Dam temperature control system and method based on optical fiber sensing and intelligent algorithm

Abstract

This invention relates to a dam body temperature control system and method based on fiber optic sensing and intelligent algorithms, belonging to the field of temperature control in concrete dam construction for hydraulic engineering. It addresses the problems of traditional temperature control methods, such as lagging monitoring, coarse control, low heat exchange efficiency, and susceptibility to cracking. The system comprises a distributed temperature measurement module, an edge computing unit, an artificial intelligence (AI) temperature control decision-making module, and a cooling water execution unit. The distributed temperature measurement module collects dam body temperature data and transmits it to the edge computing unit. After data refinement, the data is uploaded to the AI ​​temperature control decision-making module. The AI ​​module predicts temperature trends and generates control commands through an LSTM network. The cooling water execution unit uses a water-blocking block staggered cooling pipe system in conjunction with intelligent valves to achieve precise cooling, forming a measurement-calculation-control-feedback closed loop. This invention achieves real-time temperature monitoring, intelligent prediction, and precise control, improving heat exchange efficiency and temperature control accuracy, reducing energy consumption and manual intervention, effectively suppressing concrete cracking, and ensuring dam construction safety and structural durability.

Description

Technical Field

[0001] This invention relates to the field of temperature control technology in the construction of concrete dams for water conservancy projects, and particularly to a dam body temperature control system and method based on fiber optic sensing and intelligent algorithms. Background Technology

[0002] In the field of water conservancy engineering construction, concrete dams, as key hydraulic structures, are directly affected by temperature control during construction, which in turn impacts the safety and stability of the dam body. Concrete dam construction is characterized by large pouring areas, complex construction procedures, and difficulties in internal heat dissipation. Among these challenges, the excessive temperature rise caused by the exothermic reaction of concrete hydration is particularly prominent. High temperatures not only induce stress concentration within the concrete but can also lead to penetrating cracks, seriously threatening the long-term safe operation of the dam.

[0003] For over sixty years, effectively controlling the temperature during concrete dam construction to prevent cracking has been a critical technical challenge in this field. Traditionally, temperature control in concrete dams has relied primarily on embedding cooling channels and implementing insulation measures, with data collection and manual adjustments made using equipment such as valves and mercury thermometers. However, this method has significant drawbacks: the intervals for manually adjusting cooling water flow are long, data collection is inefficient and involves large volumes, leading to significant waste of cooling water resources and hindering precise control.

[0004] With the development of distributed fiber optic sensing technology, its application in monitoring the temperature field of concrete structures has provided new possibilities for real-time and precise temperature control. Distributed fiber optic temperature measurement systems, through fiber optic loops embedded within the concrete, can achieve reliable, accurate, and real-time monitoring of the concrete temperature field, significantly improving the spatial resolution and monitoring accuracy. For example, by embedding fiber optic loops in different pouring layers, temperature data from multiple depths and regions within the concrete can be collected, providing rich data support for the formulation of temperature control strategies.

[0005] Nevertheless, existing technologies still have significant shortcomings. Although distributed fiber optic temperature measurement technology can achieve real-time monitoring of the concrete temperature field, it lacks intelligent control algorithms to dynamically respond to and precisely adjust the temperature field under complex working conditions. Related patents exist, such as CN103835506B, which discloses a concrete water cooling control method based on a numerical model. This method improves the stability of temperature control by simulating and predicting the temperature field and automatically adjusting the water flow rate according to a flow-temperature closed-loop control. However, this method relies on complex simulation calculations and preset model parameters, making it difficult to respond to on-site temperature disturbances in real time, and it has significant shortcomings in data processing efficiency and energy consumption optimization.

[0006] Furthermore, CN119509734B discloses a monitoring system and method for regulating concrete cooling water flow rate based on image technology. By embedding cooling water flow state factors into the concrete temperature field calculation process, the intelligent level of the concrete cooling process is improved. However, this method focuses on image monitoring and regulation of cooling water flow rate, fails to fully utilize the high-precision monitoring capabilities of the distributed fiber optic temperature measurement system, and lacks comprehensive, real-time intelligent perception and dynamic prediction of the concrete temperature field.

[0007] Similarly, CN121326019A discloses a temperature control system, method, equipment, and storage medium for large-volume concrete. This system uses a multi-dimensional sensor module and a non-contact monitoring module to collaboratively collect contact and non-contact temperature data, achieving comprehensive and accurate monitoring of the internal and surface temperatures of the concrete. It also uses a controller to generate a three-dimensional temperature field distribution and predict temperature change trends. However, although this system integrates multiple monitoring methods, there is still room for improvement in intelligent control algorithms and real-time closed-loop regulation, especially when facing complex and changing construction conditions, making it difficult to achieve rapid and accurate temperature control response.

[0008] In summary, current technologies lack an intelligent solution capable of real-time monitoring of the internal temperature of concrete dams, dynamic decision-making, and closed-loop regulation of the cooling system, thus failing to fully utilize multi-source sensor data to achieve high-precision and high-efficiency temperature control. Specifically, existing technologies suffer from the following defects and shortcomings: 1. Single monitoring method or insufficient data coverage: Traditional methods rely on manual data collection, which is inefficient and has limited spatial coverage; although some technologies have introduced distributed fiber optic temperature measurement, they have failed to be effectively integrated with other monitoring methods, resulting in insufficient data comprehensiveness and real-time performance.

[0009] 2. Lack of intelligent control algorithms: Existing control methods are mostly based on preset parameters or simple closed-loop control, which makes it difficult to respond to changes in field temperature in real time and lacks dynamic prediction and self-optimization capabilities based on deep learning of multi-source data.

[0010] 3. Low system integration: Most existing technologies only focus on monitoring or controlling a certain link, lacking an integrated system of "real-time sensing - online analysis - dynamic prediction - closed-loop control", making it difficult to achieve efficient, energy-saving and accurate intelligent temperature control.

[0011] Therefore, developing a dam temperature control system based on fiber optic sensing and intelligent algorithms, integrating functions such as distributed high-precision monitoring, real-time edge computing, artificial intelligence dynamic prediction, and closed-loop precise execution, has significant practical implications and broad application prospects. This invention addresses the aforementioned technical problems by proposing an intelligent and precise temperature control system and method for concrete dams based on distributed fiber optics and artificial intelligence, thereby overcoming the shortcomings of existing technologies and achieving efficient, energy-saving, and precise temperature control during concrete dam construction. Summary of the Invention

[0012] The technical problem to be solved by this invention is to provide a dam body temperature control system and method based on fiber optic sensing and intelligent algorithms, which solves the technical problems of lagging temperature monitoring, rough control, low heat exchange efficiency and easy cracking in concrete dam construction, and realizes integrated temperature control with real-time temperature field perception, intelligent prediction, precise regulation and closed-loop self-optimization.

