Phase change efficiency evaluation method, device, equipment and medium for microchannel heat exchange module

Abstract

The present invention relates to the technical field of heat exchange modules, and discloses a phase change efficiency evaluation method, device, equipment and medium for a microchannel heat exchange module. The method comprises: collecting data from a microchannel heat exchange module to obtain a multidimensional sensor data set, and performing nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation on the multidimensional sensor data set to obtain a nonlinear multi-order coupled phase change heat transfer model; performing empirical mode decomposition and feature extraction to obtain target phase change feature data; establishing an optimization objective function including phase change efficiency calculation items and heat exchange performance calculation items, and performing particle swarm optimization calculation based on the optimization objective function to obtain coupling parameter optimization data; inputting the target phase change feature data and the coupling parameter optimization data into a phase change efficiency prediction model for processing to obtain phase change efficiency prediction data. The present invention realizes the optimal configuration of the phase change efficiency calculation items and the heat exchange performance calculation items, and realizes all-round monitoring of the performance of the heat exchange module.

Description

Technical Field

[0001] The present invention relates to the technical field of heat exchange modules, and in particular to a method, device, equipment and medium for evaluating the phase change efficiency of a microchannel heat exchange module. Background Art

[0002] Since the phase change process in the microchannel has strong nonlinear and multi-scale characteristics, traditional phase change efficiency evaluation methods are difficult to accurately capture complex thermodynamic behaviors. Current phase change efficiency evaluation methods mainly rely on empirical models and simplified assumptions, and cannot effectively deal with local fluctuations and global evolution characteristics in the microchannel. Especially under high heat flux density conditions, traditional evaluation methods have problems such as insufficient accuracy and narrow scope of application, making it difficult to comprehensively and accurately evaluate the performance of microchannel heat exchange modules. In addition, the existing evaluation methods lack in-depth analysis of the coupling effects of multiple physical quantities in the phase change process, resulting in a large deviation between the evaluation results and the actual operating status. Summary of the invention

[0003] The present invention provides a phase change efficiency evaluation method, device, equipment and medium for a microchannel heat exchange module. The present invention realizes the optimal configuration of phase change efficiency calculation items and heat exchange performance calculation items, and realizes all-round monitoring of the performance of the heat exchange module.

[0004] In a first aspect, the present invention provides a method for evaluating the phase change efficiency of a microchannel heat exchange module, the method comprising:

[0005] Data is collected on the microchannel heat exchange module to obtain a multi-dimensional sensing data set, and nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation are performed on the multi-dimensional sensing data set to obtain a nonlinear multi-order coupling phase change heat transfer model;

[0006] Performing empirical mode decomposition and feature extraction on the multi-dimensional sensor data set to obtain target phase change feature data;

[0007] According to the nonlinear multi-order coupled phase change heat transfer model and the target phase change characteristic data, an optimization objective function including a phase change efficiency calculation item and a heat exchange performance calculation item is established, and a particle swarm optimization calculation is performed based on the optimization objective function to obtain coupling parameter optimization data;

[0008] The target phase change characteristic data and the coupling parameter optimization data are input into a phase change efficiency prediction model for processing to obtain phase change efficiency prediction data.

[0009] In a second aspect, the present invention provides a phase change efficiency evaluation device for a microchannel heat exchange module, the phase change efficiency evaluation device for a microchannel heat exchange module comprising:

[0010] An acquisition module is used to collect data from the microchannel heat exchange module to obtain a multi-dimensional sensing data set, and perform nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation on the multi-dimensional sensing data set to obtain a nonlinear multi-order coupling phase change heat transfer model;

[0011] A decomposition module, used for performing empirical mode decomposition and feature extraction on the multi-dimensional sensor data set to obtain target phase change feature data;

[0012] A calculation module, used to establish an optimization objective function including a phase change efficiency calculation item and a heat exchange performance calculation item according to the nonlinear multi-order coupled phase change heat transfer model and the target phase change characteristic data, and perform particle swarm optimization calculation based on the optimization objective function to obtain coupling parameter optimization data;

[0013] The processing module is used to input the target phase change characteristic data and the coupling parameter optimization data into the phase change efficiency prediction model for processing to obtain phase change efficiency prediction data.

[0014] A third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned phase change efficiency evaluation method of the microchannel heat exchange module.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for evaluating the phase change efficiency of a microchannel heat exchange module.

[0016] In the technical solution provided by the present invention, by establishing a nonlinear multi-order coupled phase change heat transfer model, combined with empirical mode decomposition and deep learning methods, the phase change heat transfer characteristics in the microchannel are accurately described, overcoming the accuracy loss problem caused by simplified assumptions in traditional evaluation methods. Adaptive median filtering and modal reconstruction technology are used to effectively eliminate thermal noise interference in the phase change process, extract phase change characteristics with physical significance, and ensure the reliability of the evaluation results. The particle swarm optimization algorithm is introduced to optimize the coupling parameters, achieve the optimal configuration of the phase change efficiency calculation items and the heat exchange performance calculation items, and enhance the applicability of the evaluation model. The time convolution network and time series converter are integrated to construct a multi-scale feature extraction framework, which solves the problem of collaborative analysis of local features and global features in the phase change process. Through digital twin technology, a two-layer evaluation system of microscopic phase change analysis and macroscopic heat exchange analysis is established to achieve all-round monitoring of the performance of the heat exchange module. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 A schematic diagram of the steps of a method for evaluating the phase change efficiency of a microchannel heat exchange module according to an embodiment of the present invention;

[0019] Figure 2 It is a structural schematic diagram of a phase change efficiency evaluation device of a microchannel heat exchange module in an embodiment of the present invention;

[0020] Figure 3 It is a schematic block diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Embodiments of the present invention provide a method, device, equipment and medium for evaluating the phase change efficiency of a microchannel heat exchange module. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the phase change efficiency evaluation method of the microchannel heat exchange module in the embodiment of the present invention includes:

[0023] Step S1, collecting data from the microchannel heat exchange module to obtain a multi-dimensional sensing data set, and performing nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation on the multi-dimensional sensing data set to obtain a nonlinear multi-order coupling phase change heat transfer model;

[0024] It is understandable that the execution subject of the present invention may be a phase change efficiency evaluation device for a microchannel heat exchange module, or a terminal or a server, which is not specifically limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0025] Specifically, a high-precision sensor network is arranged at key positions of the microchannel heat exchange module, including installing a temperature sensor on the heat exchange wall to collect real-time temperature data, and installing a pressure sensor and a flow sensor at the fluid inlet and fluid outlet to collect real-time pressure data and real-time flow data, and a multi-dimensional sensing data set is formed through the various physical quantities collected by the sensors. At the same time, the surface contour of the inner wall of the microchannel is scanned by a three-dimensional topography scanning device to obtain the three-dimensional scanning data of the microchannel, and the scanning data is processed using adaptive mesh generation technology to construct the geometric structure data of the microchannel. In the mesh generation process, the adaptive technology can dynamically adjust the grid density according to the complexity of the microchannel geometry, thereby optimizing the calculation efficiency while ensuring the calculation accuracy. According to the obtained microchannel geometry data, a multi-physics field coupled heat transfer equation group including convective heat transfer, conductive heat transfer and phase change heat transfer is established. The construction of the heat transfer equation group is based on the basic principle of energy conservation, combined with fluid mechanics, heat conduction theory and phase change dynamics, to comprehensively describe the complex heat transfer process in the microchannel. In this process, the key physical quantities in the equations are quantified by matrix representation, where the thermal conductivity coefficient matrix characterizing the thermal conductivity is used to describe the heat conduction process in the solid wall, the phase change latent heat matrix characterizing the phase change characteristics is used to quantify the release and absorption effects of latent heat in the phase change process, and the boundary condition matrix characterizing the boundary conditions is used to describe the heat exchange behavior between the microchannel wall and the fluid. These matrices together constitute the initial heat transfer model, which is used to characterize the basic thermal physical behavior of the microchannel heat exchange module. In order to improve the accuracy of the model, the nonlinear term coefficients and parameter fitting of the initial heat transfer model are carried out in combination with the real-time temperature data in the multi-dimensional sensor data set. The nonlinear model parameters are corrected by optimization algorithms such as the least squares method or genetic algorithm to obtain the nonlinear coupling coefficient that characterizes the complex thermal physical relationship, reflecting the real heat transfer and phase change characteristics in the microchannel. Based on the nonlinear coupling coefficient, the heat transfer process between the solid wall and the fluid is quantitatively analyzed by establishing the heat conduction relationship equation between the microchannel wall and the fluid. This process involves the dynamic coupling of the wall thermal conductivity characteristics and the fluid heat transfer characteristics to obtain the thermal conduction coupling data. Based on the same nonlinear coupling coefficient, by establishing the phase change process equation of the fluid in the microchannel, the release and absorption of the phase change latent heat are calculated and analyzed in detail. This equation needs to combine the phase change temperature range, phase change heat flux density and fluid dynamics behavior to accurately describe the spatiotemporal distribution of heat during the phase change process and obtain the phase change coupling data. The heat conduction coupling data is integrated with the phase change coupling data to complete the coupling calculation of the heat conduction process and the phase change process. The coupling calculation uses nonlinear iterative solution methods, such as the finite element method or the finite difference method, to comprehensively simulate the heat transfer and flow coupling effects in the microchannel, and finally constructs a nonlinear multi-order coupled phase change heat transfer model. This model can accurately predict the heat transfer law in the microchannel and can quantitatively evaluate the phase change efficiency.

