Method and system for temperature control of corrosion resistant refrigeration tube with intelligent temperature control coating

Abstract

The application discloses a temperature control method and system of a corrosion-resistant refrigeration pipe with an intelligent temperature control coating, and relates to the technical field of refrigeration process control. The method collects the temperature, flow rate, external wind speed and temperature in each electric cooling section pipe during the refrigeration process, uses the coupled heat balance equation of Fourier's law and convective heat transfer to quantize the radial and axial heat conduction, and calculates the heat absorption amount. A physically embedded meta-learning graph neural network is designed to map the heat absorption amount to a target voltage. The influence of the energized voltage of the electric cooling section on the remaining electric cooling sections is considered to design a voltage coupling conduction function, and a group intelligence algorithm is used to iteratively determine the optimal energized voltage. The temperature difference feedback is used to optimize the parameters. The system includes a collection module, a regulation and control module and a feedback module, which respectively perform data collection, voltage regulation and control, and parameter optimization decision, so as to accurately control the temperature, eliminate the interference between sections and adapt to the cooling demand under multiple working conditions.

Description

Technical Field

[0001] This invention relates to the field of refrigeration process control technology, and more specifically to a temperature control method and system for corrosion-resistant refrigeration pipes with intelligent temperature control coating. Background Technology

[0002] As a core component in cold chain transportation and industrial precision cooling, refrigeration pipes must simultaneously meet the dual requirements of corrosion resistance and high-precision temperature control. Existing technologies still have significant shortcomings in temperature control: Firstly, current temperature control methods for calculating heat loss in refrigeration pipes do not comprehensively consider key factors such as the flow rate of the medium inside the pipe and external environmental factors like wind speed and temperature, leading to large deviations in heat loss calculations and an inability to accurately reflect the actual heat absorption. Secondly, while some technologies introduce electrocooling sections, they use fixed algorithms or simple neural networks to determine the energizing voltage, failing to quantify the mutual influence of voltages in each electrocooling section. This can easily lead to voltage and heat absorption mismatch due to inter-section coupling interference, resulting in localized overcooling or insufficient cooling capacity, making it difficult to meet the high-precision requirements of precision instrument cooling and thus limiting the application boundaries of corrosion-resistant refrigeration pipes. Summary of the Invention

[0003] The purpose of this invention is to provide a temperature control method and system for corrosion-resistant refrigeration pipes with intelligent temperature control coatings. By coupling the heat balance equation to calculate the heat absorption, physically embedding the learning graph neural network to map the voltage, and using the voltage coupling conduction function to quantify the inter-segment influence and determine the optimal energizing voltage, the invention achieves the goals of precise temperature control, elimination of inter-segment interference, and adaptation to cooling requirements under multiple operating conditions.

[0004] The technical solution to achieve the objective of this invention is as follows:

[0005] On one hand, the present invention provides a temperature control method for a corrosion-resistant refrigeration tube with an intelligent temperature control coating, comprising the following steps:

[0006] During the cooling process, the tube temperature and flow rate of each electrocooling section are collected, and the external wind speed and external temperature are measured simultaneously.

[0007] Based on Fourier's law and the convective heat transfer formula, a coupled heat balance equation is constructed, and the heat absorption of each electrocooling segment is calculated. The heat absorption of each electrocooling segment is converted and mapped to the target voltage through a physical embedded element learning graph neural network. With the goal of satisfying the target voltage of each electrocooling segment, the voltage coupling transfer function is designed by considering the influence of the voltage of the electrocooling segment on the other electrocooling segments. The optimal voltage is determined iteratively through a swarm intelligence algorithm and corresponding control is applied.

[0008] Calculate the absolute temperature difference between the tube temperature of each electrocooling section and the preset target temperature, and compare the absolute temperature difference with the temperature difference threshold to decide whether to simulate and optimize to obtain the optimal parameters of the coupled thermal balance equation and the physical embedded element learning graph neural network.

[0009] Furthermore, by combining physical embedding meta-learning graph neural networks with the collaborative calculation of voltage-coupled conduction functions, the heat absorption is converted into the optimal energizing voltage, including the following steps:

[0010] The heat absorption, pipe temperature, pipe flow rate, external wind speed, and external temperature of each electrocooling section calculated based on the coupled heat balance equation are input into a physical embedding meta-learning graph neural network pre-trained with historical data from multiple operating conditions. This neural network has been pre-trained using operating data covering different scenarios of pipe flow rate, pipe temperature, external wind speed, and external temperature. Furthermore, physical constraint terms obtained by transforming the coupled heat balance equation are embedded during network construction. Finally, the network output layer linearly maps the node feature vectors to the target voltage of each electrocooling section without considering inter-segment coupling.

[0011] Based on the thermal conductivity of the corrosion-resistant base tube and the electro-cooling coating, the axial length of the electro-cooling section, the coating resistance, and the axial thermal contact area, the axial equivalent thermal conductivity is first calculated by combining the axial thermal cross-sectional areas of the base tube and the coating. The axial equivalent thermal conductivity is obtained by multiplying the thermal conductivity of the base tube by the sum of the axial thermal cross-sectional areas of the base tube and the coating by the thermal conductivity of the coating, and then dividing by the sum of the axial thermal cross-sectional areas of the base tube and the coating. Then, based on Fourier's law of heat conduction and Ohm's law of electricity, the inter-segment voltage influence attenuation coefficient is introduced to calculate the voltage coupling coefficient of a certain electro-cooling section to the temperature of other arbitrary electro-cooling sections. Based on this voltage coupling coefficient, the voltage coupling conduction function is further constructed.

[0012] Subsequently, the initial energizing voltage of each electrocooling segment, with the initial value initially set based on the target voltage, was substituted into the voltage coupling conduction function mentioned above. The external coupling voltage division generated by each electrocooling segment due to the influence of the energizing voltage of all other electrocooling segments was calculated. Among them, the electrocooling segments closer to the target segment had a larger attenuation coefficient due to the inter-segment voltage influence, and their external coupling voltage division value was higher. The electrocooling segments farther away had a smaller attenuation coefficient, and their voltage division influence on the target segment was significantly reduced.

