Super-thick wall concrete crack control method

Abstract

The invention belongs to the technical field of building construction, and provides a super-thick wall concrete crack control method, which comprises the following steps: acquiring a wall three-dimensional solid model, dividing the wall three-dimensional solid model into grid units to obtain a finite element model, monitoring the pressure of a heat dissipation steel pipe by using a pressure sensor network, and judging whether the bond stress of concrete to the steel pipe is abnormal or not; if the grid units are normal, temperature parameters of the grid units are calculated through a temperature sensor network, whether a cooling circulation system is started or not is judged, if yes, temperature control targets of the grid units are simulated and calculated, and prediction time periods are marked for the grid units according to the cooling parameters of the cooling circulation system; constructing a hybrid prediction model to predict the unit temperature of the grid units in the prediction time period to obtain a predicted unit temperature sequence of each grid unit, dynamically calculating an adjustment value of the cooling parameter through a PID algorithm, and performing feedback iterative calculation according to the adjustment value to obtain an actual adjustment value of each cooling parameter; and the cooling parameter is adjusted according to the actual adjustment value.

Description

Technical Field

[0001] The invention belongs to the technical field of building construction, in particular to a method for controlling cracks in ultra-thick wall concrete. Background Art

[0002] In large-scale construction projects, the construction of ultra-thick concrete walls is a challenging task. With the continuous development of construction technology, the performance requirements for ultra-thick concrete wall structures are increasing, especially crack control, which is directly related to the safety, durability, and usability of the building.

[0003] Existing technologies have not built a multi-sensor network covering the entire wall area. After the ultra-thick wall concrete is poured, the hydration heat causes the internal temperature to rise sharply. Temperature difference stress is the main cause of cracks. Traditional methods rely on empirical formulas to estimate temperature development and lack real-time and accurate monitoring of the three-dimensional temperature field inside the wall. Cooling measures are often passively initiated only after the temperature is abnormal, resulting in delayed cooling and uncontrolled temperature gradients. Existing technologies lack real-time monitoring of the bond strength of heat dissipation steel pipes and are unable to promptly detect hidden dangers such as debonding and voids at the interface between the steel pipe and concrete, resulting in reduced heat dissipation efficiency or even failure of the steel pipe. The bond strength of the heat dissipation steel pipe directly affects the cooling efficiency and structural safety. In traditional construction, there is a lack of effective monitoring methods after the heat dissipation steel pipe is installed, and it is impossible to distinguish between bond strength abnormalities such as concrete voids and sensor failures, which can easily lead to cooling system failure or unnecessary grouting due to misjudgment.

[0004] Existing cooling systems mostly use cooling water circulation with fixed flow and temperature, or rely on manual experience to adjust cooling parameters. A closed-loop control mechanism based on real-time data has not been established. For example, the time difference between cooling water reaching steel pipes in different areas, namely the lag effect, is not incorporated into the control model, resulting in the cooling effect and temperature changes being out of sync. Traditional temperature prediction relies on a single physical model or data model, which does not fully integrate the thermal conductivity characteristics of concrete with real-time monitoring data. For complex boundary conditions, a single model cannot accurately capture temperature dynamics, resulting in large prediction errors and delayed adjustment of cooling parameters.

[0005] On the other hand, existing crack control technologies often only focus on local temperature changes and cooling effects, while ignoring the performance of the wall as a whole structure. Existing technologies lack the full application of three-dimensional solid models and finite element models of the wall, and cannot comprehensively and accurately simulate the temperature and stress field changes of the wall under different working conditions, making it difficult to formulate a scientific and reasonable crack control plan.

[0006] In view of the above problems, the present invention proposes a method for controlling cracks in ultra-thick wall concrete. Summary of the Invention

[0007] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0008] The technical solution adopted by the present invention to solve the technical problem is: a method for controlling cracks in ultra-thick wall concrete, comprising:

[0009] The 3D solid model of the wall is divided into grid cells to obtain a finite element model. A pressure sensor network is constructed to monitor the pressure of the heat dissipation steel pipe to determine whether the bond strength of the concrete to the steel pipe is normal. If normal, the wall temperature data is collected through the temperature sensor network and the temperature parameters of each grid cell are calculated based on the finite element model.

