Hydration heat monitoring and dynamic control method and system during mass concrete pouring process

Abstract

The present invention discloses a method and system for monitoring and dynamically controlling the hydration heat during large-volume concrete pouring, relating to the field of concrete pouring technology. The method comprises pre-embedding a plurality of dual-mode temperature measuring probes within the concrete to collect real-time dual-mode temperature field data within the concrete and simultaneously collect environmental parameters; pre-processing the acquired dual-mode temperature field data, dynamically assigning weights, and fusing the data; constructing a finite element heat conduction model based on a non-steady-state heat transfer equation and a hydration heat model, and inverting the total hydration heat and temperature difference trends from the measured temperature field data; combining an LSTM neural network model to predict the temperature difference between the concrete core and surface, and the temperature difference between the surface and the environment, within a set period of time in the future; and closed-loop controlling the cooling water flow rate and cooling water temperature based on the predicted temperature difference results. By combining dual-mode sensor temperature measurement with heat inversion, the hydration heat temperature difference within the concrete can be predicted in advance, and the cooling scheme can be controlled in real time to avoid the occurrence of temperature stress cracks.

Description

Technical Field

[0001] The present invention relates to the technical field of concrete pouring, and in particular to a method and system for monitoring and dynamically controlling hydration heat during a mass concrete pouring process. Background Art

[0002] During the pouring process of large-volume concrete caps, such as continuous rigid frame aqueducts, there is a risk of severe hydration heat release and temperature cracks due to their large volume and one-time casting.

[0003] Currently, traditional methods for monitoring the temperature of large-volume concrete caps rely primarily on single-point temperature sensors, most commonly resistance temperature probes and vibrating wire temperature probes. Resistance temperature probes measure temperature through changes in resistance, offering low cost but high accuracy due to environmental interference. Vibrating wire temperature probes measure temperature through changes in frequency, offering high accuracy but not directly correlating to heat.

[0004] It can be seen that the technical problems with single-point temperature sensors are: on the one hand, only monitoring the local temperature through the temperature sensor cannot directly quantify the total hydration heat inside the concrete, resulting in delayed heat control; on the other hand, it is difficult to accurately judge the heat distribution by relying solely on temperature data, and it is impossible to adjust the cooling plan in real time, such as the water pipe flow rate and cooling water temperature.

[0005] Furthermore, traditional temperature control systems, driven by a single temperature data source, primarily operate by controlling the opening and closing of cooling water pipes based on pre-set temperature thresholds. This temperature control method suffers from significant lag and lacks predictive power, as it relies solely on currently measured temperature data and lacks the ability to predict dynamic heat changes within the concrete. During the rapid release of hydration heat in large-volume concrete caps, the cooling water pipes are not activated until the temperature reaches the threshold. By then, significant heat has already accumulated within the concrete, potentially causing damage to the concrete structure and failing to effectively prevent the formation of temperature cracks. Summary of the Invention

[0006] In a first aspect of the present invention, in order to solve the above technical problems, a method for monitoring and dynamically controlling hydration heat during mass concrete pouring is provided, the method comprising:

[0007] Embedding multiple dual-mode temperature probes in the concrete to collect real-time dual-mode temperature field data inside the concrete and simultaneously collect environmental parameters; wherein the dual-mode temperature probes integrate resistive and vibrating wire sensors or resistive and fiber Bragg grating sensors;

[0008] Preprocessing the acquired dual-mode temperature field data to eliminate electromagnetic interference and vibration noise, and dynamically assigning weights and fusing the dual-mode temperature field data to output calibrated measured temperature field data;

[0009] Based on the unsteady-state heat transfer equation and the hydration heat model, a finite element heat conduction model is constructed, and the total hydration heat and temperature difference trend are inverted using the measured temperature field data; the temperature difference between the concrete core and the surface, and the temperature difference between the surface and the environment, are predicted in the future set time period using the LSTM neural network model;

[0010] According to the temperature difference results predicted by the LSTM neural network model, the cooling water flow rate and cooling water temperature are closed-loop regulated through a dynamic optimization strategy to ensure that the temperature difference fluctuation is stable within a safe threshold.

[0011] Furthermore, the temperature difference result predicted by the LSTM neural network model triggers hierarchical regulation:

[0012] When the predicted temperature difference exceeds the first level threshold, the cooling water flow rate is increased according to the PID control algorithm;

[0013] When the predicted temperature difference exceeds the secondary threshold, the refrigerator is started to lower the cooling water temperature.