[0013] To achieve the above technical objectives, the present invention adopts the following technical solution: The system of this invention mainly consists of a distributed temperature measurement module, an edge computing unit, an artificial intelligence temperature control decision module, and a cooling water flow execution unit. The distributed temperature measurement module is connected to the edge computing unit to complete the real-time acquisition and transmission of spatiotemporal temperature sequence data inside the concrete dam. The edge computing unit is connected to the artificial intelligence temperature control decision module to complete the preprocessing, feature extraction, and data refinement of multi-source monitoring data. The artificial intelligence temperature control decision module is connected to the cooling water flow execution unit to realize temperature trend prediction, temperature control strategy generation, and control command issuance. The cooling water flow execution unit is connected back to the edge computing unit to realize closed-loop regulation of cooling water flow and temperature and operational status feedback. The four components work together to form an intelligent temperature control closed loop of measurement-computation-control-feedback.

[0014] The distributed temperature monitoring module uses high-temperature and corrosion-resistant distributed optical fibers, laid synchronously with the cooling pipe network in a serpentine pattern. The spacing between temperature monitoring points is no more than 1 meter, forming a continuous temperature monitoring network covering the entire cross-section of the dam, which can stably acquire multi-point temperature data inside the concrete. The edge computing unit is deployed on the on-site industrial control computer, using a multi-core processor architecture. The data acquisition frequency is 1-5 minutes / time. Data cleaning is completed through sliding window anomaly detection, Z-Score outlier determination, and linear interpolation. After normalization and principal component analysis compression, the data is refined and uploaded within 3 seconds. At the same time, the time dimension, spatial dimension, and thermodynamic characteristics are extracted to calculate the short-term temperature trend and comprehensive risk assessment index.

[0015] The AI-powered temperature control decision-making module adopts a multi-layered AI architecture, using a stacked LSTM network as the core for temperature prediction. It takes in multi-source data such as the internal temperature of concrete, ambient temperature and humidity, cooling water parameters, dam water level, and construction progress as inputs. It uses RMSE as the loss function and Adam as the optimizer to complete model training and outputs predicted temperatures of key concrete points for future periods. At the same time, it integrates a large-scale language model to intelligently parse construction logs, technical specifications, and historical operation and maintenance data, transforming expert experience into quantifiable control parameters, generating temperature control diagnostic reports, strategy explanations, and operation and maintenance suggestions, and achieving self-optimization of control strategies through reinforcement learning.

[0016] The cooling water flow actuator adopts a staggered cooling pipe with water-blocking blocks arranged at intervals of 20-40cm. The pipe material is pressure-resistant stainless steel or HDPE. Combined with intelligent valves and variable frequency water pumps, multiple parallel circulation pipelines are formed through water distributors and collectors, which can extend the water flow path, enhance the turbulent heat exchange effect, and significantly improve cooling efficiency. The actuator dynamically adjusts the flow rate and water temperature according to AI commands. When the temperature is higher than the threshold, the cooling flow rate of the corresponding area is increased and the inlet water temperature is reduced. When the temperature is close to the threshold, the flow rate is reduced to save energy.

[0017] The method of this invention includes five steps: synchronous deployment of cooling pipe network and distributed temperature measurement optical fiber, multi-source data acquisition and edge computing refinement, AI temperature prediction and intelligent decision-making, closed-loop water cooling, real-time feedback and model self-correction. The entire process requires no manual intervention and can dynamically respond to on-site temperature disturbances. Under the premise of ensuring the safety of the dam structure, it can significantly improve the temperature control accuracy and heat exchange efficiency, and effectively reduce the risk of concrete cracking.

[0018] The dam temperature control system and method based on fiber optic sensing and intelligent algorithms provided by this invention have the following beneficial effects: 1. This invention designs an intelligent and precise temperature control system for concrete dams based on distributed optical fiber and artificial intelligence, which effectively solves the temperature control problem in the construction process of concrete dams in the field of water conservancy engineering. In particular, it addresses the problems of insufficient real-time monitoring capability, lag in dynamic response and low temperature control accuracy in existing technologies, and achieves high-precision and high-efficiency temperature control.

[0019] 2. This invention employs a distributed fiber optic temperature measurement network, enabling real-time monitoring of the temperature of concrete dams, significantly improving temperature control accuracy, while reducing the intensity of manual intervention and resource consumption, ensuring that the dam temperature is controlled within the design range.

[0020] 3. This invention introduces an edge computing unit to deploy a local artificial intelligence computing model, which performs real-time feature extraction and preliminary trend analysis on the collected raw data, completes data refinement, provides high-quality, low-latency input to the central decision-making module, and improves data processing efficiency.

[0021] 4. This invention integrates an LSTM prediction model, which captures the long-term time dependence of physical processes such as concrete hydration heat release and environmental heat exchange, thereby achieving dynamic prediction of temperature changes and improving the accuracy of temperature prediction.

[0022] 5. This invention designs a multi-layered AI architecture artificial intelligence temperature control decision module, which combines traditional machine learning, deep learning and large-scale language models to realize control strategy generation and system self-optimization, thereby improving the accuracy and intelligence level of decision-making.

[0023] 6. This invention employs a cooling water flow execution unit, which, based on AI decision-making results, adjusts the water flow and temperature of each cooling pipe network in real time through intelligent valves and variable frequency water pumps, thereby achieving precise cooling of high-temperature areas and improving temperature control performance.

[0024] 7. This invention innovatively introduces water-blocking block staggered cooling pipes, which significantly improves heat exchange efficiency by extending the cooling water flow path and enhancing the turbulent heat transfer effect, achieving faster and more uniform concrete cooling and effectively reducing the temperature difference between the inside and outside of the dam.

[0025] 8. This invention integrates large-scale language models, such as DeepSeek and ChatGPT, which endow the system with natural language processing and knowledge reasoning capabilities. It can intelligently parse text data such as construction logs, technical specifications, and historical maintenance reports to generate temperature control diagnostic reports, strategy explanations, and operation and maintenance suggestions.

[0026] 9. Through data monitoring and effect evaluation in actual engineering applications, this invention has verified the innovation and practicality of the system in the field of concrete dam temperature control, ensuring the safety and long-term durability of the dam structure.