[0026] Step S2, performing empirical mode decomposition and feature extraction on the multi-dimensional sensor data set to obtain target phase change feature data;

[0027] Specifically, the multi-dimensional sensor data set is input into the empirical mode decomposition algorithm, and multiple intrinsic mode functions are generated by decomposition. These functions correspond to signal components in different frequency ranges, thereby revealing the multi-scale characteristics of heat transfer and flow during the phase change process. Since thermal noise and redundant information are inevitably present in the original sensor data, the adaptive window division technology is introduced into the decomposed IMF, and the division scale is dynamically adjusted according to the local fluctuation characteristics of the signal, and the dynamic adjustment method of the filter window is combined to optimize the window parameters. The optimal window parameters are used to guide subsequent data processing operations so that the processing results can more accurately reflect the key physical characteristics of the phase change process. According to the optimal window parameters, the adaptive median filtering technology is applied to each intrinsic mode function to remove high-frequency thermal noise and abnormal spike signals, thereby retaining the main characteristics and trend information of the signal. The median filtering operation can effectively reduce the random fluctuations caused by sensor errors or environmental interference, while maintaining the integrity and accuracy of the signal edge. The modal function after eliminating thermal noise is modally reconstructed. The reconstructed signal data is generated by recombining the filtered IMFs according to the frequency domain characteristics. The reconstructed signal data is input into the temporal convolutional network, and its deep convolutional structure is used to extract local features of the signal. During the convolution operation, the short-term pattern changes of the signal are captured by sliding the convolution kernel, revealing the local phase change characteristics, including microscopic heat flow fluctuations, local temperature anomalies, etc. The pooling operation further compresses the signal dimension and extracts local features that are representative of the phase change behavior, thereby reducing the computational complexity and improving the generalization ability of the model. The obtained local phase change feature data reflects the short-term characteristic changes of the phase change process. In order to capture the timing-dependent features in the phase change process, the local phase change feature data is input into the time series converter, and the global features of the signal are extracted using its built-in attention mechanism. The time series converter identifies the dynamic correlation between different moments in the phase change process by focusing on the feature weight distribution of different time steps, and constructs the global phase change feature data. The global feature data can reveal the overall change trend of the phase change state and reflect the interaction and causal relationship between local features. The attention mechanism can dynamically adjust the feature weights at different time points so that the extracted global features are more in line with the actual physical laws of the phase change process. The local phase change feature data and the global phase change feature data are complementary in physical meaning and time scale. The two are combined through feature weighted fusion operation to generate phase change feature fusion data containing multiple layers of information. The feature weighted fusion operation uses weighted average or other statistical methods to combine the contribution of different features to uniformly process the data to ensure that the fused data contains both local detail information and reflects the global trend. Based on the phase change feature fusion data, the temperature field distribution, fluid flow parameters and phase change state during the phase change process are analyzed, and the corresponding feature vectors are constructed. The feature vector construction process combines the spatial distribution characteristics and the time evolution characteristics to form highly generalized target phase change feature data.

[0028] The number of windows for multiple intrinsic mode functions is determined and calculated according to the optimal window parameters to form a sliding window sequence. The generation of the sliding window sequence needs to comprehensively consider the frequency characteristics and local fluctuation characteristics of each intrinsic mode function. The number of windows is dynamically adjusted through the optimal window parameters to ensure that the sliding window can effectively capture the changing trend of the signal, while minimizing the noise residue or signal distortion caused by improper window size selection. The median is calculated for the data in each window in the sliding window sequence, and the thermal noise in multiple intrinsic mode functions is filtered out as the core operation to obtain the first-order denoised modal data. The abnormal data points in the window are smoothed by median calculation, which significantly suppresses the influence of high-frequency noise and spikes caused by sensor errors or environmental interference. The thermal noise variance matrix is ​​constructed based on the first-order denoised modal data. By analyzing the statistical characteristics of the thermal noise distribution in different modal functions, the filter parameters are adaptively adjusted to generate adaptive filtering parameters. These parameters are dynamically optimized according to the local variance and frequency distribution of the signal, so that the filter can flexibly adapt to the characteristics of different modal functions, thereby improving the noise reduction effect while maintaining signal integrity. The adaptive filtering parameters are input into the second-order median filter, and the first-order denoised modal data is subjected to secondary filtering to obtain the modal function after the thermal noise is eliminated. The modal function after the thermal noise is eliminated is subjected to phase modulation analysis based on Hilbert transform to extract phase characteristic data. Hilbert transform is a signal processing tool that can convert real signals into complex signals to calculate important information such as instantaneous phase and instantaneous amplitude. Through phase modulation analysis, the timing characteristics and dynamic change laws related to the phase change process in the modal function are revealed. Based on the phase characteristic data, the modal function after the thermal noise is eliminated is subjected to phase reconstruction operation to generate reconstructed signal data. Phase reconstruction combines the instantaneous phase and amplitude information of the modal function to more realistically reflect the heat transfer and phase change characteristics in the microchannel.

[0029] Step S3, establishing an optimization objective function including phase change efficiency calculation items and heat transfer performance calculation items according to the nonlinear multi-order coupled phase change heat transfer model and the target phase change characteristic data, and performing particle swarm optimization calculation based on the optimization objective function to obtain coupling parameter optimization data;

[0030] Specifically, a systematic phase change efficiency calculation is performed based on the heat conduction coefficient matrix and phase change coefficient matrix in the nonlinear multi-order coupled phase change heat transfer model. The heat conduction coefficient matrix describes the heat transfer characteristics between the solid wall and the fluid in the microchannel heat exchange module, while the phase change coefficient matrix characterizes the absorption and release behavior of latent heat of the fluid during the phase change process. By solving these matrices, the initial value of the phase change efficiency of the microchannel under the initial conditions is calculated, providing the model with a preliminary quantification of the phase change performance. The temperature field distribution data, fluid flow parameter data and phase change state data in the target phase change characteristic data are weighted and calculated to generate characteristic weight coefficients. According to the degree of influence of each type of data on the phase change efficiency, by analyzing its sensitivity and contribution in the model, the weight value is dynamically adjusted, so that the subsequent phase change efficiency calculation is more in line with the actual physical process. The initial value of the phase change efficiency is combined with the characteristic weight coefficient to construct the phase change efficiency calculation item, and the calculation item is numerically normalized to obtain the normalized phase change efficiency data. At the same time, based on the heat transfer equations in the nonlinear multi-order coupled phase change heat transfer model, the heat transfer performance calculation is performed to obtain the heat transfer coefficient data. The heat transfer equations describe the thermodynamic performance of the heat exchange module by coupling convection, heat conduction and phase change heat transfer processes. After the heat transfer coefficient data is numerically normalized, the normalized heat transfer performance data is generated so that it can be calculated on the same scale as the normalized phase change efficiency data. The normalized phase change efficiency data is linearly combined with the normalized heat transfer performance data to construct the final optimization objective function, which contains the phase change efficiency calculation term and the heat transfer performance calculation term. The objective function is optimized to reflect the comprehensive performance of the microchannel heat exchange module under specific working conditions. Its design can balance the relationship between phase change efficiency and heat transfer performance, thereby achieving global optimization of performance. After constructing the objective function, the initialization parameters of the particle swarm optimization algorithm are set, including the initial position, velocity, acceleration constant and inertia weight of the particle, to generate the particle swarm initialization data. Each particle represents a possible parameter combination, and the initialization of its position and velocity gives the particle swarm the ability to explore the search space. During the optimization process, the algorithm performs iterative optimization calculations on the particle swarm initialization data. In each iteration, the particle position is adjusted according to the particle's current velocity and position update rules, and the fitness value of the current solution is calculated. In each round of iteration, the historical optimal position and global optimal position of each particle are recorded, and the acceleration constant and inertia weight are dynamically adjusted to balance the ability of global search and local search, so as to gradually converge to the global optimal solution. The particle swarm optimization trajectory data records the movement process of all particles in the search space. These trajectories reflect the algorithm's search path and global convergence trend. By analyzing the particle optimization trajectory data, the global optimal solution is extracted, and the coupling parameters are mapped based on the optimal solution to obtain the coupling parameter optimization data.