[0014] Furthermore, the adjustment step of the cooling water flow rate is ±0.2 m / s; the adjustment step of the cooling water temperature is ±2°C.

[0015] Furthermore, the cooling water flow rate is increased according to the PID control algorithm, which satisfies the expression:

[0016]

[0017] Where, is the deviation between the predicted temperature difference and the target value; is the current cooling water flow rate; 、 、 They are proportional coefficient, integral coefficient and differential coefficient respectively, and the above 、 、 Parameter tuning of .

[0018] Furthermore, the dynamic optimization strategy includes rolling time domain control and exception handling mechanism, wherein:

[0019] The rolling time domain control is a control sequence that optimizes the cooling water flow rate and cooling water temperature at future rated times in a fixed cycle;

[0020] The abnormal handling mechanism is that if the temperature difference drop rate after three or more consecutive adjustments is less than <5%, the expert system diagnosis will be triggered.

[0021] Furthermore, the objective function of the rolling horizon control is:

[0022]

[0023] Where, To control the number of times; is the temperature difference; is the cooling water flow rate; is the cooling water temperature.

[0024] Furthermore, the preprocessing of the acquired dual-mode temperature field data specifically includes:

[0025] Unify the sampling time of dual-mode sensors based on interpolation method;

[0026] The 3σ criterion was used to eliminate data points exceeding ±3 times the standard deviation;

[0027] Wavelet noise reduction and electromagnetic interference compensation are performed on the resistive sensor data; median filtering and vibration compensation are performed on the vibrating string sensor data.

[0028] Furthermore, the dynamic weight allocation and data fusion of the dual-mode temperature field data specifically include:

[0029] Calculate the confidence index of the dual-mode sensor integrated in the dual-mode temperature measuring probe respectively: ,in, is the confidence index of the resistive sensor in the dual-mode temperature probe, is a confidence index of the vibrating wire sensor in the dual-mode temperature probe; is the short-term temperature standard deviation of the resistive sensor in the dual-mode temperature probe, is the short-term temperature standard deviation of the vibrating wire sensor in the dual-mode temperature probe;

[0030] Determine the fusion coefficient of dual-mode data according to the intensity of environmental interference : ;

[0031] Calculate the fusion temperature of the dual-mode data, which satisfies the expression:

[0032]

[0033] Where, 、 are the dual-mode temperature field data collected respectively.

[0034] Furthermore, the finite element heat conduction model is constructed by the following steps:

[0035] The concrete structure is divided using non-uniform hexahedral meshes, and tetrahedral meshes are used to refine the mesh near the cooling pipes.

[0036] Embedded in the hydration heat model , which satisfies the expression:

[0037]

[0038] Where, is the total hydration heat, in J / m³; Degree of hydration: ,in, are relevant material parameters and are iteratively updated through the Levenberg-Marquardt algorithm to minimize the residual error between the measured temperature field data and the predicted temperature field data.

[0039] Furthermore, the step of inverting the total hydration heat by the measured temperature field data is as follows: inputting the measured temperature field data into the unsteady heat transfer equation, inverting the total hydration heat , where the unsteady heat transfer equation satisfies the expression:

[0040]

[0041] Where, is the density of concrete, in kg / m³; is the specific heat capacity of concrete, the unit is J / (kg·K); is the thermal conductivity of concrete, the unit is W / (m·K); is the boundary heat exchange capacity, the unit is W / m³.

[0042] Furthermore, the training method of the LSTM neural network model includes:

[0043] Its input includes the historical temperature series of the concrete core and surface, the environmental parameters, the cooling water flow rate and the cooling water temperature;

[0044] Its output is the temperature forecast value for the set period in the future;

[0045] The loss function of the LSTM neural network model adopts mean square error.

