Intelligent temperature control method and system

Abstract

The invention discloses an intelligent temperature control method and system, and belongs to the technical field of temperature control, and the method specifically comprises the steps: collecting the related data of a target region in real time; according to the relevant data of the target area, a four-field coupling dynamic prediction model is constructed, the temperature of the target area is predicted, the four-field coupling dynamic prediction model is constructed based on a four-field coupling model and the relevant data, collected in real time, of the target area, and the four-field coupling model is obtained based on simultaneous coupling of a basic field variable model; determining a cooling water pipe flow control strategy according to the predicted temperature of the target area; cooling the target area until the target temperature is reached; the temperature monitoring precision, the prediction accuracy and the control intelligence level of the mass concrete structure in the construction process are remarkably improved, the crack risk is effectively reduced, and the construction quality and the safety guarantee capacity are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of temperature control, and in particular relates to an intelligent temperature control method and system. Background Art

[0002] In water conservancy infrastructure construction, large-volume concrete structures (thickness greater than 1 meter) are widely used in key load-bearing areas such as dam bodies, gate piers, and foundations. However, during the concrete hydration and hardening process, a large amount of hydration heat accumulates within the structure, causing the core area to rapidly rise in temperature. This creates a large internal and external temperature difference and temperature gradient, which can easily induce thermal cracks and strength degradation, compromising the safety and durability of the project.

[0003] The following temperature control technologies are currently commonly used in engineering practice: distributed fiber optic sensors for temperature monitoring, PID controllers for adjusting cooling water valves or flow rates, and cooling water pipe networks laid out inside the concrete to control the cooling rate through circulating cold water.

[0004] The defects of existing temperature control technology are: large errors occur deep in the structure, and the error rate of traditional temperature field prediction models is generally large, resulting in large errors in subsequent temperature adjustment. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a temperature intelligent control method and system.

[0006] To achieve the above object, the present invention provides the following technical solutions: A temperature intelligent control method, comprising: Collect relevant data of the target area in real time; Based on the relevant data of the target area, a four-field coupling dynamic prediction model is constructed to predict the temperature of the target area; Determine the cooling water pipe flow control strategy based on the predicted temperature of the target area; The target area is cooled until the target temperature is reached.

[0007] Specifically, the four-field coupling dynamic prediction model is constructed based on the relevant data of the target area, including: Establish a basic field variable model based on historical data of the target area; The basic field variable equations are coupled together to obtain a four-field coupling model; Based on the four-field coupling model and the real-time collected target area related data, a four-field coupling dynamic prediction model is constructed to predict the temperature of the target area.

[0008] Specifically, the basic field variable model is established based on the historical data of the target area, including: a hydrothermal model, a temperature field model, a stress field model and a humidity field model.

[0009] Specifically, the basic field variable equations are coupled simultaneously to obtain a four-field coupling model, including: Set the temperature field as the main controlled variable; Divide the target area into three-dimensional voxel units or finite element nodes, and establish a state vector containing four variables at each node; Constructing a coupling path between the four variables at each node and determining an influence weight of each coupling path; At each time step and each node, a state update is performed based on the state of the previous time step and the influence weight; The node states are combined into a four-field coupling model, wherein the state vector of each node is a time series state vector, and each dimension of the four-field coupling model represents a spatial position, a time point, and four variables.

[0010] Specifically, constructing a coupling path between the four variables at each node and determining the influence weight of each coupling path includes: Analyze the physical interaction path between each variable and construct a coupling path dominated by the temperature field; Identify the types of influencing factors between pairs of variables in the coupling path; Construct a coupling path mapping matrix, in which rows represent affected variables and columns represent influencing variables. Based on the coupling path mapping matrix, an initial influence weight is assigned to each coupling path and dynamically adjusted according to the temporal state changes of the variables.

[0011] Specifically, in each time step and each node, the state update is performed according to the state of the previous time step and the influence weight, including: Extract the state vector of each node from the previous time step; Based on the constructed coupling path and its influence weight, the state vector of each node is updated; The updated state vector is passed to the next moment for iterative evolution.

[0012] Specifically, the four-field coupled dynamic prediction model is constructed to predict the temperature of the target area, including: Combined with the real-time relevant data of the target area, the four-field coupling model outputs time series data of four variables; Construct a multimodal sequence neural network model, set the loss function, and train the constructed multimodal sequence neural network model; The time series data of the four variables are input into the trained multimodal sequence neural network, and the predicted temperature of the target area is output.