[0010] Specifically, a three-dimensional solid model of the wall is obtained, the three-dimensional solid model of the wall is meshed, the three-dimensional solid model of the wall is evenly discretized into tetrahedral mesh units, the vertex coordinates of each mesh unit are marked as node coordinates, all node coordinates are integrated to form a node coordinate set, and a finite element model is obtained;

[0011] The pressure value of each pressure collection point is collected in real time. For each pressure collection point, the pressure change value between the current moment and the previous sampling moment is calculated. A cumulative flag value is set for each pressure collection point and its initial value is 0. If the pressure value of any pressure collection point at the current moment is less than the pressure threshold, or the pressure change value of the pressure collection point is greater than the change standard value, the corresponding cumulative flag value is self-incremented. Otherwise, the corresponding cumulative flag value is reset to 0.

[0012] The locations where pressure sensors are arranged on the heat dissipation steel pipes are marked as pressure collection points, and the cumulative mark values ​​of each pressure collection point are obtained. For any pressure collection point, if the cumulative mark value is less than the preset stable mark value and there is an abnormal mark at the pressure collection point, the abnormal mark of the pressure collection point is cleared; if the cumulative mark value is greater than or equal to the preset stable mark value, an abnormal mark is generated for the pressure collection point;

[0013] If there are no abnormal marks at all pressure collection points, it is judged that the bond strength of concrete on the steel pipe is normal;

[0014] The system determines whether to start the cooling cycle system based on the temperature parameters of the grid cells. If so, the temperature control target of each grid cell is simulated and calculated. The prediction period is marked for each grid cell based on the cooling parameters of the cooling cycle system. A hybrid prediction model is constructed to predict the predicted unit temperature sequence of the grid cell within the prediction period.

[0015] Specifically, the diameter of the heat dissipation steel pipe is obtained to calculate the flow area of ​​the steel pipe, and the cooling water flow rate is calculated in combination with the cooling water flow rate in the cooling parameters at the current moment;

[0016] Based on any grid cell, the shortest distance between the grid cell and the heat dissipation steel pipe is calculated, and the position on the heat dissipation steel pipe with the shortest distance to the grid cell is marked as the cooling control point of the grid cell. The length of the heat dissipation steel pipe between the cooling control point and the cooling water inlet is obtained. The required time for the cooling water to reach the cooling control point is calculated based on the cooling water flow rate. The period starting from the current time and lasting for the required time is marked as the predicted period of the grid cell.

[0017] Construct a hybrid prediction model that includes physical branches and data branches, calculate dynamic weights based on the prediction accuracy of the physical branches and data branches, perform weighted fusion of the prediction results of the physical branches and data branches based on the dynamic weights, calculate the predicted unit temperature of the grid cells within the prediction period, and integrate the predicted unit temperatures according to the time series to obtain the predicted unit temperature sequence of the grid cells;

[0018] Among them, physical branches and data branches include:

[0019] In the physics branch, the cooling effect is equivalent to the cooling and heat dissipation term embedded in the finite element heat conduction equation in the finite element model, and the finite element heat conduction equation is modified. For the modified finite element heat conduction equation, the predicted temperature parameters of each grid cell in the prediction period are solved by the explicit difference method, and the physical prediction unit temperature of the grid cell in the prediction period is calculated;

[0020] In the data branch, an LSTM network is constructed, and the cell temperature sequence of each grid cell is obtained to train the LSTM network. The cell temperature of the grid cell is predicted using the trained LSTM network to obtain the data-predicted cell temperature of the grid cell within the prediction period.

[0021] Among them, the unit temperature series is obtained as follows:

[0022] Obtain the temperature parameters of each grid cell for data processing, calculate the temperature of the center point of each grid cell, and mark it as the cell temperature of the grid cell. For any grid cell, sort out all the cell temperatures of the grid cell from the moment the concrete pouring is completed to the current moment, and integrate the cell temperatures according to the time sequence to obtain the cell temperature sequence of the grid cell;

[0023] Based on the temperature control target of each grid unit and the predicted unit temperature sequence, the PID algorithm is used to dynamically calculate the adjustment value of each cooling parameter of the cooling water cycle. The actual adjustment value of each cooling parameter is obtained based on the feedback calculation of the adjustment value, and the cooling parameter is adjusted according to the actual adjustment value;

[0024] Obtain the temperature control target and the predicted unit temperature of the grid unit at the end of the corresponding prediction period, perform data processing on the temperature control target and the predicted unit temperature, and obtain the global average deviation at the current moment;

[0025] Based on the PID algorithm, the global average deviation is used as input, the comprehensive adjustment amount is output, and a linear parameter mapping model is constructed. The adjustment value of the cooling parameter is obtained through the comprehensive adjustment amount mapping. The cooling water flow rate is calculated according to the feedback of the cooling parameter adjustment value, and the prediction period of each grid unit is updated. According to the updated prediction period, the hybrid prediction model and the PID algorithm are used to calculate the actual adjustment value of the cooling parameter.

