An intelligent temperature control method and device for construction of a fan foundation in an alpine region

Abstract

The present application relates to the field of construction temperature regulation, and more particularly to an intelligent temperature control method and device for fan foundation construction in high-cold regions. The method comprises the following steps: collecting real-time concrete temperature monitoring parameters based on distributed temperature sensors, performing time series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution, and constructing a temperature fluctuation distribution evolution graph; extracting concrete surface environmental parameters, performing heat exchange simulation processing on the temperature fluctuation distribution evolution graph, and performing multiple coupling correlation analysis to construct a temperature-environment quantitative relationship model; obtaining historical meteorological logs of the construction area and meteorological environmental data of the construction area, performing similar meteorological condition matching calculation, and performing short-term meteorological trend prediction to generate short-term meteorological trend prediction features. The present application ensures the safe construction of fan foundation concrete in high-cold environments and improves construction quality through dynamic temperature regulation requirements.

Description

Technical Field

[0001] This invention relates to the field of intelligent temperature control, and more particularly to an intelligent temperature control method and device for wind turbine foundation construction in high-altitude and cold regions. Background Technology

[0002] In the construction of high-power wind turbines in high-altitude and cold regions, temperature monitoring and control of the concrete foundation are crucial for ensuring structural safety and construction quality. Because concrete hydration is slow in low-temperature environments and is prone to frost damage, abnormal temperature fluctuations not only affect concrete strength formation but can also cause structural damage such as cracks and spalling, thus threatening the long-term stability of the wind turbine foundation. Traditional construction temperature management relies heavily on manual inspections and simple temperature measurement equipment. These methods are limited and outdated, making it difficult to achieve precise and dynamic management of the internal temperature of the concrete, and thus unable to meet the complex temperature control requirements of extreme high-altitude and cold environments.

[0003] Furthermore, concrete temperature is influenced by multiple factors, including environmental climate change, construction techniques, and material properties, making temperature control extremely complex and requiring real-time monitoring. Existing temperature control technologies are typically based on fixed parameter settings, lacking the ability to intelligently identify and adaptively adjust to dynamic temperature changes at the construction site. This results in unsatisfactory temperature control performance, low energy efficiency, and difficulty in responding to sudden environmental changes, failing to effectively prevent low-temperature freezing damage and related quality accidents. Meanwhile, the complex and variable construction environment in high-altitude and cold regions makes temperature monitoring equipment susceptible to external interference, challenging data accuracy and stability and increasing construction risks. Therefore, there is an urgent need for a comprehensive monitoring and control system based on advanced sensing technology, intelligent data analysis, and adaptive temperature control strategies. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes an intelligent temperature control method and device for wind turbine foundation construction in high-altitude and cold regions, thereby solving at least one of the aforementioned technical problems.

[0005] To achieve the above objectives, this invention provides an intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions, comprising the following steps:

[0006] Based on the real-time concrete temperature monitoring parameters collected by distributed temperature sensors, the time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution are carried out to construct a temperature fluctuation distribution evolution map.

[0007] Environmental parameters of the concrete surface were extracted, and the heat exchange simulation was performed on the temperature fluctuation distribution evolution diagram. Multivariate coupling correlation analysis was conducted to construct a quantitative relationship model between temperature and environment.

[0008] Historical meteorological logs and meteorological environment data of the construction area are obtained, similar meteorological conditions are matched and calculated, and short-term meteorological trend prediction is performed to generate short-term meteorological trend prediction features.

[0009] Based on the short-term meteorological trend prediction characteristics, the temperature-environment quantitative relationship model is used to predict the short-term multi-regional temperature situation and fit the regional temperature change distribution, and to construct the natural temperature change prediction distribution field.

[0010] Based on the predicted distribution field of natural temperature changes, we conduct differentiated regional heating power demand analysis, carry out preventive temperature regulation, and generate regional differentiated temperature regulation instructions.

[0011] The temperature control of the heater is executed based on regionally differentiated temperature control commands, and iterative control learning is performed to build an optimized temperature control model for the heater.

[0012] This specification provides an intelligent temperature control device for wind turbine foundation construction in cold regions, used to execute the intelligent temperature control method for wind turbine foundation construction in cold regions as described above, including:

[0013] The temperature fluctuation distribution module is used to collect real-time concrete temperature monitoring parameters based on distributed temperature sensors, perform time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution, and construct a temperature fluctuation distribution evolution map.

[0014] The coupling and correlation module is used to extract environmental parameters of concrete surface, perform heat exchange simulation processing on temperature fluctuation distribution evolution diagram, and conduct multivariate coupling and correlation analysis to construct a quantitative relationship model between temperature and environment.

[0015] The meteorological trend prediction module is used to obtain historical meteorological logs and meteorological environment data of the construction area, perform similar meteorological condition matching calculations, and perform short-term meteorological trend prediction to generate short-term meteorological trend prediction features.

[0016] The temperature trend prediction module is used to perform short-term multi-regional temperature trend prediction and regional temperature change distribution fitting based on the temperature-environment quantitative relationship model according to the short-term meteorological trend prediction characteristics, and to construct a natural temperature change prediction distribution field.

[0017] The differentiated temperature control module is used to perform differentiated regional heating power demand analysis based on the predicted distribution field of natural temperature changes, and to perform preventive temperature control, generating regional differentiated temperature control commands.

[0018] The intelligent temperature control optimization module is used to execute the temperature control of the heater based on regionally differentiated temperature control commands, and to perform iterative control learning to build a temperature control optimization model for the heater.

[0019] The beneficial effects of this invention are as follows: By forming a temperature sensing network covering all layers and locations within the structure through distributed temperature sensors (such as fiber optic DTS or multi-point thermocouples), the temperature change trend and distribution pattern during the curing process of concrete can be captured in real time. Using sequence analysis and spatial interpolation techniques, local overcooling, overheating, and abrupt temperature change points can be accurately identified, constructing a complete temperature fluctuation evolution map. This provides fundamental support for subsequent differentiated thermal control, significantly improving the real-time visualization and control response to the temperature field evolution. By synchronously collecting environmental factors (air temperature, wind speed, radiation, humidity, etc.) on the concrete surface and inputting them into a heat exchange simulation model, the transmission path and influence intensity of external meteorological conditions on internal thermal field changes can be simulated. Combining machine learning algorithms for multi-factor coupling analysis establishes a quantitative correlation between "environment-concrete temperature" in cold environments, forming the core of model-driven prediction. This model can explain the environmental drivers behind temperature changes, improving the predictability and targeting of subsequent temperature control adjustments.

[0020] By accessing the local historical meteorological database and combining it with current measured meteorological data for multi-dimensional feature similarity matching, this method can quickly identify the most similar weather evolution scenarios between the current weather and the past, constructing local meteorological trend forecasts for the near term (1–6 hours). This method overcomes the problem of insufficient accuracy of traditional prediction models in remote areas, achieving refined trend prediction of local microclimates and providing dynamic, real-time, and scenario-adaptive environmental prior information for preventive temperature control. Using the short-term meteorological trend forecast results as driving input, combined with the temperature-environment quantification model constructed in the second step, high-resolution, multi-regional trend predictions of the thermal field evolution of concrete structures can be performed over a future period. This prediction process not only considers the current state but also reflects the feedforward influence of climate abrupt changes, ultimately generating a temperature change distribution field under natural conditions. This distribution field serves as a "baseline prediction before heating," providing objective quantitative evidence for formulating differentiated temperature control schemes and verifying temperature control effectiveness.

[0021] Based on the predicted rate and magnitude of temperature decrease in different regions of the distribution field, the system can intelligently analyze the required heating power level and duration for different regions, enabling on-demand allocation of thermal control resources and avoiding uniform heating in a "one-size-fits-all" manner. By generating precise regional control commands (such as adjusting heating cable power and directional heating activation sequence), the system achieves refined, energy-saving, and intelligent construction temperature control, significantly improving insulation efficiency and energy consumption ratio, and effectively reducing the risk of frost damage and construction costs. After implementing differentiated temperature control strategies, the system continuously monitors the deviation between the control effect and the actual temperature response, and combines this with weather change feedback to construct a closed-loop control mechanism, thereby enabling model self-iterative learning and parameter optimization. This mechanism gradually endows the system with "experience memory" and "strategy evolution" capabilities, continuously improving the adaptability and intelligence of temperature control decisions. Ultimately, this forms a set of warm air blower temperature control optimization models with self-learning and self-adjustment capabilities, continuously ensuring the safe curing and construction quality of the blower foundation concrete in cold environments. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the steps of an intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to the present invention.

[0023] Figure 2 This is a detailed flowchart illustrating the implementation steps for collecting real-time concrete temperature monitoring parameters based on distributed temperature sensors, performing time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution, and constructing a temperature fluctuation distribution evolution map.

[0024] Figure 3 A detailed flowchart illustrating the implementation steps for extracting concrete surface environmental parameters, performing heat exchange simulation processing on the temperature fluctuation distribution evolution diagram, conducting multivariate coupling correlation analysis, and constructing a temperature-environment quantitative relationship model.

[0025] Figure 4 A detailed flowchart illustrating the implementation steps for obtaining historical meteorological logs and meteorological environment data of the construction area, performing similar meteorological condition matching calculations, and generating short-term meteorological trend prediction features is provided. Detailed Implementation

[0026] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0027] This application provides an intelligent temperature control method and device for wind turbine foundation construction in high-altitude and cold regions. The executing entities of the intelligent temperature control method and device for wind turbine foundation construction in high-altitude and cold regions include, but are not limited to, mechanical equipment, data processing platforms, cloud server nodes, and network upload devices equipped with the system, which can be considered as general computing nodes of this application. The data processing platform includes, but is not limited to, at least one of an audio-visual management system, an information management system, and a cloud-based data management system.

[0028] Please see Figures 1 to 4 This invention provides an intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions, comprising the following steps:

[0029] Based on the real-time concrete temperature monitoring parameters collected by distributed temperature sensors, the time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution are carried out to construct a temperature fluctuation distribution evolution map.

[0030] Environmental parameters of the concrete surface were extracted, and the heat exchange simulation was performed on the temperature fluctuation distribution evolution diagram. Multivariate coupling correlation analysis was conducted to construct a quantitative relationship model between temperature and environment.

[0031] Historical meteorological logs and meteorological environment data of the construction area are obtained, similar meteorological conditions are matched and calculated, and short-term meteorological trend prediction is performed to generate short-term meteorological trend prediction features.

[0032] Based on the short-term meteorological trend prediction characteristics, the temperature-environment quantitative relationship model is used to predict the short-term multi-regional temperature situation and fit the regional temperature change distribution, and to construct the natural temperature change prediction distribution field.

[0033] Based on the predicted distribution field of natural temperature changes, we conduct differentiated regional heating power demand analysis, carry out preventive temperature regulation, and generate regional differentiated temperature regulation instructions.

[0034] The temperature control of the heater is executed based on regionally differentiated temperature control commands, and iterative control learning is performed to build an optimized temperature control model for the heater.