[0031] Step S4: input the target phase change characteristic data and the coupling parameter optimization data into the phase change efficiency prediction model for processing to obtain phase change efficiency prediction data.

[0032] Specifically, the target phase change feature data is input into the empirical mode decomposition module of the phase change efficiency prediction model. The module consists of three layers of screening networks. Each layer of the screening network uses a Butterworth filter to separate the signal and gradually extract the modal components in different frequency ranges. The design of the Butterworth filter ensures that the separation process has good signal fidelity and stability, and can effectively suppress high-frequency noise and low-frequency interference to generate multi-scale modal component data. The multi-scale modal component data is input into the temporal convolutional network module of the phase change efficiency prediction model. The module contains four dilated convolutional layers, and the dilation rate of each convolutional layer is set to 1, 2, 4 and 8 respectively to capture the features on different time scales. The temporal convolutional network uses a small convolution kernel with a convolution kernel size of 3 for calculation, and applies the ReLU activation function and residual connection mechanism in each convolution layer to enhance the expression ability of the model and the efficiency of gradient propagation. By dilating the convolution structure, the network can expand the receptive field, thereby capturing the long-distance time-dependent features without significantly increasing the computational complexity. The output local temporal feature data describes in detail the change law of the target data on a short time scale. The local time series feature data is input into the time series converter module, which contains 6 transformer encoding layers. Each encoding layer integrates an 8-head attention mechanism and a feedforward neural network to efficiently model complex time series dependencies. The attention mechanism dynamically adjusts the importance of each time step feature by calculating the correlation of features at different time steps, thereby improving the model's sensitivity to global dependency information. At the same time, the hidden layer dimension of the feedforward neural network is set to 256 to ensure that the model can handle large-scale input data and extract deep features. After being processed by the time series converter module, the global time series feature data is obtained, which reflects the dynamic change trend of the target data in the entire time domain. In order to enhance the feature expression ability, the local time series feature data and the global time series feature data are feature fused to generate multi-scale fused feature data. Feature fusion integrates feature information at different time scales by weighting or splicing to form a more comprehensive and accurate feature representation. The multi-scale fused feature data and the coupled parameter optimization data are feature spliced ​​to generate a combined feature vector. The combined feature vector is subjected to time series dependency analysis to generate time series dependency feature data by capturing the evolution pattern of features over time. The time-dependent feature data is input into the decoder network of the phase change efficiency prediction model, which consists of three transposed convolutional layers. Each transposed convolutional layer uses the LeakyReLU activation function and batch normalization operation to ensure the stability and nonlinear expression of the signal during the decoding process. Upsampling is achieved through transposed convolution, and the time series dimension is gradually restored, so that the decoded output data is consistent with the input time series, and the phase change efficiency mapping data is generated.The phase change efficiency mapping data is input into the output layer of the phase change efficiency prediction model. The output layer performs numerical normalization on the prediction results through the sigmoid function to compress the output value to the range between 0 and 1, and generates phase change efficiency prediction data to reflect the phase change efficiency of the microchannel heat exchange module.

[0033] The fluctuation characteristics of the phase change efficiency prediction data are analyzed, and the key fluctuation characteristic values ​​are extracted by capturing the dynamic change characteristics of the efficiency over time during the phase change process. These fluctuation characteristic values ​​reflect the stability and sensitivity during the phase change process. The fluctuation characteristic values ​​are nonlinearly mapped to generate a nonlinear mapping matrix, revealing the implicit relationship between the fluctuation characteristic values ​​and the thermodynamic and fluid dynamic behaviors in the microchannel. Based on the nonlinear mapping matrix, a microscopic phase change analysis model is constructed, which focuses on the detailed description of the local physical field during the phase change process. Through this model, the temperature field gradient, fluid velocity distribution and pressure field distribution are accurately calculated. These micro-level evaluation data reveal the local mechanism of heat and momentum transfer in the microchannel. At the same time, based on the nonlinear mapping matrix, a macroscopic heat exchange analysis model is established to analyze the overall performance of the heat exchange module from the system level. The model generates macroscopic evaluation data by calculating the heat transfer coefficient distribution, heat flux density change and total heat transfer. These data provide a global perspective on the comprehensive performance of the heat exchange module, which complements the microscopic analysis. A Markov state transition model is established based on the micro-level evaluation data and the macro-level evaluation data. The Markov model defines the state transition probability, describes the dynamic evolution law of phase change efficiency between different states, and generates the efficiency evolution feature sequence. The efficiency evolution feature sequence is input into the digital twin model, and a virtual image of the microchannel heat exchange module is constructed through real-time data update and physical quantity mapping. The digital twin model combines real operating data and physical simulation to dynamically reflect the operating state of the microchannel heat exchange module under actual working conditions and generate digital twin operating data. Fault mode identification is performed on the digital twin operating data. By matching the operating data with the preset phase change abnormal mode library and heat exchange abnormal mode library, the abnormal state is quickly located and the relevant abnormal state data is extracted. Combined with these data, the phase change efficiency reduction threshold and the heat exchange performance attenuation threshold are set to construct a multi-level early warning mechanism. The early warning mechanism responds in a graded manner according to the degree of abnormality, and can provide real-time alarms and response strategies for minor performance fluctuations, potential faults, and serious faults. Generate multi-level efficiency evaluation data.

[0034] In the embodiment of the present invention, by establishing a nonlinear multi-order coupled phase change heat transfer model, combined with empirical mode decomposition and deep learning methods, the phase change heat transfer characteristics in the microchannel are accurately described, overcoming the accuracy loss problem caused by simplified assumptions in traditional evaluation methods. Adaptive median filtering and modal reconstruction technology are used to effectively eliminate thermal noise interference in the phase change process, extract phase change characteristics with physical significance, and ensure the reliability of the evaluation results. The particle swarm optimization algorithm is introduced to optimize the coupling parameters, achieve the optimal configuration of the phase change efficiency calculation items and the heat exchange performance calculation items, and enhance the applicability of the evaluation model. The time convolution network and time series converter are integrated to construct a multi-scale feature extraction framework, which solves the problem of collaborative analysis of local features and global features in the phase change process. Through digital twin technology, a two-layer evaluation system of microscopic phase change analysis and macroscopic heat exchange analysis is established to achieve all-round monitoring of the performance of the heat exchange module.

[0035] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0036] Temperature sensors, pressure sensors and flow sensors are installed on the heat exchange wall, fluid inlet and fluid outlet of the microchannel heat exchange module to collect real-time temperature data, real-time pressure data and real-time flow data to obtain a multi-dimensional sensor data set;

[0037] Performing a three-dimensional topographic scan on the surface profile of the inner wall of the microchannel to obtain three-dimensional scanning data, and performing adaptive meshing on the three-dimensional scanning data to obtain microchannel geometric structure data;

[0038] Based on the microchannel geometry data, a heat transfer equation group including convective heat transfer, conductive heat transfer and phase change heat transfer is established. Based on the heat transfer equation group, the thermal conductivity coefficient matrix representing the thermal conductivity characteristics, the phase change latent heat matrix representing the phase change characteristics and the boundary condition matrix representing the boundary characteristics are calculated to obtain the initial heat transfer model.

