Heat dissipation prediction correction method and system based on salt spray deposition model

Abstract

The present application relates to the field of material physical performance analysis and testing technology, and specifically discloses a heat dissipation prediction correction method and system based on a salt mist deposition model, which accurately simulates the influence of salt mist deposition on the thermal performance of a heat sink by constructing a deposition kinetics model and a multi-physical field coupling model.The present application first collects salt mist environment parameters to establish a deposition model, and then constructs a multi-physical field model including heat conduction, moisture migration and conductance effect, performs heat transfer simulation to generate an initial prediction curve; at the same time, the actual heat dissipation performance change curve is analyzed by monitoring the surface resistance impedance spectrum signal of the heat sink; the model parameters are adjusted reversely by comparing the deviation of the two curves to realize prediction correction, and finally the high-precision heat dissipation performance prediction result is output.The present application improves the analysis accuracy and efficiency, shortens the test period, and provides strong support for product reliability design and optimization.

Description

Technical Field

[0001] This invention belongs to the field of material physical property analysis and testing technology, and relates to a heat dissipation prediction and correction method and system based on salt spray deposition model. Background Technology

[0002] In high-salt-fog environments such as oceans and coastal areas, the reliability of electronic equipment, energy facilities, and precision instruments faces severe challenges, making the thermal management system of their core components particularly critical. As a key component of the thermal management system, the heat sink's surface material's physical properties directly determine its heat dissipation efficiency. Salt spray deposition is a complex physicochemical process that alters the surface geometry, roughness, and thermophysical parameters of the heat sink. Therefore, accurately analyzing and studying the impact of salt spray deposition on the thermal performance of heat sink materials is of great significance for ensuring the long-term stable operation of equipment.

[0003] In existing technologies, the analysis of material thermal properties under salt spray conditions typically employs two approaches. The first is the physical experimental method, which involves placing the material or radiator sample in an artificial salt spray chamber for prolonged accelerated aging tests, periodically removing the sample for offline thermal resistance testing, and using extensive experimental data to summarize the performance degradation patterns. The second method is the numerical simulation method, where researchers, based on experience or limited experimental observations, add a simplified deposition layer with uniform geometry and physical properties to the radiator model in thermal simulation software, and then analyze its static impact on heat dissipation performance through calculations.

[0004] However, all of the aforementioned existing analytical methods have inherent technical limitations. While physical experiments can reflect real-world conditions, they are time-consuming, costly, and the results are only valid for specific experimental conditions, making it difficult to develop universally applicable analytical models. Numerical simulations, by oversimplifying the complex salt spray deposition process and neglecting the non-uniformity, porosity, and dynamic hygroscopic properties of the deposition layer under temperature and humidity cycles, often result in significant discrepancies between their analytical results and actual conditions, failing to provide accurate guidance for product reliability design. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a heat dissipation prediction and correction method and system based on salt spray deposition model are proposed.

[0006] The objective of this invention can be achieved through the following technical solution: The first aspect of this invention provides a heat dissipation prediction and correction method based on a salt spray deposition model, including: S1, collecting fluid dynamic parameters and droplet physical property parameters in a salt spray environment, and establishing a deposition dynamic model based on the fluid dynamic parameters and droplet physical property parameters to generate deposition seed characteristic parameters.

[0007] S2. Based on the characteristic parameters of the sediment seed, a multi-physics coupling model that couples thermal conduction, moisture migration and electrical conductivity effects is constructed.

[0008] S3 drives a multiphysics coupling model to perform heat transfer simulation and generates an initial prediction curve for heat dissipation performance.

[0009] S4. Obtain the electrical impedance spectrum signal of the surface deposited layer of the radiator when it is running in a salt spray environment, and analyze the actual change curve of heat dissipation performance based on the electrical impedance spectrum signal.

[0010] S5. Based on the deviation between the initial predicted curve and the actual change curve, the key control parameters in the sedimentation kinetic model are adjusted in reverse, and the corrected sedimentation seed characteristic parameters are obtained by recalculating based on the adjusted key control parameters.

[0011] S6. Update the multiphysics coupling model using the corrected sedimentation seed characteristic parameters, and output the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model.

[0012] The second aspect of the present invention provides a heat dissipation prediction and correction system based on a salt spray deposition model, comprising: an environment and deposition sensing module, which collects fluid dynamic parameters and droplet physical property parameters in a salt spray environment, establishes a deposition dynamics model based on the fluid dynamic parameters and droplet physical property parameters, and generates deposition seed characteristic parameters.

[0013] The multiphysics modeling module constructs a multiphysics coupled model that combines thermal conduction, moisture migration, and electrical conductivity effects based on the characteristic parameters of the sediment seed.

[0014] The multiphysics heat transfer simulation module drives the multiphysics coupled model to perform heat transfer simulation and generates an initial prediction curve of heat dissipation performance.

[0015] The online status monitoring and analysis module acquires the electrical impedance spectrum signal of the surface deposited layer of the radiator when it is running in a salt spray environment, and analyzes a curve showing the actual change in heat dissipation performance based on the electrical impedance spectrum signal.

[0016] The model calibration and evolution module adjusts the key control parameters in the sedimentation kinetics model in reverse based on the deviation between the initial predicted curve and the actual change curve, and recalculates based on the adjusted key control parameters to obtain the corrected sedimentation seed characteristic parameters.