[0013] The core optimization objective is to approximate the target voltage output by the physical embedded element learning graph neural network with the superposition value of the actual energized voltage and the externally coupled voltage divider. At the same time, an energy consumption constraint term based on Joule's law is introduced to construct a complete voltage optimization objective function. In addition, the upper and lower limits of the safe operating voltage of the electrocooling coating are set as constraints. The objective function is iteratively updated using a swarm intelligence algorithm until the objective function value tends to stabilize and the deviation between the superposition value of the actual energized voltage and the externally coupled voltage divider and the target voltage meets the preset accuracy requirements. The actual energized voltage of each electrocooling section obtained at this time is the optimal energized voltage.

[0014] Furthermore, the coupled heat balance equations are constructed and the heat absorption of each electrocooling section is calculated, including the following steps:

[0015] Based on the total length of the refrigeration tube and the axial length of a single electrocooling section, the refrigeration tube is divided into several independent electrocooling sections, and the heat of each electrocooling section is transferred radially and axially.

[0016] Calculate the total radial thermal resistance and the axial coupling thermal resistance, i.e. the obstacles to heat transfer in the radial and axial directions, respectively; and calculate the radial heat transfer and the axial net heat transfer based on the relationship between thermal resistance and temperature difference.

[0017] According to the principle of energy conservation, the heat absorption required by the electrocooling coating must completely offset the total heat loss of the section. Based on this, a coupled heat balance equation is constructed and the heat absorption that each electrocooling section needs to compensate is calculated.

[0018] Furthermore, considering the problem that purely data-driven neural networks are prone to violating physical laws, a physical embedding meta-learning graph neural network is constructed to generate the target voltage, including the following steps:

[0019] The heat absorption, tube temperature, tube flow rate, external wind speed, and external temperature of each electrocooling section are integrated into an input feature vector.

[0020] The coupled thermal balance equation is mathematically transformed to obtain a physical constraint term that reflects energy conservation. This physical constraint term is then added to the input feature vector to generate physical enhancement features.

[0021] Each electrocooling segment is used as a node in a graph neural network, and the axial thermal conductivity between adjacent electrocooling segments is used as the edge weight between nodes. This edge weight reflects the heat transfer correlation strength between segments. Based on this, an adjacency matrix is ​​constructed to complete the topology construction of the graph neural network. Among them, the thermal conductivity is inversely proportional to the axial coupling thermal resistance.

[0022] Each electrocooling segment is used as a node in the graph neural network, and the axial thermal conductivity between segments is used as the edge weight between nodes. Based on this, an adjacency matrix is ​​constructed to complete the topology construction of the graph neural network.

[0023] Historical multi-condition data covering different pipe flow rates, pipe temperatures, external wind speeds, and external temperatures were used to pre-train the constructed graph neural network. The pre-training was used to fit the common patterns of heat absorption-voltage mapping under each condition, and the pre-training meta-parameters of the network were generated.

[0024] The physical enhancement features are input into a graph neural network with pre-trained meta-parameters. Through a three-layer feature propagation mechanism, each layer updates node features based on edge weights to achieve inter-segment information interaction. Multiple rounds of iterative training are performed to optimize the network parameters.

[0025] The trained and fitted graph neural network linearly maps the feature vectors of the last layer nodes through the output layer to obtain the target voltage of each electrocooling segment without considering inter-segment coupling.

[0026] Furthermore, the heat in each electrocooling section is transferred radially and axially. Radial transfer is a unidirectional steady-state transfer along the fluid inside the pipe, the corrosion-resistant base pipe, the electrocooling coating, and the external air, with heat diffusing from inside the pipe to the outside or penetrating from the outside into the pipe. Axial transfer is carried out through the corrosion-resistant base pipe and the electrocooling coating between two adjacent electrocooling sections, with the transfer direction from the section with higher temperature to the section with lower temperature.

[0027] Furthermore, the total radial thermal resistance is formed by connecting four parts in series: the internal convective thermal resistance, the radial conductive thermal resistance of the corrosion-resistant base tube, the radial conductive thermal resistance of the electro-cooling coating, and the external convective thermal resistance. The total resistance of the series thermal resistance is the sum of the individual resistance values. The internal convective thermal resistance is calculated using the Dittus-Boelter empirical formula to determine the internal convective heat transfer coefficient and the internal convective heat transfer area, i.e., the contact area between the fluid inside the tube and the inner wall of the corrosion-resistant base tube. The internal convective thermal resistance is obtained by taking the reciprocal of the product of the heat transfer coefficient and the heat transfer area. The radial conductive thermal resistance of the corrosion-resistant base tube is determined based on the thermal conductivity characteristics of the cylindrical wall, utilizing the inner and outer diameters of the base tube and the thermal conductivity of the base tube material. The radial thermal resistance of the corrosion-resistant base tube is calculated using the formula for the thermal resistance of a cylindrical wall. Since the thickness of the electro-cooling coating is much smaller than the outer diameter of the corrosion-resistant base tube, its radial thermal conductivity can be simplified according to the thermal conductivity characteristics of a flat wall. The flat wall thermal resistance, i.e., the radial thermal resistance of the electro-cooling coating, is obtained by dividing the coating thickness by the product of the coating thermal conductivity and the coating heat transfer area. When calculating the external convective thermal resistance, the external convective heat transfer coefficient is first calculated using the Churchill-Bernstein empirical formula. Then, the external convective heat transfer area is determined, which is the contact area between the outer wall of the electro-cooling coating and the external air. The reciprocal of the product of the thermal resistance heat transfer coefficient and the heat transfer area is taken to obtain the external convective thermal resistance.