[0027] 10. This invention enables full-area, continuous, and real-time monitoring of the internal temperature of the dam. The distributed fiber optic temperature measurement has high spatial resolution and stable and reliable data, completely changing the shortcomings of traditional single-point manual temperature measurement, which is lagging and has large errors.

[0028] 11. This invention adopts an architecture that combines edge computing and artificial intelligence, which has low data processing latency and fast response. It can dynamically capture the long-term temporal characteristics of concrete hydration heat, predict the risk of temperature rise in advance, and improve the dynamic response capability of the system.

[0029] 12. This invention forms a complete closed loop of measurement-calculation-control, automatically adjusting the cooling water flow and temperature without the need for frequent manual operation, greatly reducing the intensity of manual intervention, water consumption and energy consumption, and realizing refined and intelligent management and control.

[0030] 13. This invention optimizes the control strategy through reinforcement learning, which significantly improves the temperature control accuracy and energy efficiency, reduces energy consumption, and meets the temperature control requirements of roller-compacted concrete construction in high-temperature environments.

[0031] 14. This invention utilizes a distributed optical fiber temperature measurement module to monitor the internal temperature of a concrete dam in real time. By collecting temperature data from multiple depths and regions through optical fiber loops embedded inside the concrete, it achieves temperature field monitoring with high spatial resolution and monitoring accuracy.

[0032] 15. This invention uses an edge computing unit to perform real-time feature extraction and preliminary trend analysis on the collected raw temperature data, ambient temperature and humidity, cooling water parameters, etc., and completes localized intelligent data refinement, providing strong support for the central decision-making module.

[0033] 16. This invention utilizes a large-scale language model to enhance the system's natural language processing and knowledge reasoning capabilities, improves the interpretability of the decision-making process and the practicality of operation and maintenance suggestions, and provides engineers with a more scientific and reasonable basis for decision-making.

[0034] 17. This invention effectively prevents the formation of through cracks by dynamically responding to changes in concrete temperature, ensuring the safety of the dam body and improving the overall quality and durability of the concrete dam. Attached Figure Description

[0035] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram of the dam body temperature control system of the present invention; Figure 2 This is a schematic diagram of the distributed optical fiber temperature measurement network deployment of the present invention; Figure 3 This is a schematic diagram of the water-blocking block staggered cooling pipe structure of the present invention; Figure 4 This is a flowchart of the LSTM prediction algorithm of the present invention. Detailed Implementation

[0036] The technical solutions of the present invention will be further described below with reference to the embodiments and accompanying drawings: Example 1 This embodiment provides a dam temperature control system based on fiber optic sensing and intelligent algorithms, such as... Figure 1 As shown, the intelligent and precise temperature control system for concrete dams includes a distributed temperature measurement module, an edge computing unit, an artificial intelligence temperature control decision module, and a cooling water execution unit. The distributed temperature measurement module connects to the edge computing unit to complete the real-time acquisition and transmission of spatiotemporal temperature sequence data within the concrete dam body. The edge computing unit connects to the artificial intelligence temperature control decision module to complete the preprocessing, feature extraction, and data refinement of multi-source monitoring data. The artificial intelligence temperature control decision module connects to the cooling water execution unit to realize temperature trend prediction, temperature control strategy generation, and command issuance. The cooling water execution unit connects back to the edge computing unit to realize closed-loop regulation of cooling water flow and temperature, as well as operational status feedback, jointly completing the intelligent and precise temperature control system of measurement-calculation-control-feedback.

[0037] During dam construction, the pre-layout of the cooling pipe network is carried out first. The cooling pipes used have a diameter of approximately 30-50mm, and multiple high-strength water-blocking blocks are embedded inside the pipes. The water-blocking blocks are arranged in a staggered pattern at intervals of approximately 20-40cm. Both the pipes and water-blocking blocks are made of corrosion-resistant materials such as pressure-resistant stainless steel or HDPE (High-Density Polyethylene). To ensure uniform cooling of all parts of the dam, the cooling pipes are laid out in a crisscross pattern within the concrete dam cross-section: evenly laid in both vertical and horizontal directions, with a longitudinal spacing of approximately 1-2 meters between each layer. The pipes are usually arranged along construction joints or design gaps for easy installation and inspection. During installation, the cooling pipes are firmly fixed to the reinforcing steel frame, using methods such as binding, to prevent displacement or damage to the pipes during concrete pouring. The inlet and outlet of each cooling pipe network are connected by a water distributor and a water collector. The water distributor is responsible for distributing the cooling water to each group of cooling pipes according to demand. The water collector collects the return water from each pipe network and sends it to the return water system, thus forming a multi-parallel circulating pipe system.

[0038] like Figure 2 As shown, the distributed temperature measurement module uses high-temperature and corrosion-resistant distributed optical fiber cables, which are laid synchronously with the cooling pipe network in a serpentine pattern. Each layer of loops is interconnected to form a continuous structure, covering the entire dam's temperature field. Each optical fiber loop is distributed along the thickness and length of the concrete, continuously measuring the temperature distribution of the corresponding concrete layer, achieving real-time temperature monitoring at multiple points in space. In this embodiment, the optical cables are evenly distributed along the horizontal and vertical directions of the dam, with the spacing between temperature measurement points not exceeding 1 meter, forming a continuous temperature monitoring network covering the entire cross-section of the dam. During cable laying, the optical cables are fixed to the steel reinforcement frame using conduits or channels. Both ends of the optical cables are connected to the data acquisition system of the edge computing center through optical fiber temperature demodulators, enabling real-time transmission of temperature data.

[0039] The edge computing unit is the core component of the system for achieving low-latency, high-reliability data processing. Deployed in the industrial control computer at the dam site, it directly connects to the distributed fiber optic temperature measurement network, environmental sensors, and cooling water monitoring system. The hardware employs a multi-core processor architecture, equipped with large-capacity memory and a high-speed solid-state drive. The software environment is based on a Windows operating system, integrating a Python development framework and machine learning libraries. It interconnects via industrial Ethernet or fieldbus protocols, with a data acquisition frequency of 1–5 minutes per acquisition. The edge computing unit receives data such as the internal temperature of the concrete, ambient temperature and humidity, inlet and outlet temperatures and flow rates of the cooling water, dam water level, pouring time, and slab size. Data cleaning is performed using sliding window anomaly detection, Z-Score outlier determination, and linear interpolation. After Min-Max (Minimum-Maximum) normalization, lightweight temporal convolutional network feature extraction, exponentially weighted moving average trend prediction, and comprehensive risk assessment, the 28-dimensional feature vector is reduced and compressed using principal component analysis. This data is then packaged into a JSON (JavaScript Object Notation) data packet and uploaded within 3 seconds. The specific implementation process is as follows: The raw data received by the edge computing unit includes: (1) concrete internal temperature data collected by distributed optical fiber, with a measurement point spacing of no more than 1 meter; (2) environmental temperature and humidity sensor data; (3) cooling water inlet temperature, outlet temperature and real-time flow rate of each pipeline; (4) dam water level data; (5) construction progress parameters, such as pouring time, silo size and boundary conditions.