[0039] Based on the real-time temperature data in the multi-dimensional sensor data set and the initial heat transfer model, the nonlinear term coefficient is identified and the parameter is fitted to obtain the nonlinear coupling coefficient;

[0040] Based on the nonlinear coupling coefficient, by establishing the heat conduction relationship equation between the microchannel wall and the fluid, the heat transfer process between the wall and the fluid is quantitatively calculated and numerically analyzed to obtain the heat conduction coupling data;

[0041] Based on the nonlinear coupling coefficient, by establishing the phase change process equation of the fluid in the microchannel, the phase change latent heat release and absorption process is quantitatively calculated and numerically analyzed to obtain the phase change coupling data;

[0042] According to the heat conduction coupling data and phase change coupling data, the heat conduction process and the phase change process are coupled and calculated to obtain a nonlinear multi-order coupled phase change heat transfer model.

[0043] Specifically, a temperature sensor is installed on the heat exchange wall of the microchannel heat exchange module to monitor the operating status of the microchannel heat exchange module in real time. At the same time, a pressure sensor and a flow sensor are installed at the inlet and outlet of the fluid to collect real-time temperature data, real-time pressure data and real-time flow data. These sensor data constitute a multi-dimensional sensing data set, in which the temperature data represents the temperature change of the fluid in different regions, the pressure data reflects the resistance of the fluid flow, and the flow data represents the volume or mass flow of the fluid through the microchannel per unit time. The inner wall of the microchannel is scanned with high precision in three dimensions to obtain its surface profile information. The three-dimensional scanning data is processed by adaptive meshing technology to generate accurate microchannel geometric structure data. The core of adaptive meshing is to dynamically adjust the density of the grid according to the complexity of the geometric structure, such as increasing the grid density at the corners of the microchannel or in areas with large surface roughness, and using larger grids in smooth areas. Ensure the calculation accuracy while reducing the demand for computing resources. Based on the geometric structure data of the microchannel, a group of heat transfer equations is established to describe various heat transfer mechanisms in the microchannel. The heat transfer equations include convection heat transfer, conduction heat transfer and phase change heat transfer. The core expression is:

[0044] ;

[0045] in, is the fluid density, is the specific heat capacity, is the temperature, is the fluid velocity vector, is the thermal conductivity, is the latent heat of phase change, is the time step, is the Dirac function representing the heat release near the phase transition temperature. This set of equations comprehensively considers the heat convection term, the heat conduction term, and the heat released or absorbed by the phase transition. After constructing the set of equations, the heat conduction coefficient matrix related to heat conduction is extracted respectively. , the phase change latent heat matrix related to the phase change process , and the boundary condition matrix describing the boundary conditions For example, the elements of the heat transfer coefficient matrix are expressed as:

[0046] ;

[0047] in and is the basis function, is the computational domain of the microchannel. These matrices together constitute the initial heat transfer model, which is used to preliminarily describe the heat transfer behavior in the microchannel. Combined with the real-time temperature data in the multi-dimensional sensor data set, the nonlinear term coefficients of the initial heat transfer model are identified and fitted through optimization methods such as the least squares method to generate the nonlinear coupling coefficients. The nonlinear coupling coefficient characterizes the complex thermodynamic and fluid dynamic interactions in the microchannel. , construct the heat conduction relationship equation between the microchannel wall and the fluid, and quantitatively analyze the heat transfer process between the wall and the fluid. The relationship is expressed as:

[0048] ;

[0049] in, is the wall heat flux, is the convective heat transfer coefficient, and are the wall temperature and the fluid temperature respectively. The heat conduction coupling data is obtained by calculation to reveal the thermal conductivity characteristics of the wall. At the same time, by establishing a thermodynamic model of the fluid phase change process, the release and absorption of the phase change latent heat are quantitatively analyzed. The phase change process equation is expressed as:

[0050] ;

[0051] in, is the enthalpy, Describe the local temperature range of phase change. These calculations generate phase change coupling data to characterize the heat exchange of the fluid during the phase change process. The heat conduction coupling data is coupled with the phase change coupling data and solved by nonlinear iterative methods to generate a nonlinear multi-order coupled phase change heat transfer model. For example, the Newton-Raphson iterative method is introduced in the solution process to iteratively update the heat transfer coefficient matrix and the phase change latent heat matrix, and gradually converge to a stable solution.

[0052] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0053] Perform empirical mode decomposition on multi-dimensional sensor data sets to obtain multiple intrinsic mode functions, and perform adaptive window division and dynamic adjustment of filter windows on multiple intrinsic mode functions to obtain optimal window parameters;

[0054] Adaptive median filtering is performed on multiple intrinsic mode functions according to optimal window parameters to obtain mode functions after thermal noise is eliminated, and mode reconstruction is performed on the mode functions after thermal noise is eliminated to obtain reconstructed signal data;

[0055] The reconstructed signal data is input into the time convolution network for convolution and pooling operations to obtain local phase change feature data, and the local phase change feature data is input into the time series converter, and the time series dependent features are extracted through the attention mechanism in the time series converter to obtain the global phase change feature data;

[0056] A feature weighted fusion operation is performed on the local phase change feature data and the global phase change feature data to obtain the phase change feature fusion data, and based on the phase change feature fusion data, a feature vector is constructed for the temperature field distribution, fluid flow parameters and phase change state of the phase change process to obtain the target phase change feature data.

[0057] Specifically, empirical mode decomposition (EMD) is performed on the multi-dimensional sensor data set to decompose the original data into multiple intrinsic mode functions (IMFs). These IMFs are components with decreasing frequencies, where each mode function Two conditions are met: first, the difference between the number of local extreme points and the number of zero crossing points of the signal is at most 1, and second, the average value at any time is zero. Through the EMD method, the original data is decomposed into components of different frequency scales:

[0058] ;

[0059] in, is the original signal, is the total number of IMF, is the residual component, which represents the low-frequency trend in the signal that cannot be decomposed. Each IMF The dynamic characteristics corresponding to a specific frequency range can reflect the multi-scale heat transfer and phase change behavior in the microchannel heat exchange module. Adaptive window division and dynamic adjustment of the filter window are performed on multiple IMFs to determine the optimal window parameters. The window division is based on the frequency and energy distribution of the signal, and the window size is dynamically adjusted to capture the changes in local features. The optimization goal of window division is to minimize the mean square error of signal distortion during the filtering process, and the formula is:

[0060] ;

[0061] in, is the IMF after filtering. By optimizing the MSE value, the optimal window parameters are determined for adaptive median filtering. Median filtering eliminates high-frequency noise and spike anomalies by calculating the median of the data in the window and replacing the center value, and obtains the modal function after eliminating thermal noise. After filtering is completed, modal reconstruction is performed on multiple modal functions after eliminating thermal noise. Modal reconstruction is to perform weighted superposition of filtered IMFs to restore the overall picture of the signal and reconstruct the signal. The formula is:

[0062] ;

[0063] in, is the modal function after noise reduction. The reconstructed signal data can reflect the global dynamic characteristics of the microchannel heat exchange module, while effectively reducing the interference of noise on signal analysis. The reconstructed signal data is input into the time convolution network for deep feature extraction. The time convolution network consists of multiple layers of dilated convolution layers, and the dilation rate of each convolution layer increases exponentially (such as 1, 2, 4, 8) to expand the network's receptive field and capture long-term dependent features. The calculation formula of time convolution is:

[0064] ;

[0065] in, For output, is the convolution kernel weight, is the convolution kernel size, is the expansion rate, is the input data. Through convolution and pooling operations, the temporal convolutional network outputs local phase change feature data, revealing the heat transfer and phase change patterns in a short time scale. The local phase change feature data is input into the time series converter model, and the temporal dependency of the features is analyzed through its built-in attention mechanism. The feature representation of each moment in the converter is:

[0066] ;

[0067] in, For time The characteristic output of is the feature representation of the input sequence, is the attention weight, indicating time About Time Through the 6-layer encoder module and multi-head attention mechanism, the converter generates global time series feature data to capture the phase change behavior over a long time scale. The local phase change feature data is weightedly fused with the global time series feature data to generate phase change feature fusion data. The fusion formula is:

[0068] ;

[0069] in, is a local feature, is a global feature, is the weighting coefficient, and the fusion result is obtained by the optimal weight distribution. Based on the fusion feature data, the temperature field distribution, fluid flow parameters and phase change state of the microchannel heat exchange module are constructed with feature vectors to generate the target phase change feature data.

