Intelligent temperature monitoring, early warning and regulating method and system for mass concrete

Abstract

The invention discloses an intelligent temperature monitoring, early warning and regulating method and system for mass concrete. The method comprises the following steps: constructing a long-short-term memory environment prediction model based on historical climate data, and then dynamically predicting environment parameters such as air temperature, humidity and wind speed to obtain an environment prediction data set; constructing a temperature-humidity-chemical-mechanical multi-field coupling analysis model, and correcting the model by using monitoring data, so as to comprehensively predict the temperature, humidity and mechanical response state in combination with an environment prediction result, and obtaining a mass concrete state prediction data set; further, an intelligent early warning system based on crack risk grade evaluation is constructed, a prediction result is comprehensively analyzed, the risk grade of concrete is judged, and a PID control system is used for implementing temperature regulation and control; and finally, an integrated visual display platform is established, and the whole process of monitoring, prediction, early warning and regulation is dynamically displayed and managed.

Description

Technical Field

[0001] This invention relates to the field of concrete construction technology, and in particular to a method and system for intelligent temperature monitoring, early warning and control of large-volume concrete. Background Technology

[0002] With the increasing demands for quality and safety in the construction industry regarding large-volume concrete construction, effectively monitoring and controlling changes in multiple physical fields such as temperature, humidity, and stress during the construction process has become a key factor in ensuring structural quality and safety. Temperature changes significantly impact concrete performance (such as strength and shrinkage) during construction, especially in large-volume concrete construction. Due to the significant temperature gradient difference between the interior and exterior of the concrete, temperature stress accumulation can easily lead to cracks, thus affecting the structure's durability and safety. Therefore, temperature control and monitoring are crucial for large-volume concrete construction.

[0003] Traditional methods for monitoring the temperature of large-volume concrete typically rely on manual inspection and the deployment of single temperature sensors. These methods cannot comprehensively and in real-time reflect changes in physical quantities such as temperature, humidity, and stress both inside and outside the concrete. With the development of sensing and intelligent technologies, various sensors are used to monitor physical quantities of concrete such as temperature, humidity, stress, and displacement, as well as climatic conditions at the construction site (such as temperature, humidity, and wind speed). However, existing monitoring systems often cannot achieve comprehensive analysis and prediction of multiple physical quantities, and lack timely early warning and control measures for problems such as cracking caused by temperature changes.

[0004] To improve the safety and quality of large-volume concrete construction, researchers have recently explored intelligent methods for temperature monitoring, early warning, and control during the concrete construction process. For example, internal sensors are used to collect concrete state data, which is then combined with an intelligent cloud platform to achieve real-time monitoring and early warning of temperature, humidity, and stress after the structure is poured. Another example is using finite element analysis (FEM) techniques to simulate and analyze the concrete state, determining changes in various physical fields within the structure after pouring, thereby guiding preventative measures for high-risk areas.

[0005] However, existing technologies have certain limitations in practical applications. For example, current methods lack the ability to adapt in real time to complex environmental changes at construction sites, and there are still some human intervention factors involved in the optimization of temperature control schemes. Therefore, how to effectively combine performance monitoring data, predictive models, and temperature control systems to improve the accuracy and intelligence of temperature regulation remains an urgent problem to be solved. Summary of the Invention

[0006] This invention provides a method for intelligent temperature monitoring, early warning, and control of large-volume concrete, solving the problems existing in the prior art. It includes the following operational steps:

[0007] Internal sensors are installed inside the concrete at the construction site, and environmental sensors are installed outside the concrete. The internal sensors are used to monitor the temperature, humidity, stress, and displacement inside the concrete, and the external environmental sensors are used to monitor the temperature, humidity, and wind speed of the external environment.

[0008] We collected climate and environmental data from the construction site and local historical climate and environmental data, and built a dynamic environmental prediction model based on a long short-term memory neural network (LSTM) to obtain a construction environment prediction dataset.

[0009] Using the construction environment data monitoring dataset and the dataset collected by internal sensors, a multi-field coupling prediction model for large-volume concrete is constructed based on the embedded concrete thermo-humidification-chemical-mechanical multi-field coupling control equation and the physical information neural network (PINN) algorithm with finite boundary conditions.

[0010] The multi-field coupling prediction model predicts the state of concrete based on the construction environment monitoring dataset, and obtains the predicted results of internal temperature, humidity, stress and displacement of large-volume concrete.

[0011] The predicted concrete condition is analyzed using an early warning system based on crack risk levels, and the risk level is output.

[0012] When the risk level exceeds the preset threshold, the PID control system is activated. Under intelligent decision-making or manual intervention mode, the PID parameters are adaptively adjusted and the heat preservation and cooling device is linked to regulate the temperature of the concrete.

[0013] The monitoring results, prediction results, early warning information and temperature control data generated by the above operation steps are input into the visualization platform for display, and the visualization platform enables real-time management of the monitoring, prediction and control process.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] 1. This invention constructs a closed-loop system encompassing on-site monitoring, environmental prediction, model analysis, and intelligent temperature control execution for crack risk early warning. This system integrates monitoring, decision-making, and execution, proactively identifying and addressing temperature crack risks during the construction phase, significantly improving the safety and reliability of the construction process. Furthermore, by organically combining a data-driven dynamic environmental prediction model with a physics-driven multi-field coupling model, the system achieves good on-site adaptability and prediction accuracy while maintaining physical consistency, thus overcoming the inherent limitations of single data or single physical models.