[0028] Furthermore, axial coupling thermal resistance reflects the obstruction of axial heat transfer between adjacent sections. Its calculation method is based on the principle of parallel thermal resistance, including the axial thermal resistance of the corrosion-resistant base tube and the axial thermal resistance of the electro-cooling coating:

[0029] The axial coupling thermal resistance is calculated by dividing the distance between segments by the product of the axial thermal conductivity of the base tube and the axial thermal cross-sectional area of ​​the base tube. The axial thermal cross-sectional area of ​​the base tube refers to the cross-sectional area of ​​the base tube perpendicular to the axial direction, which is the effective area for axial heat transfer of the base tube.

[0030] The axial thermal resistance of the electrocooling coating is calculated by dividing the distance between segments by the product of the coating's axial thermal conductivity and its axial thermal cross-sectional area. The axial thermal cross-sectional area of ​​the coating refers to the cross-sectional area of ​​the coating perpendicular to the axial direction, which is the effective area for axial heat transfer.

[0031] Since the base tube and the coating are parallel heat transfer paths, the reciprocal of the total axial coupling thermal resistance is equal to the sum of the reciprocal of the axial thermal resistance of the base tube and the reciprocal of the axial thermal resistance of the coating. Furthermore, since the structure of all electrocooled sections, namely the base tube size and the coating thickness, is consistent, the total axial coupling thermal resistance between all adjacent sections is equal.

[0032] Furthermore, radial heat transfer is calculated by subtracting the external temperature from the internal temperature and then dividing by the total radial thermal resistance. A positive result indicates that the fluid inside the pipe is releasing heat to the external environment, requiring the electro-cooling coating to absorb heat to compensate. A negative result indicates that the external environment is transferring heat to the fluid inside the pipe, requiring the electro-cooling coating to adjust its heat absorption intensity to maintain the target temperature. For the electro-cooled section from the second to the last segment, the axial net heat transfer is calculated by subtracting the heat transferred from the adjacent rear segment to the heat transferred to the adjacent front segment. The transferred or outgoing heat is calculated by subtracting the target segment temperature from the adjacent segment temperature and then dividing by the axial coupling thermal resistance. Heat is transferred from the segment with higher temperature to the segment with lower temperature. For the first segment, since there is no adjacent segment on the left, the heat transferred from the left is zero. For the last segment, since there is no adjacent segment on the right, the heat transferred to the right is zero.

[0033] Furthermore, the corrosion-resistant refrigeration tube with intelligent temperature control coating includes, from the inside out, a corrosion-resistant base tube, an electro-cooling coating, and an insulating protective layer. The electro-cooling coating is uniformly attached to the outer wall of the corrosion-resistant base tube, and electrode groups are arranged at preset distances along the axial direction of the refrigeration tube, forming the electro-cooling section between adjacent electrode groups.

[0034] In a second aspect, the present invention provides a temperature control system for a corrosion-resistant refrigeration tube with an intelligent temperature control coating, comprising a data acquisition module, a control module and a feedback module;

[0035] The data acquisition module collects the tube temperature and flow rate of each electrocooling section in each cycle, and simultaneously measures the external wind speed and external temperature.

[0036] In each cycle, the control module constructs a coupled heat balance equation based on Fourier's law and the convective heat transfer formula, calculates the heat absorption of each electrocooling segment, and converts the heat absorption of each electrocooling segment into a target voltage through a physical embedded element learning graph neural network. With the aim of satisfying the target voltage of each electrocooling segment, the module collaboratively considers the influence of the voltage of the electrocooling segment on the other electrocooling segments to design a voltage coupling transfer function. The optimal voltage is determined iteratively through a swarm intelligence algorithm and corresponding control is applied.

[0037] The feedback module compares the absolute temperature difference between the temperature of each electrocooling segment and the preset target temperature with the temperature difference threshold in each cycle to decide whether to simulate and optimize the optimal parameters of the coupled thermal balance equation and the physical embedding element learning graph neural network in this cycle.

[0038] Compared with the prior art, the significant advantages of this invention are:

[0039] 1. Taking into account the parameters inside and outside the tube, the heat absorption of each electrocooling section is quantified by radial and axial heat conduction through coupled heat balance equations, and the model parameters are dynamically adjusted according to the actual operating conditions.

[0040] 2. By using physical embedded meta-learning graph neural networks and voltage-coupled conduction functions in synergistic calculation, the heat absorption and energizing voltage are accurately mapped and inter-segment coupling interference is eliminated, effectively reducing the temperature fluctuation amplitude inside the tube and significantly improving temperature control accuracy. Attached Figure Description

[0041] Figure 1 A flowchart illustrating the temperature control method for corrosion-resistant refrigeration pipes with intelligent temperature control coating;

[0042] Figure 2 This is a flowchart illustrating the construction of the coupled heat balance equation and the calculation of heat absorption in this invention;

[0043] Figure 3 This is a flowchart illustrating the determination of the optimal energizing voltage in this invention.

[0044] Figure 4 This is a flowchart of the temperature difference comparison and parameter optimization decision-making process in this invention. Detailed Implementation

[0045] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0046] Example 1

[0047] like Figure 1 As shown, this invention discloses a temperature control method for a corrosion-resistant refrigeration tube with an intelligent temperature control coating, comprising the following steps:

[0048] During the cooling process, the tube temperature and flow rate of each electrocooling section are collected, and the external wind speed and external temperature are measured simultaneously.

[0049] Based on Fourier's law and the convective heat transfer formula, a coupled heat balance equation is constructed, and the heat absorption of each electrocooling segment is calculated. The heat absorption of each electrocooling segment is converted and mapped to the target voltage through a physical embedded element learning graph neural network. With the goal of satisfying the target voltage of each electrocooling segment, the voltage coupling transfer function is designed by considering the influence of the voltage of the electrocooling segment on the other electrocooling segments. The optimal voltage is determined iteratively through a swarm intelligence algorithm and corresponding control is applied.

[0050] Calculate the absolute temperature difference between the tube temperature of each electrocooling section and the preset target temperature, and compare the absolute temperature difference with the temperature difference threshold to decide whether to simulate and optimize to obtain the optimal parameters of the coupled thermal balance equation and the physical embedded element learning graph neural network.