[0035] In the embodiments of the present invention, see Figure 1 This is a flowchart illustrating the steps of an intelligent temperature control method for wind turbine foundation construction in cold regions according to the present invention. In this example, the steps of the intelligent temperature control method for wind turbine foundation construction in cold regions include:

[0036] Based on the real-time concrete temperature monitoring parameters collected by distributed temperature sensors, the time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution are carried out to construct a temperature fluctuation distribution evolution map.

[0037] In this embodiment, real-time concrete temperature monitoring parameters are collected based on distributed temperature sensors. Time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution are performed to construct a temperature fluctuation distribution evolution map. A multi-point distributed temperature sensor array is deployed in the concrete structure of the wind turbine foundation, using PT100 platinum resistance temperature sensors to achieve high-precision temperature measurement of ±0.1℃. The sensor nodes are arranged in a three-dimensional grid, with core monitoring points (0.5m spacing) inside the concrete and boundary monitoring points (0.3m spacing) on ​​the surface, forming a three-dimensional monitoring network covering the entire concrete structure. Real-time acquisition of sensor data is achieved through LoRa wireless communication technology, with a sampling frequency set to once every 60 seconds to ensure stable operation of the sensor network in an extreme environment of -40℃. Discrete trend analysis is performed on the collected temperature data using time series analysis methods. A moving average algorithm is used to process temperature data noise, with a window length set to 10 minutes. The temperature change rate is calculated using the first-order difference method to identify the periodic characteristics of temperature fluctuations. Based on a spatial interpolation algorithm, the Kriging interpolation method is used to spatially reconstruct the temperature distribution between monitoring points, with a grid resolution of 0.1m × 0.1m, generating a three-dimensional temperature field distribution map. By calculating the temperature gradient, high-risk areas with internal temperature differences exceeding 5°C in concrete are identified, and a temperature fluctuation intensity assessment index is established to quantify the temperature stability of different areas.

[0038] Environmental parameters of the concrete surface were extracted, and the heat exchange simulation was performed on the temperature fluctuation distribution evolution diagram. Multivariate coupling correlation analysis was conducted to construct a quantitative relationship model between temperature and environment.

[0039] In this embodiment, environmental parameters of the concrete surface are extracted, and the heat exchange simulation of the temperature fluctuation distribution evolution diagram is performed. Multivariate coupling correlation analysis is then conducted to construct a quantitative temperature-environment relationship model. Environmental parameter monitoring equipment is deployed, including an anemometer (measurement range 0-60 m / s, accuracy ±0.1 m / s), a humidity sensor (measurement range 0-100% RH, accuracy ±2% RH), and a solar radiometer (measurement range 0-2000 W / m²). 2 Accuracy ±5W / m 2 A barometer (measuring range 300-1100 hPa, accuracy ±0.1 hPa) and a thermal barometer were used. A mathematical model of heat exchange on the concrete surface was established. Based on Fourier's law of heat conduction and Newton's law of cooling, the convective heat transfer coefficient between the concrete surface and the environment was calculated. This coefficient varied from 15-45 W / (m²) under wind speeds of 2-8 m / s. 2(K). A finite element method (FEM) was used to simulate heat exchange between concrete and the environment. A three-dimensional model of the heat exchange was established, with tetrahedral elements used for mesh generation, and element sizes controlled within 0.05 m. Boundary conditions were set as time-varying functions of ambient temperature and convective heat transfer coefficient. Multiple linear regression analysis was used to establish a quantitative relationship between temperature change and environmental parameters. The regression equation is: ΔT=a1×Wind+a2×Humidity+a3×Radiation+a4×Pressure+b, where the coefficients were solved using the least squares method, and the correlation coefficient R0 is calculated. 2 The value reached above 0.85. Principal component analysis was used to identify key environmental factors affecting temperature changes. Contribution rate analysis showed that wind speed and humidity had weights of 35% and 28% respectively on temperature changes. An environmental parameter-temperature response time lag model was established, and cross-correlation analysis was used to determine the time delay of the influence of different environmental factors on temperature. The delay of wind speed was approximately 15 minutes, and the delay of humidity was approximately 30 minutes.

[0040] Historical meteorological logs and meteorological environment data of the construction area are obtained, similar meteorological conditions are matched and calculated, and short-term meteorological trend prediction is performed to generate short-term meteorological trend prediction features.

[0041] In this embodiment, historical meteorological logs and meteorological environmental data of the construction area are acquired, similar meteorological conditions are matched and calculated, and short-term meteorological trend predictions are performed to generate short-term meteorological trend prediction features. Historical meteorological data of the construction area for the past 10 years are collected, including parameters such as daily average temperature, maximum and minimum temperatures, wind speed, humidity, precipitation, and air pressure. The data comes from local meteorological stations and satellite remote sensing data. A meteorological similarity matching algorithm is established, using a combination of Euclidean distance and cosine similarity to calculate the similarity between the current meteorological conditions and historical meteorological patterns. The similarity threshold is set to 0.8, and the matching window length is 72 hours. Through cluster analysis, historical meteorological data are divided into typical meteorological pattern categories, including six typical patterns such as sunny high-pressure type, cold air intrusion type, and warm and humid airflow type. Each pattern includes parameters such as temperature change characteristics, duration, and transition probability. A Long Short-Term Memory (LSTM) neural network model is used for short-term meteorological prediction. The network structure contains three hidden layers, each with 128 neurons, a learning rate of 0.001, and a training cycle of 500 rounds. The input features include the time series of meteorological parameters over the past 72 hours, and the output is the predicted meteorological parameters for the next 24-72 hours. Model training uses 5 years of historical data, with a validation set of 20% and a test set of 10%. Prediction accuracy is evaluated using root mean square error (RMSE) and mean absolute error (MAE), with the RMSE for temperature prediction controlled within 1.5℃ and the MAE for wind speed prediction controlled within 0.8 m / s. Key features of meteorological change are extracted through trend analysis, including the rate of temperature change, the magnitude of wind speed change, and the cycle of humidity fluctuations, to construct a short-term meteorological trend prediction feature vector. This vector has 15 dimensions, covering the trend parameters of major meteorological elements such as temperature, wind speed, and humidity.

[0042] Based on the short-term meteorological trend prediction characteristics, the temperature-environment quantitative relationship model is used to predict the short-term multi-regional temperature situation and fit the regional temperature change distribution, and to construct the natural temperature change prediction distribution field.

[0043] In this embodiment, based on short-term meteorological trend prediction features, a temperature-environment quantification relationship model is used to predict short-term multi-regional temperature trends and fit regional temperature change distributions, constructing a natural temperature change prediction distribution field. The short-term meteorological trend prediction features generated in step S3 are input into the temperature-environment quantification relationship model established in step S2, and the predicted temperature response values ​​for different regions of the concrete are calculated through the model. Monte Carlo simulation is used to handle prediction uncertainties, with 1000 random samplings set to account for the impact of meteorological prediction errors on temperature prediction, generating a confidence interval for temperature prediction with a confidence level set to 95%. A multi-regional temperature trend prediction model is established, dividing the concrete structure into three main regions: a core region, a transition region, and a boundary region. Each region is further subdivided into 2×2×2 sub-regional grids, totaling 24 prediction units. Spatial autocorrelation analysis is used to calculate the correlation of temperature changes between different regions, with correlation coefficients between 0.6 and 0.9. A temperature propagation model between regions is established, with propagation coefficients obtained through regression of historical data. A Gaussian process regression method was used to fit the regional temperature change distribution, with the radial basis function (RBF) chosen as the kernel function. The length scale parameter was optimized to 0.5m through cross-validation. A mathematical expression for the predicted temperature change distribution field was established, and a three-dimensional spatial interpolation method was used to expand the discrete prediction points into a continuous temperature distribution field, with a resolution of 0.1m × 0.1m × 0.1m. Using a spatiotemporal interpolation algorithm, the hourly temperature distribution field for the next 24 hours was generated, forming a four-dimensional (x, y, z, t) predicted natural temperature change distribution field. A temperature prediction accuracy evaluation mechanism was established, and historical data validation showed that the prediction accuracy reached 85% within 6 hours, 78% within 12 hours, and 70% within 24 hours.

[0044] Based on the predicted distribution field of natural temperature changes, we conduct differentiated regional heating power demand analysis, carry out preventive temperature regulation, and generate regional differentiated temperature regulation instructions.

[0045] In this embodiment, differentiated regional heating power demand analysis is performed based on the predicted distribution field of natural temperature changes, and preventative temperature control is implemented to generate differentiated regional temperature control commands. Based on the predicted distribution field of natural temperature changes constructed in step S4, the time points and durations when the temperature in each region is below a set threshold (5°C) are analyzed to identify key periods and areas requiring heating intervention. A heating power demand calculation model is established based on the concrete heat capacity (approximately 2.4 × 10⁻⁶ m³ / h). 6 J / (m 3 Given the target temperature difference and time constraints (K), the required heating power density is calculated using the formula: P = ρ × c × V × ΔT / Δt, where ρ is the density of concrete (2400 kg / m³). 3Where c is the specific heat capacity of 1000 J / (kg·K). A multi-objective optimization algorithm is used for heating power allocation. The objective functions include maximizing temperature control accuracy, minimizing energy consumption cost, and maximizing equipment utilization. Constraints include a maximum power limit for a single heating unit (5 kW / m²). 2 The system is constrained by both the total power budget and the overall power requirements. A multi-objective optimization problem is solved using a genetic algorithm, with a population size of 100, 300 generations, a crossover probability of 0.8, and a mutation probability of 0.1. A preventative temperature control strategy is established, starting the heating system 1-3 hours in advance based on temperature predictions to avoid temperature fluctuations caused by passive responses. A tiered heating control strategy is designed, dividing the heating power into low-power preheating modes (1-2 kW / m²). 2 ), medium power sustaining mode (2-4kW / m) 2 ) and high-power rapid heating mode (4-5kW / m 2 The system dynamically switches heating modes based on predicted temperature trends. A regionally differentiated control command generation mechanism is established: a constant temperature control strategy is used in the core area, with a target temperature range of 8-12℃; a gradient control strategy is used in the transition area, with a temperature range of 5-10℃; and an anti-freeze control strategy is used in the boundary area, maintaining a minimum temperature above 2℃. The generated temperature control commands include parameters such as heating area coordinates, power setpoint, start time, duration, and priority.

[0046] The temperature control of the heater is executed based on regionally differentiated temperature control commands, and iterative control learning is performed to build an optimized temperature control model for the heater.