[0046] A second aspect of the present invention provides a system for monitoring and dynamically controlling hydration heat during mass concrete pouring, comprising:

[0047] A dual-mode temperature probe, embedded in the concrete cap, integrating a resistive and vibrating-wire sensor or a resistive and fiber grating sensor;

[0048] A dual-mode data fusion module is used to receive the raw data from the dual-mode temperature measuring probe and output calibrated measured temperature field data through a dynamic weighted fusion algorithm;

[0049] Finite element heat conduction model, which is based on the unsteady-state heat transfer equation to construct a three-dimensional heat conduction model of concrete;

[0050] a heat inversion module that uses an optimization algorithm to iteratively update relevant parameters in the finite element heat conduction model and, based on the optimized parameters, combines an LSTM neural network model to predict the temperature difference between the concrete core and surface, and the temperature difference between the surface and the environment, for a set period in the future; and

[0051] The dynamic control module is configured to trigger hierarchical control of the cooling water flow rate and the cooling water temperature according to the comparison result of the predicted temperature difference result and the threshold value, and trigger the finite element heat conduction model to recalculate the temperature field.

[0052] Furthermore, the three-dimensional heat conduction model of concrete satisfies the expression:

[0053]

[0054] Where, is the density of concrete, in kg / m³; is the specific heat capacity of concrete, the unit is J / (kg·K); is the thermal conductivity of concrete, the unit is W / (m·K); is the boundary heat exchange capacity, the unit is W / m³; is the total hydration heat, in J / m³; These are material-related parameters and are preset according to the concrete type.

[0055] Preferably, the dual-mode temperature probe is provided with a stainless steel armored shell and is encapsulated with epoxy resin, and a vibrating string signal line and an RTD lead are led out from the inside to the outside at one end of the stainless steel armored shell; the dual-mode sensors are separated by a stainless steel partition wall, wherein the vibrating string sensor is provided with an Invar frame.

[0056] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0057] The present invention combines dual-mode sensor temperature measurement with heat inversion to achieve early prediction of the hydration heat temperature difference inside the concrete during the concrete cap pouring process, and then adjusts the cooling plan in real time, greatly avoiding the occurrence of temperature stress cracks. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0059] Figure 1 A flowchart of the disclosed embodiment of the present invention;

[0060] Figure 2 This is an overall framework diagram disclosed in an embodiment of the present invention;

[0061] Figure 3 A flow chart of the system architecture disclosed in an embodiment of the present invention;

[0062] Figure 4 This is a schematic structural diagram of a dual-mode temperature measuring probe disclosed in an embodiment of the present invention.

[0063] In the picture:

[0064] 100. Dual-mode temperature probe; 101. Stainless steel armored housing; 102. Epoxy resin filler; 103. Stainless steel partition wall; 104. 4-core waterproof cable;

[0065] 110, platinum resistance wire; 111, ceramic substrate; 112, 2-core RTD lead wire;

[0066] 120. Excited body; 121. Excitation / pickup coil; 122. Steel string; 123. 2-core vibrating string signal line; 124. Invar frame. DETAILED DESCRIPTION

[0067] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0068] The present invention aims to provide a method and system for monitoring and dynamically controlling the hydration heat during the pouring of large-volume concrete, so as to solve the problems in the traditional large-volume concrete pedestal pouring process, namely, that on the one hand, the total hydration heat inside the concrete cannot be directly quantified; on the other hand, the hydration heat temperature difference cannot be predicted in advance and the cooling scheme cannot be adjusted in real time.

[0069] See also Figure 1-2The method for monitoring and dynamically controlling hydration heat during mass concrete pouring provided by the present invention comprises the following steps:

[0070] S1. Embed several dual-mode temperature probes within the concrete to collect real-time dual-mode temperature field data within the concrete and simultaneously collect environmental parameters. The dual-mode temperature probes integrate resistive and vibrating-wire sensors, or resistive and fiber Bragg grating sensors. Environmental parameters include wind speed, humidity, and sunlight intensity.

[0071] The following description takes resistive and vibrating-wire sensors as examples.

[0072] The resistive sensor preferably uses a Pt100 platinum resistor with a range of -50°C to 150°C. It is enclosed in a 316L stainless steel housing and is resistant to alkaline environments such as concrete. The resistive sensor must be calibrated in an ice-water mixture (0°C) and boiling water (100°C) before use.

[0073] The preferred vibrating wire material of the vibrating wire sensor is tungsten steel, and its frame is invar steel (thermal expansion coefficient ≤ 1.2×10⁻ 6 / °C) to reduce non-target thermal strain interference; a built-in triaxial MEMS accelerometer with a ±5g range is used for vibration compensation. The vibrating wire sensor is fixed to the steel frame, and the wire must be pre-tensioned according to the manufacturer's requirements before installation to ensure the frequency output is within the calibration range.