[0051] Furthermore, the corrosion-resistant refrigeration pipe with intelligent temperature control coating consists of, from the inside out, a corrosion-resistant base pipe, an electro-cooling coating, and an insulating protective layer. The structure and arrangement of each part are as follows:

[0052] The electro-cooling coating uses a heat-absorbing material known in the art, the specific material composition and basic working principle of which are existing technologies and will not be elaborated here. The electro-cooling coating is uniformly applied to the outer wall of the corrosion-resistant base tube, spaced at predetermined intervals along the axial direction of the refrigeration tube. Electrode groups are set up to achieve segmented energization control of the electrocooling coating, and an electrocooling section is formed between two adjacent electrode groups.

[0053] Temperature sensors are embedded in the inner wall of the corrosion-resistant base pipe corresponding to each electrocooling section to monitor the pipe temperature of that section. A flow meter is connected in series at the input end of the cooling pipe to collect the fluid flow rate inside the pipe. Meanwhile, wind speed sensors and ambient temperature sensors are installed at predetermined intervals outside the refrigeration pipes to measure the wind speed and temperature of the external environment and record them as the external wind speed. and external temperature .

[0054] like Figure 2 As shown, further, a coupled heat balance equation is constructed based on Fourier's law and the convective heat transfer formula to calculate the heat absorption of each electrocooling section, including the following steps:

[0055] Based on the total length of the refrigeration pipe and the axial length of a single electro-cooling section Refrigeration pipes can be divided into The first section of the electrocooling section, denoted as the fluid delivery end within the refrigeration pipe, is... The electrocooling section near the receiving end is In the In each electrocooling section Heat transfer includes radial and axial heat transfer. Radial heat transfer is a unidirectional steady-state transfer along the fluid inside the pipe, the radial direction of the corrosion-resistant base pipe, the radial direction of the electro-cooling coating, and the external air. Axial heat transfer is carried out between adjacent sections through the axial direction of the corrosion-resistant base pipe and the axial direction of the electro-cooling coating. The direction of heat transfer is from the electro-cooling section with a higher temperature to the electro-cooling section with a lower temperature.

[0056] Calculate the total radial thermal resistance, where the first... Total radial thermal resistance of each electrocooling section The calculation formula is as follows:

[0057] ,

[0058] in, For the convection thermal resistance inside the tube, To improve the radial thermal resistance of the corrosion-resistant base tube, For the radial thermal resistance of the electrocooled coating, External convective thermal resistance, internal convective thermal resistance The calculation formula is as follows:

[0059] ,

[0060] in, For the first The convective heat transfer coefficient within the pipe section can be calculated using the Dittus-Boelter empirical formula. For the first The radial heat exchange area of ​​the corrosion-resistant base pipe is the contact area between the fluid inside the pipe and the inner wall of the corrosion-resistant base pipe. , For the inner diameter of the corrosion-resistant base pipe, the first Convection heat transfer coefficient within the pipe section The calculation formula is as follows:

[0061] ,

[0062] in, The Reynolds number of the fluid inside the pipe. The Prandtl number of the fluid inside the pipe is obtained from the fluid property table. Let be the thermal conductivity of the fluid inside the pipe. The inner diameter of the corrosion-resistant base pipe, where, , For fluid density, For fluid dynamic viscosity, The fluid velocity inside the pipe. Given the flow rate inside the pipe, based on Fourier's law of heat and the differential equation of heat conduction in cylindrical coordinates, a formula for heat conduction based on the cylinder wall can be derived, which is the radial thermal resistance of the corrosion-resistant base pipe. The calculation formula is as follows:

[0063] ,

[0064] in, For corrosion-resistant base pipe outer diameter With corrosion-resistant base pipe inner diameter The natural logarithm of the ratio quantifies the effect of the difference between the inner and outer diameters of the cylinder wall on the heat conduction path and thermal resistance. The larger the ratio of the outer diameter to the inner diameter of the corrosion-resistant base tube, the larger the value of the natural logarithm, and the greater the corresponding thermal resistance. The thermal conductivity of the corrosion-resistant base pipe material is affected by the thickness of the electro-cooling coating. Much smaller than the outer diameter of the corrosion-resistant base pipe The electrothermal coating can be approximated as a flat wall, therefore the radial thermal resistance of the electrothermal coating is low. The calculation formula is as follows:

[0065] ,

[0066] in, For the thickness of the electrothermal coating, The radial thermal conductivity of the electrocooling coating is... For the first The radial heat transfer area of ​​the corrosion-resistant base pipe-electrothermal coating section, i.e., the contact area between the outer wall of the corrosion-resistant base pipe and the electrothermal coating, varies depending on the thickness of the electrothermal coating. Much smaller than the outer diameter of the corrosion-resistant base pipe The outer wall area of ​​the electrocooling coating is approximately equal to External convection thermal resistance The calculation formula is as follows:

[0067] ,

[0068] in, For the first The external convective heat transfer coefficient of the section was calculated using the Churchill-Bernstein empirical formula.

[0069] Calculate axial coupling thermal resistance That is, the first Section and the Inter-segment heat transfer is hindered. Axial heat between segments is transferred simultaneously through the corrosion-resistant base tube and the electro-cooling coating, which are connected in parallel. When heat is transferred through multiple parallel paths, the calculation method for the total thermal resistance is similar to that for parallel resistance. This is because the reciprocal of thermal resistance is similar to electrical conductance. In the case of parallel connection, the total thermal conductance is the sum of the thermal conductances of each branch. Therefore, the formula for calculating axial coupling thermal resistance is as follows:

[0070] ,

[0071] in, To improve the axial thermal resistance of the corrosion-resistant base tube, The axial thermal resistance of the electrocooled coating is, where , The axial heat-conducting cross-sectional area of ​​the corrosion-resistant base pipe is the effective area for heat transfer between sections through the axial direction of the corrosion-resistant base pipe. Thermal conductivity of corrosion-resistant base pipe material, axial thermal resistance of electro-cooling coating , The axial heat-conducting cross-sectional area of ​​the electrocooling coating is the effective area through which heat is transferred axially between sections. The thermal conductivity of the electrothermal coating material is given. Since all sections have the same structure, the axial coupling thermal resistance between all adjacent sections is equal. And let the axial coupling thermal resistance between adjacent segments be denoted as . ;