[0016] 2. This invention employs a multi-parameter monitoring system, which, compared to methods that only monitor temperature, more directly reflects the stress-strain evolution process within concrete, providing a more reliable physical basis for crack risk assessment and proactive control. Furthermore, the proposed intelligent temperature control system, based on crack risk levels and combined with an adaptive PID control system, can reduce temperature gradients and crack risk while ensuring control stability, minimizing human intervention and misoperation, and improving control effectiveness and energy efficiency.

[0017] 3. This invention incorporates multiple methods such as Bayesian optimization, cross-validation, and field data calibration in model fitting and parameter calibration, which improves the generalization ability and robustness of the model, enabling the model to maintain high reliability even in complex field environments and with incomplete data.

[0018] In summary, this invention integrates multi-source monitoring, environmental parameter prediction, model coupling analysis, risk classification prediction, and adaptive control to achieve precise perception and proactive control of the concrete state during the construction of large-volume concrete, thereby effectively reducing the probability of early cracking and improving project quality and construction safety.

[0019] Preferably, the internal sensors and external environmental sensors are used to collect concrete condition monitoring datasets and construction environment datasets, respectively. The specific operation steps are as follows:

[0020] The internal sensors are used to monitor the temperature, humidity and stress changes at various internal measuring points of the concrete during the pouring and curing process, and to obtain concrete condition monitoring data.

[0021] The external environment sensor acquires environmental data at the construction site during concrete pouring and curing.

[0022] Preferably, the concrete condition data includes the internal temperature, humidity, stress, and displacement of the concrete after pouring and during curing.

[0023] The construction environment data includes external temperature, humidity, and wind speed during concrete construction.

[0024] Preferably, historical climate and environmental data are collected to construct a dynamic prediction model; the dynamic prediction model is corrected using the construction environment data to obtain a construction environment prediction dataset; then, the construction environment data and the concrete condition monitoring dataset are combined to construct a multi-field coupled prediction model for large-volume concrete; finally, the construction environment prediction dataset is input into the prediction model to obtain a large-volume concrete condition prediction dataset. Specifically, this includes:

[0025] Historical climate and environmental data of the construction site were collected; a dynamic prediction model was constructed using an LSTM model as a framework.

[0026] The construction environment data is analyzed using the dynamic prediction model to obtain a construction environment prediction dataset;

[0027] Based on the equations of energy conservation, mass conservation, and momentum conservation, a constitutive equation for concrete is introduced to construct a multi-field coupled analysis model of temperature, humidity, chemistry, and mechanics.

[0028] Based on the construction environment data and the concrete condition monitoring dataset, the analysis model is fitted using a hybrid method of numerical simulation and data-driven approach to obtain the final multi-field coupled prediction model for large-volume concrete. The construction environment prediction dataset is then input into the prediction model for calculation to obtain the large-volume concrete condition prediction dataset.

[0029] Preferably, historical climate and environmental data of the construction site are collected; a dynamic prediction model is constructed using the historical climate and environmental data and an LSTM model as a framework, and the construction environment prediction data is obtained through the model. The specific operation steps are as follows:

[0030] The historical climate and environmental data are collected to construct a historical climate and environmental database;

[0031] Time series processing is performed on the historical climate and environment database to extract key climate features and analyze their dynamic change patterns.

[0032] A feature matrix and target variables are constructed based on the dynamic changes described above;

[0033] Using the LSTM model as a framework for machine learning, the historical climate and environment database is divided into training and validation sets.

[0034] The training set is input into the LSTM model for training, and an optimization algorithm is used to improve the model's convergence performance and prediction accuracy.

[0035] The model is then validated using a validation set to obtain a dynamic prediction model that can be applied to construction sites.

[0036] Finally, the construction environment data is input into the dynamic prediction model, and the construction environment data prediction dataset is output.

[0037] Preferably, based on the principles of energy conservation, mass conservation, and hydration reaction kinetics, an elastoplastic constitutive relation for concrete is introduced to construct a multi-field coupled analysis model of temperature, humidity, chemistry, and mechanics. This multi-field coupled analysis model is then fitted using on-site measured temperature, humidity, and stress monitoring data to obtain the final multi-field coupled prediction model for large-volume concrete. The specific operation steps are as follows:

[0038] Based on the equations of energy conservation, mass conservation, and momentum conservation, the constitutive relation equation of concrete is introduced, and a multi-field coupled analysis model of temperature-humidity-chemical-mechanical is constructed to obtain the partial differential control equations of multi-field coupling.

[0039] The parameters in the coupled partial differential equations of thermo-humidification-mechanics multi-field coupling of the above-mentioned large-volume concrete were determined by actual construction measurements. The parameters include influencing factors such as density, thermal conductivity, specific heat capacity, humidity diffusion coefficient, thermal expansion coefficient, shrinkage coefficient, hydration heat release function, boundary condition function, and elastic modulus.

[0040] Using a physical information neural network as the machine learning framework, a multi-field coupled partial differential equation of heat-humidification-chemical-mechanical multifield is embedded as the physical loss function for machine learning, and finite element technology is used to handle the complex boundary conditions of the model, thereby establishing a multi-field coupled prediction model framework for large-volume concrete.

[0041] By combining the parameter values, the construction environment data, and the concrete condition monitoring dataset, the framework of the multi-field coupling prediction model for large-volume concrete is trained and fitted to obtain the final multi-field coupling prediction model for large-volume concrete, and the large-volume concrete condition prediction dataset is output.

[0042] Preferably, the large-volume concrete state prediction dataset is predicted using the aforementioned multi-field coupled prediction model. The specific operation steps are as follows:

[0043] The construction environment data prediction dataset is input into the large-volume concrete multi-field coupling prediction model, and the large-volume concrete state prediction dataset is output.