[0074] Dual-mode temperature probes are arranged in a three-dimensional grid within the concrete cap. Optionally, spacing is 0.5 m in the core area (≥1.5 m from the surface) and 1.0 m in the surface area. The dual-mode temperature probes are connected to the data acquisition module via a four-core waterproof cable 104. The resistive sensor uses a four-wire connection method and is driven by a constant current source with a voltage resolution of ≤1 μV. The vibrating wire sensor is electromagnetically excited (frequency 800-1200 Hz), and the signal is transmitted to the readout via a shielded cable with a frequency resolution of ≤0.1 Hz.

[0075] S2. Preprocess the acquired dual-mode temperature field data to eliminate electromagnetic interference and vibration noise, and perform dynamic weight allocation and data fusion on the dual-mode temperature field data to output calibrated measured temperature field data.

[0076] The process of preprocessing the dual-mode temperature field data mainly includes the following steps:

[0077] The sampling time of the dual-mode sensor is unified based on the interpolation method; the 3σ criterion is adopted to eliminate data points exceeding ±3 times the standard deviation; wavelet noise reduction and electromagnetic interference compensation are performed on the resistive sensor data; and median filtering and vibration compensation are performed on the vibrating string sensor data.

[0078] Regarding the dynamic weight allocation and data fusion of dual-mode temperature field data, specifically including:

[0079] S21. Calculate the confidence index of the dual-mode sensor integrated by the dual-mode temperature probe respectively: ,in, It is the confidence index of the resistive sensor in the dual-mode temperature probe. It is the confidence index of the vibrating wire sensor in the dual-mode temperature probe; is the short-term temperature standard deviation of the resistive sensor in the dual-mode temperature probe, is the short-term temperature standard deviation of the vibrating wire sensor in the dual-mode temperature probe.

[0080] S22. Determine the fusion coefficient of dual-mode data based on the intensity of environmental interference : .

[0081] S23. Calculate the fusion temperature of the dual-mode data, which satisfies the expression:

[0082]

[0083] Where, 、 They are the collected dual-mode temperature field data respectively.

[0084] S3. Based on the unsteady-state heat transfer equation and hydration heat model, a finite element heat conduction model is constructed, and the total hydration heat and temperature difference trend are inverted through the measured temperature field data; combined with the LSTM neural network model, the temperature difference between the concrete core and surface and the temperature difference between the surface and the environment in the future set time period are predicted.

[0085] In a further embodiment of the present invention, the construction of the finite element heat conduction model includes the following steps:

[0086] S31. Use non-uniform hexahedral mesh to divide the concrete structure, and use tetrahedral mesh to encrypt the area near the cooling pipe.

[0087] S32, embedded hydration heat model , which satisfies the expression:

[0088]

[0089] Where, is the total hydration heat, in J / m³; Degree of hydration: ,in, The relevant material parameters are iteratively updated through the Levenberg-Marquardt algorithm to minimize the residual error between the measured temperature field data and the predicted temperature field data.

[0090] The finite element model provides an accurate forward calculation tool for the heat inversion algorithm by adjusting the material parameters. and hydration heat parameters , which can iteratively approximate the measured temperature field data.

[0091] In this embodiment, the forward calculation of the finite element heat conduction model uses the Crank-Nicolson format for time discretization, and GPU parallel computing is used to improve the calculation speed.

[0092] In a further embodiment of the present invention, the process of inverting the total heat of hydration from the measured temperature field data includes the following steps:

[0093] S33. Input the measured temperature field data into the unsteady heat transfer equation to invert the total hydration heat. , where the unsteady heat transfer equation satisfies the expression:

[0094]

[0095] Where, is the density of concrete, in kg / m³; is the specific heat capacity of concrete, the unit is J / (kg·K); is the thermal conductivity of concrete, the unit is W / (m·K); is the boundary heat exchange capacity, the unit is W / m³.

[0096] S34. Call the finite element heat conduction model to calculate the temperature field .

[0097] S35. Calculate residuals ,If the residual is greater than 1.0℃, the formula parameters are updated using the Levenberg-Marquardt algorithm.

[0098] S36. Repeat the iteration until the residual converges or the maximum number of iterations is reached.