[0013] Specifically, determining the cooling water pipe flow control strategy based on the predicted temperature of the target area includes: Average all predicted temperatures within the target area and construct a temperature trend curve; Risk level classification of target areas based on temperature trend curves; Based on reinforcement learning, the current state of the target area is input and the optimal flow rate of the cooling water pipe is output; A PLC control signal is generated according to the optimal flow rate to adjust the actual flow rate of the cooling water pipe.

[0014] A temperature intelligent control system for implementing the aforementioned temperature intelligent control method, comprising: a data acquisition module, a prediction module, a flow control module and a cooling module; The data acquisition module is used to collect relevant data of the target area in real time; The prediction module is used to construct a four-field coupling dynamic prediction model based on the relevant data of the target area to predict the temperature of the target area; The flow control module is used to determine the cooling water pipe flow control strategy based on the predicted temperature of the target area; The cooling module is used to cool the target area until the target temperature is reached.

[0015] Specifically, the prediction module includes: a coupling unit and a prediction unit; The coupling unit is used to establish a basic field variable model, and to perform simultaneous coupling on the basic field variable equations to obtain a four-field coupling model; The prediction unit constructs a four-field coupling dynamic prediction model based on the four-field coupling model and the target area related data collected in real time to predict the temperature of the target area.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention proposes a temperature intelligent control method and system. By integrating multimodal sensing, a four-field coupled dynamic prediction model, and reinforcement learning control, it significantly improves the temperature monitoring accuracy, prediction accuracy, and intelligent control level of large-volume concrete structures during the construction process, realizes dynamic optimization and adjustment of the cooling flow, breaks through the bottleneck of the temperature control rate of ultra-thick structures, effectively reduces the risk of cracks, and improves construction quality and safety assurance capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a temperature intelligent control method provided by the present invention; Figure 2 Schematic diagram of the four-field coupling model provided by the present invention; Figure 3 This is an architecture diagram of a temperature intelligent control system provided by the present invention. DETAILED DESCRIPTION

[0018] The present application is described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but are not intended to limit the present application in any form. It should be noted that those skilled in the art may make several variations and improvements without departing from the scope of the present application. These all fall within the scope of protection of the present application.

[0019] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0020] It should be noted that, if there is no conflict, the various features in the embodiments of the present application can be combined with each other and are all within the scope of protection of the present application. In addition, although the functional modules are divided in the device schematic and the logical order is shown in the flow chart, in some cases, the steps shown or described can be performed in a different order than the module division in the device or the order in the flow chart. In addition, the words "first", "second", "third", etc. used in this application do not limit the data and execution order, but only distinguish between the same items or similar items with basically the same functions and effects.

[0021] Unless otherwise defined, all technical and scientific terms used in this specification have the same meanings as those commonly understood by those skilled in the art to which this application belongs. The terms used in this specification and in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The term "and / or" as used in this specification includes any and all combinations of one or more of the relevant listed items.

[0022] Example 1 See also Figure 1-Figure 2 The present invention provides an embodiment of a temperature intelligent control method, which is applied to the interior of concrete in a water conservancy project and includes the following specific steps: Step S1: real-time collection of relevant data of the target area, the target area being the surface and interior of the concrete of the water conservancy project, the relevant data of the target area including temperature, stress, humidity and auxiliary information.

[0023] In this embodiment, the target areas include but are not limited to: key sections inside concrete construction bodies (such as dam bodies, gate piers, and foundation cushions); surface areas prone to water loss or heat; cooling water pipe layout areas and temperature stress concentration areas; and locations where temperature differences and crack risks are known or predicted to exist.

[0024] The auxiliary information includes: ambient temperature and humidity, cooling water pipe flow, boundary conditions, etc.

[0025] Step S2: constructing a four-field coupling dynamic prediction model based on the relevant data of the target area.

[0026] The specific steps of step S2 are: Step S201: Establish a basic field variable model based on historical data of the target area.

[0027] In this embodiment, the basic field variable model includes hydration heat field, temperature field, stress field and humidity field. The hydration heat field is a heat source function model based on the hydration reaction rate; the temperature field is based on the conduction equation of non-steady-state heat and introduces anisotropic thermal conductivity; the stress field adopts a nonlinear viscoelastic-plastic constitutive relationship and couples the thermal expansion and contraction stress induced by temperature; the humidity field adopts a vapor migration-capillary diffusion joint model to reflect the transformation process of free water and adsorbed water.

[0028] Step S202: performing simultaneous coupling on the basic field variable equations to obtain a four-field coupling model.