[0072] Calculate the radial heat transfer and axial coupled heat transfer of each segment. Radial heat transfer refers to the heat transferred radially outward. Based on the relationship between total radial thermal resistance and temperature difference, the first... The formula for calculating radial heat transfer in a segment is shown below:

[0073] ,

[0074] in, For radial heat transfer, i.e., the heat transfer rate per unit time... The heat released by the segment to the external environment through the radial heat transfer link. For the first Temperature inside the pipe section For the external temperature, if the first pipe section temperature greater than the external temperature , If the temperature inside the tube is positive, meaning it's an exothermic process, and the temperature inside the tube is lower than the external temperature, then... A negative value indicates an endothermic process. Without interfering with the cooling transmission of the refrigerant tube, the greater the transmission distance, the more heat the refrigerant tube releases, and the lower the temperature of the fluid inside the refrigerant tube. This leads to the following: The axial coupling heat transfer between the segment and the adjacent segment is , ,in, For axial heat transfer, i.e. from the first The first segment is passed to the first Axial heat of the segment For axial heat transfer, i.e. from the first The first segment is passed to the first The axial heat flow rate of a section follows Fourier's law, is directly proportional to the temperature difference between adjacent sections and inversely proportional to the axial coupling thermal resistance. The calculation formula is shown below:

[0075] ,

[0076] ,

[0077] Among them, when At that time, there was no adjacent segment on the left. ,when At that time, there was no adjacent segment on the right. , For the first The temperature inside the pipe section For the first The temperature inside the pipe section, if , It is positive, that is, the first Duan Xiangdi Sectional heat transfer, if , It is positive, that is, the first Duan Xiangdi Section heat transfer, therefore the first Total heat transfer of the section The sum of radial heat transfer and axial net heat transfer is calculated using the following formula:

[0078] ,

[0079] in, For the first unit of time The total heat loss outwards is equal to the net heat loss axially. Subtract axial heat transfer That is, the heat lost axially outward from this section. A positive value indicates that there is heat loss in this section, which needs to be compensated by the heat absorption of the electrocooling coating.

[0080] Construct a coupled heat balance equation, calculate the heat absorption in each segment, and according to the law of conservation of energy, the first... The heat absorption provided by the segment electrocooling coating through the Peltier effect The total heat loss in this section must be completely offset. To maintain the temperature inside the tube Stable, for the first In this context, the energy balance relationship is as follows:

[0081] ,

[0082] The left side of the equation represents the heat absorbed by the electrocooling coating. Plus axial heat transfer The right side of the equation represents radial heat transfer. Plus axial heat transfer , sorted out Section heat absorption The calculation formula is as follows:

[0083] ,

[0084] Combining the total heat transfer formula, we can obtain the coupled heat balance equation, which is shown below:

[0085] ,

[0086] The heat absorption of each electrocooling section in the axial and radial directions can be quantified according to the coupled heat balance equation, which is also the cooling loss of each electrocooling section.

[0087] like Figure 3As shown, further, a physical embedding meta-learning graph neural network is used to convert the heat absorption of each electrocooling segment into voltage and dynamically determine the optimal energizing voltage of each electrocooling segment through voltage coupling conduction function, including the following steps:

[0088] Construct a physical embedded cloud learning graph neural network. In the network input layer, the input feature vector consists of the heat absorption and operating parameters of each electrocooling section. Segment input features The definition is as follows:

[0089] ,

[0090] in, For the first The heat absorption of the segment For the first The temperature inside the pipe section For the flow rate within the pipe, External wind speed, The input feature vector, which takes the external temperature as its reference, utilizes the multi-parameter coupling principle of the heat transfer process. By integrating the heat absorption, the temperature inside the pipe, and the internal and external boundary conditions, including the flow rate inside the pipe, the external wind speed, and the external temperature, it ensures that the network can fully capture the dynamic operating status of the refrigeration pipe.

[0091] To avoid physical inconsistencies in purely data-driven models, i.e., predictions that violate energy conservation, a mathematical transformation is used to embed the coupled thermal balance equation into the feature vector, generating physically enhanced features. The expression is as follows:

[0092] ,

[0093] in, For the first The total radial thermal resistance of the segment, This is a normalization coefficient, whose value is based on the magnitude analysis of each parameter in the feature vector. It ensures that the numerical range of the physical constraint term is consistent with other features, thus avoiding gradient imbalance during network training. This is a physical constraint term, derived from the transformation of the heat balance equation. The introduction of this constraint term ensures that the network satisfies the law of conservation of energy during the learning process, making the prediction results physically reasonable.

[0094] Considering the significant axial heat conduction within the electrocooling chamber, and the fact that the heat transfer processes in each section are not independent, a graph neural network is used to model the coupling relationships between the sections. Each electrocooling section is considered as shown in the diagram. For each node, the axial thermal conduction intensity between segments is used as the edge weight between nodes to construct an adjacency matrix. The expression is as follows:

[0095] ,

[0096] in, For the first Sectional electro-cooling section, The reciprocal of the inter-segment axial coupling thermal resistance Characterizing the thermal conductivity between adjacent segments, the lower the thermal resistance, the stronger the conductivity, and the larger the edge weight. Graph neural networks update node features through a three-layer propagation mechanism, realizing the interaction of coupled information between segments. The feature update formula for the layer is shown below:

[0097] ,

[0098] in, , For the first The section in Feature vectors of the layer, initial layer For physical enhancement features, For the first The weight matrix of the layer, For the first The bias vector, weight matrix, and bias vector of a layer are learnable parameters of the network, which are optimized through training to capture the non-linear relationships between features. The ReLU activation function is used to introduce nonlinear transformation capabilities, enabling the network to fit complex heat transfer-voltage mapping relationships. The summation term... To perform weighted aggregation of the features of adjacent nodes, the weight is the inter-segment heat conduction capability, thereby realizing the transmission of inter-segment coupling information;