[0026] The beneficial effects of the present invention are as follows:

[0027] 1. By comprehensively acquiring basic wall information and constructing a precise finite element model, this invention meticulously simulates changes in the wall's temperature field. Using a network of pressure sensors to monitor the pressure of heat-dissipating steel pipes in real time, this technology proactively detects abnormalities in the concrete's bond strength with the pipes, ensuring structural stability. When the bond strength is normal, the temperature sensor network and finite element model accurately determine whether to activate the cooling circulation system, avoiding the waste of resources caused by blind activation. This effectively improves the foresight and accuracy of crack control and reduces the risk of cracks in extremely thick walls.

[0028] 2. The present invention combines the finite element model with the LSTM network to construct a hybrid prediction model, accurately predicts the grid unit temperature, and obtains the predicted unit temperature sequence. According to the temperature control target and the prediction sequence, the PID algorithm is used to dynamically calculate the cooling parameter adjustment value, and the feedback is updated during the prediction period for iterative calculation to obtain the actual adjustment value. This dynamic and precise control method enables the cooling water circulation system to adaptively adjust according to the actual temperature changes of the wall, greatly improving the cooling efficiency, effectively controlling the temperature stress of ultra-thick wall concrete, and significantly reducing the occurrence of cracks. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention will be further described below with reference to the accompanying drawings.

[0030] Figure 1 This is a flowchart of the steps of a method for controlling cracks in ultra-thick wall concrete according to an embodiment of the present invention;

[0031] Figure 2 This is a flow chart of the steps for obtaining the predicted unit temperature sequence in a method for controlling cracks in ultra-thick wall concrete described in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0033] Example 1

[0034] See also Figure 1As shown, a method for controlling cracks in ultra-thick wall concrete according to an embodiment of the present invention includes the following steps:

[0035] S1: Obtain basic wall information, design the layout of heat dissipation steel pipes, install and build a cooling water circulation system, and build a 3D solid model of the wall based on the basic wall information;

[0036] Before the construction of ultra-thick wall concrete, basic wall information is obtained, including the three-dimensional geometric dimensions of the wall and concrete design parameters. The three-dimensional geometric dimensions of the wall include the length, width, and height of the wall, and the concrete design parameters include the strength grade, mix ratio, and pouring plan. These parameters are obtained by obtaining and extracting the architectural design drawings.

[0037] Based on the principles of heat conduction and engineering experience, the horizontal and vertical spacing of the heat dissipation steel pipes on the wall is set. The heat dissipation steel pipes are made of 32mm diameter hot-dip galvanized steel pipes, which have both strength and thermal conductivity. The spatial coordinates of the heat dissipation steel pipes are accurately marked through CAD 3D modeling, and a heat dissipation steel pipe position coordinate matrix containing hundreds of positioning points is generated to obtain the heat dissipation steel pipe layout design. According to the layout design, the heat dissipation steel pipes are fixed to the wall steel skeleton. Steel support brackets or special clamps are used to ensure the heat dissipation steel pipes are accurately and firmly positioned. After the heat dissipation steel pipes are installed, they are connected to the cooling water circulation system, including water pumps, water pipes, flow control valves and other equipment, to confirm that there are no leaks in the pipelines and the connection is reliable.

[0038] Using Revit software, a 3D solid model of the wall was constructed by combining basic wall information with the coordinate matrix of the heat dissipation steel pipe positions. Material properties were assigned to each component in the 3D solid model, enabling the 3D solid model to perform thermomechanical coupling analysis. Initial values ​​for the cooling parameters of the cooling water circulation system were set. These parameters included the cooling water inlet temperature, cooling water flow rate, and cooling water circulation pressure. These initial values ​​were integrated into the 3D solid model to obtain a 3D solid model of the wall that served as a digital twin of the actual wall.

[0039] It should be noted that the purpose of this step is to obtain wall foundation information and design the distribution of heat dissipation steel pipes to ensure the scientific and reasonable layout of the heat dissipation system. Revit is used to build a three-dimensional solid model of the wall with thermal-mechanical coupling analysis capabilities. This provides a precise digital foundation for subsequent temperature simulation, thermal-mechanical analysis, and actual construction monitoring, improves the accuracy and scientific nature of pre-construction planning, and lays the foundation for crack control from the source.

[0040] S2: Divide the three-dimensional wall model into grid cells to obtain a finite element model. Use a pressure sensor network to monitor the pressure of the heat dissipation steel pipe to determine whether the bond strength of the concrete to the steel pipe is abnormal. If normal, collect wall temperature data through a temperature sensor network and calculate the temperature parameters of each grid cell based on the finite element model.