[0044] Accordingly, this application also provides an intelligent temperature monitoring, early warning and control system for large-volume concrete, characterized by comprising: a data acquisition module; an analysis module; an early warning and control module; and a visualization module;

[0045] The acquisition module is used to install internal sensors and external environmental sensors on the concrete at the construction site; and to collect concrete condition monitoring data and construction environment data using the internal sensors and external environmental sensors, respectively.

[0046] The analysis module is used to collect historical climate and environmental data to construct a dynamic prediction model; input the construction environment data into the dynamic prediction model to obtain a construction environment data prediction dataset; construct a multi-field coupled prediction model for large-volume concrete using the construction environment data and concrete condition monitoring data; and predict the concrete condition parameter set using the multi-field coupled prediction model for large-volume concrete to obtain a large-volume concrete condition prediction dataset.

[0047] The early warning and control module is used to analyze the construction environment data prediction dataset and the large volume concrete state prediction dataset through the early warning system. When the concrete is abnormal, the temperature is controlled by the PID control system.

[0048] The visualization module inputs the above operation steps into a visualization display platform, and the visualization platform enables real-time management of the monitoring, prediction, and control process.

[0049] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages:

[0050] Analysis of the above-mentioned intelligent temperature monitoring, early warning and control method and system for large-volume concrete provided by the present invention shows that, in specific applications, firstly, internal sensors and external environmental sensors are set up at the construction site to collect monitoring data on the internal state of the concrete after pouring and external environmental data during the concrete construction period; in this way, the changes in physical quantities such as internal temperature, humidity, stress, and displacement of the concrete after pouring, as well as the changes in environmental parameters such as ambient temperature, humidity, and wind speed during the concrete construction period can be monitored.

[0051] Furthermore, collecting historical climate data from the construction site forms the basis for a dynamic prediction model, which helps to understand and predict the performance of concrete under different environmental conditions. The hardening process and performance of large-volume concrete are significantly affected by environmental factors. Using an LSTM model as a framework, climate data from a past period is received, and the data is learned through multiple LSTM layers. Finally, the prediction results are output to obtain a dynamic prediction model, which can accurately predict future environmental variables such as temperature and humidity. Based on theory, a multi-field coupling analysis model is constructed to couple temperature, humidity, chemical reactions, and the mechanical properties of concrete. Through multi-field coupling analysis, the performance of concrete can be described and predicted more accurately. By fitting the multi-field coupling analysis model, the final multi-field coupling prediction model for large-volume concrete is generated to ensure the predictive ability of the model in practical applications. Finally, the multi-field coupling prediction model for large-volume concrete is used to make predictions based on the construction environment data prediction dataset to obtain a large-volume concrete state prediction dataset.

[0052] Furthermore, the construction environment data prediction dataset and the large-volume concrete condition prediction dataset are input into the prediction system to predict temperature cracks in large-volume concrete. Based on the pre-set crack risk level, graded early warnings are issued, and environmental parameters are adjusted based on the early warning results. All data predictions and operation steps are input into the visualization platform, and the operation platform is monitored and adjusted in real time. Attached Figure Description

[0053] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. The accompanying drawings below illustrate some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0054] Figure 1 A flowchart of an intelligent temperature monitoring, early warning and control method for large-volume concrete provided in Embodiment 1 of the present invention;

[0055] Figure 2 A flowchart illustrating the acquisition of a large-volume concrete state prediction dataset for the intelligent monitoring, early warning, and control method for large-volume concrete provided in this embodiment of the invention.

[0056] Figure 3 A flowchart of an intelligent temperature monitoring, early warning and control system for large-volume concrete provided in Embodiment 2 of the present invention;

[0057] Labels: Data Acquisition Module 10; Analysis Module 20; Early Warning and Control Module 30; Visualization Module 40. Detailed Implementation

[0058] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0060] Example 1

[0061] like Figure 1 As shown, this invention proposes a method for intelligent temperature monitoring, early warning, and control of large-volume concrete, including the following operational steps:

[0062] S1: Install internal sensors and external environment sensors on the concrete at the construction site; use the internal sensors and external environment sensors to collect concrete condition monitoring datasets and construction environment data, respectively.

[0063] The internal sensors collect monitoring data on the state of the concrete after pouring.

[0064] The construction environment data is collected through the external environment sensors.

[0065] The concrete condition monitoring dataset includes internal temperature, humidity, stress, and displacement of the concrete after pouring.

[0066] The environmental data includes ambient temperature, humidity, and wind speed during concrete construction;

[0067] It should be noted that, firstly, sensor measuring points are set up inside the concrete according to the design drawings to monitor changes in physical quantities such as internal temperature, humidity, stress, and displacement before and after concrete pouring (i.e., monitoring the material's own hydration reaction, hardening process, and concrete state (such as changes in temperature and stress)); secondly, environmental monitoring sensors are set up near the project to monitor changes in environmental parameters such as ambient temperature, humidity, and wind speed during concrete construction (i.e., monitoring the impact of the surrounding environment on the concrete surface and overall behavior (such as changes in evaporation rate caused by wind speed)).

[0068] S2: Collect historical climate and environmental data to construct a dynamic prediction model; input the construction environment data into the dynamic prediction model to obtain a construction environment data prediction dataset; use the construction environment data dataset and the concrete condition monitoring dataset to construct a multi-field coupled prediction model for large-volume concrete; predict the concrete condition parameters through the multi-field coupled prediction model for large-volume concrete to obtain a large-volume concrete condition prediction dataset.