[0099] To address the adaptability of refrigerant pipes under various operating conditions, the MAML meta-learning framework is introduced. This allows the network to quickly adjust parameters using a small amount of new data to adapt to changing operating conditions. Initial parameters shared across all operating conditions were obtained through pre-training on a large amount of historical operating data, including operational data for different pipe flow rates, pipe temperatures, external wind speeds, and external temperatures, characterizing the common patterns of each operating condition and the task parameters. For the first The adaptation parameters for this new operating condition are obtained by rapidly fine-tuning the original parameters; task parameters. and loss function The calculation formula is as follows:

[0100] ,

[0101] ,

[0102] in, For learning rate, For the first The inner loop loss function for this operating condition. Predict voltage for the network. This is the historical measured voltage. For the inner loop loss function with respect to the meta-parameters The gradient;

[0103] The output layer of a physics-embedded meta-learning graph neural network is used to predict the target voltage that meets the heat absorption requirements of this segment. The feature vector of the last layer of the graph neural network is then used to predict the target voltage. The target voltage of each electrocooling stage is obtained through linear mapping. The expression is as follows:

[0104] ,

[0105] in, The output layer parameters are obtained through meta-learning optimization. To quantify the influence of voltage from other segments on the voltage division of this segment during actual operation, a voltage coupling transfer function is constructed, and the external coupling voltage division is calculated. The calculation formula is as follows:

[0106] ,

[0107] in, For other segments of voltage to the first The coupling voltage division generated by the segment For the first The initial voltage applied to the segment, For the first segment voltage for the first The voltage coupling coefficient of the segment is derived based on Fourier's law of heat conduction and Ohm's law of electricity, and the calculation formula is shown below:

[0108] ,

[0109] in, The axial equivalent thermal conductivity of the electrocooling section is calculated from the thermal conductivity of the corrosion-resistant base tube and the electrocooling coating. The calculation formula is as follows: , For the thermal conductivity of the corrosion-resistant base pipe material, For the axial heat conduction cross-sectional area of ​​the corrosion-resistant base pipe, The radial thermal conductivity of the electrocooling coating is... The axial thermal conductivity cross-sectional area of ​​the electrocooling coating is... This refers to the axial length of a single electrocooling section. For electrocooled coating resistors, This is the inter-segment coupling attenuation coefficient, whose value decreases as the inter-segment distance increases. The attenuation factor was determined through fitting experimental data. The axial thermal contact area is denoted as , where The outer diameter of the corrosion-resistant base pipe;

[0110] The core objective is to approximate the target voltage output by the graph neural network with the superposition of the actual voltage and the externally coupled voltage divider, while also constraining energy consumption. The objective function is... The definition is as follows:

[0111] ,

[0112] Among them, the first item The second term characterizes the deviation between the actual voltage plus the externally coupled voltage divider and the target voltage output by the physical constraint learning graph neural network. This is an energy consumption term based on Joule's law, used to constrain excessive voltage increases. The weighting coefficients, determined experimentally, prioritize minimizing voltage deviation. To clarify the inter-segment coupling relationship, substituting the external coupling voltage divider formula into the objective function yields:

[0113] ,

[0114] To ensure the safe operation of the electrothermal coating, the voltage must meet the upper and lower limits, as shown below:

[0115] ,

[0116] in, This is the minimum voltage, corresponding to the non-operating state. The maximum voltage is determined based on the voltage withstand characteristics of the coating material to avoid damage to the coating due to overvoltage.

[0117] The Lagrange multiplier method is employed, incorporating constraints into the objective function through Lagrange multipliers to construct a Lagrange augmented function. A search space is then built using the energizing voltage of each electrocooling segment. With the objective of minimizing the Lagrange augmented function, a swarm intelligence algorithm is used to iteratively search within the search space to determine the optimal energizing voltage for each electrocooling segment. This swarm intelligence algorithm includes ant colony optimization, particle swarm optimization, bat optimization, wolf pack optimization, and fruit fly optimization. Since swarm intelligence algorithms are existing technologies and not the focus of this application, they will not be elaborated upon further here.

[0118] like Figure 4 As shown, further, comparing the absolute temperature difference and temperature difference threshold between the temperature of each electrocooling section and the preset target temperature to decide whether to simulate and optimize to obtain the optimal parameters of the coupled thermal balance equation and the meta-learning graph neural network includes the following steps:

[0119] The absolute temperature difference between the internal temperature of each electrocooling section and the preset target temperature is calculated using the following formula:

[0120] ,

[0121] in, For the first The absolute temperature difference of the section reflects the degree of deviation between the temperature inside the pipe and the preset target temperature. For the first time collected Temperature inside the pipe section Preset target temperature;

[0122] By comparing the absolute temperature difference of each segment with the temperature difference threshold, a three-level decision logic is formed. If the absolute temperature difference of all segments is less than or equal to the temperature difference threshold, it is determined that the current parameters are suitable for the working condition and the operation is maintained without optimization. If the absolute temperature difference of less than 1 / 3 of the total number of segments is greater than the temperature difference threshold, it is determined that there is a local parameter mismatch. Only the task parameters of the meta-learning graph neural network are fine-tuned. That is, the operating data based on the current working condition data, including the flow rate in the pipe, the temperature in the pipe, the external wind speed, and the external temperature, are updated quickly without adjusting the coupled heat balance equation. If the absolute temperature difference of more than or equal to 1 / 3 of the segments is greater than the temperature difference threshold, it is determined that there is an overall parameter mismatch, and the global optimization of the coupled heat balance equation and the meta-learning graph neural network is triggered simultaneously.