[0047] In this embodiment, the temperature control of the heater is executed based on regionally differentiated temperature control commands, and iterative control learning is performed to construct an optimized temperature control model for the heater. A distributed heating system execution architecture is established, including a resistance heater (power range 1-5kW, temperature response time 10-15 minutes), an infrared radiation heater (power range 2-8kW, temperature response time 5-8 minutes), and a hot air circulation system (air volume 50-200m³ / h). 3 / h, temperature response time 15-20 minutes). Centralized control of the heating equipment is achieved through fieldbus technology (CAN bus), with a communication rate set to 250kbit / s and a response time controlled within 100ms. An execution effect evaluation mechanism is established, which monitors the heating effect in real time through the temperature monitoring network in step S1, calculates the deviation between the actual temperature change and the expected temperature change, and triggers adjustment of control parameters when the deviation exceeds ±1℃. An iterative control learning model is constructed using reinforcement learning algorithm, selecting the Q-learning algorithm with a learning rate of 0.1, a discount factor of 0.9, and an exploration rate using an ε-greedy strategy with an initial ε value of 0.3 and a decay rate of 0.995. The state space is defined as a combination of temperature distribution, environmental parameters, and heating equipment state, and the action space includes operations such as heating power adjustment, heating time adjustment, and heating area selection. A reward function is established, comprehensively considering factors such as temperature control accuracy, energy efficiency, and equipment lifespan. The reward function is in the form: R = α × accuracy score - β × energy cost - γ × equipment loss, where α = 0.6, β = 0.3, and γ = 0.1. A temperature control optimization model for the warm air blower was trained using historical control data. The training data included temperature control records from 100 construction cycles. The model training employed a batch update method with a batch size of 32 and a training cycle of 1000 rounds. Model performance evaluation metrics were established, including temperature control accuracy (target ±0.5℃), energy efficiency (energy saving of over 15% compared to traditional methods), and prediction accuracy (over 85%). An adaptive parameter adjustment mechanism was constructed to dynamically adjust model parameters based on seasonal changes, construction progress, and equipment status, enabling continuous improvement and performance enhancement of the warm air blower temperature control optimization model.

[0048] In this embodiment, see Figure 2 The specific steps for constructing the temperature fluctuation distribution evolution based on the real-time concrete temperature monitoring parameters collected by distributed temperature sensors are as follows:

[0049] Real-time concrete temperature monitoring parameters in the wind turbine foundation construction area are collected using distributed temperature sensors.

[0050] Abnormal parameter filtering is performed on the real-time concrete temperature monitoring parameters to obtain optimized abnormal temperature monitoring parameters.

[0051] Perform individual sensor spatial positioning calculations on the distributed temperature sensors to extract the spatial position coordinates of multiple sensors;

[0052] Perform time-series temperature dispersion trend analysis on abnormal optimized temperature monitoring parameters to generate time-series temperature dispersion trend curves;

[0053] Based on the spatial location coordinates of multiple sensors, the spatial temperature fluctuation distribution evolution of the discrete trend curve of time-series temperature is carried out, and a temperature fluctuation distribution evolution map is constructed.

[0054] In this embodiment, during the concrete construction of high-power wind turbine foundations in cold regions, the ambient temperature is generally between -15℃ and -30℃. The heat released by the hydration reaction inside the concrete is easily absorbed by the surrounding low-temperature environment, leading to insufficient early strength of the concrete and even frost damage. Therefore, real-time concrete temperature monitoring using a distributed temperature sensor system is particularly crucial. This system typically consists of high-precision fiber optic sensing lines, a demodulator, a signal processing module, and a data transmission system. The sensing fibers are deployed along key structural parts of the wind turbine foundation, such as the center of the base plate, the lower middle part of the side walls, and around the anchor bolts, with a common spacing of 0.5–1 meter and a depth controlled between 0.3–1.5 meters. The sampling frequency can be dynamically adjusted according to the construction stage, typically set to once every 5 minutes, with a temperature acquisition accuracy of ±0.2℃. The system has long-distance synchronous multi-point acquisition capabilities, supporting high-density deployment over a range of hundreds of meters, and uploads data to the construction management platform in real time via wireless or wired methods, ensuring that construction managers can monitor the evolution trend of the internal temperature field of the concrete 24 hours a day. During system operation, protection performance such as freeze resistance, pressure resistance, and waterproofing should also be considered, especially during the early curing stage of concrete in winter. This type of sensor system must be able to operate stably in an environment of -40℃. This step provides the raw data foundation for all subsequent temperature control adjustments and temperature trend analysis, playing an irreplaceable and crucial role. Due to the complex environment at the construction site, sensors may be affected by electromagnetic interference, mechanical damage, temperature control equipment, or human error, leading to abnormalities such as abrupt changes, drift values, and signal loss in the collected temperature data. If used directly for analysis without processing, it can easily mislead the judgment of the hydration heat behavior of concrete. Therefore, it is necessary to identify and optimize the abnormal parameters of the collected real-time temperature data. In this step, a multiple detection mechanism based on statistical methods is adopted, specifically including the sliding window mean method (setting window lengths of 11 and 21 points), the median deviation method (MAD, setting a threshold of 2 times the standard deviation), and Z-score standardization (setting a threshold of ±3), etc., to screen the temperature data one by one. For sensor node data that is lost more than three times (within 15 minutes), linear interpolation or local spline fitting is used to complete the data. Sampling data with a significant deviation of more than 5°C from neighboring nodes are marked as "abnormal abrupt changes" and corrected using trend fitting. In a real-world project, during a period of drastic nighttime temperature fluctuations, multiple sensor data points exhibited anomalies with instantaneous temperature drops exceeding 8°C. Analysis revealed this was caused by unstable battery power in the data acquisition unit. After processing, the optimized data became more stable and accurately reflected the true temperature change trend. The resulting "Abnormal Optimized Temperature Monitoring Parameters" dataset provides a reliable data foundation for subsequent temporal trend analysis and spatial evolution modeling.

[0055] To achieve spatial visualization analysis of temperature distribution within concrete structures, precise spatial positioning of each temperature sensor is essential, establishing its position model in a three-dimensional coordinate system. Before construction, the theoretical layout path of the sensors is typically preset in a BIM (Building Information Modeling) platform and calibrated using coordinate positioning points. During actual on-site operation, a total station or high-precision RTK GPS device is used to measure the fiber optic sensing path point by point, especially in critical thermal control areas (such as anchor bolt areas and the center of the base slab), recording a spatial node every 0.5–1 meter and assigning it spatial coordinates (X, Y, Z). To further improve accuracy, after concrete pouring, a laser scanner or structured light scanning device can be used to create a point cloud model of the foundation surface. Spatial deviation correction is performed by fitting the model and comparing it with the sensing path, ensuring that the positioning error is controlled within ±5mm. Furthermore, the curved layout of the fiber optic sensing lines must be considered, especially around structural corners and columns, where path reconstruction should be performed using curve fitting and a three-dimensional path matching algorithm. Finally, a high-precision sensor coordinate database is obtained, recording the spatial location of each sampling point and its corresponding sensor number. This database will serve as the core spatial index for 3D interpolation and thermal field reconstruction of time-series temperature data, enabling precise spatial mapping of data from the acquisition end to the analysis end. After obtaining the optimized temperature data, time-series analysis is required to identify the main stages and key inflection points of the concrete hydration heat process. Temperature changes during concrete hardening can be roughly divided into three stages: the temperature rise period, the plateau period, and the cooling period. To extract the characteristics of these stages, this step uses wavelet transform combined with a weighted moving average method to denoise and smooth the temperature series, reducing random disturbances caused by environmental fluctuations. Temperature change curves are plotted for each sensor location based on its time series, forming a "time-series temperature discrete trend curve." In practical applications, such as the curve collected by the center sensor of the foundation slab in a wind farm project, the temperature rises rapidly in the first 6 hours after concrete pouring, with an average heating rate of 3.2℃ / h; the temperature reaches a peak of approximately 41.7℃ at the 18th hour; and drops to 28.5℃ at the 36th hour, then tends to stabilize. By comparing the curve characteristics of multiple sensor points, it is possible to determine whether the hydration heat response of concrete at different locations is synchronous and whether there are abnormal areas (such as delayed temperature rise or heat accumulation). Furthermore, the first and second derivatives of time can be calculated to extract the rate of temperature change and acceleration indices, which can be used to further analyze the response efficiency of insulation measures and the rationality of temperature control strategies. Time-series temperature curves provide important data for concrete quality assessment and intelligent temperature control, and are also a prerequisite for subsequent dynamic analysis of the spatial temperature field.

[0056] By combining the spatial coordinates of each sensor with the corresponding temporal temperature trend data, a three-dimensional temperature fluctuation evolution map of the concrete structure throughout the entire curing cycle can be constructed using interpolation and thermal field reconstruction algorithms. This step uses Kriging interpolation for spatial temperature field reconstruction. Kriging comprehensively considers the covariance and distance relationships between spatial locations, making it suitable for modeling the heterogeneous temperature distribution within concrete. Frame-by-frame reconstruction is performed at each moment during the interpolation process, ultimately forming a four-dimensional temperature distribution map with temporal evolution capabilities. The image can be displayed in the three-dimensional model as isothermal surfaces or colored thermal zones, clearly showing the distribution of high-temperature zones, cooling zones, and abnormal areas. For example, in a high-altitude wind power project, the constructed temperature fluctuation evolution map revealed that the temperature in the outer area of ​​the foundation slab was consistently more than 10°C lower than that in the core area, and the upward heat diffusion was slow, indicating the need for enhanced insulation measures. This map can also be used for time-lapse animation playback to analyze the thermal field evolution path and its correspondence with external temperature and construction time. The graphical results can be uploaded to a remote construction control platform, combined with an AI intelligent analysis system to determine whether measures such as electric heating, infrared insulation, or ventilation cooling are necessary. Temperature fluctuation distribution evolution diagrams can not only improve the accuracy of temperature control, but also provide long-term data support for the health assessment of concrete structures in the later stage. They are an important pillar tool for realizing intelligent and digital concrete construction.

[0057] In this embodiment, the specific steps for filtering out abnormal parameters from real-time concrete temperature monitoring parameters to obtain optimized temperature monitoring parameters are as follows:

[0058] Define a sliding time window of fixed length;

[0059] Based on the sliding time window, a sliding window fitting analysis is performed on the real-time temperature monitoring parameters to extract temperature monitoring parameters from multiple time windows.

[0060] Calculate the fitting residuals, temperature slope, and temperature change rate of the temperature monitoring parameters;

[0061] Temperature jumps between adjacent time windows are detected based on temperature slope and temperature change rate, and abnormal temperature jump points are marked.

[0062] Based on the fitted residuals, slope change rate analysis is performed, and slope deviation is gradually identified to mark abnormal temperature drift points.

[0063] Abnormal temperature jump points and abnormal temperature drift points are identified as abnormal temperature monitoring points.

[0064] Perform sensor monitoring error analysis on abnormal temperature monitoring points, and extract error monitoring points and abnormal outlier monitoring points;

[0065] Calculate the multi-point average temperature monitoring parameters of the real-time concrete temperature monitoring parameters;

[0066] Based on the multi-point average temperature monitoring parameters, the error monitoring points are replaced by mean interpolation, and outlier monitoring points are filtered out to obtain abnormal optimized temperature monitoring parameters.