[0069] It should be noted that collecting historical climate data from the construction site, including information on temperature, humidity, precipitation, and wind speed, is fundamental to building a dynamic prediction model. This data helps establish a comprehensive environmental context, aiding in understanding and predicting the performance of concrete under different environmental conditions. The hardening process and performance of mass concrete are significantly affected by environmental factors (such as temperature, humidity, and chemical reactions); therefore, historical climate data provides essential background information for the model.

[0070] The LSTM model is used to capture long-term dependencies in time series data. By inputting historical climate and environmental data into the LSTM model, the system can dynamically predict future environmental change trends based on past data. The construction environment is dynamic, and climatic factors such as temperature and humidity have temporal correlations. LSTM can capture this time series characteristic, making it an ideal tool for climate prediction.

[0071] After using a dynamic prediction model based on an LSTM model to predict environmental data at the construction site, a "construction environment prediction dataset" is generated. This dataset contains predicted values ​​of the construction environment (such as temperature and humidity) for a future period. This provides accurate environmental input data for subsequent multi-field coupling analysis. Accurate environmental prediction is a prerequisite for large-volume concrete prediction models. The physical properties and chemical reactions of concrete vary significantly under different environmental conditions, thus requiring precise environmental data for subsequent analysis.

[0072] A multi-field coupled analysis model was constructed using a construction environment data prediction dataset to couple temperature, humidity, chemical reactions (such as cement hydration), and the mechanical properties of concrete (such as stress and strain). This model can comprehensively consider the interactions between multiple physical fields, such as the influence of temperature on humidity and humidity on chemical reactions. The behavior of large-volume concrete is affected by a variety of interrelated factors. For example, temperature affects the rate of concrete hydration, humidity affects the drying rate, and these, in turn, affect mechanical properties. Through multi-field coupled analysis, the performance of concrete can be described and predicted more accurately.

[0073] By fitting a multi-field coupling analysis model, a final multi-field coupling prediction model for large-volume concrete is generated. This model can accurately predict the behavior of large-volume concrete under different environmental conditions during construction (such as stress, temperature, and humidity distribution), taking into account the interactions between multiple factors. The fitting process allows the model to better adapt to actual construction site data, ensuring the model's predictive ability in practical applications. This step is crucial for combining the theoretical model with the actual environment; fitting improves the model's accuracy and reliability.

[0074] S3: The construction environment data prediction dataset and the large volume concrete state prediction dataset are analyzed through the early warning system. When the concrete is abnormal, the temperature is controlled by the PID control system.

[0075] It should be noted that the two datasets obtained from the data analysis module (construction environment data prediction dataset and large volume concrete state prediction dataset) are imported into the pre-trained intelligent early warning subsystem. Through data analysis, regional and hierarchical early warnings are issued for possible temperature cracks in large volume concrete, and early warnings are also issued for climate change situations that may increase the risk of temperature cracks in the region in the future.

[0076] In response to the early warning of temperature crack risk, the intelligent control subsystem is activated. Based on existing predictive model data, the optimal temperature control scheme is derived through analysis, and the PID control system is activated to achieve automatic and precise temperature control.

[0077] The above steps involve two subsystems. The first is the intelligent early warning subsystem, whose working steps are as follows: Based on real-time monitoring data and machine learning model prediction data, a three-level early warning mechanism of temperature gradient / tensile stress / displacement is generated through threshold setting, thereby classifying and issuing early warnings for the risk of temperature cracks in large-volume concrete, and displaying the risk map in the user interface module; the second is the intelligent control subsystem, whose working steps are as follows:

[0078] Based on the intelligent early warning subsystem, a loss function is constructed that includes temperature uniformity, cooling rate, and energy consumption cost (where temperature uniformity is measured by the maximum temperature difference at the same time, i.e.) The objective is to minimize U; the cooling rate is expressed as the rate of temperature decrease per unit time, i.e. The objective is to minimize -R; energy cost is calculated by dividing total energy consumption by unit energy cost, denoted as P. Then, through normalization of the three indicators, we obtain... Then, the three factors are weighted and summed to construct the loss function L, i.e. The optimal temperature control scheme for large-volume concrete was derived using a deep reinforcement learning strategy. The relevant parameters were then substituted into the PINN large-volume concrete model for retraining to predict the subsequent temperature control effect, thereby further optimizing the control scheme and displaying it visually.

[0079] Then, by using a PID controller to link with equipment such as a water storage and maintenance device and a misting spray system (for areas where water storage and maintenance cannot be carried out), precise temperature control can be achieved.

[0080] S4: Input the above operation steps into the visualization platform, and display the construction environment data prediction structure, concrete state prediction results, crack risk level monitoring results, and temperature control process through the visualization platform, so as to display and manage the entire process of monitoring, prediction, early warning and control in real time.

[0081] It should be noted that data integration and visualization platform display are steps to integrate the datasets from the first three steps and import them into the user interface module for data visualization. Specifically: 1) Integrate data from the data collection module, data analysis module, and intelligent early warning and control module; 2) Import the data into the visualization platform, render the data, and display it to users through multi-terminal synchronization.

[0082] Note that the visualization function of the user interface module refers to building a 3D visualization platform by integrating BIM models, real-time monitoring data, and machine learning models. This platform enables 4D dynamic simulation of the internal temperature / stress field of concrete, providing users with multi-type and multi-dimensional visualizations, including 3D images of the building entity, temperature / stress cloud maps, temperature / stress change trend maps, and temperature crack risk distribution maps. The machine learning model is used to compensate for the lack and distortion of real-time monitoring data.