[0040] The edge computing unit employs a layered processing architecture at the data computation level, performing real-time computation and feature mining on the received multi-source monitoring data. After data reception is complete, the unit first initiates data cleaning computation, using a sliding window anomaly detection algorithm with a window size set to 10 consecutive sampling points. For temperature data, its rate of change is calculated: (1); In the formula, The rate of temperature change; The time interval is (h). , They are respectively , Temperature at any given time (°C).

[0041] Within 3 consecutive sampling periods (8℃ / h is the threshold for temperature change rate; exceeding this threshold is considered abnormal.) The data was identified as outlier. Simultaneously, the deviation of each measurement point's data was calculated using the Z-Score method. (2); In the formula, Standard scores; For the first Original temperature data (°C) at each measuring point; The average temperature (°C) at all measuring points; The standard deviation of temperature (°C) for all measuring points; Refers to a sliding time window.

[0042] when In this case, the data point is considered an outlier and a linear interpolation algorithm is used for data repair.

[0043] After data cleaning, the unit performs multi-source data normalization calculations, standardizing monitoring parameters of different dimensions separately. Temperature data uses the Min-Max normalization algorithm: (3); In the formula, Normalized temperature value (mapped to) (interval); The original temperature value (°C); , These represent the maximum and minimum values ​​(°C) of the temperature series, respectively.

[0044] Traffic data is normalized to the maximum designed traffic volume: (4); In the formula, Normalized flow values ​​(mapped to) (interval); The original temperature value ( ); To design the maximum flow rate ( ).

[0045] Map all parameters uniformly to The interval is used to eliminate the influence of dimensions. Next, the real-time feature extraction calculation stage begins, where the unit extracts multi-dimensional feature vectors, including time-dimensional features, through a lightweight temporal convolutional network model. (5); In the formula, for Moment, time scale The temperature change characteristics over time, i.e., the instantaneous rate of temperature change, characterizes the rate at which the concrete temperature rises or falls within the corresponding time span. For the present Real-time temperature monitoring value (°C) at concrete measuring point; For the present Time Rewind Historical temperature monitoring values ​​(°C) over a long period of time; The time window / time step is set to 10, 30, and 60 minutes respectively (used to extract the temperature time series change trends at short, medium, and multi-scales), corresponding to the change trends at different time scales.

[0046] Spatial dimensional characteristics reflect the regional temperature uniformity: (6); In the formula, The regional temperature uniformity coefficient ranges from 0 to 1. The closer the value is to 1, the more uniform the concrete temperature distribution in the monitored area of ​​the dam body is, and the better the temperature control effect is. The lower the value, the greater the regional temperature difference and the higher the risk of cracking. For the monitoring area Real-time temperature values ​​(°C) of a distributed fiber optic temperature measurement point; , These are the maximum and minimum temperatures (°C) of all measuring points within the current monitoring area, respectively. This represents the average temperature (°C) of all measuring points within the entire monitoring area.

[0047] Thermodynamic characteristics, comprehensively evaluating the heat exchange efficiency of the cooling system: (7); In the formula, The comprehensive heat exchange efficiency (thermodynamic characteristic value) is dimensionless and ranges from 0 to 1; the higher the value, the stronger the overall heat exchange capacity of the cooling pipe network and the higher the temperature control efficiency. , These are the inlet and outlet temperatures of the cooling water in the cooling pipes (°C). Real-time flow rate of cooling water in the cooling pipes ( ); The measured average temperature (°C) inside the dam's concrete was obtained from a distributed fiber optic temperature measurement module. The effective heat exchange contact area of ​​a single group / area cooling pipe ( ).

[0048] In the trend analysis and risk assessment calculation phase, the unit uses an exponentially weighted moving average combined with linear regression to predict short-term temperature trends. The calculation formula is as follows: (8); In the formula, , respectively , Predicted temperature (°C) of concrete measuring point at any given time; For a smooth system, dimensionless, with a range of values ​​of 1 / 2. In this embodiment, the value is 0.3; For the present Real-time temperature monitoring value (°C) at concrete measuring point; The historical trend weighting coefficient is dimensionless. This is a trend adjustment coefficient, dimensionless. For the temperature change trend term ( The risk assessment index is calculated based on the linear regression slope of the most recent six sampling points. Simultaneously, the comprehensive risk assessment index is calculated in real time. (9); In the formula, The temperature control comprehensive risk assessment index is dimensionless. The larger the value, the higher the risk level of temperature control cracking and temperature exceeding the standard in the concrete of that area of ​​the dam. These are the component weighting coefficients, corresponding to the risk proportion weights of the three dimensions: temperature deviation, temperature rise rate, and temperature uniformity. They comprehensively assess temperature control risks and are dimensionless. In this embodiment, the value is taken as... ,satisfy ; The current measured average temperature (°C) of concrete in the monitoring area; The allowable temperature threshold / standard control upper limit temperature (°C) for concrete temperature control design is the compliant temperature control benchmark value preset for dam construction; This represents the percentage deviation of the current temperature from the threshold, indicating the extent to which the current temperature exceeds the design limit. The maximum temperature change rate (maximum temperature rise rate) within the region characterizes the rate of rapid temperature increase in concrete hydration heat. The temperature uniformity coefficient for the region, as defined above, takes a value of 0 to 1; The value represents the temperature unevenness deviation in the region. The larger the value, the greater the temperature difference in the internal region of the dam and the higher the risk of structural stress cracking.

[0049] Finally, data refinement and compression calculations were performed. Principal component analysis was applied to reduce the dimensionality of the extracted 28-dimensional feature vectors, retaining 95% of the original information and compressing the data volume to less than 35% of the original. The refined feature dataset includes core temperature features, thermodynamic indicators, risk assessment indices, and trend prediction values. These, along with metadata such as timestamps and device identifiers, are encapsulated into a standardized JSON data package and uploaded in real time through redundant network channels. This ensures that the entire process from raw data to refined features is completed within 3 seconds, providing high-quality, low-latency input to the central decision-making module.