[0070] In a specific embodiment, the execution step performs an adaptive median filtering operation on multiple intrinsic mode functions according to the optimal window parameters to obtain the mode function after eliminating thermal noise, and performs modal reconstruction on the mode function after eliminating thermal noise to obtain the reconstructed signal data. The process may specifically include the following steps:

[0071] The number of windows is determined and calculated for multiple intrinsic mode functions according to the optimal window parameters to obtain a sliding window sequence;

[0072] Calculate the median of the data in each window in the sliding window sequence, and filter out the thermal noise in multiple intrinsic mode functions to obtain first-order denoised modal data;

[0073] A thermal noise variance matrix is ​​constructed based on the first-order denoised modal data, and the filter parameters are adaptively adjusted to obtain adaptive filter parameters, and the adaptive filter parameters are input into a second-order median filter to perform secondary filtering on the first-order denoised modal data to obtain a modal function after eliminating thermal noise;

[0074] A phase modulation analysis based on Hilbert transform is performed on the modal function after thermal noise elimination to obtain phase characteristic data, and a phase reconstruction operation is performed on the modal function after thermal noise elimination based on the phase characteristic data to obtain reconstructed signal data.

[0075] Specifically, the number of windows is determined and calculated for multiple intrinsic mode functions (IMFs) according to the optimal window parameters to form a sliding window sequence. The number and size of the sliding windows are determined by the frequency characteristics and noise distribution of the signal, and its goal is to capture the key characteristics of each mode function through local data partitioning. The calculation formula for the number of sliding windows is expressed as:

[0076] ;

[0077] in, is the total number of windows, is the signal length, The size of a single window. Window size The optimal window parameters are dynamically adjusted so that the window can effectively cover the local characteristics of the modal signal while avoiding information loss caused by too small a window. After dividing the sliding window, the median of the data in each window is calculated to replace the outliers or noise points to achieve preliminary thermal noise filtering. The core calculation formula of the median filter is:

[0078] ;

[0079] in, For the The median filtering result of data points is to is the data point in the window, and the window width is The advantage of median filtering is its strong suppression ability for high-frequency noise and spike interference, ensuring the smoothness and authenticity of the modal signal. After median filtering, the first-order denoised modal data is obtained, and its noise content is significantly reduced. Based on the first-order denoised modal data, the thermal noise variance matrix is ​​constructed to quantify the distribution characteristics of the residual noise. The elements of the thermal noise variance matrix are defined as:

[0080] ;

[0081] in, For the The modal function is in The variance of the window, is the modal data after first-order denoising, is the mean of the data in the window, is the number of data points in the window. According to the thermal noise variance matrix, the filter parameters are adaptively adjusted to generate adaptive filter parameters. These parameters are optimized by minimizing the trade-off function between noise variance and signal fidelity. The objective function is expressed as:

[0082] ;

[0083] in, To balance the weight, is the time-varying rate of the signal. The optimized adaptive filtering parameters are input into the second-order median filter, and the first-order denoised modal data is subjected to secondary filtering to eliminate the remaining low-frequency thermal noise. The second-order median filter extends the capability of the first-order filter, and its dynamically adjusted filter window can better adapt to the changes in the local characteristics of the signal and generate the modal function after eliminating the thermal noise. After the denoising process is completed, the phase modulation analysis based on the Hilbert transform is applied to the modal function after the thermal noise is eliminated to extract the phase characteristics of the signal. The Hilbert transform converts the real signal Convert to complex signal , calculate the instantaneous phase and amplitude:

[0084] ;

[0085] in, is the Hilbert transform, and the calculation formula is:

[0086] ;

[0087] The instantaneous phase and amplitude are defined as:

[0088] ;

[0089] Through phase modulation analysis, the extracted phase characteristic data can reflect the periodicity and nonlinear characteristics of the signal. Based on the phase characteristic data, the phase reconstruction operation is performed on the modal function after eliminating thermal noise to generate reconstructed signal data. Phase reconstruction adjusts the phase and amplitude of the signal so that the reconstructed signal can accurately restore the physical characteristics of the original signal while eliminating noise interference. The expression of the reconstructed signal is:

[0090] .

[0091] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0092] Based on the heat conduction coefficient matrix and phase change coefficient matrix in the nonlinear multi-order coupled phase change heat transfer model, the system phase change efficiency is calculated to obtain the initial value of the phase change efficiency.

[0093] Performing weight distribution calculation on the temperature field distribution data, fluid flow parameter data and phase change state data in the target phase change characteristic data to obtain a characteristic weight coefficient;

[0094] A phase change efficiency calculation item is constructed according to the initial value of the phase change efficiency and the characteristic weight coefficient, and the phase change efficiency calculation item is numerically normalized to obtain normalized phase change efficiency data;

[0095] The heat transfer performance is calculated based on the heat transfer equations in the nonlinear multi-order coupled phase change heat transfer model to obtain the heat transfer coefficient data, and the heat transfer coefficient data is numerically normalized to obtain the normalized heat transfer performance data;

[0096] The normalized phase change efficiency data and the normalized heat transfer performance data are linearly combined to construct an optimization objective function, wherein the optimization objective function includes a phase change efficiency calculation item and a heat transfer performance calculation item;

[0097] The particle swarm initialization parameters are set for the optimization objective function, including particle position, velocity, acceleration constant and inertia weight, to obtain the particle swarm initialization data, and the particle swarm initialization data is iteratively optimized and calculated. The particle position and velocity are updated in each iteration, and the optimal position and global optimal position of each particle are recorded to obtain the particle optimization trajectory data.

[0098] The global optimal solution is extracted based on the particle optimization trajectory data, and the coupling parameter mapping is performed on the global optimal solution to obtain the coupling parameter optimization data.

[0099] Specifically, the heat conduction coefficient matrix is ​​constructed based on the nonlinear multi-order coupled phase change heat transfer model. And the phase change coefficient matrix . Heat transfer coefficient matrix Reflects the heat conduction behavior between the microchannel wall and the fluid, which is defined as:

[0100] ;

[0101] in, is the thermal conductivity, and is the basis function, is the computational domain of the microchannel. The phase change coefficient matrix The latent heat release and absorption behavior of the fluid during phase change is characterized in the form of:

[0102] ;

[0103] in, is the latent heat of phase change, is the Dirac function, indicating that at the phase transition temperature The heat release distribution near . Combining these two matrices, the initial value of the phase change efficiency of the system is obtained by numerical solution , whose expression is:

[0104] ;

[0105] in, is the total input heat flow in the microchannel. The temperature field distribution data, fluid flow parameter data and phase change state data in the target phase change characteristic data are weighted to generate the characteristic weight coefficient The basis for weight distribution is the influence of different characteristics on phase change efficiency, and its weight is determined by sensitivity analysis:

[0106] ;

[0107] in, Indicates Class feature data, is the partial derivative of the phase change efficiency with respect to this feature. and feature weight coefficient Combined, construct the phase change efficiency calculation term:

[0108] ;

[0109] Phase change efficiency calculation term Numerical normalization is performed to eliminate the deviation caused by dimension differences. The normalization formula is:

[0110] ;

[0111] in, and Based on the heat transfer equations in the nonlinear multi-order coupled phase change heat transfer model, the heat transfer performance indicators are calculated, including the heat transfer coefficient data. , whose expression is:

[0112] ;

[0113] in, To exchange heat, is the heat transfer area, is the temperature difference of the fluid. Perform normalization processing to generate normalized heat transfer performance data The normalized phase change efficiency data Normalized heat transfer performance data Perform linear combination and construct optimization objective function :

[0114] ;

[0115] in, and is the weight coefficient of the linear combination, which is used to balance the optimization objectives of phase change efficiency and heat transfer performance. The particle swarm optimization (PSO) algorithm is used to optimize the objective function Perform global optimization. In PSO, set the particle swarm initialization parameters, including particle positions ,speed , acceleration constant and , and the inertia weight The particle update rule after initialization is:

[0116] ;

[0117] ;

[0118] in, and Respectively The particle in The velocity and position in the iteration, is the individual optimal position of the particle, is the global optimal position, and is a random number. Through iterative optimization, the optimal position of each particle and the global optimal position are recorded to generate particle optimization trajectory data. The global optimal solution is extracted based on the optimized trajectory. , and perform coupling parameter mapping on it to generate the final coupling parameter optimization data.