[0083] Specifically, such as Figure 2 As shown, in step S2, historical climate and environmental data are collected to construct a dynamic prediction model; the construction environment data is input into the dynamic prediction model to obtain a construction environment data prediction dataset; a multi-field coupled prediction model for large-volume concrete is constructed using the construction environment dataset and concrete condition monitoring data; the concrete condition parameters are predicted using the multi-field coupled prediction model for large-volume concrete to obtain a large-volume concrete condition prediction dataset, including:

[0084] Historical climate and environmental data of the construction site are collected; a dynamic prediction model is constructed using the historical climate and environmental data and an LSTM model as a framework; the construction environment data is analyzed using the dynamic prediction model to obtain a construction environment data prediction dataset; a temperature-humidity-chemical-mechanical multi-field coupling analysis model is established by combining the construction environment dataset with the concrete condition monitoring dataset; the temperature-humidity-chemical-mechanical multi-field coupling analysis model is fitted to obtain the final multi-field coupling prediction model for large-volume concrete; the construction environment data prediction dataset is input into the multi-field coupling prediction model for large-volume concrete to obtain a large-volume concrete condition prediction dataset. The specific operation steps are as follows:

[0085] S21: Collect the historical climate and environmental data to construct a historical climate and environmental database;

[0086] It should be noted that historical climate and environmental data were collected from the local construction sites to build a historical climate and environmental database.

[0087] S22: Extract historical climate features from the historical climate environment database by feature engineering according to time series, and analyze the dynamic change characteristics of the historical climate features;

[0088] The feature matrix and target variables are constructed based on the aforementioned dynamic changes.

[0089] Using the LSTM model as the framework for the machine learning model, the historical climate and environment database is divided into a training set and a validation set. The training set is input into the LSTM model for training, and the hyperparameters and the prediction results of the LSTM model are adjusted using the Adam optimizer and the mean squared error loss function. The LSTM model is then validated using the validation set to obtain the final dynamic prediction model.

[0090] It should be noted that feature engineering is performed on historical climate data (including variables such as temperature, humidity, wind speed, and precipitation) in time series to extract key features that reflect the patterns of climate change, and to analyze their dynamic changes based on the time series data. The purpose of feature engineering is to transform the raw data into a format that the model can understand and learn effectively. This usually includes preprocessing of the time series data, feature selection, and identification of change patterns.

[0091] The feature matrix output from the analysis of dynamic change characteristics is usually composed of climate data over a period of time, while the target variable is the future climate variable that we want to predict (such as temperature or humidity in the next few days).

[0092] LSTM (Long Short-Term Memory) is a neural network model particularly well-suited for processing time series data. LSTM can capture long-term dependencies in time series data, i.e., how past climate data influences future predictions. In this model, LSTM receives input data (such as climate data over a past period), learns from the data through multiple LSTM layers, and finally outputs prediction results.

[0093] The Adam optimizer (an efficient optimization algorithm that adaptively adjusts the learning rate of each parameter) and the Mean Squared Error (MSE) loss function (which measures the difference between the model's predictions and actual values; the smaller the MSE, the closer the model's predictions are to the actual values) optimize model performance and improve prediction accuracy during model training by adjusting hyperparameters (such as the number of LSTM layers, the number of neurons per layer, and the learning rate). Ultimately, the trained model can be used for real-time predictions. By combining real-time climate data updates, the model can be continuously adjusted to make high-precision predictions of future environmental variables such as temperature and humidity.

[0094] S23: Input the construction environment data into the dynamic prediction model and output the construction environment data prediction dataset;

[0095] S24: Based on the conservation of energy, conservation of mass, and the constitutive equation of concrete, a multi-field coupled analysis model of temperature-humidity-chemical-mechanical fields is constructed to couple the temperature field, humidity field, chemical field, and mechanical field.

[0096] The temperature-humidity-chemical-mechanical multi-field coupled analysis model is based on the conservation of energy, the conservation of mass, and the constitutive equation of concrete to calculate the relevant changes in the temperature field, humidity field, chemical field, and mechanical field, thereby constructing the temperature-humidity-chemical-mechanical multi-field coupled analysis model;

[0097] The temperature field control equation:

[0098] ;

[0099] in, Represented as density; Expressed as specific heat capacity; Expressed as heat released during hydration; Represented as the humidity-thermal coupling coefficient; Relative humidity (0~1);

[0100] Humidity field governing equation:

[0101] ;

[0102] in, It is expressed as the humidity diffusion coefficient (related to porosity); Expressed as the thermal and humid diffusivity coefficient; This is represented as a humidity source term caused by the hydration reaction (hydration consumes water).

[0103] Chemical field governing equations:

[0104]

[0105] in: It is expressed as a reaction rate function (related to porosity); It is expressed as activation energy; Expressed as the gas constant; Represented as temperature;

[0106] The governing equations of mechanics first calculate temperature strain, humidity strain, and creep effect:

[0107]

[0108] Stress-strain relationship:

[0109]

[0110] in, Represented as thermal strain, The coefficient of thermal expansion; This is expressed as humidity shrinkage strain. The shrinkage coefficient; Creep strain is expressed as the Kelvin-Voigt creep model, which characterizes the viscoelastic behavior of concrete. Its constitutive relation can be expressed as a linear superposition of stress, elastic strain, and viscous strain rate. The creep strain is calculated based on this model.