[0050] The prediction model is the core of the artificial intelligence temperature control decision module. It achieves high-precision prediction of future temperature change trends through deep learning of the time series data of the temperature field of the concrete dam. The specific implementation process is as follows: The distributed optical fiber temperature measurement network continuously collects temperature data of the measuring points inside the concrete dam at intervals of no more than 1 meter. At the same time, the system integrates other key monitoring data to form the multi-dimensional input time series variables of the model, specifically covering: (1) the historical temperature sequence inside the concrete; (2) the ambient temperature; (3) the inlet temperature of the cooling water; (4) the real-time water flow of each cooling pipe; (5) the water level in front of and behind the dam. After these data are transmitted to the edge computing unit, data cleaning is first carried out to remove abnormal values ​​caused by instantaneous sensor failures. Then, the Z-Scor normalization method is used to normalize the multi-source heterogeneous data, eliminate the influence of dimensions, and convert each parameter into a distribution with a mean of 0 and a standard deviation of 1. The calculation formula is: (10); In the formula, Standard scores; This is the original data. The mean, The standard deviation is denoted as .

[0051] The preprocessed data is organized into a supervised learning sequence according to the time step, forming the basic dataset for model training and prediction.

[0052] To effectively capture the long-term time dependence of concrete temperature changes, the prediction model constructed in this system adopts a stacked structure. For example... Figure 4 The flowchart of the LSTM (Long Short-Term Memory) prediction algorithm of this invention is shown. The temperature prediction model of the artificial intelligence temperature control decision module is a two-layer stacked LSTM structure. The first layer of LSTM has 128 neurons and is used for initial learning of time series features. The second layer of LSTM has 64 neurons and is used to further extract higher-order time series patterns. This structure effectively enhances the model's ability to understand complex nonlinear temperature evolution processes. The output layer is a fully connected layer using a linear activation function, and the output is a scalar, i.e., the next time step. The predicted average temperature for a specific point or area.

[0053] The prediction model includes a forget gate, an input gate, and an output gate. The forget gate determines the cell state from the previous time step. To determine which information to discard, it uses the Sigmoid (Sigmoid Activation Function) to output a value between 0 and 1, calculated as follows: (11); In the formula, The output value of the forget gate is in the range of [0, 1]. The closer the value is to 0, the more it means that the cell state information of the previous moment is completely discarded. The closer the value is to 1, the more it means that the historical cell state information is completely retained. It is used to control how much past temperature time sequence information needs to be forgotten. This represents the S-type activation function, which maps and compresses the calculation result to the 0~1 range to achieve the logic control effect of a gate switch; The trainable weight matrix for the forget gate is a parameter that is continuously optimized during model training; for The LSTM hidden layer state at time (the previous sampling time) carries the temperature feature information of all previous historical time moments; For the present The network input vector at any given time is the preprocessed and normalized multi-source monitoring feature data of the dam body temperature control. To compensate for the bias term of the forget gate, the model can be trained with offset parameters to adjust the activation threshold of the function.

[0054] The input gate determines which new information is stored in the cell state, while the Sigmoid layer determines which values ​​are updated. (12); In the formula, The input gate output value ranges from [0, 1]. The closer the value is to 1, the higher the proportion of the newly added input information at the current moment that needs to be retained in the cell state. The closer the value is to 0, the more the new input information is blocked and discarded. is the trainable weight matrix of the input gate, and is the core learning parameter for iterative optimization during model training; As the bias term of the input gate, the model can be trained with offset parameters to adjust the trigger threshold of function activation and improve the model's fitting ability.

[0055] The Tanh layer generates a new candidate value vector: (13); In the formula, for The candidate cell state at time step is a temporary candidate memory vector generated at the current time step and awaiting update, with values ​​ranging from [...]. [1, 1] represents the newly added alternative temperature feature information generated by the combination of the current input and the historical state; This represents the hyperbolic activation function, a non-linear activation function that compresses the output to [...]. The interval [1, 1] is used to generate effective features for candidate memories, enhancing the model's nonlinear fitting ability; The trainable weight matrix for candidate states is the core learning parameter that is iteratively optimized during the training of the LSTM model and is used to perform linear transformation on the input features. As a bias term for candidate cell states, the model can be trained with offset parameters to adjust the output baseline of the function activation, thereby improving the model's expressive and adaptive capabilities.

[0056] Both were completed together.

[0057] Cell state update, changing the old state Combined with the outputs of the forget gate and the input gate, the cell state is updated to a new state: (14); In the formula, , They are respectively , The constantly updated cell state (long-term memory unit) stores the long-term temporal features ultimately retained by the LSTM network. In dam temperature control prediction, it is used to retain deep memory information such as the historical trend of concrete temperature changes and periodic characteristics. The output gate determines the output at any given moment based on the current input and the updated cell state. The output gate is calculated as follows: (15); In the formula, The output value is the gating output value of the output gate, with a value range of [0, 1]. It is used to control how much information about the cell's current state can be output as a hidden state. The closer the value is to 1, the higher the proportion of information that the cell's memory can output. The output gate trainable weight matrix is ​​the core learning parameter that is iteratively optimized during the training of the LSTM model. As the output gate bias term, the model can be trained with offset parameters to adjust the function activation threshold and improve the model's fitting and adaptability.

[0058] The final hidden state is: (16); In the formula, for The final hidden layer state output at time step is the final temporal feature result output by the LSTM unit. It carries all the temperature time series and operating condition features integrated at the current time step. It can be directly used for dam concrete temperature prediction, temperature control risk level determination, and water flow regulation decision output. At the same time, it will also be used as the input for the calculation of the next time step to participate in the loop operation.

[0059] This hidden state will be passed to the next time step and the next layer of the network.

[0060] Before model training, the preprocessed time series data was divided into training and test sets in an approximately 6:4 ratio, and the root mean square error was selected as the loss function. (17); In the formula, The root mean square error is measured in the same dimension as the temperature data (unit: °C), which directly reflects the average deviation between the temperature prediction result and the actual temperature. This refers to the total number of samples involved in the error calculation, i.e., the total number of temperature measurement data points. The sample number is a number ranging from 1 to 1. Total number of temperature measurement data points; For the first The model prediction value for each sample, that is, the predicted concrete temperature value output by the LSTM model; For the first The true measured value of each sample, that is, the true value of the actual concrete temperature obtained by on-site fiber optic temperature measurement.