[0041] Based on the obtained three-dimensional solid model of the wall, the three-dimensional solid model of the wall is meshed using finite element pre-processing software, and the three-dimensional solid model of the wall is evenly discretized into a number of tetrahedral grid units to obtain a finite element model. The vertex coordinates of each grid unit are marked as node coordinates, and all node coordinates are integrated to form a node coordinate set V;

[0042] V={v1,v2,......,v n ,......,v N};

[0043] where v n represents the nth node coordinate in the finite element model, n = 1, 2, ..., N, N represents the number of all node coordinates in the finite element model;

[0044] It should be noted that the mesh cells in the finite element model are tetrahedral in shape. In the finite element model, adjacent mesh cells share vertices and edges. Therefore, there is no one-to-one correspondence between the number of mesh cells and the number of mesh cell vertices. Therefore, the number of mesh cells is different from the number of node coordinates, and the number of node coordinates is less than the number of mesh cells.

[0045] Temperature sensors are placed at each node coordinate to build a temperature sensor network. Pressure sensors are evenly placed on heat dissipation steel pipes to build a pressure sensor network. After the ultra-thick wall concrete is constructed, the collection period is marked with the moment when the concrete pouring is completed. A sampling cycle is set within the collection period. The interval between adjacent sampling moments is a sampling cycle, and the starting point of the collection period is the first sampling moment in the collection period.

[0046] Through the pressure sensor network evenly distributed on the heat dissipation steel pipe, the locations where the pressure sensors are distributed on the heat dissipation steel pipe are marked as pressure collection points, and the pressure collection points are marked with serial numbers. The pressure value of each pressure collection point is collected in real time at the sampling time. For each pressure collection point, the pressure change value between the current sampling time and the previous sampling time is calculated;

[0047] Set a cumulative flag value flag for each pressure collection point r And assign its initial value to 0, where r represents the serial number of the pressure collection point. If the pressure value of the pressure collection point with serial number r at the current moment is less than the pressure threshold, or the pressure change value of the pressure collection point is greater than the change standard value, the corresponding cumulative flag value flag rPerform a self-add operation. If the pressure value of the pressure collection point with the serial number r at the current moment is greater than or equal to the pressure threshold, and the pressure change value of the pressure collection point is less than or equal to the change standard value, the corresponding cumulative flag value flag is set. r Return to 0;

[0048] If the cumulative flag value flag i If the pressure is less than the preset stable mark value, it is judged that the value at the pressure collection point is normal and the bond strength of the concrete to the heat dissipation steel pipe is normal. At this time, if there is an abnormal mark at the pressure collection point, clear the abnormal mark at the pressure collection point;

[0049] If the cumulative flag value flag i If the value is greater than or equal to the preset stability mark value, it is judged that an abnormal value has appeared at the pressure collection point, and the gripping force of the heat dissipation steel pipe may be abnormal, and an abnormal mark is generated for the pressure collection point;

[0050] For pressure collection points with abnormal marks, two adjacent pressure collection points with abnormal marks are classified into the same collection point group. The number of pressure collection points with abnormal marks in each collection point group is obtained and compared with the stable mark value.

[0051] If the number of pressure collection points is less than the stability mark value, it is determined that the reason why the pressure collection point in the collection point group generates an abnormal mark is that the pressure sensor is faulty. The pressure collection point is overwritten and marked as a faulty collection point, and the pressure value collected by the pressure sensor corresponding to the faulty collection point is no longer trusted.

[0052] If the number of pressure collection points is greater than or equal to the stability mark value, it is determined that the reason for the abnormal mark generated by the pressure collection point in the collection point group is that the bond strength of the concrete to the heat dissipation steel pipe is abnormal, resulting in a gap between the heat dissipation steel pipe and the concrete. A grouting reinforcement warning is generated, and the construction personnel are notified to perform grouting reinforcement at the gap.

[0053] If there are no abnormal marks at all pressure collection points, it is judged that the cooling water circulation system can work normally. The wall temperature data TS (t i );

[0054] TS(t i )={T1(t i ), T2(t i ),......,T n (t i ),......,T N (t i )};

[0055] Among them, t i represents the i-th sampling time, T n (ti ) represents the temperature value collected at the i-th sampling moment at the coordinate of the n-th node in the finite element model;

[0056] Construct the finite element heat conduction equation based on Fourier's heat conduction law:

[0057]