[0111] It should be noted that, based on the above governing equations, this application constructs a multi-physics coupled model of chemical-temperature-humidity-mechanical systems for large-volume concrete. The coupling mechanism is as follows: In the chemical field, the hydration reaction releases heat, affecting the temperature change and thermophysical properties in the temperature field. At the same time, the hydration products change the pore structure, affecting humidity diffusion. The degree of reaction is represented by the degree of hydration α. ​​In the temperature field, the temperature changes due to the heat release from cement hydration and temperature transfer, which simultaneously affects the hydration reaction rate in the chemical field, the humidity diffusion coefficient in the humidity field, and the mechanical property parameters of the stress field, and also causes the concrete to thermally expand and deform. Its characteristic index is temperature T. In the humidity field, the humidity changes under the influence of the temperature field and the chemical field, generating a humidity gradient, which affects heat conduction (temperature field) while causing shrinkage deformation (stress field). Its characteristic index is relative humidity H. The stress field is used to calculate the mechanical effects of concrete under the influence of temperature and humidity.

[0112] Chemical-temperature coupling via heat source term :

[0113] ;

[0114] In the formula, This represents the heat of hydration released at the degree of hydration α. Indicates the final heat of hydration;

[0115] Specific heat capacity :

[0116] ;

[0117] In the formula This indicates the specific heat capacity of concrete. This indicates the specific heat capacity of cement after hydration. This indicates the specific heat capacity of cement. Indicates the specific heat capacity of aggregate. This indicates the specific heat capacity of water. , , These represent the amounts of cement, aggregate, and water used per cubic meter, respectively.

[0118] thermal conductivity :

[0119] ;

[0120] In the formula Indicates the final thermal conductivity;

[0121] Arrhenius equation (i.e., the effect of reaction temperature on the rate of a chemical reaction):

[0122] ;

[0123] Chemical-humidity coupling occurs via a humidity consumption term (water consumption due to hydration reactions):

[0124] ;

[0125] In the formula, k is the humidity consumption coefficient of the hydration reaction;

[0126] Hydration threshold:

[0127] ;

[0128] In the formula, The threshold of relative humidity for the hydration reaction;

[0129] Chemical-mechanical coupling is expressed by the elastic modulus formula:

[0130] ;

[0131] In the formula, It is the limiting elastic modulus (i.e., the ability of a material to resist elastic deformation). These represent the final degree of hydration and the initial degree of hydration, respectively. It is an exponential constant;

[0132] Tensile strength formula:

[0133] ;

[0134] In the formula, It is the ultimate tensile modulus (i.e. the maximum stress that a material can withstand in a tensile test). It is an exponential constant;

[0135] Temperature-humidity coupling:

[0136] Thermal gradient-driven humidity diffusion (Soret effect): ;

[0137] Humidity affects temperature changes, and is considered a humidity change influence term in the temperature control equation: ;

[0138] Temperature-mechanical coupling occurs through thermal expansion and strain:

[0139] ;

[0140] Humidity-mechanical coupling through humidity contraction strain:

[0141] ;

[0142] The above-mentioned mutual coupling relationship is characterized by the multi-field coupled partial differential equation of thermo-humidification-mechanics of large-volume concrete, and the measured values ​​of influencing factors including density, thermal conductivity, specific heat capacity, humidity diffusion coefficient, thermal expansion coefficient, shrinkage coefficient, hydration heat release function, boundary condition function, and elastic modulus are determined based on actual measurements.

[0143] S25: Using physical information neural networks and the finite element method, a multi-field coupled prediction model for large-volume concrete with embedded thermal-humidification-chemical-mechanical multi-field coupled partial differential equations was established and fitted.

[0144] The construction environment dataset and the concrete condition monitoring dataset were used to fit the multi-field coupled prediction model for large-volume concrete.

[0145] It is important to note that a multi-field coupled prediction model framework for large-volume concrete, embedding thermo-humidification-mechanical multi-field coupled partial differential equations, is established using the PINN algorithm (Physical Information Neural Network (PINN) is a deep learning technique that solves partial differential equations (PDEs) related to physical systems through neural networks. The core idea of ​​PINN is to input physical laws (such as equations, boundary conditions, etc.) as additional information into the training process of the neural network, enabling the neural network to not only perform data-driven learning but also follow physical laws.) and the FEM algorithm (Finite Element Method (FEM) is a numerical method used to solve boundary value problems and partial differential equations, particularly applicable to fields such as structural mechanics, heat conduction, and electromagnetism. The basic idea of ​​FEM is to divide the continuous physical domain into discrete elements and describe the relationships between these elements through mathematical models, thereby obtaining approximate solutions.). FEM is used to handle boundary conditions, while PINN technology is used to solve the solutions within the large-volume concrete domain. Its core idea is to divide the computational domain into two regions: the boundary region (ΩFE, applying FEM) and the interior region (ΩNN, applying PINNs). In the boundary region, thermal boundary conditions (adiabatic boundary, convective heat dissipation, etc.), humidity boundary conditions, and mechanical boundary conditions (fixed boundary, free boundary, etc.) are set according to engineering realities. A local mesh is constructed using the finite element method (FEM), and the solution is obtained directly using standard FEM techniques through shape functions and node values. In the interior region, PINN is used for approximate solution. This domain decomposition strategy retains the precise control of the boundaries by FEM while utilizing PINN's flexible modeling capabilities for complex physical fields.