[0029] like Figure 2 As shown, the specific steps of step S202 are: Step S2021: Set the temperature field as the main controlled variable.

[0030] In this embodiment, temperature is determined to be the core driving variable. The reason is that the release of hydration heat directly affects the temperature rise, the temperature change causes the material to expand and contract, thereby generating stress, and the temperature change affects the evaporation and migration of water, thereby affecting the humidity field. Therefore, temperature is used as the main axis, and its changes drive the linkage response of the other three fields.

[0031] Step S2022: Divide the target area into three-dimensional voxel units or finite element nodes, and establish a state vector containing four variables at each node.

[0032] In this embodiment, a three-dimensional voxel is a pixel block or cubic grid unit in three-dimensional space, similar to a two-dimensional image composed of many pixels. A three-dimensional structure can be divided into many tiny cubes, which are voxels. Each voxel is used to represent a set of states in the space, such as temperature, moisture, etc.; finite element nodes. In engineering modeling, complex structures (such as concrete dams) are usually divided into many small units, and the corner points of each unit are called finite element nodes.

[0033] Specifically, concrete is a large-volume, inhomogeneous material with highly uneven internal temperature, humidity, and stress distribution. After dividing the overall structure into small areas, a state vector containing four variables is established for each divided unit. In essence, each area is modeled separately to reflect the differences in the structure's behavior at different locations.

[0034] The four variables in the state vector include: hydration heat release rate, nodal temperature, nodal stress state, and nodal humidity.

[0035] Step S2023: constructing a coupling path between the four variables at each node, and determining an influence weight of each coupling path.

[0036] The specific steps of step S2023 are: Step S20231: Analyze the physical interaction path between each variable and construct a coupling path dominated by the temperature field.

[0037] In this embodiment, the coupling path refers to the path of how one variable affects another variable, and in this application, it refers to how the temperature variable affects the other three variables.

[0038] Specifically, in this step, the temperature variable is first clearly defined as the dominant variable, that is, in the entire four-field coupling model, changes in temperature are the main driving force that causes changes in other variables. Through physical mechanism analysis, the following coupling path is constructed: temperature → hydration heat release: increased temperature will accelerate the hydration reaction rate of cement, thereby causing an increase in heat; temperature → humidity migration: increased temperature will promote the evaporation and migration of internal moisture, reducing local humidity; temperature → stress evolution: temperature changes will cause thermal expansion and contraction of the material, and then form internal stress under the influence of boundary conditions; hydration heat → temperature: hydration heat release itself will further push up local temperature; humidity → hydration heat: cement hydration depends on available moisture, and a decrease in humidity will in turn slow down the release of hydration heat; stress → change in structural thermal conductivity: microcracks will form in high stress areas, changing thermal conductivity.

[0039] Step S20232: Identify the types of influencing factors between variable pairs in the coupling path.

[0040] Specifically, the variable pairs in each coupling path are analyzed and their impact mechanism types are identified, for example: accelerating impact: such as increasing temperature accelerating the release of hydration heat; inhibiting impact: such as decreasing humidity slowing down the release of hydration heat; threshold triggering: such as stress exceeding a certain critical value will affect the heat conduction path; feedback impact: such as hydration heat release in turn increasing the temperature, forming a positive feedback; asymmetric impact: such as temperature has a significant impact on humidity, but humidity has a weaker impact on temperature; this classification is not only used for qualitative description, but will also be reflected as a dynamic weighting factor in the subsequent impact weight calculation.

[0041] Step S20233: constructing a coupling path mapping matrix, wherein the rows represent the affected variables and the columns represent the influencing variables.

[0042] Specifically, based on the variable paths and influencing factors identified in the previous step, a coupling path mapping matrix is ​​constructed. The rows of this matrix represent the affected variables, the columns represent the influencing variables, and the matrix elements are the path type labels or initial weight ratios. The simplified coupling path mapping matrix is ​​shown in Table 1. Table 1

[0043] Step S20234: Based on the coupling path mapping matrix, an initial influence weight is assigned to each coupling path, and dynamically adjusted according to the temporal state change of the variable.

[0044] In this embodiment, an initial influence weight is first assigned to each pair of variable paths based on empirical values, literature parameters, or a priori models. Dynamic adjustments are then made, with the following specific logic: reading the four variable states of each node at the current moment; determining the variable change rate, such as the temperature rise rate or humidity decrease rate; and calling a preset adjustment rule library for adjustment based on the current state and change trend. After all path weights are output, the four-field coupling model state of the node is updated.