[0123] For the coupled thermal balance equation, the least squares algorithm is used to optimize the radial conduction thermal resistance and convective heat transfer coefficient. For the meta-learning graph neural network, the Adam optimizer is used to update the network weights, biases and meta-learning rate through the loss function of voltage prediction error. After optimization, the generalization ability needs to be verified through the current operating condition data. After the optimized parameters are deployed, if the temperature control requirements are still not met after three consecutive sampling periods, the system will automatically backtrack to the parameters before optimization and record the current operating condition data for subsequent offline training to ensure the system stability and the effectiveness of parameter optimization.

[0124] Example 2

[0125] This invention discloses a temperature control system for a corrosion-resistant refrigeration pipe with an intelligent temperature control coating, comprising a data acquisition module, a control module, and a feedback module;

[0126] The data acquisition module collects the tube temperature and flow rate of each electrocooling section in each cycle, and simultaneously measures the external wind speed and external temperature.

[0127] In each cycle, the control module constructs a coupled heat balance equation based on Fourier's law and the convective heat transfer formula, calculates the heat absorption of each electrocooling segment, and converts the heat absorption of each electrocooling segment into a target voltage through a physical embedded element learning graph neural network. With the aim of satisfying the target voltage of each electrocooling segment, the module collaboratively considers the influence of the voltage of the electrocooling segment on the other electrocooling segments to design a voltage coupling transfer function. The optimal voltage is determined iteratively through a swarm intelligence algorithm and corresponding control is applied.

[0128] The feedback module compares the absolute temperature difference between the temperature of each electrocooling section and the preset target temperature with the temperature difference threshold in each cycle to decide whether to simulate and optimize the optimal parameters of the coupled heat balance equation and the meta-learning graph neural network in this cycle.

[0129] This invention discloses a temperature control method and system for corrosion-resistant refrigeration pipes with intelligent temperature control coatings. It aims to solve the problems of large calculation deviations in heat loss calculations and insufficient temperature control accuracy caused by inter-segment voltage coupling interference in existing refrigeration pipes. By collecting the temperature inside the pipe, the flow rate inside the pipe, and the external wind speed and temperature in each electro-cooled section, and based on Fourier's law and convective heat transfer formulas, the radial total thermal resistance and axial coupling thermal resistance are calculated. Combined with the thermal resistance and temperature difference, the radial and axial net heat transfer is calculated. Based on energy conservation, a coupled heat balance equation is constructed to obtain the heat absorption that needs to be compensated for in each segment. Subsequently, a physical embedded element learning graph neural network is used, combined with the voltage coupling conduction function... The system focuses on approximating the target voltage by superimposing the actual voltage and coupled voltage divider. Energy consumption constraints and upper and lower limits for coating safety voltage are added. A swarm intelligence algorithm iteratively updates the voltage until the target function is stable and the deviation meets the standard, obtaining and applying the optimal energizing voltage. The Peltier effect is used to compensate for cooling losses. Finally, the absolute temperature difference and temperature difference threshold between each section of the tube and the target temperature are compared. A three-level logical decision-making process optimizes the coupled thermal balance equation and the parameters of the physical embedded element learning graph neural network. The system includes acquisition, control, and feedback modules, which respectively perform data acquisition, voltage control, and parameter decision-making, achieving precise temperature control of the refrigeration tube.

[0130] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A temperature control method for a corrosion-resistant refrigeration pipe with an intelligent temperature control coating, characterized in that, Includes the following steps: During the cooling process, the tube temperature and flow rate of each electrocooling section are collected, and the external wind speed and external temperature are measured simultaneously. Based on Fourier's law and the convective heat transfer formula, a coupled heat balance equation is constructed, and the heat absorption of each electrocooling segment is calculated. The heat absorption of each electrocooling segment is converted and mapped to the target voltage through a physical embedded element learning graph neural network. With the goal of satisfying the target voltage of each electrocooling segment, the voltage coupling transfer function is designed by considering the influence of the voltage of the electrocooling segment on the other electrocooling segments. The optimal voltage is determined iteratively through a swarm intelligence algorithm and corresponding control is applied. Calculate the absolute temperature difference between the tube temperature of each electrocooling section and the preset target temperature, and compare the absolute temperature difference with the temperature difference threshold to decide whether to simulate and optimize to obtain the optimal parameters of the coupled thermal balance equation and the physical embedded element learning graph neural network.

2. The temperature control method for the corrosion-resistant refrigeration pipe with intelligent temperature control coating as described in claim 1, characterized in that, By combining physical embedding meta-learning graph neural networks with the collaborative calculation of voltage-coupled conduction functions, the heat absorption is converted into the optimal energizing voltage, including the following steps: The heat absorption, tube temperature, tube flow rate, external wind speed, and external temperature of each electrocooling section are collected and input into the trained physical embedding meta-learning graph neural network to obtain the target voltage. Based on the thermal conductivity of the corrosion-resistant base tube and the electro-cooling coating, the axial length of the electro-cooling section, the coating resistance, and the axial heat conduction contact area, combined with Fourier's law of heat conduction and Ohm's law of electricity, the voltage coupling coefficient of a certain voltage segment to the temperature of other segments is calculated, and the voltage coupling conduction function is determined. Considering the mutual interference of voltages between different electrocooling sections, the external coupling voltage division caused by the energizing voltage of other electrocooling sections on the current electrocooling section is calculated by the voltage coupling conduction function. The core objective is to approximate the target voltage output by the physical embedded element learning graph neural network with the superposition value of the actual energized voltage and the externally coupled voltage divider. At the same time, an energy consumption constraint based on Joule's law is introduced to construct the objective function. Combined with the upper and lower limits of the safe operating voltage of the electrocooling coating, a swarm intelligence algorithm is used to guide the voltage iterative update until the objective function tends to stabilize and the voltage deviation meets the preset requirements, thus obtaining the optimal energized voltage of each electrocooling segment.