[0017] The prediction and design optimization module updates the multiphysics coupling model using the corrected deposition seed characteristic parameters, and outputs the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model.

[0018] Compared with existing technologies, the embodiments of the present invention have at least the following advantages or beneficial effects: The present invention realizes the analysis and characterization of the thermal performance degradation process of materials under salt spray environment by constructing a dynamic closed-loop system that combines physical models with online measurements. By establishing a simulation model that couples multiple physical fields of heat, humidity, and electricity, and continuously calibrating the model using real-time acquired electrical impedance spectrum signals, a high degree of consistency between the analysis results and physical reality is ensured, overcoming the huge deviation caused by model simplification in traditional analysis methods.

[0019] This invention transforms the traditionally long-term, destructive physical environment testing into an efficient, non-invasive online analysis and prediction method. By combining short-term calibration with a self-evolutionary model, the system can quickly converge and accurately extrapolate the long-term performance evolution trend of materials, greatly shortening the performance evaluation cycle of materials and components under specific environments and reducing testing costs and development time.

[0020] This invention provides a novel proactive design analysis capability that surpasses traditional passive performance testing. Utilizing validated high-fidelity models, it enables virtual simulation analysis of the microstructures of different material surfaces, allowing for in-depth exploration of their intrinsic mechanisms affecting salt spray deposition processes and thermal properties. This enables optimized material design early in product development, proactively enhancing its resistance to environmental impacts, rather than relying solely on post-hoc testing and verification. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0023] Figure 2 This is a diagram showing the system module composition of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figure 1The first aspect of the present invention provides a heat dissipation prediction and correction method based on a salt spray deposition model, comprising: S1, collecting fluid dynamic parameters and droplet physical property parameters in a salt spray environment, and establishing a deposition dynamic model based on the fluid dynamic parameters and droplet physical property parameters to generate deposition seed characteristic parameters.

[0026] In a specific embodiment of the present invention, a deposition dynamics model is established based on fluid dynamics parameters and droplet physical property parameters to generate deposition seed characteristic parameters, including: acquiring airflow velocity field data and temperature and humidity data in an observation channel connected to the radiator environment to form fluid dynamics parameters.

[0027] Within the observation channel, data on the droplet size distribution and charge characteristics of salt spray droplets are acquired to form droplet physical property parameters.

[0028] The fluid dynamics parameters and droplet physical properties parameters are input into the Lagrange model used to simulate the transport and attachment of particles in the flow field. The spatial distribution of attachment efficiency and the initial pore structure characteristics are calculated, and the spatial distribution of attachment efficiency and the initial pore structure characteristics are used together as the deposition seed characteristic parameters.

[0029] Specifically, the purpose of this step is to accurately capture the key physical characteristics that determine the initial stage of salt spray deposition through high-frequency, high-dimensional online environmental monitoring, thereby providing the basic physical input for the subsequent construction of a high-fidelity multiphysics coupling model. This process replaces the traditional reliance on macroscopic deposition data obtained through long-term experiments, achieving a quantitative characterization of the initial driving force of the deposition process.

[0030] To achieve this objective, the system first deploys a high-precision thermistor, a capacitive humidity sensor, and a miniature hot-wire anemometer within a micro-observation channel connected to the main heat dissipation channel. This continuously collects ambient temperature, relative humidity, and local airflow velocity at a sampling frequency of at least 1 Hz. After data filtering and fusion, time-seriesd fluid dynamic parameters are generated. Simultaneously, a laser diffraction particle size analyzer and an electrostatic induction measurement unit are integrated into this micro-observation channel. The laser diffraction particle size analyzer acquires real-time data on the size distribution of salt spray droplets within the channel, with a particle size measurement range covering 1 to 100 micrometers. The electrostatic induction measurement unit analyzes the weak charge signals generated when droplets pass over a pair of sensing electrodes, resolving the droplet's charge characteristics. The collected size distribution data and charge characteristic data together constitute the droplet's physical property parameters.

[0031] Subsequently, the system inputs the real-time acquired hydrodynamic parameters and droplet physical property parameters into a preset deposition kinetics model. This model, based on the Lagrangian framework, simulates deposition behavior by solving the following adhesion efficiency function.

[0032]

[0033] in, Represents adhesion efficiency. The Stokes number characterizes the inertial effect of the droplet, and its value is determined by the droplet's relaxation time. Local airflow velocity and the characteristic length of the heat sink fins The calculations show that the droplet relaxation time is directly related to the size distribution data in the salt spray droplet parameters. This is an electrophoretic force parameter that characterizes the electrostatic attraction effect; its value depends on the amount of droplet charge resolved from the charge property data. Local electric field strength and the fluid drag force on the droplet . ... The model generates a spatially non-uniform adhesion efficiency distribution by calculating the adhesion efficiency distribution. Simultaneously, based on this distribution, the model simulates the initial nucleation and accumulation of salt crystals, statistically analyzing the resulting microporosity and connectivity, quantifying them as initial pore structure characteristics. Finally, the adhesion efficiency distribution and initial pore structure characteristics are combined as deposition seed feature parameters, providing precise initial conditions for the next modeling step.