[0067] In this embodiment, during the construction of high-power wind turbine foundation concrete in cold regions, temperature changes are affected by ambient temperature fluctuations, the heat release process of concrete hydration, and insulation measures, exhibiting distinct temporal characteristics and stages. Therefore, to achieve dynamic tracking and analysis of temperature change trends, a sliding time window mechanism needs to be introduced. A sliding time window is a method of rolling data analysis within a fixed time length, which in this scenario can be defined as a fixed time period such as 5 hours, 6 hours, or 8 hours. The selection of the time window length should be comprehensively set based on the duration of the concrete's pyrolysis peak, the sensor sampling frequency (e.g., once every 5 minutes), and the rate of temperature change. Taking 5 hours as an example, one window contains 60 temperature sampling points, which captures the stages of concrete heating, peak temperature, and cooling while avoiding noise interference from excessive detail. The sliding step size can be set to 30 minutes or 1 hour to ensure good continuity and overlap on the time axis. This windowing method provides a clear temporal structure for subsequent operations such as local temperature trend fitting and abnormal fluctuation identification, enabling precise analysis and control across different time periods. This is a fundamental prerequisite for refined management of concrete temperature control. After dividing the sliding time window, the temperature data within each window needs to be fitted and analyzed to extract its trend characteristics. The core purpose of the fitting analysis is to establish a mathematical model representing the temperature change for each time period, typically using a linear regression model or a local polynomial fitting model (such as second- or third-order polynomial fitting). In cold environments, concrete temperature often exhibits a continuous upward or slow downward trend; therefore, linear model fitting has high interpretability and applicability. Taking a 5-hour window as an example, the least squares method is applied to linearly fit every 60 data points to obtain parameters such as the slope, intercept, and fitting residual of the temperature change during that time period. Simultaneously, the deviation between the fitted value and the measured value at each point is recorded to evaluate the fitting accuracy. During the fitting process, it is also necessary to consider the external interference factors unique to construction in cold regions (such as temporary interruption of insulation, sudden start of electric heating, etc.), which may result in a nonlinear trend in some windows. In this case, it is appropriate to switch to a third-order fitting model for correction.

[0068] After completing the sliding window fitting, it is necessary to further extract key numerical features representing the thermal behavior of concrete in each time window, namely the fitting residual, temperature slope, and temperature change rate (ΔT / Δt). The fitting residual is an important indicator for evaluating the accuracy of the fitted model. It is calculated by subtracting the fitted value from the actual temperature value at each point. If the residual fluctuates drastically or has an abnormal distribution, it usually means that there may be external disturbances or sensor anomalies in that window. The temperature slope represents the rate of heating or cooling per unit time; its positive or negative value can determine whether the concrete is currently in the heating, plateau, or cooling stage. In experimental measurements, the slope is often higher than 0.8℃ / h in the initial stage of pouring, approaches 0℃ during the peak plateau period, and has a negative slope during the cooling period, approximately -0.3℃ / h. The temperature change rate further refines the rate of change of the slope between continuous windows and is a core indicator for dynamically monitoring the sensitivity to temperature changes. By jointly analyzing these three parameters, a comprehensive understanding of the internal thermal field fluctuations of concrete can be achieved, and potential anomalies, such as sudden slope changes or residual spikes, can be preliminarily identified, providing direct numerical basis for subsequent jump and drift judgments. Temperature jumps refer to drastic temperature changes within adjacent time periods, commonly caused by external intervention, uncontrolled heating, or abnormal sensor signals. In practical applications, potential temperature jump points can be identified by comparing the temperature slope and rate of change of adjacent sliding windows to see if there are significant abrupt changes. Specifically, this involves setting a temperature slope jump threshold (e.g., 0.6℃ / h) and a rate of change threshold (e.g., 0.8℃ / h). 2 When the slope change between two adjacent windows exceeds a certain threshold, it is marked as a candidate point for a temperature jump. On the first night of wind turbine foundation concrete pouring, a sudden drop in external temperature typically causes a momentary cooling in the outer concrete area, resulting in a significant temperature jump with the slope changing directly from positive to negative (e.g., from +0.5℃ / h to -0.4℃ / h). Jump detection effectively identifies the timing and location of these discontinuous changes and automatically marks them, preventing subsequent analysis from misinterpreting these drastic changes as normal trends, thereby improving the sensitivity and stability of the temperature control system.

[0069] Besides temperature jumps, a more subtle anomaly needs to be identified—temperature drift. Drift refers to the phenomenon where the temperature data collected by a sensor slowly but continuously deviates from the true value across multiple time windows, often caused by sensor aging, poor contact, or local insulation failure. To detect drift trends, slope change rate analysis can be performed based on the fitted residuals, i.e., calculating the direction and magnitude of slope changes across multiple consecutive windows. If the slope shows a unidirectional increase or decrease across multiple consecutive windows (e.g., more than 5), and the fitted residuals continue to expand (the mean residual increases by more than 1.5℃), it can be preliminarily identified as a temperature drift point. For example, in a certain engineering project, a sensor on the outer edge of the concrete showed a slow temperature rise starting from the 24th hour, while the temperature of other sensors at the same location had stabilized, with the slope showing a continuous increase of 0.1℃ / h, and the residual expanding from 0.4℃ to 2.1℃, ultimately being identified as an abnormal drift point. Marking drift points can effectively eliminate misleading data and prevent it from negatively impacting temperature control strategies or concrete quality assessments. The temperature jump points and temperature drift points identified in the first two steps are collectively categorized as "abnormal temperature monitoring points" for subsequent data cleaning and correction. This categorization is based on a logical judgment model to ensure that the time and location nodes of all temperature anomalies can be accurately located and distinguished. Jump points are mostly sudden anomalies, requiring high system sensitivity; drift points, on the other hand, represent trend deviations and require a certain amount of data accumulation to identify. Both types of anomalies share the characteristic of deviating from the normal temperature evolution pattern of hydration heat release. If not removed, they can cause the average temperature trend to be artificially inflated or deflated, misleading early curing judgments or the implementation of control measures for concrete. To achieve comprehensive coverage, the system also sets an anomaly identification reliability index, evaluating factors such as the number of identified time periods and the magnitude of changes to eliminate interfering misjudgments. This step forms a complete set of anomaly points, which serves as direct input for subsequent error analysis and data repair.

[0070] After identifying the abnormal monitoring points, further analysis of their causes is needed to distinguish whether they are caused by sensor hardware errors, environmental interference, or data processing problems. By comparing the abnormal point spatially with neighboring sensor points (e.g., within 0.5 meters), the temperature difference is calculated to see if it exceeds 3℃. Simultaneously, combining this with historical stability analysis of the sensor points (fluctuation standard deviation exceeding twice the mean), error monitoring points can be preliminarily identified. If a point exhibits isolated anomalies in both spatial and temporal dimensions, without any continuous trend support, it is identified as an outlier. For example, in actual data, a sensor point recorded a temperature of 47℃ in the 32nd hour, while several adjacent points recorded temperatures of 36–38℃. Considering there were no similar temperature increases recorded in the previous 24 hours, it was ultimately marked as an outlier. This step, by constructing an anomaly structure classification system, provides a classification basis for subsequent processing, ensuring that useful information is not excessively removed while critical errors are not overlooked during data cleaning. To further improve the rationality and accuracy of data repair, data from several spatially adjacent normal points are used as a benchmark to calculate their average temperature at the corresponding time points. Multi-point average temperature parameters not only reflect the overall temperature level of a local area but also effectively balance the bias caused by individual anomalies. In practice, typically 4–6 normal sensing points within 1 meter of the anomaly point are selected, and their average temperature at the same time is calculated as a reference. Experiments have shown that this method is highly adaptable to local temperature fluctuations, especially in basic corners or edge areas, effectively avoiding the impact of single-point errors on the overall judgment. This average temperature can also be used to fill in missing anomaly data, ensuring continuity and accuracy.

[0071] For error monitoring points, multi-point average temperature is used for interpolation replacement to ensure the interpolation is reasonable both spatially and temporally. The interpolation process can employ a weighted average method, assigning values ​​to multiple neighboring points based on distance weights and taking their weighted average as the replacement result. Outlier monitoring points, which are typically single-point anomalies without continuity, are unsuitable for interpolation correction and are therefore directly removed, i.e., their data is masked in the analysis. After the above processing, the final optimized set of abnormal temperature monitoring parameters will no longer be affected by abrupt changes, drift, or outliers, and will better reflect the actual thermal field changes in concrete. This dataset is crucial for subsequent intelligent temperature control strategy formulation and construction control model input, significantly improving the temperature control accuracy and concrete quality assurance level of wind power foundation construction in high-altitude and cold regions.

[0072] In this embodiment, see Figure 3 The specific steps for extracting concrete surface environmental parameters, performing heat exchange simulation processing on the temperature fluctuation distribution evolution diagram, and conducting multivariate coupled correlation analysis to construct a temperature-environment quantitative relationship model are as follows:

[0073] Based on the temperature fluctuation distribution evolution map, the internal temperature distribution of concrete is mined and the internal temperature distribution pattern is extracted.

[0074] Define a time period, perform logical calculations on the temperature fluctuation distribution evolution diagram for multiple periods of temperature change, and generate temperature change patterns;

[0075] Numerical real-time simulation is performed based on the internal temperature distribution pattern and temperature change law to generate a real-time temperature distribution field;

[0076] Adaptive frequency parameters are collected in the construction area of ​​the wind turbine foundation, and environmental parameters of the concrete surface are extracted.

[0077] Based on the environmental parameters of the concrete surface, heat exchange simulation processing of the real-time temperature distribution field is performed to generate simulation data of environmental-internal temperature heat exchange.

[0078] Multivariate coupling correlation analysis was performed on the simulation data of environmental-internal temperature heat exchange, and the temperature change response was quantified to construct a quantitative relationship model between temperature and environment.