[0058] Among them, ρ represents the density of concrete, c represents the specific heat capacity of concrete, T represents the temperature, t represents the time, the left side of the equation represents the rate of change of the heat capacity of unit volume of concrete with time, reflecting the trend of temperature change with time, k represents the thermal conductivity, and the right side of the equation Describes the heat conduction process inside the concrete and reflects the diffusion effect of heat conduction. h represents the endogenous heat source term, Q h The calculation formula is:

[0059]

[0060] Where θ(t) represents the temperature rise per unit volume of concrete due to hydration heat at time t, which is calculated using the hydration heat formula:

[0061]

[0062] Where Q represents the total hydration heat, that is, the heat released when the unit mass of cement is completely hydrated, m c It is expressed as the amount of cement, which is determined by the concrete mix ratio, and β is the heat release coefficient, which indicates the rate of hydration heat release;

[0063] Using finite element analysis software, the finite element heat conduction equation is input into the finite element model, and the real-time wall temperature data is imported into the finite element model. Transient thermal analysis is performed on the finite element model, and the finite element heat conduction equation is solved for each grid cell. After multiple rounds of iterative calculations, the temperature parameters of each grid cell are output. The temperature parameters include the node temperature and the temperature gradient between each grid cell.

[0064] It should be noted that this step is to monitor the heat dissipation steel pipe pressure in real time through the pressure sensor network, accurately determine the bond strength of the concrete on the steel pipe, promptly remove misjudged abnormality marks or generate grouting warnings, ensure the accuracy of the cooling water circulation system's working status, ensure the reliability of the subsequently collected wall temperature data, and effectively identify defects in the bond between the steel pipe and concrete, avoiding structural hazards caused by insufficient bond strength and ensuring the safety of the wall structure.

[0065] S3: Determine whether to start the cooling cycle system based on the temperature parameters of the grid cells. If so, simulate and calculate the temperature control target of each grid cell, and mark the prediction period for each grid cell based on the cooling parameters of the cooling cycle system. Build a hybrid prediction model combining the finite element model and the LSTM network to predict the unit temperature of the grid cell in the prediction period and obtain the predicted unit temperature sequence of each grid cell.

[0066] like Figure 2 As shown, the specific steps for obtaining the prediction unit temperature sequence are as follows:

[0067] Obtain the temperature parameters of each grid cell, calculate the temperature of the center point of each grid cell, mark it as the cell temperature of the grid cell, and for any grid cell, calculate the cell temperature difference between the current time and the previous sampling time of the grid cell to obtain the temperature rise value of the grid cell;

[0068] Set a flag value flag and assign it a value of 0. If at the current moment, the unit temperature of any grid unit is greater than the unit temperature standard value, or the temperature rise value of any grid unit is greater than the unit, assign the flag value to 1. When the flag value is 1, the cooling water circulation system is turned on, and the cooling parameter is set to the preset cooling parameter initial value. The period when the flag value is continuously 1 is marked as the cooling period;

[0069] If the cooling period begins, the temperature control target T of each grid unit in the cooling period is generated by finite element model simulation based on the concrete thermal stress safety threshold determined by the material thermal performance experiment. tar，x (t m ), where x represents the grid cell number, t m Indicates the time within the cooling period;

[0070] Obtain the diameter of the heat dissipation steel pipe to calculate the flow area of ​​the steel pipe, obtain the cooling water flow rate in the cooling parameters at the current moment and perform ratio processing on the flow area of ​​the steel pipe to obtain the cooling water flow rate;

[0071] Based on any grid cell, the shortest distance between the grid cell and the heat dissipation steel pipe is calculated, and the position on the heat dissipation steel pipe with the shortest distance to the grid cell is marked as the cooling control point of the grid cell. The length of the heat dissipation steel pipe between the cooling control point and the cooling water inlet is calculated. Combined with the cooling water flow rate at the current moment, the required time for cooling water in the cooling water circulation system to reach the cooling control point of the grid cell is calculated. The period starting from the current moment and lasting for the required time is marked as the predicted period of the grid cell.

[0072] It should be noted that the forecast periods for different grid cells are not necessarily the same;

[0073] For any grid unit, sort out all the unit temperatures of the grid units from the start of the acquisition period to the current moment, and integrate the unit temperatures into a unit temperature sequence according to the time sequence;

[0074] During the cooling period, a hybrid prediction model is constructed by combining the finite element model and the LSTM network to predict the unit temperature of the grid cells during the prediction period. The hybrid prediction model consists of a physical branch and a data branch. The physical branch is the finite element model during the cooling period, and the data branch is the LSTM network trained based on the unit temperature series.