[0119] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0120] The target phase change characteristic data is input into the empirical mode decomposition module of the phase change efficiency prediction model. The empirical mode decomposition module includes three layers of screening networks. Each layer of the screening network uses a Butterworth filter to separate signals and obtain multi-scale modal component data.

[0121] The multi-scale modal component data is input into the temporal convolutional network module of the phase change efficiency prediction model. The temporal convolutional network module contains 4 dilated convolutional layers. Each dilated convolutional layer uses the ReLU activation function and residual connection. The convolution kernel size is 3, and the dilation rates are 1, 2, 4, and 8 respectively, to obtain local time series feature data.

[0122] The local time series feature data is input into the time series converter module of the phase change efficiency prediction model. The time series converter module contains 6 transformer encoding layers. Each transformer encoding layer contains an 8-head attention mechanism and a feedforward neural network. The hidden layer dimension is 256 to obtain the global time series feature data.

[0123] Perform feature fusion on local time series feature data and global time series feature data to obtain multi-scale fusion feature data, and perform feature splicing on the multi-scale fusion feature data and coupling parameter optimization data to obtain a combined feature vector;

[0124] Perform time-dependent analysis on the combined feature vector to obtain time-dependent feature data, and input the time-dependent feature data into the decoder network of the phase change efficiency prediction model. The decoder network contains three transposed convolutional layers. Each transposed convolutional layer uses the LeakyReLU activation function and batch normalization to restore the time dimension through upsampling to obtain the phase change efficiency mapping data.

[0125] The phase change efficiency mapping data is input into the output layer of the phase change efficiency prediction model, and numerical normalization is performed through the sigmoid function to obtain the phase change efficiency prediction data.

[0126] Specifically, the target phase change feature data is input into the empirical mode decomposition (EMD) module, which contains three layers of screening networks. Each layer of the screening network uses a Butterworth filter to separate the signal and extract the modal components in different frequency ranges. The Butterworth filter ensures the high fidelity of the separated modal signal with its smooth amplitude-frequency response characteristics while avoiding signal distortion. Assume that the input signal is , after filtering, the output multi-scale modal component data is expressed as:

[0127] ;

[0128] in, For the The intrinsic mode functions, is the number of modal components, is the residual component. Each layer of the screening network recursively refines the frequency range to generate multi-scale modal component data. The multi-scale modal component data is input into the temporal convolutional network (TCN) module. The TCN contains 4 dilated convolution layers, each of which uses the ReLU activation function and residual connection. The purpose of the dilated convolution is to expand the receptive field and capture feature changes at different time scales. The calculation formula for the dilated convolution is:

[0129] ;

[0130] in, is the convolution output, is the convolution kernel weight, is the convolution kernel size (here 3), is the expansion rate (1, 2, 4, 8 respectively), The output of each convolution layer is added to the input through a residual connection to prevent the gradient from disappearing and improve the efficiency of model training. After multiple layers of dilated convolution, TCN extracts local time series feature data, which can capture dynamic change information within a short time range. The local time series feature data is input into the time series converter module. The converter contains 6 transformer encoding layers, each of which consists of an 8-head multi-head attention mechanism and a feedforward neural network. The multi-head attention mechanism calculates the attention score through the following formula:

[0131] ;

[0132] in, , , They are query vector, key vector and value vector respectively. is the dimension of the key vector. Each encoding layer also contains a feedforward neural network with a hidden layer dimension of 256 for nonlinear transformation. Through multi-layer encoding, the converter captures the global time series feature data over a long period of time and reflects the complex dependencies of the signal in the entire time series. The local time series feature data is fused with the global time series feature data to generate multi-scale fused feature data. The fusion process is achieved through weighted linear combination:

[0133] ;

[0134] in, and are local and global features respectively, and is the weight coefficient. The fused feature data contains both short-term detail information and long-term trend changes. The multi-scale fusion feature data and the coupling parameter optimization data are spliced ​​to obtain a combined feature vector. The combined feature vector is expressed as:

[0135] ;

[0136] in, The data for coupling parameter optimization represents the optimized physical constraints. The combined feature vector is subjected to temporal dependency analysis to extract the dynamic characteristics between sequences. The temporal dependency feature data is input into the decoder network. The decoder contains three transposed convolutional layers. Each convolutional layer uses the LeakyReLU activation function and batch normalization to gradually restore the temporal dimension through upsampling. The transposed convolution operation is implemented by the following formula:

[0137] ;

[0138] in, is the transposed convolution output, For input, is the convolution kernel, is the stride. Through layer-by-layer transposed convolution, the decoder maps the extracted features back to the time series domain to generate phase change efficiency mapping data. The phase change efficiency mapping data is input into the output layer of the model, and the sigmoid function is used for numerical normalization to obtain the phase change efficiency prediction data. The form of the Sigmoid function is:

[0139] ;

[0140] Its output ranges from 0 to 1, ensuring the interpretability and stability of the phase change efficiency prediction data.

[0141] In a specific embodiment, the process of executing the above-mentioned method for evaluating the phase change efficiency of the microchannel heat exchange module further includes the following steps:

[0142] Performing fluctuation characteristic analysis on phase change efficiency prediction data to obtain fluctuation characteristic values, and performing nonlinear mapping calculation on the fluctuation characteristic values ​​to obtain a nonlinear mapping matrix;

[0143] A microscopic phase change analysis model is constructed based on the nonlinear mapping matrix, and the temperature field gradient, fluid velocity distribution and pressure field distribution are calculated during the phase change process to obtain microscopic level evaluation data;

[0144] A macro heat transfer analysis model is constructed based on the nonlinear mapping matrix, and the heat transfer coefficient distribution, heat flux density change and total heat transfer are calculated to obtain macro-level evaluation data;

[0145] A Markov state transition model is established based on the micro-level evaluation data and the macro-level evaluation data. The efficiency evolution characteristic sequence is obtained by calculating the transition probability of the phase change efficiency between different states.

[0146] The efficiency evolution feature sequence is input into the digital twin model, and a virtual image of the microchannel heat exchange module is established through real-time data update and physical quantity mapping to obtain the digital twin operation data;

[0147] Fault mode identification is performed on the digital twin operation data, and abnormal state data is obtained by feature matching with the preset phase change abnormal mode library and heat exchange abnormal mode library. By setting the phase change efficiency reduction threshold and the heat exchange performance attenuation threshold, a multi-level early warning mechanism is established based on the abnormal state data to obtain multi-level efficiency evaluation data.