[0146] PINN-Enhanced FEM: By introducing PINN's physical constraints, the accuracy and stability of traditional FEM methods are improved. In traditional finite element methods, solving equations may require manual adjustment or the addition of extra physical constraints. PINN can learn these physical constraints through neural networks, further improving the quality of the solution. FEM-Enhanced PINN: Combining with the finite element method, its advantages in large-scale computation can be leveraged. The discretization of the finite element mesh improves the computational efficiency of PINN, especially when solving complex structural problems. By adding PINN constraints to the FEM model, the model can be more effectively coupled and solved between multiple physics fields.

[0147] S26: Input the construction environment data and concrete condition monitoring data into the large-volume concrete multi-field coupling prediction model framework, and then fit the large-volume concrete multi-field coupling prediction model framework embedded with the thermo-humidification-chemical-mechanical multi-field coupling partial differential equation through Bayesian optimization and cross-validation to obtain the final large-volume concrete multi-field coupling prediction model.

[0148] It should be noted that, based on the advantages of PINN physical regularization and FEM boundary conditions, Bayesian optimization and cross-validation are introduced in the large-volume concrete multi-field coupled prediction model that inputs the prior database into the thermo-humidification-chemical-mechanical multi-field coupled partial differential equation.

[0149] Bayesian optimization is an optimization method based on Bayesian theory, typically used to optimize functions that are computationally expensive or difficult to solve directly. Bayesian optimization guides the selection of subsequent experiments by updating the understanding of the objective function after each experiment, thereby finding the optimal solution with a smaller number of trials.

[0150] In this model, Bayesian optimization can be used to optimize the model's hyperparameters (such as the learning rate and number of layers in a neural network), thereby improving the model's predictive performance and efficiency.

[0151] Cross-validation is a method for evaluating model performance, typically used to avoid overfitting the model to the training data. It assesses the model's generalization ability on unseen data by dividing the dataset into multiple subsets and training and validating the model on different subsets. The purpose of introducing cross-validation is to ensure that the model does not overfit during training, i.e., to ensure that the model has good predictive ability in real-world applications.

[0152] By introducing Bayesian optimization and cross-validation, overfitting of the model is avoided, thus obtaining the final multi-field coupled prediction model for large-volume concrete.

[0153] S27: Input the construction environment data prediction dataset into the large-volume concrete multi-field coupling prediction model and output the large-volume concrete state prediction dataset.

[0154] Research has revealed that temperature cracks in concrete can be addressed by manually adjusting temperature parameters. Specifically, this provides users with an option for manual intervention and control, allowing them to determine material selection, curing measures, and other construction plans based on project requirements. The user-friendly interactive function refers to a visual display platform that provides parameter setting panels, enabling users to modify material parameters, environmental variables, or temperature control parameters, and to monitor and predict curve changes in real time, achieving user autonomy and intelligent feedback. The multi-terminal synchronization function refers to a system built on a cloud platform and local terminals (PC, mobile, etc.) to meet diverse customer operational needs, allowing users to view project status anytime, anywhere.

[0155] The following example demonstrates the specific steps for modifying concrete environmental parameters:

[0156] (1) Users call the parameter adjustment interface reserved by the visualization platform and input the climate environment parameters to be adjusted, such as ambient temperature, wind speed, humidity, etc.

[0157] (2) Substitute the adjusted parameters into the data processing module, call the multi-field coupled machine learning model for large-volume concrete to perform concrete state prediction and change analysis, and output the result dataset.

[0158] (3) Import the dataset from the previous step into the intelligent early warning and control module, call the intelligent early warning subsystem, and obtain the temperature crack risk dataset after the user modifies the parameters;

[0159] (4) Integrate the datasets generated in steps (2) and (3) and import them into the visualization platform. Use data visualization to show users the corresponding changes in the large-volume concrete model after parameter modification.

[0160] (5) The user confirms the parameter changes again and a change record is generated. The entire system will then continue to run based on the modified parameters.

[0161] Example 2

[0162] like Figure 3 As shown, the present invention also proposes an intelligent temperature monitoring, early warning and control system for large-volume concrete, comprising: a data acquisition module 10; an analysis module 20; an early warning and control module 30; and a visualization module 40.

[0163] The acquisition module 10 is used to set internal sensors and external environment sensors for the concrete at the construction site; and to use the internal sensors and external environment sensors to acquire concrete condition monitoring data and construction environment data respectively.

[0164] The analysis module 20 is used to collect historical climate and environmental data to construct a dynamic prediction model; input the construction environment data into the dynamic prediction model to obtain a construction environment data prediction dataset; construct a large-volume concrete multi-field coupled prediction model using the construction environment dataset and the concrete condition monitoring dataset; and predict the concrete condition parameters based on the construction environment data prediction dataset using the large-volume concrete multi-field coupled prediction model to obtain a large-volume concrete condition prediction dataset.

[0165] The early warning and control module 30 is used to analyze the construction environment data prediction dataset and the large volume concrete state prediction dataset through the early warning system. When the concrete is abnormal, the temperature is controlled by the PID control system.

[0166] The visualization module 40 inputs the results of the above operation steps into the visualization display platform, and realizes real-time management of the entire process of monitoring, prediction and control through the visualization platform.

[0167] In summary, the intelligent temperature monitoring, early warning, and control method and system for large-volume concrete proposed in this invention demonstrates that, in practical applications, internal sensors and external environmental sensors are first installed inside and outside the concrete to collect data on the concrete state after pouring and construction environment data during the concrete construction process. This allows for the monitoring of changes in physical quantities such as internal temperature, humidity, stress, and displacement of the concrete after pouring, as well as changes in environmental parameters such as ambient temperature, humidity, and wind speed during concrete pouring and curing.