[0045] The regulation rule base includes: if the temperature rises rapidly, the weight of hydration heat and humidity will be enhanced; if the humidity drops below a certain critical value, the support weight for hydration heat will be automatically suppressed; if the stress accumulation approaches the warning value, the feedback regulation factor of the temperature control path will be activated.

[0046] Step S2024: In each time step and each node, perform state update based on the state of the previous time step and the influence weight.

[0047] The specific steps of step S2024 are: Step S20241: Extract the state vector of each node from the previous time step.

[0048] In this embodiment, the data at the current time t is processed in chronological order. For each node in the three-dimensional grid, its state vector recorded at the previous time point t-1 is read, including the current hydration heat release rate, the current node temperature, the current node stress state, and the current node humidity.

[0049] Step S20242: Based on the constructed coupling path and its influence weight, the state vector of each node is updated.

[0050] In this embodiment, temperature is used as the main controlled variable, and its current value and rate of change will affect the other three variables. The value of each variable at time t is iteratively calculated using the coupling path and influence weight, and the state vector at time t is updated based on the calculated value.

[0051] Step S20243: Pass the updated state vector to the next moment for iterative evolution.

[0052] Specifically, the latest state vector of each node at the current moment is stored in the state buffer as the input of the next time step t+1, and the loop is iterated in sequence.

[0053] The benefits of this step are: an explicit update is performed on the state vector of each node at each time step. By introducing variable coupling paths and weight mechanisms, dynamic linkage between temperature, hydration heat, humidity and stress is achieved, and it has time continuity, state evolution and predictive adaptability, thereby providing a basis for subsequent flow regulation and risk warning. This state update mechanism can significantly improve the response speed and prediction accuracy of the temperature control system in actual engineering.

[0054] Step S2025: combining the node states into a four-field coupling model, wherein the state vector of each node is a time series state vector, and each dimension of the four-field coupling model represents a spatial position, a time point, and four variables.

[0055] like Figure 2 As shown, in Figure 2 In the figure, the X, Y, and Z axes represent the spatial dimensions, indicating the three-dimensional division of the concrete structure. Figure 2 The dots in the figure represent each voxel or finite element node in the concrete structure. The → below represents the time axis, which describes the continuous evolution of the state vector of each node over time. The variable dimensions are (H, T, S, M): Each node records four basic field variables at each moment to form a state vector V(t) = [H, T, S, M], where H represents the hydration heat release rate, T represents the node temperature, S represents the node stress state, and M represents the node humidity.

[0056] The benefits of the steps are: by setting temperature as the main controlled variable and constructing a coupled hydration heat, humidity and stress variable state system, high-precision dynamic prediction and modeling of multi-physical fields are achieved through node division in space, state evolution in time, and path weight linkage in variables. The proposed four-field coupling model has engineering practicality and can support the intelligent, precise and automated development direction of concrete temperature control in water conservancy projects, solving technical problems such as the inability of traditional temperature control models to dynamically adapt, coupling failure, and response delay.

[0057] Step S203: Based on the four-field coupling model and the real-time collected target area related data, a four-field coupling dynamic prediction model is constructed to predict the temperature of the target area.

[0058] The specific steps of step S203 are: Step S2031: In combination with the real-time relevant data of the target area, the four-field coupling model outputs time series data of four variables.

[0059] Step S2032: construct a multimodal sequence neural network model, set a loss function, and train the constructed multimodal sequence neural network model.

[0060] In this embodiment, the network structure design includes: input, four groups of time series as four modal channels of network input; main structure, using a neural network structure based on time series modeling, such as GRU, LSTM or Transformer; modal fusion, fusing the sequence features of the four variables through an attention mechanism or a multi-channel gating unit; output, the temperature prediction value of the target node in the next several time steps.

[0061] The loss function uses the mean square error (MSE) loss function to evaluate the difference between the predicted temperature and the actual temperature.

[0062] Training of the multimodal sequence neural network model: Use historical measurement data as the training set; organize samples using a sliding window method to improve the model's ability to identify short-term trends; perform multiple rounds of iterative training, optimize weight parameters, complete model fitting, and obtain a trained multimodal sequence neural network model.

[0063] Step S2033: Input the time series data of the four variables into the trained multimodal sequence neural network and output the predicted temperature of the target area.

[0064] Step S3: Determine the cooling water pipe flow control strategy based on the real-time predicted temperature data.