[0148] Specifically, the fluctuation characteristics of the phase change efficiency prediction data are analyzed to extract the fluctuation amplitude, periodicity and randomness information of the data in the time series. Assuming that the prediction data is in the form of a time series The fluctuation characteristic value is calculated by signal statistical characteristics such as standard deviation and peak factor, for example, standard deviation It is expressed as:

[0149] ;

[0150] in, is the number of time steps, is the average value of the sequence. The frequency domain characteristics such as power spectrum density are extracted by fast Fourier transform to identify the main frequency characteristics of the fluctuation. By analyzing these characteristics, the fluctuation eigenvalues ​​that can reflect the law of phase change efficiency are obtained. The nonlinear mapping calculation of the fluctuation eigenvalues ​​is performed to generate a nonlinear mapping matrix, revealing the relationship between the eigenvalues ​​and the complex behavior of the system. Let the fluctuation eigenvalue vector be , through the nonlinear mapping function , the mapping matrix The elements of are represented as:

[0151] ;

[0152] in, is the mapping scale parameter, and are two eigenvalues ​​respectively. The mapping matrix captures the nonlinear correlation between features through the Gaussian kernel function. Based on the nonlinear mapping matrix, a microscopic phase change analysis model is constructed to calculate the temperature field gradient, fluid velocity distribution and pressure field distribution in the microchannel during the phase change process. The core of the microscopic analysis model is to numerically solve the heat transfer and fluid mechanics equations. For example, the temperature field gradient is described by the heat conduction equation:

[0153] ;

[0154] in, is the temperature, is the heat source term, is the thermal conductivity. The velocity distribution is described by the Navier-Stokes equation:

[0155] ;

[0156] in, is the velocity field, is the pressure field, is the dynamic viscosity, is the density. These equations are solved by numerical methods (such as finite element or finite difference) to generate micro-level evaluation data. At the same time, a macro heat transfer analysis model is constructed based on the nonlinear mapping matrix to evaluate the heat transfer performance index. The heat transfer coefficient distribution calculation formula of the macro model is:

[0157] ;

[0158] in, is the heat transfer coefficient, For heat flow, is the heat exchange area, is the temperature difference. The total heat transfer is:

[0159] ;

[0160] Macro-level evaluation data is generated by analyzing the distribution of heat transfer coefficients, changes in heat flux density, etc. Combining micro- and macro-evaluation data, a Markov state transition model is established to quantify the dynamic changes of phase change efficiency between different states. The state transition probability matrix of the Markov model is defined as:

[0161] ;

[0162] in, For time status, From the state Transfer to state The probability of the phase change efficiency changing over time is generated by iteratively calculating the state transition. The efficiency evolution feature sequence is input into the digital twin model, and the virtual image of the microchannel heat exchange module is established by combining real-time data update and physical quantity mapping. The digital twin continuously synchronizes the real operation data and dynamically adjusts the boundary conditions and parameters of the virtual model to generate the digital twin operation data. The virtual image can reflect the operation status of the system in real time and support predictive maintenance and optimization design. The fault mode recognition is performed on the digital twin operation data, and the possible abnormal state is identified by matching the features with the preset phase change abnormal mode library and heat exchange abnormal mode library. The abnormal state data is classified by setting the phase change efficiency reduction threshold and the heat exchange performance attenuation threshold, and is used to build a multi-level early warning mechanism. The early warning mechanism provides gradually increasing alarms and countermeasures from slight performance fluctuations to serious failures through graded response. Multi-level efficiency evaluation data is generated to comprehensively evaluate the health status and operation performance of the microchannel heat exchange module.

[0163] The above describes the phase change efficiency evaluation method of the microchannel heat exchange module in the embodiment of the present invention. The following describes the phase change efficiency evaluation device of the microchannel heat exchange module in the embodiment of the present invention. Figure 2 , an embodiment of a phase change efficiency evaluation device for a microchannel heat exchange module in an embodiment of the present invention comprises:

[0164] An acquisition module is used to collect data from the microchannel heat exchange module to obtain a multi-dimensional sensing data set, and perform nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation on the multi-dimensional sensing data set to obtain a nonlinear multi-order coupling phase change heat transfer model;

[0165] A decomposition module is used to perform empirical mode decomposition and feature extraction on the multi-dimensional sensor data set to obtain target phase change feature data;

[0166] A calculation module is used to establish an optimization objective function including phase change efficiency calculation items and heat transfer performance calculation items according to a nonlinear multi-order coupled phase change heat transfer model and target phase change characteristic data, and to perform particle swarm optimization calculation based on the optimization objective function to obtain coupling parameter optimization data;

[0167] The processing module is used to input the target phase change characteristic data and the coupling parameter optimization data into the phase change efficiency prediction model for processing to obtain the phase change efficiency prediction data.

[0168] Through the collaborative cooperation of the above components, by establishing a nonlinear multi-order coupled phase change heat transfer model, combined with empirical mode decomposition and deep learning methods, the phase change heat transfer characteristics in the microchannel are accurately described, overcoming the accuracy loss caused by simplified assumptions in traditional evaluation methods. Adaptive median filtering and modal reconstruction technology are used to effectively eliminate the thermal noise interference in the phase change process, extract the phase change characteristics with physical significance, and ensure the reliability of the evaluation results. The particle swarm optimization algorithm is introduced to optimize the coupling parameters, realize the optimal configuration of the phase change efficiency calculation items and the heat transfer performance calculation items, and enhance the applicability of the evaluation model. The time convolution network and time series converter are integrated to construct a multi-scale feature extraction framework, which solves the problem of collaborative analysis of local features and global features in the phase change process. Through the digital twin technology, a two-layer evaluation system of microscopic phase change analysis and macroscopic heat transfer analysis is established to achieve all-round monitoring of the performance of the heat exchange module.

[0169] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0170] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0171] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0172] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0173] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0174] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0175] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the phase change efficiency of a microchannel heat exchange module, characterized in that: The method comprises: Data collection is performed on the microchannel heat exchange module to obtain a multi-dimensional sensing data set, and nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation are performed on the multi-dimensional sensing data set to obtain a nonlinear multi-order coupling phase change heat transfer model; specifically comprising: installing temperature sensors, pressure sensors and flow sensors on the heat exchange wall, fluid inlet and fluid outlet of the microchannel heat exchange module respectively, collecting real-time temperature data, real-time pressure data and real-time flow data to obtain a multi-dimensional sensing data set; performing a three-dimensional morphology scan on the surface contour of the inner wall of the microchannel to obtain three-dimensional scanning data, and performing adaptive meshing on the three-dimensional scanning data to obtain microchannel geometric structure data; based on the microchannel geometric structure data, a heat transfer equation group including convective heat transfer, conductive heat transfer and phase change heat transfer is established, and based on the heat transfer equation group, the heat conduction heat transfer and phase change heat transfer heat transfer are respectively calculated. coefficient matrix, a phase change latent heat matrix characterizing phase change characteristics, and a boundary condition matrix characterizing boundary characteristics to obtain an initial heat transfer model; based on the real-time temperature data in the multidimensional sensing data set and the initial heat transfer model, nonlinear term coefficient identification and parameter fitting are performed to obtain a nonlinear coupling coefficient; based on the nonlinear coupling coefficient, by establishing a heat conduction relationship equation between the microchannel wall and the fluid, the heat transfer process between the wall and the fluid is quantitatively calculated and numerically analyzed to obtain heat conduction coupling data; based on the nonlinear coupling coefficient, by establishing a phase change process equation of the fluid in the microchannel, the phase change latent heat release and absorption process is quantitatively calculated and numerically analyzed to obtain phase change coupling data; according to the heat conduction coupling data and the phase change coupling data, the heat conduction process and the phase change process are coupled and calculated to obtain a nonlinear multi-order coupled phase change heat transfer model; Performing empirical mode decomposition and feature extraction on the multi-dimensional sensor data set to obtain target phase change feature data; According to the nonlinear multi-order coupled phase change heat transfer model and the target phase change characteristic data, an optimization objective function including a phase change efficiency calculation item and a heat exchange performance calculation item is established, and a particle swarm optimization calculation is performed based on the optimization objective function to obtain coupling parameter optimization data; The target phase change characteristic data and the coupling parameter optimization data are input into a phase change efficiency prediction model for processing to obtain phase change efficiency prediction data.

2. The phase change efficiency evaluation method of the microchannel heat exchange module according to claim 1, characterized in that: The performing empirical mode decomposition and feature extraction on the multi-dimensional sensing data set to obtain target phase change feature data includes: Performing empirical mode decomposition on the multi-dimensional sensor data set to obtain a plurality of intrinsic mode functions, and performing adaptive window division and dynamic filter window adjustment on the plurality of intrinsic mode functions to obtain optimal window parameters; Performing an adaptive median filtering operation on the multiple intrinsic mode functions according to the optimal window parameters to obtain a mode function after eliminating thermal noise, and performing modal reconstruction on the mode function after eliminating thermal noise to obtain reconstructed signal data; Inputting the reconstructed signal data into a time convolution network for convolution and pooling operations to obtain local phase change feature data, and inputting the local phase change feature data into a time series converter, extracting time-dependent features through an attention mechanism in the time series converter, and obtaining global phase change feature data; A feature weighted fusion operation is performed on the local phase change feature data and the global phase change feature data to obtain phase change feature fusion data, and based on the phase change feature fusion data, a feature vector is constructed for the temperature field distribution, fluid flow parameters and phase change state of the phase change process to obtain target phase change feature data.