[0034] It should be noted that, in conjunction with the local parameter construction module, the algorithm equations and formula calculation steps are explained in detail as follows:

[0035] 1. The formula and analysis of the Stokes number, and its physical meaning: The Stokes number ( This is used to characterize the inertial effect of salt spray droplets (particles) in a flow field. It determines whether the droplets will follow the airflow around the radiator fins or deviate from the streamline due to inertia and impact and deposit on the radiator surface. Calculation formula: Detailed explanation of formula parameters: Stokes number, a dimensionless parameter. Local airflow velocity (unit: The anemometer was measured by a miniature hot-wire anemometer. Characteristic length of radiator fins (unit: ), which is a macroscopic geometric boundary parameter describing the impedance area of ​​the heated structure inside the flow field channel. Relaxation time of droplets (unit: This reflects the degree of hysteresis in the droplet's response to changes in airflow velocity. Its calculation relies on the size distribution data within the salt spray droplet parameters, and the specific expansion formula is as follows: ,in, For droplet density, The equivalent diameter of the droplet is measured by a laser diffraction particle size analyzer. This represents the aerodynamic viscosity coefficient. The calculation essentially reflects the interplay between the momentum of salt particles and the inertia of the heat-transferring fluid. When... At this time, the droplets have high inertia and are very likely to break away from the streamlines, impacting and depositing on the fin wall; when At this time, droplets tend to be expelled with the airflow.

[0036] 2. Calculation formulas and analysis of electrophoretic force parameters, physical meaning: Electrophoretic force parameters ( This is used to characterize the adsorption capacity of a local electrostatic field for charged salt spray droplets. It measures the relative strength between the electrostatic attraction and the fluid drag (resistance), determining the probability of positive adsorption and adhesion of micron-sized droplets on the radiator surface. Calculation formula: Detailed explanation of formula parameters: Electrophoretic force parameters, dimensionless parameters (characterizing the ratio of electric force to drag force). The amount of charge carried by a droplet (unit: The value (coulomb) is derived by analyzing the weak charge signal generated when the droplet passes over the electrode using the electrostatic induction measurement unit. Local electric field intensity on and around the radiator surface (unit: Their product This refers to the electrostatic attraction experienced by the droplet. Fluid drag force on a droplet (unit: (Newton). In low Reynolds number fluid environments with tiny particles, this drag force is typically expressed based on the Stokes-Reynolds drag law as: ,in, For fluid dynamic viscosity, Where is the droplet diameter, Let be the relative sliding velocity between the airflow and the droplet. The computational essence is explained as follows: a coupled solution is used. (Mechanical inertial impact) and (Electrostatic adsorption) The deposition kinetics model can comprehensively calculate the final adhesion efficiency under the combined physical field. This provides rigorous mathematical and physical support for the "seed characteristic parameters" used to generate virtual salt deposits.

[0037] It should also be noted that the surface energy parameter This is a dimensionless factor used to quantify the interaction strength between a droplet and the solid surface of a heat sink, typically ranging from 0 to 1. The physical essence of this parameter originates from wetting theory in materials science, specifically the solid-liquid-gas three-phase contact angle described by Young's equations. Decision. In the sedimentation kinetics model, Through a preset mapping function and contact angle Association, for example, using The calculation is performed in the form of [formula missing]. Wherein, the contact angle [formula missing]. It is an inherent parameter of the material that is pre-set based on the surface material of the radiator and its surface treatment process. The setting of the value directly affects the adhesion efficiency function. Calculation: When When the angle approaches 1 (corresponding to a hydrophilic surface with a small contact angle), it indicates that the droplets easily spread on the surface, have strong adhesion, and are conducive to deposition; when When the value approaches 0 (corresponding to a hydrophobic surface with a large contact angle), it indicates that the droplets tend to remain spherical, with weak adhesion, which is unfavorable for deposition. By introducing this parameter, the model can accurately distinguish the resistance of different surface materials and coatings to salt spray deposition, providing key physical input for subsequent anti-deposition optimization design.

[0038] S2. Based on the characteristic parameters of the sediment seed, a multi-physics coupling model that couples thermal conduction, moisture migration and electrical conductivity effects is constructed.

[0039] In a specific embodiment of the present invention, a multiphysics coupling model that couples thermal conduction, moisture migration and electrical conductivity effects is constructed based on the sedimentation seed characteristic parameters, including: defining the equivalent porous media network parameters of the sedimentation layer based on the initial pore structure characteristics in the sedimentation seed characteristic parameters.

[0040] A capillary adsorption model is established based on equivalent porous media network parameters to describe the behavior of liquid films in pores, generating parameters for moisture storage and migration.

[0041] By combining moisture storage and migration parameters with an ion ionization model, the equivalent conductivity distribution parameters of the deposited layer were calculated.

[0042] The equivalent porous media network parameters, moisture storage and migration parameters, and equivalent conductivity distribution parameters are integrated with the geometric model of the heat sink to form a multiphysics coupling model.

[0043] Specifically, the purpose of this step is to transform the abstract sedimentary seed characteristic parameters obtained in the previous stage into an equivalent solid model that can be calculated in a simulation environment and can characterize the complex physical properties of the sedimentary layer. This process is the core link in realizing the coupled simulation of thermal, humidity, and electrical multiphysics fields, mapping the microscopic sedimentation mechanism to macroscopic engineering calculation parameters.

[0044] The system first receives the sedimentation seed feature parameters from the previous stage and extracts the initial pore structure features, which include the average porosity and pore tortuosity of the initial stage of the sedimentation layer. Based on this, the system calls a preset physical model to transform these microstructure parameters into equivalent porous media network parameters. These equivalent porous media network parameters mainly consist of permeability tensor and effective thermal conductivity, and are used to characterize the flow resistance of fluids in the porous media and the heat transfer capacity of the solid skeleton in subsequent simulations.