[0099] In a further embodiment, the training method of the LSTM neural network model includes: its input includes the historical temperature series of the concrete core and surface, environmental parameters, cooling water flow rate and cooling water temperature; its output is the temperature prediction value of the future set time period; wherein, the loss function of the LSTM neural network model adopts mean square error.

[0100] S4. Based on the temperature difference results predicted by the LSTM neural network model, the cooling water flow rate and cooling water temperature are closed-loop controlled through a dynamic optimization strategy to ensure that the temperature difference fluctuation is stable within the safety threshold.

[0101] In this embodiment, hierarchical control is triggered based on the temperature difference results predicted by the LSTM neural network model:

[0102] When the predicted temperature difference exceeds the first-level threshold (ΔT ≥ 22° C. in this embodiment), the cooling water flow rate is increased according to the PID control algorithm.

[0103] When the predicted temperature difference exceeds the secondary threshold (ΔT ≥ 25° C. in this embodiment), the refrigerator is started to reduce the cooling water inlet temperature.

[0104] When the predicted temperature difference reaches the threshold of the fuse mechanism (ΔT ≥ 28°C in this embodiment), pouring is suspended, liquid nitrogen emergency refrigeration is started, and an alarm signal is pushed to the monitoring terminal.

[0105] Optionally, the adjustment step of the cooling water flow rate is ±0.2 m / s; the adjustment step of the cooling water temperature is ±2°C.

[0106] In a further embodiment of the present invention, the cooling water flow rate is increased according to a PID control algorithm, which satisfies the expression:

[0107]

[0108] Where, is the deviation between the predicted temperature difference and the target value; is the current cooling water flow rate; 、 、 They are proportional coefficient, integral coefficient and differential coefficient respectively, and are calculated according to the Ziegler-Nichols method. 、 、 Parameter tuning of .

[0109] In this embodiment, the dynamic optimization strategy mainly includes rolling time domain control and exception handling mechanism.

[0110] Rolling horizon control: This is to optimize the control sequence of cooling water flow rate and cooling water temperature for the future rated time at a fixed period. In this embodiment, the control sequence for the next two hours is re-optimized every 30 minutes. The objective function of rolling horizon control is:

[0111]

[0112] Where, To control the number of times; is the temperature difference; is the cooling water flow rate; is the cooling water temperature.

[0113] Abnormal handling mechanism: If the temperature difference drop rate is less than 5% after three or more consecutive adjustments, the expert system diagnosis will be triggered. Possible causes include sensor failure, material parameter deviation, etc.

[0114] Through the above-mentioned closed-loop control system, the engineering-level temperature control goals of temperature difference prediction accuracy of ±1.5°C, control response time of <15 minutes, and maximum temperature difference fluctuation of ≤3°C can be achieved.

[0115] See also Figure 3-4 The present invention also provides a hydration heat monitoring and dynamic control system for the large-volume concrete pouring process using the above-mentioned method, which mainly includes: a dual-mode temperature measuring probe, a dual-mode data fusion module, a finite element heat conduction model, a heat inversion module, a dynamic control module, an edge computing module and a digital twin platform.

[0116] The dual-mode temperature measuring probe is pre-buried in the concrete foundation, and the dual-mode temperature measuring probe integrates a resistive and vibrating string sensor or a resistive and fiber grating sensor.

[0117] The dual-mode data fusion module is used to receive the raw data from the dual-mode temperature measurement probe and output the calibrated measured temperature field data through a dynamic weighted fusion algorithm.

[0118] The finite element heat conduction model is based on the unsteady-state heat transfer equation to construct a three-dimensional heat conduction model of concrete. The three-dimensional heat conduction model of concrete satisfies the expression:

[0119]

[0120] Where, is the density of concrete, in kg / m³; is the specific heat capacity of concrete, the unit is J / (kg·K); is the thermal conductivity of concrete, the unit is W / (m·K); is the boundary heat exchange capacity, the unit is W / m³; is the total hydration heat, in J / m³; These are material-related parameters and are preset according to the concrete type.

[0121] The heat inversion module uses an optimization algorithm to iteratively update the relevant parameters in the finite element heat conduction model. Based on the optimized parameters, it combines the LSTM neural network model to predict the temperature difference between the concrete core and surface and the temperature difference between the surface and the environment in the future set time period.

[0122] The dynamic control module is configured to trigger hierarchical control of the cooling water flow rate and the cooling water temperature based on the comparison result of the predicted temperature difference result and the threshold value, and trigger the finite element heat conduction model to recalculate the temperature field.