[0075] Specifically, for the physical branch, during the cooling period, the heat dissipation steel pipe removes the heat of the concrete through forced convection, and the cooling effect is equivalent to the cooling and heat dissipation term S cool The finite element heat conduction equation embedded in the finite element model updates the finite element model. The revised finite element heat conduction equation is:

[0076]

[0077] The cooling term S cool The calculation formula is:

[0078]

[0079] Among them, m w It represents the cooling water mass flow rate, which is calculated by multiplying the cooling water density and the cooling water flow rate. w represents the specific heat capacity of water, V unit represents the grid cell volume, T in Indicates the cooling water inlet temperature, T out Indicates the cooling water outlet temperature, and the calculation formula is:

[0080]

[0081] Among them, Q abs The heat exchange between the heat dissipation steel pipe and the concrete is calculated in real time using Fourier's heat conduction law;

[0082] In the finite element model, the predicted temperature parameters of each grid cell at each sampling time in the prediction period are solved by the explicit difference method for the modified finite element heat conduction equation. Based on the obtained predicted temperature parameters, the physical predicted unit temperature of the grid cell at each sampling time in the prediction period is calculated;

[0083] For the data branch, the LSTM network uses a two-layer LSTM with 128 neurons in each layer. A fully connected layer is added after the LSTM layer. ReLU is used as the activation function and mean square error (MSE) is used as the loss function. The LSTM network is trained using the cell temperature sequence. The trained LSTM network is used to predict the cell temperature of the grid cell, obtaining the data-predicted cell temperature of the grid cell at each sampling time within the prediction period.

[0084] Obtain the most recently predicted physical prediction unit temperature and data prediction unit temperature of each grid cell at the current moment, and compare the physical prediction unit temperature and data prediction unit temperature of each grid cell with the unit temperature at the current moment. For any grid cell, if the physical prediction unit temperature is closer to the unit temperature, mark the grid cell as a physical cell; otherwise, mark it as a data cell.

[0085] The hybrid prediction model integrates the physical branch and the data branch, calculates dynamic weights based on the number of physical units and the number of data units, and performs a weighted fusion of the physical prediction unit temperature and the data prediction unit temperature of the grid unit at any sampling moment within the prediction period according to the dynamic weights. The predicted unit temperature of the grid unit at the sampling moment within the prediction period is calculated, and the predicted unit temperatures are integrated according to the time sequence to obtain the predicted unit temperature sequence of the grid unit.

[0086] The prediction unit temperature of each grid unit in the corresponding prediction period is predicted by the hybrid prediction model to obtain the prediction unit temperature sequence of each grid unit;

[0087] It should be noted that this step is used to calculate the unit temperature and temperature rise of the grid cells, promptly determine when to start the cooling period, and ensure a rapid response to temperature anomalies. A hybrid prediction model is constructed by combining the finite element model with the LSTM network, integrating physical analysis with data-driven prediction. This allows for more accurate prediction of the grid cell temperature during the cooling period, providing a reliable basis for temperature control, effectively reducing the risk of temperature cracks, and improving the foresight and accuracy of crack control.

[0088] S4: Based on the temperature control target of each grid unit and the predicted unit temperature sequence, the adjustment value of each cooling parameter of the cooling water cycle is dynamically calculated using the PID algorithm. The prediction period of each grid unit is updated based on the adjustment value feedback. The actual adjustment value of each cooling parameter is iteratively calculated based on the updated prediction period, and the cooling parameter is adjusted according to the actual adjustment value.

[0089] Get the temperature control target T of each grid unit during the cooling period tar，x (t m ), extract the grid cell at the end of the corresponding prediction period t end The temperature control target Ttar，x (t end );

[0090] Obtain the predicted unit temperature sequence of each grid unit, extract the predicted unit temperature of the grid unit at the end of the corresponding prediction period and integrate it into the predicted temperature vector T pre ;

[0091]

[0092] Wherein, x represents the sequence number of the grid unit, x=1, 2, ..., XL, XL represents the total number of grid units;

[0093] Calculate the difference between the predicted cell temperature of a single grid cell at the end of the corresponding prediction period and the temperature control target, marked as the cell deviation, and take the average of the cell deviations of all grid cells to obtain the global average deviation e at the current moment x (t now ), where t now Indicates the current moment;

[0094] Based on the PID algorithm, the global average deviation is used as input and the comprehensive adjustment amount u is used. temp Is the output, the output calculation formula is:

[0095]

[0096] Among them, τ represents the time variable, K p Represents the proportionality coefficient, K i Indicates the integral coefficient, K d represents the differential coefficient;

[0097] A linear parameter mapping model is constructed to map the calculated integrated adjustment amount to the adjusted value of the cooling parameter. The cooling water flow rate is updated according to the adjusted value of the cooling parameter. The prediction period of each grid unit is updated based on the updated cooling water flow rate. The predicted unit temperature sequence of each grid unit is re-predicted using the hybrid prediction model based on the updated prediction period. Based on the obtained predicted unit temperature sequence, the actual adjustment value of the cooling parameter is calculated using the PID algorithm.