[0045] Next, based on the pore size distribution defined in the equivalent porous media network parameters, the system establishes moisture storage and migration parameters for the capillary adsorption network. In engineering, this is achieved by introducing a capillary pressure-saturation relationship curve, which defines the amount of liquid water that can be contained in the pores of the sediment layer through capillary adsorption under different ambient relative humidity levels—that is, the saturation. Simultaneously, the unsaturated relative permeability is calculated based on this model, together forming the moisture storage and migration parameters, used to describe the dynamic behavior of moisture absorption and desaturation in the sediment layer during wet-dry cycles.

[0046] Subsequently, the system calculates the equivalent conductivity distribution parameters of the deposited layer based on the saturation determined by the moisture storage and migration parameters, and the preset salinity ionization degree. This calculation process follows the conductivity theory of porous media, and its core is characterized by the following formula: ,in, It is the final generated equivalent conductivity distribution parameter, which is a spatial distribution function. It is a coefficient related to sediment materials and is a preset value. The intrinsic conductivity of the salt solution in the pores is determined by the preset salt concentration. It is the porosity of the deposited layer, derived from the equivalent porous media network parameters. It is the water saturation, determined by moisture storage and migration parameters under the current ambient humidity. and These are empirical indices related to pore structure and saturation state, which are either preset or obtained through calibration.

[0047] It should be noted that, in the equivalent conductivity distribution parameters In the calculations, each preset parameter in the formula has a clear physical meaning and engineering basis for its value. Material coefficient This coefficient is a dimensionless geometric factor used to correct for the contribution of the conductivity of the salt crystal (typically sodium chloride) itself to the overall equivalent conductivity. In this embodiment, since the salt crystal can be considered an insulator in a dry state, its conductivity is much lower than that of an aqueous electrolyte solution; therefore, its typical value is [value missing]. This value is based on the "electrolyte-dominant assumption," which assumes that charge transport within the deposition layer occurs entirely through the liquid electrolyte composed of dissolved salt ions within the pores, and the contribution of the solid framework is negligible. Intrinsic conductivity This parameter represents the state of full saturation ( The conductivity of the salt solution within the pores (unit: S / m). Its value is primarily determined by the ion concentration and temperature of the solution. In this embodiment, the system calculates the conductivity based on the real-time ambient temperature and assuming the salt concentration (mainly NaCl) in the salt spray droplets is the standard seawater concentration (approximately 3.5%), by referring to a pre-defined table of Korlausch's conductivity law or empirical formulas. At a typical operating temperature of 25°C, its typical value is... S / m. This value is based on the principle of "standard marine environment simulation" to ensure that the model can reflect the electrochemical characteristics under typical marine environments in the initial stage.

[0048] Experience Index and These two dimensionless exponents are key parameters in Archie's formula and its modified form, used to describe the nonlinear effects of pore structure and water saturation on ion migration paths, respectively. Pore structure exponent This reflects the tortuosity and connectivity of the pore network. For porous media formed by salt crystallization, which have a relatively loose structure and good connectivity, the typical value is set between 1.8 and 2.2. In this embodiment, it is preferably... This value is based on statistical results from a large amount of experimental data on the conductivity of porous media. This is a classic value for unconsolidated or weakly consolidated particle systems. Saturation index. : Describes the change in water content (saturation) in pores The decrease in water content leads to a sharp reduction in conductive paths. In unsaturated porous media, as water content decreases, the liquid film becomes discontinuous, causing the rate of decrease in conductivity to be much faster than the linear decrease in water content. For a typical water-gas two-phase system, the typical value is usually set to 2.0. This value is based on the classic "series-parallel conductive path model". It can fit the conductivity-saturation relationship curves of most unsaturated soils and rocks very well, and has wide applicability.

[0049] Finally, in the simulation software, the system uses the three-dimensional geometric model of the cleanroom radiator as a base and creates a virtual computational domain of variable thickness on its salt spray-affected surface. The system assigns the calculated equivalent porous media network parameters, moisture storage and migration parameters, and equivalent conductivity distribution parameters to each grid cell of this virtual computational domain as field variables. Through this spatial mapping and superposition, a complete multiphysics coupling model capable of participating in multiphysics calculations is finally constructed.

[0050] S3 drives a multiphysics coupling model to perform heat transfer simulation and generates an initial prediction curve for heat dissipation performance.

[0051] In a specific embodiment of the present invention, driving a multiphysics coupling model to perform heat transfer simulation and generate an initial prediction curve of heat dissipation performance includes: setting a discrete time step sequence, and within each time step, using the multiphysics coupling model to calculate the instantaneous physical state of the deposition layer, including the instantaneous thickness field and the instantaneous equivalent thermal conductivity distribution.

[0052] The instantaneous physical state is applied as a dynamic thermal resistance boundary condition to the surface mesh of the heat sink, and the transient heat conduction equation is solved to obtain the predicted temperature value of the critical nodes of the heat sink at the current time step.

[0053] By concatenating the predicted temperature values ​​for all time steps in chronological order, an initial prediction curve of the heat dissipation performance as it evolves over time is generated.

[0054] Specifically, the core of the step of driving a multiphysics coupling model to perform heat transfer simulation and generate an initial prediction curve of heat dissipation performance lies in simulating the dynamic accumulation process of salt spray deposition layer from nothing to something and from thin to thick, and its gradual impact on heat dissipation performance, through time discretization.