[0079] In this embodiment, during the construction of wind power foundations in high-altitude and cold regions, concrete experiences strong spatial and temporal temperature gradients during the hardening process. To understand the distribution characteristics of its internal temperature field, it is necessary to delve into the regularities from existing temperature fluctuation distribution evolution maps. Temperature fluctuation distribution evolution maps are typically based on three-dimensional space, incorporating a temporal dimension to present a four-dimensional thermal evolution dynamic. When constructing this map, the time-series temperature data from various sensors have been mapped to the structural coordinate system. In this step, image processing and thermal field modeling algorithms are used to cluster and recognize patterns in the temperature distribution map, identifying typical temperature distribution regions, such as the central heat accumulation zone, the edge heat dissipation zone, and the high-gradient zone at structural connections. Spatial K-means clustering (k value set to 3-6) or Gaussian mixture model (GMM) can divide the temperature distribution map into several representative regional patterns. In a real-world case, a stable temperature plateau region was observed in the middle of the wind turbine tower base plate. Its heat accumulation duration is significantly longer than other regions, and it exhibits an ellipsoidal shape, indicating that this region may have concentrated heat accumulation due to excessive insulation or excessive heat release. After extracting these patterns, their spatial location, duration, temperature gradient, and other parameters can be further labeled and categorized to ultimately form an "internal temperature distribution pattern library," providing structural template references for subsequent pattern calculations and intelligent control models. Temperature changes during concrete hardening exhibit periodicity and stages, especially in cold regions where ambient temperatures fluctuate frequently, resulting in significant differences in concrete temperature response capabilities at different times. Therefore, after obtaining the temperature fluctuation evolution diagram and distribution patterns, it is necessary to divide the time axis into periods based on the actual construction period and thermodynamic characteristics, and extract the thermal change patterns within each period. The time period can be defined by segmenting according to the pouring start time, such as dividing it into an early heating period (0–12 hours), a mid-term plateau period (12–36 hours), and a late cooling period (36–72 hours), or by decomposing it into a 24-hour period based on the natural diurnal variation. Within the cycle, differential analysis and overlay calculations are performed on the temperature fluctuation evolution diagram according to time frames to calculate key parameters such as the average temperature rise rate, temperature fall rate, and thermal equilibrium stabilization time at different locations, and trend regression analysis is conducted (e.g., using moving average or exponential smoothing methods). For example, in a certain project, the average temperature rise rate in the middle of the foundation slab over 72 hours was 0.43℃ / h, with a plateau period lasting 13 hours; while the temperature plateau period at the edge locations was less than 6 hours, showing a rapid cooling trend. These analyses can reveal the temperature response patterns of the concrete structure at different locations and time periods, ultimately generating a set of temperature change patterns for subsequent numerical simulation and control model calls.

[0080] To enable the temperature control system to predict and intervene in the concrete thermal field, the extracted internal temperature distribution patterns and temperature change laws need to be numerically embedded into a real-time simulation module, thereby establishing a dynamically updated "real-time temperature distribution field" model. This model discretizes the three-dimensional structure of the wind turbine foundation into a finite element mesh (e.g., with a side length of 0.2m) and calculates the temperature value of each mesh node in real time. The simulation calculation incorporates the temperature change rate and the thermal diffusivity (typically taken as 1.0 × 10⁻⁶). -6 m 2 The model takes parameters such as the time period variation coefficient and updates the sensor data in real time as boundary condition inputs. The simulation algorithm can employ the finite difference method (FDM) or a finite element heat conduction model, combined with GPU parallel processing to ensure real-time computation. The model outputs the three-dimensional spatial distribution of the internal temperature of the entire wind turbine foundation concrete structure at each moment, which can be visualized through heat maps or isothermal surface plots. In actual measurements, the model's error in predicting the temperature distribution at the 36th hour was controlled within ±1.2℃, indicating its good dynamic response capability and spatial simulation accuracy. The real-time temperature distribution field provides fundamental support for environmental intervention simulation, intelligent heating control, and structural thermal stress analysis. The internal temperature field of concrete is not only affected by the material's heat release but also highly dependent on surface environmental factors such as wind speed, air temperature, humidity, and radiant heat. To achieve heat exchange modeling, a surface environmental parameter acquisition system needs to be constructed. This system should have adaptive frequency control capabilities, adjusting the sampling frequency according to different construction stages, diurnal variations, or sudden weather conditions. During the day, wind speeds typically fluctuate frequently, so the sampling period is set to 10 minutes; at night, when conditions are stable, this can be extended to 30 or 60 minutes. The sensor layout includes infrared surface thermometers, anemometers, hygrometers, and calorimeters, installed at a height of 1.5 meters around the foundation and on the tower base surface, respectively. In actual testing, during foundation construction at a high-altitude wind farm, the nighttime surface temperature dropped to -25℃, with a diurnal temperature range of 18℃; the instantaneous maximum wind speed reached 10.2 m / s, significantly disturbing the heat flux density on the concrete surface. The acquisition system supports edge computing and anomaly detection, ensuring timely response during periods of severe cooling or strong winds, improving the real-time performance and accuracy of the heat exchange model input. The extracted environmental parameters will serve as boundary conditions for the heat flux on the concrete surface, providing external input support for heat exchange simulation.

[0081] After extracting the surface environment parameters, a heat exchange simulation model between the environment and the internal temperature needs to be constructed to quantify the energy transfer process between them. There are three main mechanisms between the concrete structure surface and the environment: convective heat transfer, radiative heat transfer, and conduction. The mathematical modeling relies on steady-state and transient heat conduction equations. During the simulation, dynamic boundary conditions are applied to the surface nodes using data such as ambient temperature, wind speed, and radiative heat, and the heat flux per unit time is calculated based on Newton's law of cooling (Q = hAΔT). The convective heat transfer coefficient h is dynamically adjusted with wind speed (e.g., h ranges from 5 to 50 W / m for wind speeds of 0–10 m / s). 2 (·K). This simulation can be executed using multiphysics software such as COMSOL or a self-built model in Python and embedded into a real-time temperature control platform. In actual engineering, during a sudden increase in wind speed at night, simulation data showed that the cumulative heat flux at the edge of the tower base reached 8500 J / m² within 4 hours. 2 This resulted in a localized temperature drop of over 3.5℃ in the concrete. The simulation data reveals the direct impact of environmental changes on the internal temperature field of concrete, providing a quantitative basis for intelligent insulation decisions (such as localized heating or thermal film covering), and laying the foundation for further development of an environment-internal relationship model.

[0082] In this embodiment, the specific steps for acquiring adaptive frequency parameters and extracting concrete surface environmental parameters in the wind turbine foundation construction area are as follows:

[0083] A regional topology analysis was performed on the wind turbine foundation construction area to extract the topological features of the construction area.

[0084] Based on the topological characteristics of the construction area, key equipment nodes and redundant areas are identified, and core equipment areas and redundant areas are extracted.

[0085] Based on the core equipment area and the redundant area, a multi-layer environmental architecture model is constructed to build a three-layer environmental monitoring framework, which includes a core temperature monitoring layer, a gradient change sensing layer and a boundary environment monitoring layer.

[0086] Based on a three-layer environmental monitoring framework, adaptive sampling frequency adjustment is performed, and multi-dimensional surface environmental parameters are collected to extract concrete surface environmental parameters.

[0087] In this embodiment, during the foundation construction of high-power wind turbine units in cold regions, due to the large scale of the concrete structure, the complex distribution of components, and the significant differences in heat conduction paths, it is necessary to first conduct a regional topology analysis of the entire construction area to comprehensively identify its structural spatial characteristics and heat conduction relationships. The core of topology analysis is to abstract the concrete structure from a physical level into a node-edge model, identifying the relative spatial positions, contact relationships, heat propagation paths, and structural layers of each component (such as the base plate, foundation, anchor cage, and edge thickening strip). In practice, BIM models (Building Information Modeling) and on-site construction drawings are often combined, supplemented by laser scanning measurement and coordinate transformation, to construct a three-dimensional topological structure map with centimeter-level accuracy. Taking a certain type of 6.5MW wind turbine as an example, its foundation base plate has a diameter of 22 meters and a thickness of over 2.8 meters, with the anchor cages arranged within a 4-meter radius of the center. Topological analysis identified five thermally affected structural units. The anchor bolt area, due to its dense reinforcement and high heat capacity, became the heat accumulation center, while the outer edge area, close to the air-soil interface, had a short heat conduction path and rapid cooling, becoming the heat dissipation end. These structural differences will significantly impact subsequent temperature control strategies, making topological extraction fundamental for thermal distribution modeling and the construction of environmental monitoring hierarchies. After completing the topological analysis, the construction area needs to be zoned based on thermal response characteristics and functional importance to extract core equipment areas and redundant areas. This process combines multi-dimensional information such as structural location, thermal coupling strength, construction sensitivity, and subsequent operational requirements. Critical equipment nodes typically refer to locations located at the thermal center, with dense functions and extremely high requirements for temperature control accuracy, such as the anchor bolt area, around the main bearing reinforcement group, and the large-volume casting center under the foundation. These areas are highly susceptible to early cracking or strength decline when temperature control is unbalanced. Redundant areas are mostly located at the edges of concrete or along non-critical thermal paths, such as around the base slab, ventilation ducts, and non-load-bearing areas of the structure. Although coupled with the overall thermal field, their temperature fluctuations have limited impact on structural safety. In actual projects, analysis of the 72-hour average temperature rise rate and peak temperature distribution of various concrete zones revealed that the core zone's temperature rise rate reached 0.86℃ / h, significantly higher than the 0.35℃ / h of the edge zone, with a thermal response time difference exceeding 9 hours. Combining the structural diagram and thermal response results, the boundaries of various zones can be accurately delineated, providing a spatial basis for the subsequent multi-layer monitoring architecture. This zone identification allows for precise allocation of monitoring resources, improving data acquisition efficiency and avoiding redundant deployment and over-monitoring issues.

[0088] After clarifying the differences in thermal response across different regions of the structure, a hierarchical environmental monitoring framework can be constructed to achieve differentiated data collection by region and multi-scale thermal information fusion. The three-layer environmental monitoring framework proposed in this step includes a core temperature monitoring layer, a gradient change sensing layer, and a boundary environmental monitoring layer. The core temperature monitoring layer deploys high-precision, high-frequency thermocouples or fiber optic thermometers in key equipment areas to capture the central heat release rate, temperature peak occurrence time, and local thermal imbalance risk; the density of sampling points is high, and a sampling frequency of 5 minutes / time is recommended. The gradient change sensing layer is distributed in the transition area between the core and the edge, mainly used to monitor the trend of thermal gradient changes and predict the direction of temperature field diffusion. It uses medium-precision Pt100 or thermistor elements, with a sampling point spacing of 1–1.5 meters and a sampling frequency controlled at 10–15 minutes / time. The boundary environmental monitoring layer is deployed at the outer edge of the concrete, the outer wall of the formwork, and the contact points between the surface insulation layer and the surrounding air, using infrared thermometers, anemometers, radiometers, etc., to acquire boundary data of heat exchange between the concrete and the environment. In a construction project, a total of 72 sensing points were deployed across a three-layer structure, accounting for approximately 40%, 35%, and 25% of the total area, respectively. The distribution of these points was dynamically adjusted based on feedback from the structure's thermal response. This three-layer architecture not only enhanced the spatial resolution capability of the internal thermal state of the structure but also improved its dynamic response performance under external environmental disturbances, serving as the foundational framework for intelligent temperature control and thermal field prediction. After the three-layer environmental monitoring architecture was established, the system needed to further introduce an adaptive sampling frequency adjustment mechanism to adapt to the rapidly changing environmental conditions in high-altitude and cold regions and to achieve multi-dimensional acquisition of factors influencing heat exchange on the concrete surface. Adaptive frequency adjustment relies on the calculation of the amplitude and trend of real-time sensor data changes. For example, when the external temperature drops by more than 0.5℃ within 24 minutes or the instantaneous wind speed exceeds 8m / s, the system automatically increases the sampling frequency in the core area from 10 minutes / time to 3 minutes / time and the sampling frequency in the boundary layer to 5 minutes / time, ensuring data real-time performance and continuity. This mechanism uses an edge computing module for fluctuation judgment and scheduling to avoid excessive load on the central server. Meanwhile, in the multi-dimensional surface environmental parameter acquisition, the system is equipped with an infrared non-contact thermometer for surface temperature acquisition, an anemometer for convective heat transfer monitoring, and a radiometer for solar heat flux estimation. Humidity and air pressure monitoring points are also deployed to estimate parameters affecting evaporative heat transfer. In a high-altitude, cold wind field application, the lowest temperature reached -28.6℃, and the daytime temperature difference exceeded 16℃. The system dynamically responded to the drastic climate changes through adaptive sampling, obtaining surface environmental temperature data with an error controlled within ±0.7℃. The final concrete surface environmental parameter data will be used as input to the heat exchange model to dynamically simulate the energy coupling behavior between the concrete and the external thermal field, forming the prerequisite for proactive temperature control strategies.

