the construction parameters include concrete elastic modulus, concrete shrinkage and creep, main beam dead load, main beam second-stage dead load, section temperature difference, concrete internal and external temperature difference and effective prestress; A linear control module is used to establish N multidimensional state matrices according to the linear monitoring data of the first N beam segments and the linear observation times, establish a GA-grey prediction model according to the N multidimensional state matrices, obtain the linear prediction data of the N+1th beam segment according to the GA-grey prediction model, and then control the linear shape of the N+1th beam segment; the linear observation times are divided into before the hanging basket travels, after the hanging basket travels, before the block concrete pouring, after the block concrete pouring, before the prestressing and after the prestressing according to the construction process; A stress control module is used to establish a time decay LSTM model based on the construction parameter monitoring data of the first N beam segments as a time feature sequence and the stress monitoring data of the first N beam segments as a feature result, obtain the stress prediction data of the N+1th beam segment according to the time decay LSTM model, and control the stress of the N+1th beam segment; The construction parameter control module is used to establish an adaptive correction algorithm based on the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment, the linear pre-construction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment, and use the adaptive correction algorithm to dynamically adjust the adaptive learning rate to control the construction parameters of the N+1th beam segment.

2. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: The cross-sectional temperature difference is fitted according to the temperature distribution index curve of the concrete box girder in the vertical direction of the cross-sectional area. The temperature distribution index curve calculation formula is: ； in, is the temperature difference in the vertical direction of the calculation point, represents the maximum temperature difference in the vertical direction of the box girder, e represents the base of the natural logarithm, a represents the temperature parameter, and y represents the distance from the calculation point to the top surface of the box girder, and the calculation point is the installation point of the temperature sensor; The temperature difference between the inside and outside of the concrete is calculated by the concrete hydration heat temperature. The concrete hydration heat temperature includes the maximum internal hydration temperature and the maximum surface hydration temperature. The calculation formula of the maximum internal hydration temperature is: ； The calculation formula for the maximum surface hydration temperature is: ； in, Indicates the maximum internal hydration temperature, express The adiabatic temperature rise of concrete at the age of represents the temperature drop coefficient, Indicates the concrete pouring temperature, represents the maximum surface hydration temperature, represents the average atmospheric temperature, Indicates the actual thickness of the concrete. represents the calculated thickness of concrete, It represents the difference between the maximum internal hydration temperature and the average temperature of the atmosphere.

3. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: Establishing the GA-grey prediction model includes: Establishing N multidimensional state matrices according to the linear monitoring data of the first N beam segments and the linear observation measurements; According to the grey system theory, the GM (1,1) grey prediction model is established. By accumulating the state matrix of the first N beam segments and constructing the differential equation, the basic grey prediction model is obtained. The parameters of the basic grey prediction model are optimized using a genetic algorithm, and selection, crossover and mutation operations are performed through a fitness function and an initial population, and the optimal parameter combination is obtained after multiple generations of iterative optimization; The basic grey prediction model is modified according to the optimal parameter combination to obtain a GA-grey prediction model, and the GA-grey prediction model is used to predict the linear shape of the N+1th beam segment to obtain linear shape prediction data of the N+1th beam segment.

4. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: The stress monitoring data is obtained through strain sensors, which are pre-buried in the stress control section of the box girder and distributed on the upper top plate and the lower bottom plate.

5. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: Establishing the time decay LSTM model includes: Collecting the construction parameter monitoring data of the first N beam segments and the stress monitoring data of the first N beam segments to form an input feature sequence and a target output sequence; Normalize the input features and divide the data into training and validation sets; Define the time-decay LSTM model architecture, including the input layer, LSTM layer, and output layer, and set the activation function and loss function; Introduce a time decay factor in the LSTM layer and assign different weights based on the time step; Inputting the construction parameter monitoring data of the first N beam segments into the trained LSTM model to obtain the stress prediction data of the N+1th beam segment; The time decay factor calculation formula is: ； in, represents the time decay factor, t represents the time step, T represents the total time step, e represents the base of the natural logarithm, Indicates the correction factor.

6. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: The adaptive correction algorithm includes: Step 1: Obtain the linear prediction data of the N+1th beam segment and the stress prediction data of the N+1th beam segment, define the number of iterations and convergence conditions, and start the algorithm; Step 2: Calculate the linear error between the linear prediction data of the N+1th beam segment and the linear pre-construction data of the N+1th beam segment, and the stress error between the stress prediction data of the N+1th beam segment and the stress pre-construction data of the N+1th beam segment; Step 3: Determine whether the linear error and the stress error are within a preset threshold. If the linear error is less than the linear error threshold and the stress error is less than the stress error threshold, the adaptive correction algorithm converges and the iteration ends; otherwise, continue to step 4; Step 4: Calculate the adaptive learning rate according to the linear error and the stress error, introduce the historical error to correct the linear prediction data of the N+1th beam segment and the stress prediction data of the N+1th beam segment, use the corrected linear prediction data and stress prediction data as new linear prediction data and linear observation data, add 1 to the number of iterations, and return to step 2; Step 5: When the algorithm stops iterating, the construction parameters of the N+1th beam segment are corrected by using the adaptive learning rate of the last iteration.

7. A prestressed concrete cast-in-place continuous construction control system according to claim 6, characterized in that: The linear error includes an elevation error, an axis error and a deflection error, and the linear error threshold includes an elevation error threshold, an axis error threshold and a deflection error threshold.

8. A prestressed concrete cast-in-place continuous construction control system according to claim 6, characterized in that: The construction parameters of the N+1th beam section are corrected by a parameter correction formula, and the parameter correction formula is: ； in, represents the modified construction parameters of the N+1th beam segment, represents the pre-construction parameters of the N+1th beam segment, represents the adaptive learning rate of the N+1th beam segment.

9. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: Controlling the linear shape of the N+1th beam segment specifically includes: modifying the linear shape pre-construction data in the pre-construction plan of the N+1th beam segment into the linear shape prediction data of the N+1th beam segment; controlling the stress of the N+1th beam segment specifically includes: modifying the stress pre-construction data in the pre-construction plan of the N+1th beam segment into the linear shape prediction data of the N+1th beam segment; controlling the construction parameters of the N+1th beam segment specifically includes: modifying the pre-construction parameters of the N+1th beam segment into the modified construction parameters of the N+1th beam segment.

10. A prestressed concrete cast-in-place continuous construction control system according to claim 1, characterized in that: The construction control system also includes: The implementation control module is used to implement the engineering of the N+1th beam segment according to the linear prediction data of the N+1th beam segment, the stress prediction data of the N+1th beam segment and the revised construction parameters of the N+1th beam segment.

Images