[0055] The purpose of this step is to transform the static physical field model into a dynamic time-evolution process. Traditional static simulations typically calculate thermal performance under a fixed state, while salt spray deposition is a typical non-steady-state process. Therefore, the system first sets a discrete time step sequence. (For example, one step is one hour or half a day). Within each time step... Within this timeframe, the system invokes a multiphysics coupling model to perform the following calculations: Depositional growth calculation: Based on current depositional kinetic parameters (such as adhesion efficiency), calculate the mass of newly added sediment during this time period, and convert it into an instantaneous thickness field according to the porosity model. Property update calculation: Based on the current ambient humidity, the water saturation in the sediment layer is updated using a capillary adsorption model, and then the instantaneous equivalent thermal conductivity distribution of the sediment layer is updated using a mixed-medium thermal conductivity formula. .

[0056] Next, the system transforms the calculated instantaneous thickness field and thermal conductivity distribution into a dynamic thermal resistance boundary condition. Apply this boundary condition to the corresponding surface mesh of the radiator geometry. Then, solve the transient heat conduction equation: in Let represent the heat dissipation power of the heat sink. By solving this equation, the system obtains the current temperature of critical nodes of the heat sink (such as the chip junction temperature). Predicted temperature value Throughout the formula, This term describes the rate of change of heat per unit volume due to thermal conduction. It is first described by the gradient ( Calculate the direction and rate of temperature change at each point, and then multiply by the thermal conductivity. The heat flux density is obtained, and finally, the divergence is used to determine the heat flux density. Calculate the net inflow or outflow of this heat flux in space. This term is related to the heat source term. (Heat production rate per unit volume) together determine the left side (The rate of change of energy per unit volume over time) constitutes the complete energy conservation equation.

[0057] Finally, the system will simulate all time steps covered (e.g., from...). arrive Calculated in hours Connecting them in chronological order forms a continuous curve. This curve visually illustrates the theoretical trajectory of the gradual decline in the device's heat dissipation performance over time under uncorrected initial physical assumptions—that is, the initial predicted curve of heat dissipation performance. This curve serves as the baseline for subsequent comparison with measured data (obtained through impedance inversion), calculation of deviations, and driving model correction.

[0058] S4. Obtain the electrical impedance spectrum signal of the surface deposited layer of the radiator when it is running in a salt spray environment, and analyze the actual change curve of heat dissipation performance based on the electrical impedance spectrum signal.

[0059] In a specific embodiment of the present invention, obtaining the electrical impedance spectrum signal of the surface deposited layer of the radiator when it is running in a salt spray environment, and analyzing an actual change curve of heat dissipation performance based on the electrical impedance spectrum signal, includes: deploying at least one pair of sensing electrodes in a predetermined monitoring area of ​​the radiator, and applying a sweep frequency excitation signal through the sensing electrodes.

[0060] The response signal on the sensing electrode is acquired synchronously, and the complex impedance data sequence that varies with frequency is calculated. The complex impedance data sequence is then used as the impedance spectrum signal.

[0061] Specifically, the purpose of this step is to establish a non-invasive physical sensing channel for real-time acquisition of raw data reflecting the dynamic evolution of salt spray deposits in a real environment. This raw data serves as the physical benchmark for subsequent model calibration and is a crucial bridge connecting the digital twin model and the physical entity. To achieve this, the system first deploys one or more sets of interdigitated sensing electrode pairs made of corrosion-resistant metallic materials, such as gold or platinum, in areas of the heat sink most prone to salt spray deposition and significantly impacting heat dissipation, such as the leeward side at the fin roots, using thin-film deposition or microfabrication processes. An insulating layer is placed between the sensing electrode pairs and the heat sink substrate to ensure that the measurement signal reflects only the physical characteristics of the deposited layer between the electrodes. After deployment, the system periodically performs frequency sweep measurements via an electrochemical workstation or dedicated impedance measurement module connected to the sensing electrode pairs. Specifically, a small-signal sinusoidal excitation voltage with an amplitude between 10 and 50 millivolts is applied, and the scanning frequency range covers logarithmically spaced frequency points from 1 Hz to 1 MHz. At each frequency point, the amplitude and phase of the response current flowing through the electrode pairs are simultaneously measured. By performing vector operations on the excitation voltage and response current, the system calculates a series of complex impedance values. The spectral data composed of these complex impedance values ​​is the electrical impedance spectrum signal. This electrical impedance spectrum signal, as a complete data packet, is defined as an online monitoring signal. Its overall shape is modulated by the instantaneous thickness of the deposition layer, the moisture content caused by changes in environmental humidity, and the ionic conductivity caused by salt precipitation and dissolution, thus comprehensively characterizing the real-time physicochemical state of the deposition layer.

[0062] By fitting the impedance spectrum signal using an equivalent circuit model, characteristic impedance parameters strongly correlated with the accumulation state of the deposition layer are extracted.

[0063] Establish a functional model that maps characteristic impedance parameters to the core temperature of the heat sink.

[0064] The characteristic impedance parameter that changes over time is input into the functional relationship model to calculate the time series of the radiator core temperature, and the time series is used as the actual change curve of the heat dissipation performance.

[0065] Specifically, the purpose of this step is to decode and convert the raw electrical signals, which contain rich physicochemical information and were collected in the previous stage, into engineering parameters that can be directly compared with the results of thermal simulation, namely, the actual operating temperature of the core equipment. This process constitutes a key data processing link from indirect electrical sensing to direct thermal characterization.