[0061] The optimizer used is Adam (Adaptive Moment Estimation), with an initial learning rate of 0.002. Early stopping is employed to prevent overfitting. The model is trained offline on an industrial control computer in an edge computing center using the PyTorch or TensorFlow framework until the loss function converges on the test set and reaches the preset accuracy. The model parameters are then saved for online prediction.

[0062] The large-scale language intelligent evaluation model is built on large-scale language models such as DeepSeek. It deeply integrates the professional knowledge base of concrete temperature control with natural language processing capabilities, analyzes construction text data, transforms expert experience into control parameters, generates temperature control diagnostic reports, strategy explanations and operation and maintenance suggestions, and supports multi-plan simulation comparison and knowledge self-updating, so as to realize in-depth analysis and multi-dimensional evaluation of prediction results and control strategies.

[0063] Specifically, the model can automatically generate a complete analysis report including decision-making basis, expected effects, and risk assessment, significantly enhancing the interpretability of the decision-making process. It also possesses multi-plan simulation and comparative analysis capabilities, able to simulate and compare the expected effects of various control strategies based on historical data and current operating conditions, providing quantitative support for optimized decision-making. By continuously analyzing the temperature field evolution trend and equipment operating status, the intelligent model can perform multi-plan simulation and comparative analysis. Based on historical data and current operating conditions, it simulates and generates comparisons of the expected effects of various control strategies. For example: "Regarding the current temperature rise trend, the system provides three control schemes: Scheme A (rapid cooling) is expected to control the temperature within the standard range within 2 hours, but with higher energy consumption; Scheme B (stable cooling) requires 5 hours to achieve the same effect, but with lower energy consumption; Scheme C is recommended, achieving the optimal balance between energy consumption and efficiency while ensuring structural safety." Furthermore, the system has adaptive knowledge updating and experience accumulation capabilities, continuously absorbing expert experience from engineer feedback adjustments and abnormal operating condition handling records, constantly optimizing its knowledge base and reasoning logic.

[0064] Based on AI decision-making results, the system can dynamically cool the control unit, using intelligent valves and variable frequency water pumps to adjust the water flow and temperature of each cooling pipe network in real time, achieving precise cooling of high-temperature areas. When the temperature in a certain area exceeds a set threshold, the system automatically increases the water supply to the corresponding cooling pipe or decreases the inlet water temperature to enhance the cooling effect; when the temperature is close to or below the threshold, the system appropriately reduces the water supply to save energy. As cooling and environmental factors influence the system, real-time temperature data is continuously fed back to the edge computing center. The AI ​​module compares the predicted results with the actual monitored values ​​to perform online correction of the prediction model and control strategy, achieving closed-loop self-correction of prediction and feedback to ensure control accuracy and the system's dynamic adaptive capability.

[0065] like Figure 3 As shown, the cooling water flow execution unit mainly consists of a cooling water source, a pre-cooling water tank, a circulating water pump, a water distributor-collector, intelligent regulating valves, and staggered water-blocking cooling pipes. The water-blocking blocks inside the pipes are arranged in a staggered pattern at intervals of 20-40cm. The pipe material is pressure-resistant stainless steel or HDPE. With the help of intelligent valves and variable frequency water pumps, precise regulation of zoned flow is achieved. The cooling water forms turbulence under the action of the water-blocking blocks to enhance heat transfer. The operating status is transmitted back to the edge computing unit in real time.

[0066] In this embodiment, the cooling water source is a reservoir at the construction site. After being cooled by a pre-cooling water tank, the water is pressurized by a circulating water pump and delivered to a distributor. The distributor distributes the cooling water to various cooling pipes inside the dam as needed. Each cooling pipe is equipped with an electric intelligent valve. The valve opening is dynamically adjusted by an AI control module, thereby achieving precise control of the cooling intensity in a local area. When the AI ​​algorithm determines that a certain cooling pipe needs increased flow, the corresponding intelligent valve opens, and a pressure variable frequency pump supplies cooling water to that pipe group. As the cooling water flows in the pipes, it is deflected multiple times by staggered water-blocking blocks, generating a strong turbulent heat transfer effect and accelerating the heat transfer between the pipe wall and the water body. During this process, the distributor-collector precisely controls the distribution flow of each pipe group and collects the water temperature information at the pipe outlet, feeding it back to the system. A temperature-measuring fiber continuously monitors changes in the dam's temperature field. When the monitoring results deviate from the design temperature range, the AI ​​module will automatically adjust the valve opening and pump speed parameters for the next control cycle to ensure temperature control accuracy and prevent overcooling or undercooling.

[0067] Through the above-described embodiments, the present invention can realize real-time monitoring and intelligent control of concrete dam temperature, significantly reduce energy consumption and improve temperature control accuracy, and meet the temperature control requirements of roller-compacted concrete construction in high-temperature environments.

[0068] Example 2 In another preferred embodiment, based on Embodiment 1, this embodiment provides a dam body temperature control method based on fiber optic sensing and intelligent algorithms. The method employs the dam body temperature control system based on fiber optic sensing and intelligent algorithms described in Embodiment 1 for dam body temperature control. The specific steps are as follows: Step 1: Pre-embed a crisscrossing cooling pipe network and simultaneously lay distributed temperature-measuring optical fibers to construct a dam temperature monitoring network. The longitudinal spacing of the cooling pipe network is 1-2 meters, and the pipe diameter is 30-50mm. Multiple parallel circulation pipelines are formed through water distributors and collectors. The cooling water forms turbulence under the action of staggered water-blocking blocks to enhance heat transfer. The distributed temperature-measuring optical fibers use high-temperature and corrosion-resistant optical cables and are laid in a serpentine pattern simultaneously with the cooling pipe network. The spacing between temperature measurement points is no more than 1 meter, covering the entire dam temperature field. The optical cables are fixed to the steel reinforcement frame and connected to the edge computing unit.

[0069] Step 2: The edge computing unit collects real-time data on concrete temperature, environment, cooling water, and construction progress, and completes cleaning, normalization, feature extraction, and data refinement. The edge computing unit adopts a hierarchical processing architecture to perform sliding window anomaly detection, Z-Score (standard score) outlier repair, Min-Max normalization, extract multi-dimensional features of time, space, and thermodynamics, calculate short-term temperature trends and comprehensive risk assessment index, and upload the feature vectors to the artificial intelligence temperature control decision module after dimensionality reduction and compression.

[0070] Step 3: As Figure 4As shown, the AI-powered temperature control decision-making module follows the LSTM prediction algorithm process, using a two-layer stacked LSTM model to predict temperature change trends. It then combines a large-scale language model to analyze construction data and integrates expert experience to generate the optimal temperature control strategy and an interpretable analysis report.