[0089] In this embodiment, reference Figure 4The specific steps for obtaining historical meteorological logs and meteorological environment data of the construction area, performing similar meteorological condition matching calculations, and generating short-term meteorological trend prediction features are as follows:

[0090] Acquire historical weather logs for the construction area; integrate weather radar, infrared thermal imaging, and wind speed and direction sensors to acquire meteorological environmental data for the construction area;

[0091] Historical meteorological changes were analyzed from historical meteorological logs of the construction area to extract information on historical meteorological changes.

[0092] Based on meteorological environmental data of the construction area, a time-series meteorological change analysis was conducted to obtain the characteristics of time-series meteorological changes.

[0093] Based on the characteristics of time-series meteorological changes, similarity calculations and most similar event matching are performed on historical meteorological change information to obtain the most similar historical samples for the current meteorological environment.

[0094] Based on the historical samples, short-term weather trend predictions for the construction area are generated, resulting in short-term weather trend prediction features.

[0095] In this embodiment, historical meteorological logs of the construction area need to be obtained. Historical meteorological logs typically contain multi-dimensional meteorological parameters such as temperature, humidity, wind speed, wind direction, air pressure, precipitation, and solar radiation intensity, with a time span generally covering more than five years to meet the needs of seasonal and annual climate change analysis. Data sources include the National Meteorological Administration's public database, local meteorological stations, and dedicated environmental monitoring equipment installed during wind farm construction. The collected historical meteorological data undergoes time-series cleaning processing, such as outlier removal, missing value imputation, and unified timestamp standardization, to ensure the accuracy of subsequent analysis. Taking a high-altitude, cold-climate wind farm as an example, the annual average temperature in the construction area ranges from -15℃ to 5℃, with the lowest winter temperature reaching -38℃ and the maximum wind speed exceeding 20m / s. Historical logs show that these extreme meteorological events have a significant impact on the concrete temperature field. By constructing a historical meteorological database, basic data support is provided for subsequent meteorological trend analysis and temperature control during construction. Based on a multi-sensor integration scheme, the construction site is equipped with ground-based radar, infrared thermal imagers, and wind speed and direction sensors, forming a multi-source data fusion system covering the area. Weather radar can remotely monitor cloud movement, precipitation intensity, and extent, providing spatial dynamic information for temperature change early warning. Infrared thermal imaging technology captures the temperature distribution of the ground surface and concrete surfaces in real time, helping to determine the influence of external heat and cold sources. Wind speed and direction sensors, in the form of ultrasonic or mechanical devices, are deployed at key wind gaps and open areas to collect high-resolution wind speed (0–50 m / s, accuracy ±0.1 m / s) and wind direction (360°, accuracy ±2°) data. This data is transmitted wirelessly to a central monitoring platform in real time, and a time synchronization mechanism ensures data consistency. Field experiments show that combining radar data with ground temperature sensors improves the timeliness and spatial resolution of meteorological change response, providing accurate environmental input parameters for intelligent temperature control systems.

[0096] Based on complete historical meteorological logs, time-series statistics and pattern recognition techniques were used to analyze historical meteorological changes. First, long-term trends, seasonal fluctuations, and residuals were extracted using time-series decomposition methods (such as Seasonal-Trend Decomposition (STL)) to identify the periodic meteorological characteristics and anomalous events in the construction area. Cluster analysis (such as K-means and DBSCAN) was then used to classify meteorological conditions into several typical categories, including regular, cold wave, storm, and frost. Taking a high-altitude, cold wind field as an example, the analysis revealed that frost events occur approximately 25 times per year, with an average duration of 12 hours, and cold wave events with wind speeds exceeding 10 m / s occur an average of 3 times per year. Further statistical methods were used to calculate temperature gradients, wind speed fluctuation amplitudes, and humidity change rates, forming a meteorological change information set. This information not only reflects the intensity and frequency of meteorological changes but also reveals the potential impact of meteorology on the concrete thermal environment, providing historical reference samples for similarity matching and prediction models. High-resolution time-series meteorological change analysis was then performed using real-time multi-source meteorological environmental data. The sliding time window method is used to smooth and fit key parameters such as temperature, wind speed, and radiation intensity, extracting instantaneous trends and periodic fluctuation characteristics. For example, a 10-minute sliding window is used to calculate indicators such as the rate of change of temperature and wind speed acceleration to identify meteorological abrupt changes or slow evolution processes. Autocorrelation function (ACF) and power spectrum analysis methods are employed to reveal the temporal correlation and spectral characteristics of meteorological variables. In the measured data, it was found that wind speed peaks between 9:00 and 15:00 daily, the diurnal temperature range fluctuates by 12℃, and nighttime temperatures exhibit an exponential decreasing trend. Furthermore, multivariate time series analysis models (such as Vector Autoregression, VAR) are used to analyze the mutual influence relationships between different meteorological elements, providing a foundation for dynamic prediction and correlation modeling. This temporal feature characterization helps to capture the real-time challenges posed by environmental fluctuations to concrete temperature control.

[0097] To improve the accuracy of meteorological trend forecasting, a similarity calculation method is employed to compare current time-series meteorological changes with historical meteorological information, selecting the most similar historical meteorological event samples. Commonly used similarity measures include Dynamic Time Warping (DTW) and Euclidean distance. DTW can adapt to non-linear alignment on the time axis and is suitable for comparing meteorological data with misaligned rhythms. By constructing a multi-dimensional meteorological feature vector (including temperature change rate, wind speed and direction patterns, humidity changes, etc.) and weighting each dimension, the matching priority of key meteorological factors is improved. Taking data from a specific construction day as an example, two high-altitude cold wave events from the past three years were matched as the most similar samples, with their temperature change trends and wind speed fluctuation patterns showing a consistency of over 87%. This matching result not only provides historical meteorological references for the current construction environment but also provides a training set for short-term prediction models, helping to accurately capture the impact of potential extreme weather on concrete temperature. Using the matched most similar historical samples as prior conditions, combined with real-time meteorological data, a time series prediction algorithm is applied to predict the future short-term meteorological trends in the construction area. Commonly used methods include deep learning models based on recurrent neural networks (RNNs) and long short-term memory networks (LSTMs), which are suitable for weather forecasting under extreme cold and cold climate conditions due to their excellent modeling capabilities for complex nonlinear time-series data. The training process integrates historical meteorological data with real-time monitoring data, and the model outputs temperature, wind speed, and humidity variation curves for the next 6 to 72 hours. Statistical verification of the prediction results shows errors controlled within ±1.2℃ for temperature and ±1.5m / s for wind speed, enabling early warning of potential cold waves, frost, or strong winds. By accurately grasping short-term meteorological trends, the system can dynamically adjust concrete temperature control strategies, rationally arrange heating, insulation, and ventilation measures, and ensure the structural safety and construction quality of concrete under extreme weather conditions.

[0098] In this embodiment, the specific steps for constructing a natural temperature change prediction distribution field by performing short-term multi-regional temperature situation prediction and fitting regional temperature change distribution based on the temperature-environment quantitative relationship model using short-term meteorological trend prediction characteristics are as follows:

[0099] Based on the short-term meteorological trend prediction characteristics, a deep convolutional learning model of the temperature-environment quantitative relationship is performed, and a dynamic analysis of the evolution of meteorological natural temperature is conducted to construct a twin model of temperature evolution in the construction area.

[0100] Short-term multi-regional temperature trend prediction is performed based on the twin model of temperature evolution in the construction area, generating short-term temperature trend prediction values ​​for different regions.

[0101] Calculate the temperature change amplitude and range distribution of the short-term temperature trend prediction values;

[0102] Based on the temperature change amplitude and range distribution, a regional temperature change distribution is fitted to construct a natural temperature change prediction distribution field.

[0103] In this embodiment, short-term meteorological trend prediction features (including multi-dimensional time series data such as air temperature, wind speed, and humidity) and ambient temperature data of the concrete construction area are input into a pre-established temperature-environment quantification relationship model. Multi-layer convolutional layers are used to extract spatial-temporal features from the input data, capturing complex nonlinear temperature-environment correlation features and improving the model's ability to express ambient temperature fluctuations. This convolutional learning process focuses on the dynamics of natural temperature evolution under meteorological conditions, extracting local spatial temperature gradients and temporal evolution patterns through convolutional kernels to achieve a deep understanding of the dynamic temperature field. Based on this training, a twin model of temperature evolution in the construction area is constructed. This twin model simulates the actual temperature change process of concrete under different meteorological conditions, accurately reflecting the impact of environmental changes on the concrete thermal field. In experiments, the model was validated in a typical high-altitude and cold construction environment (air temperature -30℃ to 5℃, wind speed 0~15m / s), with prediction errors controlled within ±0.8℃, demonstrating the model's high adaptability and accuracy to temperature evolution under extreme climatic conditions. The construction area was divided into multiple temperature monitoring sub-regions. A twin model was input with meteorological environmental parameters and historical temperature data for each corresponding region, and the model was used to predict the temperature change trends of the concrete interior and surface in each region over the next 6 to 72 hours. A spatial resolution of approximately 5 meters was used to ensure fine-grained dynamic temperature capture. The model outputs predicted temperature trends for each sub-region, reflecting the internal temperature change trends and potential local thermal stress concentrations. Experimental data shows that the average temperature error predicted by the twin model in different sub-regions remained within ±0.7℃, effectively supporting the precise implementation of local temperature control measures during construction. This prediction provides construction managers with a multi-dimensional view of the temperature situation, facilitating the development of differentiated intelligent heating and insulation strategies, and demonstrating significant application value, especially in high-altitude and cold environments with high wind speeds and sudden temperature drops.

[0104] Temperature variation amplitude refers to the maximum temperature fluctuation within the prediction time window, usually quantified in degrees Celsius. The calculation formula is the difference between the highest predicted temperature and the lowest predicted temperature within the region. Range distribution is determined through statistical analysis to define the spatial coverage area where temperature fluctuations reach specific thresholds (e.g., ±3℃, ±5℃). Specifically, a sliding window method is used to extract local maximum and minimum values ​​from the time-series temperature data of each sub-region, calculating the temperature difference within the time interval. Simultaneously, spatial interpolation algorithms (e.g., Kriging interpolation) are combined with temperature variation data to depict the spatial distribution of temperature fluctuations. Experimental parameters show that during the prediction of a typical winter cold wave event, the temperature variation amplitude in the core area can reach 8℃, while the peripheral area is approximately 5℃. The temperature variation range is relatively concentrated, mainly in active construction areas and windy areas. These calculation results provide a quantitative basis for the dynamic response of the construction temperature control system, helping to identify high-risk temperature difference areas and achieve precise local temperature control.