[0066] The system first receives periodically acquired impedance spectrum signals and extracts their features. Specifically, the system plots each set of complex impedance spectrum data as a Nyquist plot and performs nonlinear fitting on the plot using a pre-defined equivalent circuit model, such as a Landel circuit model that includes solution resistance, double-layer capacitance, and charge transfer resistance. The core of the fitting process is to extract the charge transfer resistance value most closely related to the physical thickness and moisture content of the deposited layer and use it as the characteristic frequency impedance value. The physical meaning of this characteristic frequency impedance value lies in characterizing the degree of obstruction for ions to pass through the deposited layer to reach the electrode surface, thus sensitively reflecting the cumulative effect of the deposited layer.

[0067] To correlate this electrical characteristic with thermal performance, the system needs to establish a pre-defined mapping model. This model is constructed through a one-time laboratory calibration. Under controlled conditions, a constant thermal load is applied to the same heat sink, and different degrees of salt spray deposition are artificially created. At each stable deposition stage, thermocouples are used to accurately measure the actual junction temperature of the heat sink, and the corresponding characteristic frequency impedance value is measured using the method of this invention. By collecting a series of data pairs, the system uses a multinomial regression or neural network algorithm to establish the mathematical relationship between the two, which can be expressed as: ,in, The desired radiator junction temperature, The characteristic frequency impedance value is measured in real time. This represents the nonlinear function relationship obtained after calibration, i.e., the preset mapping relationship model.

[0068] It should be noted that the preset mapping relationship model It is the core conversion engine connecting electrochemical microscopic sensing and macroscopic thermal performance. Its engineering purpose is to convert electrical characteristic quantities (characteristic frequency impedance values) that reflect the physical accumulation state of salt deposits into electrical characteristic quantities. Real-time decoding into the device's thermal health indicators (radiator junction temperature) Calibration Process and Data Acquisition: This model was constructed through offline laboratory calibration experiments. In a controlled environmental test chamber, a constant rated heat load (e.g., 200W) was applied to the target radiator in a clean state, and accelerated salt spray was introduced. At different stages of deposition layer growth (from clean to heavily covered), the system simultaneously used high-precision thermocouples to measure the actual junction temperature of the radiator core. The characteristic frequency impedance value at the same moment is obtained using the electrochemical impedance measurement module of the present invention. By varying the ambient humidity (e.g., from 60%RH to 95%RH) and thermal load power, a multi-dimensional data set covering the entire operating condition range of "impedance-temperature-ambient environment" is obtained. Mathematical implementation: Based on the aforementioned calibration dataset, a mapping function... This can be achieved using a multinomial regression model or a backpropagation (BP) neural network model. In an embodiment using multinomial regression, its typical expression is: .in, , , This is a calibration coefficient. Its typical value is based on the physical property that salt deposition leads to a nonlinear increase in thermal resistance; for example, in the initial deposition stage... Small changes correspond to large temperature increases, while the curve slope flattens out during the later densification stage. In embodiments employing neural networks, It is constructed as a layer containing an input layer ( and ambient humidity ), hidden layer and output layer ( A nonlinear mapping network is constructed and trained using an algorithm to capture the complex nonlinear characteristics of the contribution of salt fouling layers to thermal resistance at different water saturation levels. Physical meaning and correction logic: function The nonlinear characteristics truly reflect the coupled evolution of the "electrical-thermal" dual physical properties of the salt spray deposition layer. Due to the characteristic impedance... The charge transfer resistance is directly proportional to the effective thickness and porosity of the deposited layer, and these physical parameters directly determine the magnitude of the additional thermal resistance. It can achieve accurate inversion from electrical signals to thermal states. In actual operation, the system will collect data in real time. Input the model, output This serves as a benchmark for the "actual change curve," used for real-time comparison with the "initial predicted curve" of the simulation model, thereby driving subsequent adaptive parameter adjustment algorithms. In this way, the mapping model... It acts as a physical feedback sensor in the digital twin system, ensuring the objectivity and accuracy of the correction process.

[0069] In actual operation, the system will input the sequence of characteristic frequency impedance values ​​obtained online point by point into the preset mapping relationship model. The system performs calculations to obtain a time-varying sequence of radiator junction temperature corresponding to each measurement time point. Finally, the system visualizes or stores this time-series data, forming a continuous curve reflecting the actual performance degradation trajectory of the radiator under real salt spray conditions, for use in subsequent model calibration steps.

[0070] S5. Based on the deviation between the initial predicted curve and the actual change curve, the key control parameters in the sedimentation dynamics model are adjusted in reverse, and the corrected sedimentation seed characteristic parameters are obtained by recalculating based on the adjusted key control parameters.

[0071] In a specific embodiment of the present invention, based on the deviation between the initial prediction curve and the actual change curve, the key control parameters in the sedimentation kinetic model are adjusted in reverse, and the corrected sedimentation seed characteristic parameters are obtained by recalculating based on the adjusted key control parameters. This includes: calculating the difference between the initial prediction curve and the actual change curve at the corresponding time point and generating a deviation sequence.

[0072] The deviation sequence is input into a parameter adaptive adjustment algorithm that aims to minimize the statistical error of the deviation sequence.