[0065] The specific steps of step S3 are: Step S301: averaging all predicted temperatures within the target area and constructing a temperature trend curve; In this step, the predicted temperature values ​​of all spatial nodes in the target area are extracted from the obtained temperature prediction results, and the temperature values ​​of all voxels or finite element nodes are weighted averaged. Considering the position, material differences or structural importance, the average temperature value of the target area at each prediction time point is obtained. The average temperature values ​​are organized into a temperature trend curve in chronological order, and the curve is smoothed using methods such as spline interpolation or local weighted regression to suppress high-frequency noise. This trend curve will be used to evaluate the temperature rise rate, temperature peak, etc.

[0066] Step S302: Classify the target area into risk levels based on the temperature trend curve.

[0067] In this embodiment, the target area is classified into thermal risk levels based on key indicators in the trend curve. Key criteria include, but are not limited to: maximum temperature threshold: if the predicted temperature exceeds 75°C, the risk is determined to be high; heating rate threshold: if the average heating rate exceeds 0.7°C / h, the risk is determined to be medium-high; temperature gradient change rate: if the difference between the internal and surface temperatures exceeds 20°C, the risk is determined to be extremely high; and multi-point peak synchronization: if multiple points enter the rapid heating range simultaneously, the risk level is increased.

[0068] Based on the above conditions, the target areas are divided into four levels: Level I (low risk): normal maintenance; Level II (medium risk): mild cooling; Level III (high risk): active forced cooling; Level IV (extremely high risk): multi-point scheduling and joint control, key tracking.

[0069] Step S303: Based on reinforcement learning, the current state of the target area is input and the optimal flow rate of the cooling water pipe is output.

[0070] In this embodiment, a cooling strategy model designed based on Q-Learning or deep reinforcement learning algorithm is adopted, and the state definition is: current temperature prediction trend; current risk level; cooling water flow in the previous time step; auxiliary information such as local temperature difference and stress distribution; action definition: control the flow adjustment action of each or each group of cooling water pipes, and the action space includes increasing the flow, decreasing the flow or keeping it unchanged, or setting precision control in the continuous action space; reward function design: the reward goal is to control the predicted temperature within the safe range, minimize the cooling energy consumption, and increase the stress smoothly. If the temperature enters the safe zone, a positive reward is given; if the temperature rise accelerates or the energy consumption is too high, a penalty is given; strategy execution: the trained reinforcement learning model outputs the optimal solution for the cooling flow according to the current state, automatically updates the strategy within each prediction cycle, and supports adaptive dynamic adjustment.

[0071] Step S304: Generate a PLC control signal according to the optimal flow rate to adjust the actual flow rate of the cooling water pipe.

[0072] In this embodiment, the flow control of the cooling water pipe relies on data interaction and control instruction transmission between the programmable logic controller (PLC) and the temperature control prediction platform. First, the cooling strategy is converted into a standard PLC control instruction set, including target flow, adjustment speed, valve opening and closing actions, etc. Through the edge computing platform, the instructions are sent to the control valve of each section of the cooling water pipe in real time. The control valve realizes predictive-driven intelligent temperature control of the internal temperature of the large-volume concrete structure and adjusts the water flow according to the instructions.

[0073] The PLC uses industry-standard communication interfaces, including Modbus TCP / IP and Ethernet for configuration software; RS485 for accessing edge data acquisition gateways; and OPC UA for data connection to the cloud or control platform. These interfaces ensure data exchange with flow meters, actuators, and prediction systems, and support two-way control signal interaction.

[0074] The benefits of this step are: predictive intelligent temperature control of the internal temperature of large-volume concrete structures is achieved, flow control is transformed from empirical regulation to model decision-making management, cooling efficiency is higher and energy consumption is lower.

[0075] Step S4: Cool the target area until the target temperature is reached.

[0076] Example 2 See also Figure 3 , another embodiment provided by the present invention: a security assessment system, comprising: a data acquisition module, a prediction module, a flow control module and a cooling module; The data acquisition module is used to collect relevant data of the target area in real time; The prediction module is used to construct a four-field coupling dynamic prediction model based on the relevant data of the target area to predict the temperature of the target area; The flow control module is used to determine the cooling water pipe flow control strategy based on the predicted temperature of the target area; The cooling module is used to cool the target area until the target temperature is reached.

[0077] The prediction module includes: a coupling unit and a prediction unit; The coupling unit is used to establish a basic field variable model, and to perform simultaneous coupling on the basic field variable equations to obtain a four-field coupling model; The prediction unit constructs a four-field coupling dynamic prediction model based on the four-field coupling model and the target area related data collected in real time to predict the temperature of the target area.