[0123] The edge computing module is used to deploy the FPGA chip on one side of the dual-mode sensor to achieve a fusion computing delay of less than 50ms.

[0124] The digital twin platform is used to establish a three-dimensional temperature field mapping relationship between the dual-mode sensor and the concrete structure, correct the spatial temperature measurement deviation, and construct a three-dimensional heat conduction model of concrete.

[0125] In this embodiment, the dual-mode temperature probe 100 comprises a stainless steel armored housing 101, which is encapsulated with epoxy resin filler 102. The dual-mode sensors are separated by a stainless steel partition wall 103. A four-core waterproof cable 104 extends from one end of the stainless steel armored housing 101.

[0126] The resistive sensor includes a platinum resistance wire 110, a ceramic substrate 111, and two RTD leads 112. The vibrating wire sensor includes an excited element 120 and an excitation / pickup coil 121, arranged in parallel. These elements are arranged axially along a steel wire 122, from which two vibrating wire signal lines 123 extend. The vibrating wire sensor also includes an invar frame 124.

[0127] This invention uses fiber Bragg grating sensors instead of vibrating wire sensors to achieve higher-precision temperature measurement. This technical solution requires redesigning the data fusion algorithm, but the core heat inversion logic remains unchanged.

[0128] In addition, the present invention can also use an LSTM neural network model to replace the finite element heat conduction model to directly predict heat distribution and simplify the complexity of the physical model. Its technical solution needs to rely on a large amount of historical data training.

[0129] The core innovation of the present invention lies in the combination of dual-mode temperature measurement and heat inversion. The above-mentioned alternative solutions can achieve similar goals in specific scenarios, but the technical solution provided by the present invention has more advantages in engineering applicability and cost control.

[0130] While 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 these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring and dynamically controlling hydration heat during mass concrete pouring, characterized in that: The method comprises: Embedding multiple dual-mode temperature probes in the concrete to collect real-time dual-mode temperature field data inside the concrete and simultaneously collect environmental parameters; wherein the dual-mode temperature probes integrate resistive and vibrating wire sensors or resistive and fiber Bragg grating sensors; Preprocessing the acquired dual-mode temperature field data to eliminate electromagnetic interference and vibration noise, and dynamically assigning weights and fusing the dual-mode temperature field data to output calibrated measured temperature field data; A finite element heat conduction model is constructed based on the unsteady-state heat transfer equation and the hydration heat model, and the total hydration heat and temperature difference trend are inverted using the measured temperature field data. The temperature difference between the concrete core and the surface and the temperature difference between the surface and the environment are predicted in a future set period using an LSTM neural network model. The finite element heat conduction model is constructed by the following steps: The concrete structure is divided using non-uniform hexahedral meshes, and tetrahedral meshes are used to refine the mesh near the cooling pipes. Embedded hydration heat model , which satisfies the expression: Where, is the total hydration heat, in J / m³; is the degree of hydration: ,in, is a relevant material parameter and is iteratively updated through the Levenberg-Marquardt algorithm to minimize the residual error between the measured temperature field data and the predicted temperature field data; According to the temperature difference results predicted by the LSTM neural network model, the cooling water flow rate and cooling water temperature are closed-loop regulated through a dynamic optimization strategy to ensure that the temperature difference fluctuation is stable within a safe threshold.

2. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 1, wherein: The temperature difference result predicted by the LSTM neural network model triggers hierarchical regulation: When the predicted temperature difference exceeds the first level threshold, the cooling water flow rate is increased according to the PID control algorithm; When the predicted temperature difference exceeds the secondary threshold, the refrigerator is started to lower the cooling water temperature.

3. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 2, wherein: The adjustment step of the cooling water flow rate is ±0.2 m / s; the adjustment step of the cooling water temperature is ±2°C.

4. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 2, wherein: The cooling water flow rate is increased according to the PID control algorithm, which satisfies the expression: Where, is the deviation between the predicted temperature difference and the target value; is the current cooling water flow rate; 、 、 They are proportional coefficient, integral coefficient and differential coefficient respectively, and the above 、 、 Parameter tuning of .

5. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 1, wherein: The dynamic optimization strategy includes rolling horizon control and exception handling mechanisms, where: The rolling time domain control is a control sequence that optimizes the cooling water flow rate and cooling water temperature at future rated times in a fixed cycle; The abnormal handling mechanism is that if the temperature difference drop rate after three or more consecutive adjustments is less than <5%, the expert system diagnosis will be triggered.

6. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 5, characterized in that: The objective function of the rolling horizon control is: Where, To control the number of times; For the Maximum temperature difference between inside and outside during the first regulation; is the cooling water flow rate; is the cooling water temperature.

7. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 1, wherein: The preprocessing of the acquired dual-mode temperature field data specifically includes: Unify the sampling time of dual-mode sensors based on interpolation method; The 3σ criterion was used to eliminate data points exceeding ±3 times the standard deviation; Wavelet noise reduction and electromagnetic interference compensation are performed on the resistive sensor data; median filtering and vibration compensation are performed on the vibrating string sensor data.

8. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 1, wherein: The dynamic weight allocation and data fusion of the dual-mode temperature field data specifically includes: Calculate the confidence index of the dual-mode sensor integrated in the dual-mode temperature measuring probe respectively: ,in, is the confidence index of the resistive sensor in the dual-mode temperature probe, is a confidence index of the vibrating wire sensor in the dual-mode temperature probe; is the short-term temperature standard deviation of the resistive sensor in the dual-mode temperature probe, is the short-term temperature standard deviation of the vibrating wire sensor in the dual-mode temperature probe; Determine the fusion coefficient of dual-mode data according to the intensity of environmental interference : ; The fusion temperature of the dual-mode data is calculated, which satisfies the expression: Where, 、 are the dual-mode temperature field data collected respectively.

9. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 1, wherein: The step of inverting the total hydration heat by the measured temperature field data is as follows: inputting the measured temperature field data into the unsteady heat transfer equation, inverting the total hydration heat , where the unsteady heat transfer equation satisfies the expression: Where, is the density of concrete, in kg / m³; is the specific heat capacity of concrete, the unit is J / (kg·K); is the thermal conductivity of concrete, the unit is W / (m·K); is the boundary heat exchange capacity, the unit is W / m³.

10. The method for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 1, wherein: The training method of the LSTM neural network model includes: Its input includes the historical temperature series of the concrete core and surface, the environmental parameters, the cooling water flow rate and the cooling water temperature; Its output is the temperature forecast value for the set period in the future; The loss function of the LSTM neural network model adopts mean square error.

11. A system for monitoring and dynamically controlling hydration heat during mass concrete pouring according to any one of claims 1 to 10, characterized in that: include: A dual-mode temperature probe, embedded in the concrete cap, integrating a resistive and vibrating-wire sensor or a resistive and fiber grating sensor; A dual-mode data fusion module is used to receive the raw data from the dual-mode temperature measuring probe and output calibrated measured temperature field data through a dynamic weighted fusion algorithm; Finite element heat conduction model, which is based on the unsteady-state heat transfer equation to construct a three-dimensional heat conduction model of concrete; A heat inversion module, which uses an optimization algorithm to iteratively update the relevant parameters in the finite element heat conduction model and, based on the optimized parameters, combines an LSTM neural network model to predict the temperature difference between the concrete core and the surface, and the temperature difference between the surface and the environment, for a set period in the future; as well as The dynamic control module is configured to trigger hierarchical control of the cooling water flow rate and the cooling water temperature according to the comparison result of the predicted temperature difference result and the threshold value, and trigger the finite element heat conduction model to recalculate the temperature field.

12. The system for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 11, characterized in that: The concrete three-dimensional heat conduction model satisfies the expression: Where, is the density of concrete, in kg / m³; is the specific heat capacity of concrete, the unit is J / (kg·K); is the thermal conductivity of concrete, the unit is W / (m·K); is the boundary heat exchange capacity, the unit is W / m³; is the total hydration heat, in J / m³; These are material-related parameters and are preset according to the concrete type.

13. The system for monitoring and dynamically controlling hydration heat during mass concrete pouring according to claim 11, characterized in that: The dual-mode temperature probe is provided with a stainless steel armored shell and is potted with epoxy resin. A vibrating string signal line and an RTD lead are led out from the inside to the outside at one end of the stainless steel armored shell. The dual-mode sensors are separated by a stainless steel partition wall, wherein the vibrating string sensor is provided with an Invar frame.

Images