[0073] The adhesion efficiency coefficient and crystal growth rate coefficient in the deposition kinetics model are iteratively optimized using a parameter adaptive adjustment algorithm until the statistical error meets the convergence condition, and the corrected deposition seed characteristic parameters are output.

[0074] Specifically, the purpose of this step is to establish an automated closed-loop feedback mechanism that uses measured data from the physical world to drive the self-correction and evolution of the simulation model in the digital world, thereby ensuring the long-term accuracy of the model's predictions and its adaptability to the real environment.

[0075] The process begins with the system synchronously acquiring and comparing two data curves. The system obtains the initial predicted curve from the thermal simulation module and the actual change curve from the online monitoring and analysis module. Subsequently, the system aligns these two curves on the time axis and calculates the difference between them at each discrete time point, forming a time series, i.e., the deviation series.

[0076] Next, the system inputs the deviation sequence into a preset parameter adaptive adjustment algorithm. This algorithm can be implemented in engineering using techniques such as gradient descent, particle swarm optimization, or Kalman filtering. The core objective of the algorithm is to minimize a certain statistic of the deviation sequence, typically the root mean square error. To this end, the algorithm sets two key physical parameters from the deposition seed characteristic parameters—the adhesion efficiency coefficient and the crystallization growth rate coefficient—as adjustable variables. These two coefficients directly affect the growth rate and morphology of the deposition layer in the simulation model. The optimization process of the algorithm can be formally expressed as: ,in, This represents the optimal combination of parameters to be found. It is a parameter vector that includes the adhesion efficiency coefficient and the crystal growth rate coefficient. It is the objective function, defined as the root mean square error of the deviation sequence, where the root mean square error is a parameter. implicit functions, because Changes in this will directly lead to changes in the initial prediction curve. It is a mathematical operator, meaning "the parameter value that minimizes the objective function". The result of this expression is to make the objective function The parameter that reaches this minimum value The value of .

[0077] The algorithm optimizes itself iteratively. In each iteration, the algorithm optimizes the parameter vector. A minor adjustment is made, triggering the simulation module to regenerate a new prediction curve. Then, the new bias sequence and its root mean square error are recalculated. Based on the error trend, the algorithm determines the next adjustment to the parameter vector. The direction and step size are adjusted. This iterative process continues until the statistic of the deviation sequence, i.e., the root mean square error, is reduced to below a preset tolerance threshold, such as less than 1% of the noise level of the measured data, or until the preset maximum number of iterations is reached. When the optimization converges, the algorithm outputs the parameter vector at this point. The parameter values ​​are the corrected sedimentation seed feature parameters, which are used to update and solidify the simulation model so that its predicted behavior is aligned with physical reality to the greatest extent.

[0078] S6. Update the multiphysics coupling model using the corrected sedimentation seed characteristic parameters, and output the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model.

[0079] It should be noted that the core of the step of updating the multiphysics coupling model using the corrected deposition seed characteristic parameters lies in transforming the mathematically optimized control parameters into spatial field variables that can be identified at the physical simulation level, thereby achieving state synchronization between the digital twin and the physical entity. The purpose of this step is to eliminate prediction errors caused by parameter assumption biases in the initial model. In the preceding steps, the system has already obtained the corrected key control parameters (e.g., the corrected adhesion efficiency coefficient) through an adaptive algorithm. and crystal growth rate coefficient However, these parameters are merely numerical values ​​and cannot be directly used for heat transfer calculations; they must undergo physical mapping.

[0080] The system performs the following specific operations: Microstructure reconstruction: The system first calls the sedimentation kinetics model and inputs the corrected parameters. and The model is rerun with Lagrange particle transport calculations or a probabilistic growth algorithm. Because... Changes in the concentration of salt spray particles alter the location and quantity of salt spray particles on the radiator surface, leading to changes in the macroscopic thickness distribution of the deposited layer. Due to... The changes in the density of crystal packing within the sedimentary layer alter the microscopic porosity. Changes have occurred. This step outputs the updated initial pore structure characteristics. Macroscopic property recalculation: based on the updated porosity. The system calls the porous media physics formula to recalculate macroscopic properties: Thermal property update: Using the mixed thermal conductivity model (such as the aforementioned weighted average formula), a new equivalent thermal conductivity is calculated. For example, if the corrected porosity decreases (the deposition is denser), the proportion of heat transfer in the solid phase increases, and the equivalent thermal conductivity rises. Electrical property updates: utilizing the aforementioned... The formula is used to calculate the new equivalent conductivity. This step generates a corrected material property field that varies with spatial location. Mesh property mapping: In the simulation software, the system locks the computational domain of the virtual deposition layer on the heat sink surface. The system traverses each finite element mesh node or element in this computational domain and maps the above-calculated properties. and The values ​​are assigned to the corresponding mesh cells, overwriting the old initial values. Through this process, the multiphysics coupling model not only adjusts a global coefficient, but also completes a comprehensive physical reset from geometry to material properties. The updated model can accurately reflect the specific state of the deposition layer at the current moment (e.g., "very dense despite its thinness" or "extremely porous despite its thickness"), thus ensuring that the subsequent corrected heat dissipation performance prediction results have extremely high physical fidelity.

[0081] In a specific embodiment of the present invention, after outputting the corrected heat dissipation performance prediction result based on the updated multiphysics coupling model, the method further includes: substituting a set of different radiator surface morphology parameters into the updated multiphysics coupling model.