[0168] Furthermore, collecting historical climate data from the construction site forms the basis for a dynamic prediction model, which helps to understand and predict the performance of concrete under different environmental conditions. The hardening process and performance of large-volume concrete are significantly affected by environmental factors. Using an LSTM model as a framework, climate data from a past period is received, and the data is learned through multiple LSTM layers. Finally, the prediction results are output to obtain a dynamic prediction model, which can accurately predict environmental variables such as future temperature and humidity. A multi-field coupling analysis model is constructed using the construction environment data prediction dataset and the concrete condition monitoring dataset. Temperature, humidity, chemical reactions, and the mechanical properties of concrete are coupled and analyzed. Through multi-field coupling analysis, the performance of concrete can be described and predicted more accurately. By fitting the multi-field coupling analysis model, the final multi-field coupling prediction model for large-volume concrete is generated to ensure the predictive ability of the model in practical applications. Finally, the multi-field coupling prediction model for large-volume concrete is combined with the construction environment data prediction dataset for calculation to obtain the large-volume concrete condition prediction dataset.

[0169] Furthermore, the construction environment data prediction dataset and the large-volume concrete condition prediction dataset are used to predict temperature cracks in large-volume concrete using the prediction system. Classified early warnings are issued based on the pre-set crack risk levels, and environmental parameters are adjusted based on the early warning results. All data predictions and operation steps are input into a visualization platform, and the operation platform is monitored and adjusted in real time.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; those skilled in the art can modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent temperature monitoring, early warning, and control of large-volume concrete, characterized in that, The following steps are included: Internal sensors are installed inside the concrete at the construction site, and environmental sensors are installed outside the concrete; the internal sensors are used to monitor the internal temperature, humidity, stress, and displacement of the concrete, and the external environmental sensors are used to monitor the external ambient temperature, humidity, and wind speed. Collect on-site climate and environmental data and local historical climate and environmental data, and build a dynamic environmental prediction model based on long short-term memory neural network to obtain a construction environment prediction dataset; Using the construction environment data monitoring dataset and the dataset collected by internal sensors, a multi-field coupling prediction model for large-volume concrete is constructed based on the embedded concrete thermo-humidification-chemical-mechanical multi-field coupling control equation and the physical information neural network algorithm with finite boundary conditions. The multi-field coupling prediction model predicts the state of concrete based on the construction environment monitoring dataset, and obtains the predicted results of internal temperature, humidity, stress and displacement of large-volume concrete. The predicted concrete condition is analyzed using an early warning system based on crack risk levels, and the risk level is output. When the risk level exceeds the preset threshold, the PID control system is activated. Under intelligent decision-making or manual intervention mode, the PID parameters are adaptively adjusted and the heat preservation and cooling device is linked to regulate the temperature of the concrete. The monitoring results, prediction results, early warning information and temperature control data generated by the above operation steps are input into the visualization platform for display, and the visualization platform enables real-time management of the monitoring, prediction and control process.

2. The method according to claim 1, characterized in that, The internal sensors are arranged in layers at different depths, with each layer having measuring points evenly distributed within a plane, and a horizontal spacing of no more than 5m. They are used to collect monitoring data on the internal state of the concrete after pouring (including temperature, humidity, strain, etc.). The external environmental sensors are deployed around the foundation pit and in the curing area to collect environmental climate data during construction and curing.

3. The method according to claim 1, characterized in that, The LSTM environmental prediction model uses historical meteorological databases for feature extraction, is trained using the Adam optimizer and mean squared error loss function, and is corrected using on-site environmental datasets to predict environmental temperature, humidity, and wind speed during the future construction period.

4. The method according to claim 1, characterized in that, The multi-field coupled prediction model is constructed through the following steps: Collect historical climate and environmental data of the construction site and establish a historical meteorological database; A dynamic prediction model was built based on the LSTM framework to obtain a construction environment prediction dataset. Based on the equations of energy conservation, mass conservation, and momentum conservation, an elastic-plastic damage constitutive model of concrete is introduced to construct a multi-field coupled analysis model of temperature-humidity-chemical-mechanical. By combining the environmental monitoring dataset and concrete monitoring data, the analysis model is fitted and optimized using a hybrid method of finite element method and physical information neural network to form the final multi-field coupled prediction model for large-volume concrete.

5. The method according to claim 4, characterized in that, The multi-field coupled prediction model embeds a thermo-humidification-mechanical multi-field coupled partial differential equation and introduces prior parameters such as density, thermal conductivity, specific heat capacity, humidity diffusion coefficient, thermal expansion coefficient, contraction coefficient, hydration heat release function, boundary condition function, and elastic modulus. The model fitting and accuracy are improved through Bayesian optimization and cross-validation.

6. The method according to claim 1, characterized in that, When the risk level exceeds the threshold, the PID control system adopts an adaptive PID parameter adjustment mechanism, combined with heat preservation and cooling devices to achieve intelligent control, and supports manual intervention mode.

7. A smart temperature monitoring, early warning and control system for large-volume concrete, characterized in that, include: Data acquisition module; Analysis module; Early warning and control module; Visualization module; The acquisition module is used to acquire concrete condition monitoring datasets and environmental climate data through the internal sensors and external environmental sensors, respectively. The analysis module is used to construct an LSTM environmental prediction model and a multi-field coupled prediction model, and output the state prediction results of large-volume concrete. The early warning and control module is used to analyze the prediction results based on the crack risk level, and to control the temperature using a PID control system when the threshold is exceeded. The visualization module inputs the data generated by all the above operation steps into the visualization platform, and the visualization platform displays and manages the entire process of monitoring, prediction, early warning and control in real time.

Images