[0105] A predictive distribution field of natural temperature changes in the construction area was constructed using mathematical fitting and physical modeling methods. The fitting process employed a combination of multinomial regression and Gaussian process regression to smoothly fit the spatial temperature change field, ensuring that it reflects both local details and global consistency. Simultaneously, a temperature diffusion model was introduced, combining the thermal conductivity characteristics of concrete (approximately 1.7 W / (m·K)) and environmental boundary conditions to simulate the heat diffusion process within the concrete structure. Through iterative calculations, a continuous spatial temperature prediction field was generated, representing a distribution map of predicted temperature values ​​for various locations within the construction area over future time periods. Field verification showed that this distribution field outperformed traditional linear interpolation methods in predicting multiple high-altitude and cold weather events, reducing the average error by approximately 15%. This predicted distribution field not only provides a comprehensive dynamic reference for spatial temperature for the construction temperature control system but also supports the development of intelligent temperature regulation strategies, ensuring temperature safety during concrete construction in high-altitude and cold regions.

[0106] In this embodiment, the specific steps for performing differentiated regional heating power demand analysis based on the predicted distribution field of natural temperature changes, and for generating regional differentiated temperature control commands through preventative temperature regulation are as follows:

[0107] Identify the industrial-grade heater module array; calculate the position coordinates of each module in the industrial-grade heater module array, and match them based on the distribution field predicted by natural temperature changes to obtain matching information;

[0108] Based on the matching information, perform differentiated regional heating power demand analysis to generate heating power demand values ​​for different heating modules;

[0109] The heating time is calculated based on the heating power requirement value, and the optimal heating timing is calculated to obtain the optimal heating time point.

[0110] Based on the heating power demand and the optimal heating time, preventative temperature control is performed to generate regionally differentiated temperature control commands.

[0111] In this embodiment, the quantity and distribution of all heating modules on site are identified and confirmed through an equipment management system and sensor network. Each heating module is spatially located using Radio Frequency Identification (RFID), Wireless Sensor Network (WSN), or a positioning system (such as UWB positioning technology), accurately calculating its three-dimensional coordinates with a positioning error controlled within ±0.1 meters. After spatial coordinate acquisition, the positions of each module are mapped to the predicted distribution field of natural temperature changes in the construction area. Spatial matching algorithms (such as nearest neighbor matching and spatial interpolation) are used to analyze the temperature change characteristics of each module's location. This matching process helps identify the temperature change requirements of the area where the heating module is located, forming matching information between the heating module and the temperature prediction field. In the experiment, the module array layout covered an area of ​​approximately 500 square meters, and the spatial matching processing time was less than 30 seconds, ensuring real-time response capability. Using temperature change amplitude, change rate, and historical temperature data, combined with the heat capacity and thermal conductivity characteristics of concrete materials, the heating power requirement required for each heating module is calculated. Specifically, a thermodynamic energy balance model is used, combined with regional temperature compensation coefficients and heat loss parameters, to quantitatively analyze the energy replenishment requirements of different modules during future temperature fluctuations. For example, in areas with significant temperature fluctuations, power demand will increase substantially. This process considers the module power adjustment range (e.g., 0–5 kW) and the coordination of power distribution between modules to ensure a balanced load on the overall heating system. Experimental parameters show that in low-temperature environments (-25°C) during winter nights, the peak heating power demand in the core area can reach 4.5 kW, while the peripheral areas only require approximately 1.2 kW. This differential analysis supports subsequent precise temperature control, avoiding energy waste.

[0112] After determining the heating power requirements, the temperature rise time for each module from heating start to reaching the predetermined temperature (e.g., concrete surface temperature ≥ 5℃) is further calculated. By combining the concrete heat conduction equation and the module's thermal power characteristics, numerical simulation methods (such as the finite difference method) are used to simulate the dynamic temperature changes during the heating process and estimate the required heating time. This calculation also considers environmental heat dissipation conditions (wind speed, air temperature, etc.) and the response delay of the heating modules. Subsequently, based on the temperature rise time and short-term temperature prediction model, an optimal heating timing calculation is performed. Dynamic programming or heuristic algorithms are used to determine the best time to start heating for each module, so as to reach the predetermined temperature while reducing ineffective heating time and energy consumption. Experiments show that in typical cold nights, an advance start time of approximately 30 to 45 minutes can effectively prevent a sudden drop in concrete temperature. This calculation process achieves time-optimized control of the heating system, improving overall heating efficiency and construction safety. Combining the heating power requirements and optimal start time obtained from the previous steps, the system generates specific preventative temperature control commands. The control commands include the power setting of the heating modules, start / stop time points, and adjustment strategies, supporting time-segmented and power-segmented dynamic control of modules within the area. Commands are sent to each heating module on-site via industrial control protocols (such as Modbus and CAN bus) to achieve precise temperature regulation. Preventative control is based on temperature prediction, initiating heating in advance to avoid sudden temperature drops and reduce the risk of frost damage; simultaneously, it dynamically adjusts according to environmental parameters to prevent overheating and resource waste. In practical applications, the system response time is less than 5 seconds, supporting minute-level dynamic adjustment. This command generation strategy ensures that it is tailored to different regions, flexibly adapting to environmental changes and maximizing the uniformity and safety of concrete construction temperature.

[0113] In this embodiment, the specific steps for executing the temperature control of the heater based on regionally differentiated temperature control commands, and for iteratively learning and constructing an optimized temperature control model for the heater are as follows:

[0114] The temperature control of the heater is executed based on the regionally differentiated temperature control command, and regional temperature control feedback information is collected.

[0115] The regional temperature control feedback information is analyzed on a regional basis, and the response delay is calculated to obtain multiple regional control delay values.

[0116] Based on the regionally differentiated temperature control instructions, an analysis of the specified temperature range is performed to obtain the specified temperature range.

[0117] Based on the regional temperature control feedback information, the temperature control deviation is calculated for a specified temperature range to obtain a real-time temperature control deviation curve.

[0118] Intelligent heating parameters are optimized based on multiple regional control delay values ​​and real-time temperature control deviation curves, and iterative control learning is performed to build a temperature control optimization model for the heater.

[0119] In this embodiment, the control system sends commands to each heating module of the warm air blower via an industrial communication protocol. The modules adjust the heating power and start / stop time according to the commands, completing multi-point, multi-level temperature control within the construction area. Simultaneously, temperature sensors deployed within the construction area collect real-time data on the temperature of the concrete and its surface, forming a complete feedback information stream. The feedback information covers temperature changes, heating response status, and environmental parameter changes in each sub-area, and is uploaded to the central monitoring system in real time. To ensure data accuracy, the acquisition frequency is set to once per minute, and the sensor accuracy is required to be within ±0.2℃. During the experiment, the real-time performance and accuracy of the temperature feedback directly affect the subsequent control and optimization effect. The system response delay is controlled within 5 seconds to ensure the continuity and effectiveness of real-time control. The system performs a detailed analysis of the temperature response effect in each area based on the collected temperature control feedback information. Specifically, it compares the time of the heating command issuance with the actual time of temperature change in the corresponding area to calculate the response delay value, i.e., the time difference between the implementation of the control command and the temperature feedback change. Time series analysis methods, such as the cross-correlation function, are used to calculate the time lag between the control signal and the temperature change. Furthermore, the temperature response amplitude of each region was compared with the expected target to evaluate the control effect. In the experiment, large wind speed variations in the frigid environment led to longer response delays in some peripheral areas, with a maximum delay of up to 15 minutes, while the response delay in the core heating zone generally remained at 3-5 minutes. This response delay analysis clarified the differences in control effect among different regions, providing an important basis for subsequent optimization of temperature control parameters.

[0120] Based on regionally differentiated temperature control instructions, the specified temperature control range for each construction sub-region is analyzed and determined. The specified temperature range is typically a safe temperature range stipulated by the construction process, such as controlling the concrete surface temperature between 3°C and 8°C to ensure the normal hydration process of concrete and avoid frost damage and cracking. During the analysis, historical temperature control instruction data is combined with actual ambient temperatures, and statistical methods are used to calculate reasonable ranges for the upper and lower temperature limits, ensuring that the temperature control instructions can cover variable climatic conditions. Tolerance analysis based on interval estimation is used to dynamically adjust the specified temperature range to adapt to temperature fluctuations during construction. Experimental data shows that the specified temperature range can be appropriately increased (e.g., 5°C to 10°C) during cold nights to enhance the frost resistance of concrete. Through reasonable analysis of the temperature control range, a balance between the safety and energy efficiency of construction temperature control is achieved. Combining the aforementioned feedback temperature monitoring data with the specified temperature control range, the system calculates the temperature control deviation in real time, i.e., the deviation between the actual measured temperature and the target temperature range. By recording the deviation value at each moment, a real-time temperature control deviation curve for each region is constructed. Deviation calculations employ error functions (such as absolute error and mean square error) to reflect the accuracy and stability of temperature control. Temperature control deviation curves help identify temperature overshoot, under-adjustment, and fluctuation issues, and analyze the timeliness and effectiveness of concrete temperature control. In experiments, under typical -20℃ low-temperature conditions, the deviation curves show deviation fluctuations of 0.5–1.2℃ in some edge areas, reflecting insufficient local control. This data provides a direct basis for subsequent intelligent optimization and control, ensuring that the temperature control system can adjust in a timely manner and improve the overall temperature control level.

[0121] Based on the obtained regional regulation delay values ​​and temperature control deviation curves, machine learning methods, particularly reinforcement learning and recurrent neural networks (RNNs), are used to achieve dynamic optimization of intelligent heating parameters. The system uses historical regulation effects as training samples to establish a mapping relationship between regulation response and parameter adjustment, gradually reducing response delay and temperature control deviation through iterative learning. The model can predict future temperature change trends based on current environmental conditions and historical regulation data, proactively adjusting heating power, start-up time, and adjustment frequency to achieve pre-emptive temperature control. In actual operation, this warm air blower temperature control optimization model can reduce temperature control deviation by more than 30% and shorten response delay to 1-2 minutes, significantly improving the accuracy and real-time performance of concrete temperature control during construction. Experimental verification shows that the model adapts to complex meteorological fluctuations in high-altitude and cold regions, ensuring a safe temperature environment for concrete construction.

[0122] In this embodiment, an intelligent temperature control device for wind turbine foundation construction in cold regions is provided, which executes the intelligent temperature control method for wind turbine foundation construction in cold regions as described above, including:

[0123] The temperature fluctuation distribution module is used to collect real-time concrete temperature monitoring parameters based on distributed temperature sensors, perform time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution, and construct a temperature fluctuation distribution evolution map.