[0071] Step 4: The cooling water execution unit adjusts the cooling water flow and temperature according to the strategy to precisely cool the concrete. When the temperature control is executed, if the concrete temperature is higher than the threshold, the cooling flow in the corresponding area is increased and the inlet water temperature is reduced to enhance cooling. If the temperature is close to the threshold, the flow is reduced to save energy. The cooling water generates turbulent heat transfer under the action of the staggered water blocking blocks, which improves the cooling efficiency.

[0072] Step 5: Real-time feedback of temperature monitoring data; the AI ​​(Artificial Intelligence) module compares the predicted value with the measured value, corrects the prediction model and control strategy online, and forms a closed-loop adaptive regulation to ensure temperature control accuracy and prevent overcooling or undercooling.

[0073] In the preferred embodiment, the distributed temperature measurement module uses high-temperature and corrosion-resistant distributed optical fibers, which are laid synchronously with the cooling pipe network in a serpentine pattern. The spacing between temperature measurement points is no more than 1 meter, forming a continuous temperature monitoring network covering the entire cross-section of the dam. This configuration ensures comprehensive and accurate monitoring of the internal temperature of the concrete dam, enabling stable operation even in high-temperature, high-humidity, or corrosive environments. The high-density arrangement of temperature measurement points allows for timely capture of changes in the temperature field, providing reliable data support for subsequent temperature control decisions, thereby effectively preventing crack formation and ensuring the structural safety of the dam.

[0074] In the preferred embodiment, the edge computing unit employs a multi-core processor industrial computer, with a data acquisition frequency of 1-5 minutes per acquisition. Data cleaning is performed through sliding window anomaly detection, Z-Score outlier determination, and linear interpolation. After normalization, principal component analysis, and compression, the data is refined and uploaded within 3 seconds. These settings enable rapid data processing and efficient transmission, ensuring data accuracy and real-time performance. This provides high-quality, low-latency input to the central decision-making module, enabling the temperature control system to quickly respond to changes in on-site temperature and improving the timeliness and effectiveness of temperature control.

[0075] In the preferred embodiment, the temperature prediction model of the AI ​​temperature control decision module is a two-layer stacked LSTM structure. The input includes concrete temperature, ambient temperature and humidity, cooling water parameters, dam water level, and construction progress data. It is trained using RMSE as the loss function and Adam as the optimizer, and outputs the predicted value of concrete temperature for future periods. The above settings capture the complex patterns of concrete temperature changes through deep learning and improve prediction accuracy by combining multi-source data. This enables the temperature control system to predict temperature trends in advance and take more precise temperature control measures, effectively avoiding dam damage caused by abnormal temperatures.

[0076] In the preferred embodiment, the AI-powered temperature control decision-making module integrates a large-scale language model to parse construction text data, transform expert experience into control parameters, and generate temperature control diagnostic reports, strategy explanations, and operation and maintenance suggestions. These features enhance the system's intelligence level, making temperature control decisions more scientific and reasonable. They also improve the system's interpretability and operation and maintenance efficiency, providing strong decision support for engineers and reducing temperature control errors caused by human factors.

[0077] In the preferred embodiment, the cooling water flow execution unit adopts a staggered cooling pipe with water-blocking blocks arranged at intervals of 20-40cm inside the pipe. This, combined with intelligent valves and variable frequency water pumps, enables precise adjustment of the flow rate in different zones. The pipe material is pressure-resistant stainless steel or HDPE. This configuration, by optimizing the cooling pipe structure, improves the heat exchange efficiency of the cooling water, resulting in more uniform and rapid cooling of the concrete. At the same time, the combined use of intelligent valves and variable frequency water pumps achieves precise control of the cooling flow rate, which not only meets the temperature control requirements but also saves water resources and energy.

[0078] In the preferred embodiment, the longitudinal spacing between the cooling pipe network in step 1 is 1 to 2 meters, and the pipe diameter is 30 to 50 mm. Multiple parallel circulation pipes are formed through water distributors and collectors. The cooling water forms turbulence under the action of staggered water-blocking blocks to enhance heat exchange. The above settings ensure the uniform distribution and efficient heat exchange of cooling water inside the dam body, so that the concrete dam body can be fully cooled during the pouring process, effectively preventing cracks caused by excessive temperature, and improving the overall quality and safety of the dam body.

[0079] In the preferred embodiment, the edge computing unit in step 2 extracts time, space, and thermodynamic features, calculates short-term temperature trends and a comprehensive risk assessment index, and uploads the feature vector after dimensionality reduction and compression. The above settings, through multi-dimensional feature extraction and risk assessment, provide more comprehensive and accurate data support for the central decision-making module, making temperature control decisions more scientific and reasonable. At the same time, dimensionality reduction and compression reduce the amount of data transmission and improve the system's operating efficiency.

[0080] In the preferred embodiment, when the temperature control is executed in step 4, if the concrete temperature is higher than the threshold, the cooling flow rate in the corresponding area is increased and the inlet water temperature is reduced; if the temperature is close to the threshold, the flow rate is reduced to save energy, thus achieving dynamic and precise temperature control. The above settings dynamically adjust the cooling flow rate and inlet water temperature according to the actual internal temperature of the concrete dam, thereby achieving precise temperature control. This not only avoids damage to the dam caused by excessive temperature, but also saves water resources and energy, improving the economy and environmental friendliness of the temperature control system.

[0081] In summary, this invention proposes a dam body temperature control system and method based on fiber optic sensing and intelligent algorithms, effectively solving the temperature control problem in the construction of concrete dams in the field of water conservancy engineering, especially addressing the problems of insufficient real-time monitoring capabilities, lag in dynamic response, and low temperature control accuracy in existing technologies. Through this invention, the limitations of traditional temperature control methods, such as reliance on manual adjustment, low data processing efficiency, and serious waste of cooling water resources, are successfully overcome, achieving high-precision and high-efficiency temperature control during the construction of concrete dams.