3. The phase change efficiency evaluation method of the microchannel heat exchange module according to claim 2, characterized in that: The method of performing an adaptive median filtering operation on the multiple intrinsic mode functions according to the optimal window parameters to obtain a mode function after eliminating thermal noise, and performing mode reconstruction on the mode function after eliminating thermal noise to obtain reconstructed signal data includes: Performing window quantity determination calculation on the plurality of intrinsic mode functions according to the optimal window parameters to obtain a sliding window sequence; Calculating the median of the data in each window in the sliding window sequence, and filtering out the thermal noise in the multiple intrinsic mode functions to obtain first-order denoised modal data; Based on the first-order denoised modal data, a thermal noise variance matrix is ​​constructed, and filter parameters are adaptively adjusted to obtain adaptive filter parameters, and the adaptive filter parameters are input into a second-order median filter to perform secondary filtering on the first-order denoised modal data to obtain a modal function after eliminating thermal noise; A phase modulation analysis based on Hilbert transform is performed on the modal function after thermal noise elimination to obtain phase characteristic data, and a phase reconstruction operation is performed on the modal function after thermal noise elimination based on the phase characteristic data to obtain reconstructed signal data.

4. The phase change efficiency evaluation method of the microchannel heat exchange module according to claim 3, characterized in that: The method of establishing an optimization objective function including a phase change efficiency calculation item and a heat exchange performance calculation item according to the nonlinear multi-order coupled phase change heat transfer model and the target phase change characteristic data, and performing a particle swarm optimization calculation based on the optimization objective function to obtain coupling parameter optimization data includes: Calculate the system phase change efficiency based on the heat conduction term coefficient matrix and the phase change term coefficient matrix in the nonlinear multi-order coupled phase change heat transfer model to obtain an initial value of the phase change efficiency; Performing weight distribution calculation on the temperature field distribution data, fluid flow parameter data and phase change state data in the target phase change characteristic data to obtain a characteristic weight coefficient; Constructing a phase change efficiency calculation item according to the phase change efficiency initial value and the characteristic weight coefficient, and performing numerical normalization processing on the phase change efficiency calculation item to obtain normalized phase change efficiency data; Calculating heat transfer performance based on the heat transfer equations in the nonlinear multi-order coupled phase change heat transfer model to obtain heat transfer coefficient data, and performing numerical normalization processing on the heat transfer coefficient data to obtain normalized heat transfer performance data; Linearly combining the normalized phase change efficiency data and the normalized heat exchange performance data to construct an optimization objective function, wherein the optimization objective function includes a phase change efficiency calculation item and a heat exchange performance calculation item; Setting particle swarm initialization parameters for the optimization objective function, including particle position, velocity, acceleration constant and inertia weight, to obtain particle swarm initialization data, and performing iterative optimization calculation on the particle swarm initialization data, updating the particle position and velocity in each iteration, and recording the optimal position and global optimal position of each particle to obtain particle optimization trajectory data; A global optimal solution is extracted based on the particle optimization trajectory data, and coupling parameter mapping is performed on the global optimal solution to obtain coupling parameter optimization data.

5. The phase change efficiency evaluation method of the microchannel heat exchange module according to claim 4, characterized in that: The step of inputting the target phase change characteristic data and the coupling parameter optimization data into a phase change efficiency prediction model for processing to obtain phase change efficiency prediction data comprises: Inputting the target phase change characteristic data into an empirical mode decomposition module of a phase change efficiency prediction model, wherein the empirical mode decomposition module includes three layers of screening networks, each layer of the screening network uses a Butterworth filter to perform signal separation to obtain multi-scale modal component data; Inputting the multi-scale modal component data into the time convolution network module of the phase change efficiency prediction model, the time convolution network module includes 4 dilated convolution layers, each dilated convolution layer uses a ReLU activation function and a residual connection, the convolution kernel size is 3, and the dilation rates are 1, 2, 4, and 8 respectively, to obtain local time series feature data; Input the local time series feature data into the time series converter module of the phase change efficiency prediction model, wherein the time series converter module comprises 6 transformer encoding layers, each transformer encoding layer comprises an 8-head attention mechanism and a feedforward neural network, and the hidden layer dimension is 256, to obtain the global time series feature data; Performing feature fusion on the local time series feature data and the global time series feature data to obtain multi-scale fusion feature data, and performing feature splicing on the multi-scale fusion feature data and the coupling parameter optimization data to obtain a combined feature vector; Performing a time-dependent analysis on the combined feature vector to obtain time-dependent feature data, and inputting the time-dependent feature data into a decoder network of the phase change efficiency prediction model, wherein the decoder network comprises three transposed convolutional layers, each of which uses a LeakyReLU activation function and batch normalization, and restores the time dimension by upsampling to obtain phase change efficiency mapping data; The phase change efficiency mapping data is input into the output layer of the phase change efficiency prediction model, and numerical normalization is performed through a sigmoid function to obtain phase change efficiency prediction data.

6. The phase change efficiency evaluation method of the microchannel heat exchange module according to claim 5, characterized in that: The phase change efficiency evaluation method of the microchannel heat exchange module also includes: Performing a fluctuation characteristic analysis on the phase change efficiency prediction data to obtain a fluctuation characteristic value, and performing a nonlinear mapping calculation on the fluctuation characteristic value to obtain a nonlinear mapping matrix; A microscopic phase change analysis model is constructed based on the nonlinear mapping matrix, and the temperature field gradient, fluid velocity distribution and pressure field distribution are calculated during the phase change process to obtain microscopic level evaluation data; A macroscopic heat transfer analysis model is constructed based on the nonlinear mapping matrix, and heat transfer coefficient distribution, heat flux density change and total heat transfer are calculated to obtain macroscopic level evaluation data; A Markov state transition model is established according to the micro-level evaluation data and the macro-level evaluation data, and an efficiency evolution characteristic sequence is obtained by calculating the transition probability of phase change efficiency between different states; Inputting the efficiency evolution characteristic sequence into the digital twin model, establishing a virtual mirror image of the microchannel heat exchange module through real-time data updating and physical quantity mapping, and obtaining digital twin operation data; Fault mode identification is performed on the digital twin operation data, and abnormal state data is obtained by feature matching with a preset phase change abnormal mode library and a heat exchange abnormal mode library. A multi-level early warning mechanism is established based on the abnormal state data by setting a phase change efficiency reduction threshold and a heat exchange performance attenuation threshold to obtain multi-level efficiency evaluation data.

7. A phase change efficiency evaluation device for a microchannel heat exchange module, characterized in that: Used to perform the phase change efficiency evaluation method of the microchannel heat exchange module according to any one of claims 1 to 6, the phase change efficiency evaluation device of the microchannel heat exchange module comprises: An acquisition module is used to collect data from the microchannel heat exchange module to obtain a multi-dimensional sensing data set, and perform nonlinear zero-order heat conduction coupling calculation and first-order phase change coupling calculation on the multi-dimensional sensing data set to obtain a nonlinear multi-order coupling phase change heat transfer model; A decomposition module, used for performing empirical mode decomposition and feature extraction on the multi-dimensional sensor data set to obtain target phase change feature data; A calculation module, used to establish an optimization objective function including a phase change efficiency calculation item and a heat exchange performance calculation item according to the nonlinear multi-order coupled phase change heat transfer model and the target phase change characteristic data, and perform particle swarm optimization calculation based on the optimization objective function to obtain coupling parameter optimization data; The processing module is used to input the target phase change characteristic data and the coupling parameter optimization data into the phase change efficiency prediction model for processing to obtain phase change efficiency prediction data.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, the phase change efficiency evaluation method of the microchannel heat exchange module described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor executes the phase change efficiency evaluation method for a microchannel heat exchange module according to any one of claims 1 to 6.

Images