3. The temperature control method for the corrosion-resistant refrigeration pipe with intelligent temperature control coating as described in claim 1, characterized in that, The coupled heat balance equations are constructed and the heat absorption of each electrocooling section is calculated, including the following steps: Based on the total length of the refrigeration tube and the axial length of a single electrocooling section, the refrigeration tube is divided into several independent electrocooling sections, and the heat of each electrocooling section is transferred radially and axially. Calculate the total radial thermal resistance and the axial coupling thermal resistance, i.e. the obstacles to heat transfer in the radial and axial directions, respectively; and calculate the radial heat transfer and the axial net heat transfer based on the relationship between thermal resistance and temperature difference. According to the principle of energy conservation, the heat absorption required by the electrocooling coating must completely offset the total heat loss of the section. Based on this, a coupled heat balance equation is constructed and the heat absorption that each electrocooling section needs to compensate is calculated.

4. The temperature control method for a corrosion-resistant refrigeration pipe with an intelligent temperature control coating as described in claim 2, characterized in that, Constructing a physically embedded meta-learning graph neural network to generate the target voltage includes the following steps: The heat absorption, tube temperature, tube flow rate, external wind speed, and external temperature of each electrocooling section are integrated into an input feature vector. The coupled thermal equilibrium equation is transformed and used as a physical constraint term, which is then added to the input feature vector to generate physically enhanced features. Each electrocooling segment is used as a node in the graph neural network, and the axial thermal conductivity between segments is used as the edge weight between nodes. Based on this, an adjacency matrix is ​​constructed to complete the topology construction of the graph neural network. For the established graph neural network, historical multi-condition data covering different pipe flow, pipe temperature, external wind speed, and external temperature scenarios are used for pre-training to fit the common patterns under each condition and generate the pre-training meta-parameters of the network. The physical enhancement features are input into a graph neural network with pre-trained meta-parameters, and the node features are updated and iteratively trained through a three-layer feature propagation mechanism. The trained and fitted graph neural network, through the output layer of the graph neural network, linearly maps the feature vector of the last layer to the target voltage without considering inter-segment coupling.

5. The temperature control method for a corrosion-resistant refrigeration pipe with an intelligent temperature control coating as described in claim 3, characterized in that, Heat is transferred in both the radial and axial directions in each electrocooling section. Radial transfer is a unidirectional steady-state transfer along the fluid inside the pipe, the corrosion-resistant base pipe, the electrocooling coating, and the external air, with heat diffusing from the inside of the pipe to the outside or penetrating from the outside into the pipe. Axial transfer is carried out through the corrosion-resistant base pipe and the electrocooling coating between adjacent electrocooling sections, with the transfer direction from the section with higher temperature to the section with lower temperature.

6. The temperature control method for the corrosion-resistant refrigeration pipe with intelligent temperature control coating as described in claim 3, characterized in that, The total radial thermal resistance is formed by four parts connected in series: the internal convection thermal resistance, the radial conduction thermal resistance of the corrosion-resistant base tube, the radial conduction thermal resistance of the electro-cooling coating, and the external convection thermal resistance. The total resistance of the series thermal resistance is the sum of the individual resistance values. The internal convection thermal resistance is calculated using the Dittus-Boelter empirical formula, the external convection thermal resistance is calculated using the Churchill-Bernstein empirical formula, the radial conduction thermal resistance of the corrosion-resistant base tube is calculated based on the thermal conductivity characteristics of the cylindrical wall, and the radial conduction thermal resistance of the electro-cooling coating is simplified by calculating the thermal conductivity characteristics of a flat wall because the coating thickness is much smaller than the outer diameter of the base tube.

7. The temperature control method for a corrosion-resistant refrigeration pipe with an intelligent temperature control coating as described in claim 3, characterized in that, The axial coupling thermal resistance is formed by the parallel connection of the axial thermal resistance of the corrosion-resistant base tube and the axial thermal resistance of the electro-cooling coating. The reciprocal of the total resistance of the parallel thermal resistance is equal to the sum of the reciprocals of the individual resistances, and the axial coupling thermal resistance of all adjacent electro-cooling sections is equal due to their identical structure.

8. The temperature control method for the corrosion-resistant refrigeration pipe with intelligent temperature control coating as described in claim 3, characterized in that, Radial heat transfer is obtained by dividing the temperature difference between the inside and outside of the pipe by the total radial thermal resistance. If the temperature inside the pipe is higher than the outside temperature, it is exothermic, and vice versa. The axial net heat transfer is the axial heat transfer of a certain electrocooling section, which is equal to the axial heat transfer out of the adjacent section on the right minus the axial heat transfer into the adjacent section on the left. The first section has no adjacent section on the left, and the last section has no adjacent section on the right, so there is no heat transfer in the corresponding direction.

9. The temperature control method for a corrosion-resistant refrigeration pipe with an intelligent temperature control coating as described in claim 1, characterized in that, The corrosion-resistant refrigeration tube with intelligent temperature control coating includes, from the inside out, a corrosion-resistant base tube, an electro-cooling coating, and an insulating protective layer. The electro-cooling coating is uniformly attached to the outer wall of the corrosion-resistant base tube. Electrode groups are arranged at preset intervals along the axial direction of the refrigeration tube, and the electro-cooling section is formed between adjacent electrode groups.

10. A temperature control system for corrosion-resistant refrigeration pipes with intelligent temperature control coating, characterized in that, It includes a data acquisition module, a control module, and a feedback module; The data acquisition module collects the tube temperature and flow rate of each electrocooling section in each cycle, and simultaneously measures the external wind speed and external temperature. In each cycle, the control module constructs a coupled heat balance equation based on Fourier's law and the convective heat transfer formula, calculates the heat absorption of each electrocooling segment, and converts the heat absorption of each electrocooling segment into a target voltage through a physical embedded element learning graph neural network. With the aim of satisfying the target voltage of each electrocooling segment, the module collaboratively considers the influence of the voltage of the electrocooling segment on the other electrocooling segments to design a voltage coupling transfer function. The optimal voltage is determined iteratively through a swarm intelligence algorithm and corresponding control is applied. The feedback module compares the absolute temperature difference between the temperature of each electrocooling section and the preset target temperature with the temperature difference threshold in each cycle to decide whether to simulate and optimize the optimal parameters of the coupled heat balance equation and the meta-learning graph neural network in this cycle.

Images