[0082] In terms of technical implementation, firstly, distributed fiber optic temperature measurement technology is deeply integrated with artificial intelligence algorithms to form a measurement-calculation-control closed-loop system, achieving dynamic response and precise adjustment of the temperature field of concrete dams, bringing new application ideas to the field of concrete dam temperature control. Secondly, edge computing units are introduced to conduct low-latency, high-reliability data processing on-site in the dam area, changing the traditional data processing mode and greatly improving the system response speed and data processing efficiency. Thirdly, water-blocking block staggered cooling pipes are integrated, with a unique structural design that extends the cooling water flow path, enhances the turbulent heat transfer effect, improves heat transfer efficiency, and achieves faster and more uniform concrete cooling, differing from traditional cooling pipe design methods. Additionally, large-scale language models (such as DeepSeek and ChatGPT) are used to enhance the system's natural language processing and knowledge reasoning capabilities, enabling it to intelligently parse text data such as construction logs, technical specifications, and historical maintenance reports to generate temperature control diagnostic reports, strategy explanations, and operation and maintenance suggestions, introducing new technical means to concrete dam temperature control systems.

[0083] In terms of control strategy and system design, a multi-layered AI architecture-based artificial intelligence temperature control decision module is adopted. Combining traditional machine learning, deep learning, and large-scale language models, it achieves dynamic prediction of temperature changes, generation of control strategies, and system self-optimization, effectively improving the accuracy and intelligence of decision-making. By optimizing the control strategy through reinforcement learning and continuously adjusting and improving the temperature control method based on real-time data, the accuracy and energy efficiency of temperature control are significantly improved, while energy consumption is reduced. The designed complete closed-loop system of measurement, calculation, and control can automatically adjust the cooling water flow and temperature without frequent manual operation, greatly reducing the intensity of manual intervention, water resource consumption, and energy consumption. This achieves refined and intelligent management and control, improving construction efficiency and quality. Temperature data collected from multiple depths and regions by the distributed fiber optic temperature measurement module, combined with real-time feature extraction and preliminary trend analysis by the edge computing unit, provides high-quality, low-latency input to the central decision module. This data processing and analysis workflow design better meets the complex requirements of concrete dam temperature control.

Claims

1. A dam body temperature control system based on fiber optic sensing and intelligent algorithms, characterized in that: It includes a distributed temperature measurement module, an edge computing unit, an artificial intelligence temperature control decision module, and a cooling water flow execution unit. The distributed temperature measurement module connects to the edge computing unit to complete the real-time acquisition and transmission of spatiotemporal temperature sequence data inside the concrete dam. The edge computing unit connects to the artificial intelligence temperature control decision module to complete the preprocessing, feature extraction, and data refinement of multi-source monitoring data. The artificial intelligence temperature control decision module connects to the cooling water flow execution unit to realize temperature trend prediction, temperature control strategy generation, and command issuance. The cooling water flow execution unit connects back to the edge computing unit to realize closed-loop regulation of cooling water flow and temperature and operational status feedback, jointly completing intelligent and precise temperature control through measurement, calculation, control, and feedback.

2. The dam temperature control system based on fiber optic sensing and intelligent algorithms according to claim 1, characterized in that: The distributed temperature measurement module uses high-temperature and corrosion-resistant distributed optical fibers, which are laid in a serpentine pattern synchronously with the cooling pipe network. The spacing between temperature measurement points is no more than 1 meter, forming a continuous temperature monitoring network covering the entire cross-section of the dam.

3. The dam temperature control system based on fiber optic sensing and intelligent algorithms according to claim 1, characterized in that: The edge computing unit uses a multi-core processor industrial computer, with a data acquisition frequency of 1 to 5 minutes per time. Data cleaning is completed through sliding window anomaly detection, Z-Score outlier determination, and linear interpolation. After normalization, principal component analysis, and compression, the data is refined and uploaded within 3 seconds.

4. The dam temperature control system based on fiber optic sensing and intelligent algorithms according to claim 1, characterized in that: The temperature prediction model of the AI ​​temperature control decision module is a two-layer stacked LSTM structure. The input includes concrete temperature, ambient temperature and humidity, cooling water parameters, dam water level and construction progress data. It is trained with RMSE as the loss function and Adam as the optimizer, and outputs the predicted value of concrete temperature for future periods.

5. The dam body temperature control system based on fiber optic sensing and intelligent algorithms according to claim 1, characterized in that: The AI-powered temperature control decision-making module integrates a large-scale language model to parse construction text data, transform expert experience into control parameters, and generate temperature control diagnostic reports, strategy explanations, and operation and maintenance suggestions.

6. The dam body temperature control system based on fiber optic sensing and intelligent algorithms according to claim 1, characterized in that: The cooling water supply unit adopts a water-blocking block staggered cooling pipe. The water-blocking blocks inside the pipe are arranged in a staggered manner at intervals of 20-40cm. Together with intelligent valves and variable frequency water pumps, it can achieve precise adjustment of zoned flow. The pipe material is pressure-resistant stainless steel or HDPE.

7. A dam body temperature control method based on fiber optic sensing and intelligent algorithms, characterized in that, This method employs a dam body temperature control system based on fiber optic sensing and intelligent algorithms to achieve dam body temperature control, and includes the following steps: Step 1: Pre-embed a crisscrossing cooling pipe network and simultaneously lay distributed temperature measurement optical fibers to build a dam body temperature monitoring network; Step 2: The edge computing unit collects real-time data on concrete temperature, environment, cooling water, and construction progress, and completes cleaning, normalization, feature extraction, and data refinement. Step 3: The AI-powered temperature control decision-making module predicts temperature change trends using an LSTM model and generates the optimal temperature control strategy by combining it with a large-scale language model. Step 4: The cooling water execution unit adjusts the cooling water flow and temperature according to the strategy to precisely cool the concrete; Step 5: Real-time feedback of temperature monitoring data, online correction of prediction models and control strategies to achieve closed-loop adaptive temperature regulation.

8. The dam body temperature control method based on fiber optic sensing and intelligent algorithms according to claim 7, characterized in that: The longitudinal spacing between the cooling pipe network in step 1 is 1 to 2 meters, and the pipe diameter is 30 to 50 mm. Multiple parallel circulation pipes are formed through the water distributor and collector. The cooling water forms turbulence under the action of the staggered water blocking blocks to enhance heat exchange.

9. The dam body temperature control method based on fiber optic sensing and intelligent algorithms according to claim 7, characterized in that: In step 2, the edge computing unit extracts the time dimension, spatial dimension, and thermodynamic features, calculates the short-term temperature trend and comprehensive risk assessment index, and uploads the feature vector after dimensionality reduction and compression.

10. The dam body temperature control method based on fiber optic sensing and intelligent algorithms according to claim 7, characterized in that: When the temperature control described in step 2 is executed, if the concrete temperature is higher than the threshold, the cooling flow rate of the corresponding area will be increased and the inlet water temperature will be reduced; if the temperature is close to the threshold, the flow rate will be reduced to save energy, thereby achieving dynamic and precise temperature control.

Images