[0124] The coupling and correlation module is used to extract environmental parameters of concrete surface, perform heat exchange simulation processing on temperature fluctuation distribution evolution diagram, and conduct multivariate coupling and correlation analysis to construct a quantitative relationship model between temperature and environment.

[0125] The meteorological trend prediction module is used to obtain historical meteorological logs and meteorological environment data of the construction area, perform similar meteorological condition matching calculations, and perform short-term meteorological trend prediction to generate short-term meteorological trend prediction features.

[0126] The temperature trend prediction module is used to perform short-term multi-regional temperature trend prediction and regional temperature change distribution fitting based on the temperature-environment quantitative relationship model according to the short-term meteorological trend prediction characteristics, and to construct a natural temperature change prediction distribution field.

[0127] The differentiated temperature control module is used to perform differentiated regional heating power demand analysis based on the predicted distribution field of natural temperature changes, and to perform preventive temperature control, generating regional differentiated temperature control commands.

[0128] The intelligent temperature control optimization module is used to execute the temperature control of the heater based on regionally differentiated temperature control commands, and to perform iterative control learning to build a temperature control optimization model for the heater.

[0129] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0130] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein are implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. An intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions, characterized in that, Includes the following steps: Based on the real-time concrete temperature monitoring parameters collected by distributed temperature sensors, the time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution are carried out to construct a temperature fluctuation distribution evolution map. Environmental parameters of the concrete surface were extracted, and the heat exchange simulation was performed on the temperature fluctuation distribution evolution diagram. Multivariate coupling correlation analysis was conducted to construct a quantitative relationship model between temperature and environment. Historical meteorological logs and meteorological environment data of the construction area are obtained, similar meteorological conditions are matched and calculated, and short-term meteorological trend prediction is performed to generate short-term meteorological trend prediction features. Based on the short-term meteorological trend prediction characteristics, the temperature-environment quantitative relationship model is used to predict the short-term multi-regional temperature situation and fit the regional temperature change distribution, and to construct the natural temperature change prediction distribution field. Based on the predicted distribution field of natural temperature changes, we conduct differentiated regional heating power demand analysis, carry out preventive temperature regulation, and generate regional differentiated temperature regulation instructions. The temperature control of the heater is executed based on regionally differentiated temperature control commands, and iterative control learning is performed to build an optimized temperature control model for the heater. The specific steps for constructing a natural temperature change prediction distribution field by using a temperature-environment quantitative relationship model based on short-term meteorological trend prediction characteristics to predict short-term multi-regional temperature trends and fit regional temperature change distribution are as follows: Based on the short-term meteorological trend prediction characteristics, a deep convolutional learning model of the temperature-environment quantitative relationship is performed, and a dynamic analysis of the evolution of meteorological natural temperature is conducted to construct a twin model of temperature evolution in the construction area. Short-term multi-regional temperature trend prediction is performed based on the twin model of temperature evolution in the construction area, generating short-term temperature trend prediction values ​​for different regions. Calculate the temperature change amplitude and range distribution of the short-term temperature trend prediction values; Based on the temperature change amplitude and range distribution, a regional temperature change distribution is fitted to construct a natural temperature change prediction distribution field.

2. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 1, characterized in that, The specific steps for collecting real-time concrete temperature monitoring parameters based on distributed temperature sensors, performing time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution, and constructing the temperature fluctuation distribution evolution are as follows: Real-time concrete temperature monitoring parameters in the wind turbine foundation construction area are collected using distributed temperature sensors. Abnormal parameter filtering is performed on the real-time concrete temperature monitoring parameters to obtain optimized abnormal temperature monitoring parameters. Perform individual sensor spatial positioning calculations on the distributed temperature sensors to extract the spatial position coordinates of multiple sensors; Perform time-series temperature dispersion trend analysis on abnormal optimized temperature monitoring parameters to generate time-series temperature dispersion trend curves; Based on the spatial location coordinates of multiple sensors, the spatial temperature fluctuation distribution evolution of the discrete trend curve of time-series temperature is carried out, and a temperature fluctuation distribution evolution map is constructed.

3. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 2, characterized in that, The specific steps for filtering out abnormal parameters from real-time concrete temperature monitoring parameters to obtain optimized temperature monitoring parameters are as follows: Define a sliding time window of fixed length; Based on the sliding time window, a sliding window fitting analysis is performed on the real-time concrete temperature monitoring parameters to extract temperature monitoring parameters from multiple time windows. Calculate the fitting residuals, temperature slope, and temperature change rate of the temperature monitoring parameters; Temperature jumps between adjacent time windows are detected based on temperature slope and temperature change rate, and abnormal temperature jump points are marked. Based on the fitted residuals, slope change rate analysis is performed, and slope deviation is gradually identified to mark abnormal temperature drift points. Abnormal temperature jump points and abnormal temperature drift points are identified as abnormal temperature monitoring points. Perform sensor monitoring error analysis on abnormal temperature monitoring points, and extract error monitoring points and abnormal outlier monitoring points; Calculate the multi-point average temperature monitoring parameters of the real-time concrete temperature monitoring parameters; Based on the multi-point average temperature monitoring parameters, the error monitoring points are replaced by mean interpolation, and outlier monitoring points are filtered out to obtain abnormal optimized temperature monitoring parameters.

4. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 1, characterized in that, The specific steps for extracting concrete surface environmental parameters, simulating the heat exchange of temperature fluctuation distribution evolution diagrams, performing multivariate coupled correlation analysis, and constructing a temperature-environment quantitative relationship model are as follows: Based on the temperature fluctuation distribution evolution map, the internal temperature distribution of concrete is mined and the internal temperature distribution pattern is extracted. Define a time period, perform logical calculations on the temperature fluctuation distribution evolution diagram for multiple periods of temperature change, and generate temperature change patterns; Numerical real-time simulation is performed based on the internal temperature distribution pattern and temperature change law to generate a real-time temperature distribution field; Adaptive frequency parameters are collected in the construction area of ​​the wind turbine foundation, and environmental parameters of the concrete surface are extracted. Based on the environmental parameters of the concrete surface, heat exchange simulation processing of the real-time temperature distribution field is performed to generate simulation data of environmental-internal temperature heat exchange. Multivariate coupling correlation analysis was performed on the simulation data of environmental-internal temperature heat exchange, and the temperature change response was quantified to construct a quantitative relationship model between temperature and environment.

5. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 4, characterized in that, The specific steps for acquiring adaptive frequency parameters and extracting concrete surface environmental parameters in the wind turbine foundation construction area are as follows: A regional topology analysis was performed on the wind turbine foundation construction area to extract the topological features of the construction area. Based on the topological characteristics of the construction area, key equipment nodes and redundant areas are identified, and core equipment areas and redundant areas are extracted. Based on the core equipment area and the redundant area, a multi-layer environmental architecture model is constructed to build a three-layer environmental monitoring framework, which includes a core temperature monitoring layer, a gradient change sensing layer and a boundary environment monitoring layer. Based on a three-layer environmental monitoring framework, adaptive sampling frequency adjustment is performed, and multi-dimensional surface environmental parameters are collected to extract concrete surface environmental parameters.

6. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 1, characterized in that, The specific steps for obtaining historical meteorological logs and meteorological environment data of the construction area, performing similar meteorological condition matching calculations, and generating short-term meteorological trend prediction features are as follows: Acquire historical weather logs for the construction area; integrate weather radar, infrared thermal imaging, and wind speed and direction sensors to acquire meteorological environmental data for the construction area; Historical meteorological changes were analyzed from historical meteorological logs of the construction area to extract information on historical meteorological changes. Based on meteorological environmental data of the construction area, a time-series meteorological change analysis was conducted to obtain the characteristics of time-series meteorological changes. Based on the characteristics of time-series meteorological changes, similarity calculations and most similar event matching are performed on historical meteorological change information to obtain the most similar historical samples for the current meteorological environment. Based on the historical samples, short-term weather trend predictions for the construction area are generated, resulting in short-term weather trend prediction features.

7. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 1, characterized in that, The specific steps for analyzing differentiated regional heating power demand based on the predicted distribution field of natural temperature changes, performing preventative temperature control, and generating regional differentiated temperature control commands are as follows: Identify the industrial-grade heater module array; calculate the position coordinates of each module in the industrial-grade heater module array, and match them based on the distribution field predicted by natural temperature changes to obtain matching information; Based on the matching information, perform differentiated regional heating power demand analysis to generate heating power demand values ​​for different heating modules; The heating time is calculated based on the heating power requirement value, and the optimal heating timing is calculated to obtain the optimal heating time point. Based on the heating power demand and the optimal heating time, preventative temperature control is performed to generate regionally differentiated temperature control commands.

8. The intelligent temperature control method for wind turbine foundation construction in high-altitude and cold regions according to claim 1, characterized in that, The specific steps for executing the temperature control of the heater based on regionally differentiated temperature control commands, and for iteratively learning and constructing an optimized temperature control model for the heater are as follows: The temperature control of the heater is executed based on the regionally differentiated temperature control command, and regional temperature control feedback information is collected. The regional temperature control feedback information is analyzed on a regional basis, and the response delay is calculated to obtain multiple regional control delay values. Based on the regionally differentiated temperature control instructions, an analysis of the specified temperature range is performed to obtain the specified temperature range. Based on the regional temperature control feedback information, the temperature control deviation is calculated for a specified temperature range to obtain a real-time temperature control deviation curve. Intelligent heating parameters are optimized based on multiple regional control delay values ​​and real-time temperature control deviation curves, and iterative control learning is performed to build a temperature control optimization model for the heater.

9. An intelligent temperature control device for wind turbine foundation construction in high-altitude and cold regions, characterized in that, The intelligent temperature control method for wind turbine foundation construction in cold regions as described in claim 1 includes: The temperature fluctuation distribution module is used to collect real-time concrete temperature monitoring parameters based on distributed temperature sensors, perform time-series temperature discrete trend analysis and spatial temperature fluctuation distribution evolution, and construct a temperature fluctuation distribution evolution map. The coupling and correlation module is used to extract environmental parameters of concrete surface, perform heat exchange simulation processing on temperature fluctuation distribution evolution diagram, and conduct multivariate coupling and correlation analysis to construct a quantitative relationship model between temperature and environment. The meteorological trend prediction module is used to obtain historical meteorological logs and meteorological environment data of the construction area, perform similar meteorological condition matching calculations, and perform short-term meteorological trend prediction to generate short-term meteorological trend prediction features. The temperature trend prediction module is used to perform short-term multi-regional temperature trend prediction and regional temperature change distribution fitting based on the temperature-environment quantitative relationship model according to the short-term meteorological trend prediction characteristics, and to construct a natural temperature change prediction distribution field. The differentiated temperature control module is used to perform differentiated regional heating power demand analysis based on the predicted distribution field of natural temperature changes, and to perform preventive temperature control, generating regional differentiated temperature control commands. The intelligent temperature control optimization module is used to execute the temperature control of the heater based on regionally differentiated temperature control commands, and to perform iterative control learning to build a temperature control optimization model for the heater.

Images