[0078] In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0079] The above-described specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A temperature intelligent control method, characterized in that: include: Collect relevant data of the target area in real time; According to the relevant data of the target area, a four-field coupling dynamic prediction model is constructed to predict the temperature of the target area, wherein the four-field coupling dynamic prediction model is constructed based on a four-field coupling model and the relevant data of the target area collected in real time, and the four-field coupling model is obtained based on the simultaneous coupling of a basic field variable model; Determine the cooling water pipe flow control strategy based on the predicted temperature of the target area; The target area is cooled until the target area reaches a target temperature.

2. The intelligent temperature control method according to claim 1, wherein: The four-field coupling dynamic prediction model is constructed based on the relevant data of the target area, including: Establish a basic field variable model based on historical data of the target area; The basic field variable equations are coupled together to obtain a four-field coupling model; Based on the four-field coupling model and the real-time collected target area related data, a four-field coupling dynamic prediction model is constructed to predict the temperature of the target area.

3. The intelligent temperature control method according to claim 2, wherein: The basic field variable model is established based on the historical data of the target area, including: a hydrothermal model, a temperature field model, a stress field model and a humidity field model.

4. The method according to claim 2, wherein The four-field coupling model is obtained by simultaneously coupling the basic field variable equations, including: Set the temperature field as the main controlled variable; The target area is divided into three-dimensional voxel units or finite element nodes, and a state vector containing four variables: hydration heat, temperature, stress, and humidity is established at each node; Constructing a coupling path between the four variables at each node and determining an influence weight of each coupling path; At each time step and each node, a state update is performed based on the state of the previous time step and the influence weight; The node states are combined into a four-field coupling model, wherein the state vector of each node is a time series state vector, and each dimension of the four-field coupling model represents a spatial position, a time point, and four variables.

5. The intelligent temperature control method according to claim 4, characterized in that: The step of constructing a coupling path between the four variables at each node and determining the influence weight of each coupling path includes: Analyze the physical interaction path between each variable and construct a coupling path dominated by the temperature field; Identify the types of influencing factors between pairs of variables in the coupling path; Construct a coupling path mapping matrix, in which rows represent affected variables and columns represent influencing variables. Based on the coupling path mapping matrix, an initial influence weight is assigned to each coupling path and dynamically adjusted according to the temporal state changes of the variables.

6. The intelligent temperature control method according to claim 4, characterized in that: In each time step and each node, the state is updated according to the state and influence weight of the previous time step, including: Extract the state vector of each node from the previous time step; Based on the constructed coupling path and its influence weight, the state vector of each node is updated; The updated state vector is passed to the next moment for iterative evolution.

7. The intelligent temperature control method according to claim 2, wherein: The four-field coupled dynamic prediction model is constructed to predict the temperature of the target area, including: Combined with the real-time relevant data of the target area, the four-field coupling model outputs time series data of four variables; Construct a multimodal sequence neural network model, set the loss function, and train the constructed multimodal sequence neural network model; The time series data of the four variables are input into the trained multimodal sequence neural network, and the predicted temperature of the target area is output.

8. The intelligent temperature control method according to claim 7, characterized in that: Determining the cooling water pipe flow control strategy based on the predicted temperature of the target area includes: Average all predicted temperatures within the target area and construct a temperature trend curve; Risk level classification of target areas based on temperature trend curves; Based on reinforcement learning, the current state of the target area is input and the optimal flow rate of the cooling water pipe is output; A PLC control signal is generated according to the optimal flow rate to adjust the actual flow rate of the cooling water pipe.

9. A temperature intelligent control system, used to implement a temperature intelligent control method according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, prediction module, flow control module and cooling module; The data acquisition module is used to collect relevant data of the target area in real time; The prediction module is used to construct a four-field coupling dynamic prediction model based on the relevant data of the target area to predict the temperature of the target area; The flow control module is used to determine the cooling water pipe flow control strategy based on the predicted temperature of the target area; The cooling module is used to cool the target area until the target temperature is reached.

10. The intelligent temperature control system according to claim 9, characterized in that: The prediction module includes: a coupling unit and a prediction unit; The coupling unit is used to establish a basic field variable model, and to perform simultaneous coupling on the basic field variable equations to obtain a four-field coupling model; The prediction unit constructs a four-field coupling dynamic prediction model based on the four-field coupling model and the target area related data collected in real time to predict the temperature of the target area.

Images