[0098] Input control instructions into the cooling water circulation system and dynamically adjust the cooling parameters of the cooling water circulation system according to the actual adjustment values ​​obtained;

[0099] It should be noted that the purpose of this step is to dynamically adjust the cooling parameters based on the deviation between the predicted unit temperature and the control target through the PID algorithm, forming a closed-loop control of prediction-adjustment-feedback-optimization. This accurately balances the cooling rate and the temperature at various locations on the wall, achieving refined control of the cooling parameters, effectively suppressing the occurrence of temperature cracks, ensuring the construction quality of ultra-thick wall concrete, and improving the dynamics and accuracy of crack control.

[0100] The technical solution of an embodiment of the present invention is as follows: obtaining basic wall information, designing the layout of heat dissipation steel pipes, and installing and constructing a cooling water circulation system; constructing a three-dimensional solid model of the wall based on the basic wall information; dividing the three-dimensional solid model into grid cells to obtain a finite element model; using a pressure sensor network to monitor the pressure of the heat dissipation steel pipes to determine whether the bond strength of the concrete on the steel pipes is abnormal; if normal, collecting wall temperature data through a temperature sensor network; calculating the temperature parameters of each grid cell based on the finite element model; determining whether to activate the cooling circulation system based on the grid cell temperature parameters; if activated, simulating and calculating the temperature control target of each grid cell; marking a prediction period for each grid cell based on the cooling parameters of the cooling circulation system; constructing a hybrid prediction model based on the finite element model and an LSTM network; predicting the cell temperature of the grid cell during the prediction period to obtain a predicted cell temperature sequence for each grid cell; dynamically calculating adjustment values ​​for each cooling parameter of the cooling water circulation based on the temperature control target and the predicted cell temperature sequence for each grid cell using a PID algorithm; updating the prediction period for each grid cell based on the adjustment value feedback; iteratively calculating the actual adjustment value of each cooling parameter based on the updated prediction period; and adjusting the cooling parameters based on the actual adjustment value.

[0101] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for controlling cracks in ultra-thick wall concrete, characterized by: include: The 3D solid model of the wall is divided into grid cells to obtain a finite element model. A pressure sensor network is constructed to monitor the pressure of the heat dissipation steel pipe to determine whether the bond strength of the concrete to the steel pipe is normal. If normal, the wall temperature data is collected through the temperature sensor network and the temperature parameters of each grid cell are calculated based on the finite element model. The system determines whether to start the cooling cycle system based on the temperature parameters of the grid cells. If so, the temperature control target of each grid cell is simulated and calculated. The prediction period is marked for each grid cell based on the cooling parameters of the cooling cycle system. A hybrid prediction model is constructed to predict the predicted unit temperature sequence of the grid cell within the prediction period. According to the temperature control target of each grid unit and the predicted unit temperature sequence, the adjustment value of each cooling parameter of the cooling water cycle is dynamically calculated through the PID algorithm. The actual adjustment value of each cooling parameter is obtained according to the feedback calculation of the adjustment value, and the cooling parameter is adjusted according to the actual adjustment value.

2. The method for controlling cracks in ultra-thick wall concrete according to claim 1, characterized in that: The finite element model is obtained as follows: A three-dimensional solid model of the wall is obtained, and the three-dimensional solid model of the wall is meshed. The three-dimensional solid model of the wall is evenly discretized into tetrahedral mesh units, and the vertex coordinates of each mesh unit are marked as node coordinates. All node coordinates are integrated to form a node coordinate set to obtain a finite element model.

3. The method for controlling cracks in ultra-thick wall concrete according to claim 1, characterized in that: The method for judging whether the bond strength of the concrete to the steel pipe is normal is as follows: The locations where pressure sensors are arranged on the heat dissipation steel pipes are marked as pressure collection points, and the cumulative mark values ​​of each pressure collection point are obtained. For any pressure collection point, if the cumulative mark value is less than the preset stable mark value and there is an abnormal mark at the pressure collection point, the abnormal mark of the pressure collection point is cleared; if the cumulative mark value is greater than or equal to the preset stable mark value, an abnormal mark is generated for the pressure collection point; If there are no abnormal marks at all pressure collection points, it is judged that the bond strength of concrete on the steel pipe is normal.

4. The method for controlling cracks in ultra-thick wall concrete according to claim 3, characterized in that: The method for obtaining the cumulative flag value is: The pressure value of each pressure collection point is collected in real time. For each pressure collection point, the pressure change value between the current moment and the previous sampling moment is calculated. A cumulative flag value is set for each pressure collection point and its initial value is assigned to 0. If the pressure value of any pressure collection point at the current moment is less than the pressure threshold, or the pressure change value of the pressure collection point is greater than the change standard value, the corresponding cumulative flag value is self-added. Otherwise, the corresponding cumulative flag value is reset to 0.

5. The method for controlling cracks in ultra-thick wall concrete according to claim 1, characterized in that: The temperature parameters of each grid cell are calculated as follows: The wall temperature data is collected in real time at the sampling moment through the constructed temperature sensor network; Based on Fourier's law of heat conduction, a finite element heat conduction equation is constructed. Using finite element analysis software, the finite element heat conduction equation is input into the finite element model, and the wall temperature data collected in real time is imported into the finite element model. Transient thermal analysis is performed on the finite element model, and the finite element heat conduction equation is solved for each grid unit. After multiple rounds of iterative calculations, the temperature parameters of each grid unit are output.

6. The method for controlling cracks in ultra-thick wall concrete according to claim 1, characterized in that: The method for obtaining the prediction unit temperature sequence is as follows: A hybrid prediction model including physical branch and data branch is constructed. Dynamic weights are calculated according to the prediction accuracy of the physical branch and the data branch. The prediction results of the physical branch and the data branch are weighted and fused according to the dynamic weights. The predicted unit temperature of the grid unit within the prediction period is calculated. The predicted unit temperature is integrated according to the time series to obtain the predicted unit temperature sequence of the grid unit.

7. The method for controlling cracks in ultra-thick wall concrete according to claim 6, characterized in that: The method for obtaining the forecast period is as follows: Obtain the diameter of the heat dissipation steel pipe to calculate the flow area of ​​the steel pipe, and calculate the cooling water flow rate based on the cooling water flow in the cooling parameters at the current moment; Based on any grid unit, calculate the shortest distance between the grid unit and the heat dissipation steel pipe, and mark the position on the heat dissipation steel pipe with the shortest distance to the grid unit as the cooling control point of the grid unit. Obtain the length of the heat dissipation steel pipe between the cooling control point and the cooling water inlet, and calculate the required time for the cooling water to reach the cooling control point in combination with the cooling water flow rate. Mark the period with the current time as the starting point and the required time as the duration as the prediction period of the grid unit.

8. The method for controlling cracks in ultra-thick wall concrete according to claim 7, characterized in that: The physical branch and the data branch include: In the physics branch, the cooling effect is equivalent to the cooling and heat dissipation term embedded in the finite element heat conduction equation in the finite element model, and the finite element heat conduction equation is modified. For the modified finite element heat conduction equation, the predicted temperature parameters of each grid cell in the prediction period are solved by the explicit difference method, and the physical prediction unit temperature of the grid cell in the prediction period is calculated; In the data branch, an LSTM network is constructed, and the cell temperature sequence of each grid cell is obtained to train the LSTM network. The cell temperature of the grid cell is predicted by the trained LSTM network to obtain the data predicted cell temperature of the grid cell within the prediction period.

9. The method for controlling cracks in ultra-thick wall concrete according to claim 7, characterized in that: The unit temperature sequence is obtained as follows: The temperature parameters of each grid cell are obtained for data processing, and the temperature of the center point of each grid cell is calculated and marked as the cell temperature of the grid cell. For any grid cell, all cell temperatures of the grid cell from the moment when concrete pouring is completed to the current moment are sorted out, and the cell temperatures are integrated according to the time sequence to obtain the cell temperature sequence of the grid cell.

10. The method for controlling cracks in ultra-thick wall concrete according to claim 1, characterized in that: The actual adjustment value of each cooling parameter is obtained as follows: Obtain the temperature control target and the predicted unit temperature of the grid unit at the end of the corresponding prediction period, perform data processing on the temperature control target and the predicted unit temperature, and obtain the global average deviation at the current moment; Based on the PID algorithm, the global average deviation is used as input, the comprehensive adjustment amount is output, and a linear parameter mapping model is constructed. The adjustment value of the cooling parameter is obtained through the comprehensive adjustment amount mapping. The cooling water flow rate is calculated according to the feedback of the cooling parameter adjustment value, and the prediction period of each grid unit is updated. According to the updated prediction period, the hybrid prediction model and the PID algorithm are used to calculate the actual adjustment value of the cooling parameter.

Images