[0082] Specifically, the engineering objective of this step is to utilize the measured data to calibrate and validate the simulation model, explore the virtual design space, and proactively seek optimized design solutions that can effectively suppress the effects of salt spray deposition or delay the degradation of heat dissipation performance. This elevates the system's function from passive performance prediction to proactive guidance for innovative design.

[0083] The process begins with parameterization within the updated multiphysics coupling model. The system parameterizes key microscopic features of the radiator surface geometry, creating an adjustable set of design variables—the radiator surface topography parameters. These parameters specifically include the root mean square roughness and average width of the profile elements, defining surface roughness, as well as the dimensions of microgrooves that may be present in a particular design, such as groove width, depth, and spacing. Based on engineering experience or optimization algorithm settings, the system generates a set of discrete combinations of surface topography parameters covering the design space.

[0084] For each set of heat sink surface morphology parameters, the updated multiphysics coupling model is driven to perform long-term service simulation, generating the corresponding heat dissipation performance degradation trajectory.

[0085] In a specific embodiment of the present invention, the updated multiphysics coupling model is driven to perform long-term service simulation to generate a corresponding heat dissipation performance degradation trajectory, including: loading an accelerated environment spectrum containing periodic temperature and humidity changes as simulation input.

[0086] At each time step of the simulation process, the dynamic moisture absorption of the deposition layer is calculated using a multiphysics coupling model based on the current environmental conditions in the accelerated environment spectrum.

[0087] The equivalent thermal conductivity of the deposited layer is updated in real time based on the dynamic moisture absorption, and the instantaneous core temperature of the radiator is calculated based on the updated equivalent thermal conductivity, thus forming a heat dissipation performance degradation trajectory.

[0088] Specifically, the purpose of this step is to provide a detailed technical implementation path for the proposed virtual simulation process. It precisely defines how to simulate the complex, variable, and harsh environment of the real world in the simulation environment, and couples the calculation of the resulting changes in the physical properties of the deposition layer and their instantaneous impact on heat dissipation performance.

[0089] To achieve this, the system first loads a pre-defined accelerated environmental spectrum. This spectrum is a time-series file that defines the dynamic changes in ambient temperature, relative humidity, and the on / off state of the salt spray system over a typical 24-hour period. For example, it simulates high temperature and low humidity during the day, condensation due to cooling and high humidity at night, and intermittent exposure to high concentrations of salt spray. This spectrum computationally accelerates aging by compressing the wet-dry cycle and salt spray accumulation process that takes weeks or months to complete in nature.

[0090] Within each time step of the simulation, the system first applies the current ambient humidity value from the accelerated environment spectrum as input to the multiphysics coupling model. The multiphysics coupling model then calculates the water saturation of each grid cell within the deposition layer by solving a pre-defined capillary adsorption relationship, thus dynamically simulating the dynamic moisture absorption-drying process. Following this, the system updates the equivalent thermal conductivity at each point within the deposition layer in real time based on changes in this saturation. This calculation follows a mixed-medium thermal conductivity model, the simplified form of which can be expressed as… ,in, It is the equivalent thermal conductivity to be determined. , and These are the intrinsic thermal conductivity coefficients of the salt crystal solid framework, adsorbed water, and air within the pores, respectively, all of which are preset material property parameters. It's porosity. This represents the water saturation at the current time step. The formula describes the change in the equivalent thermal conductivity, meaning the total thermal conductivity of the deposited layer is a weighted average of the solid, liquid, and gaseous components according to their volume fractions.

[0091] It should be noted that water saturation ( The specific calculation steps are as follows: First, the system calculates the relative humidity of the environment at the current moment based on the acceleration environment spectrum. The capillary pressure (or matrix suction) within the pores of the sedimentary layer was calculated using the Kelvin equation. : in, It is the density of water. It is the ideal gas constant. It is the ambient temperature. This is the molar mass of water. Then, the calculated capillary pressure... Substituting into the inverse function of the relation, the water saturation is obtained. : in, and These are two key fitting parameters of the model, characterizing the pore structure of the sedimentary layer (such as pore size distribution and tortuosity), and were obtained through laboratory calibration using pressure plate experiments or centrifuge experiments on samples of salt-contaminated layers. Through this series of calculations, the model can accurately convert macroscopic changes in ambient humidity into dynamic changes in water content within the micropores of the sedimentary layer, providing crucial input for subsequent calculations of equivalent thermal conductivity.

[0092] It should also be noted that the intrinsic thermal conductivity ( The typical values ​​and their basis for selection are as follows: In the mixed-medium thermal conductivity model, the intrinsic thermal conductivity of each component is a parameter preset according to the standard material physical property database. The typical values ​​and their basis for selection are as follows: (Intrinsic thermal conductivity of the salt crystal solid framework): Typical value: For salt spray deposits with sodium chloride (NaCl) as the main component, the thermal conductivity in the crystalline state is approximately 6.5 W / (m·K) near room temperature. Basis for value: This value is derived from publicly available, authoritative materials physics handbooks or academic databases (such as the NIST Chemicals Database). It represents the thermal conductivity of the solid framework material itself, and is the component with the best thermal conductivity among the three phases. (Intrinsic thermal conductivity of adsorbed water): Typical value: Under standard atmospheric pressure and common operating temperature (e.g., 25°C), the intrinsic thermal conductivity of liquid water is approximately 0.6 W / (m·K). Basis for value: This value is also derived from standard thermophysical property tables (such as the International Standard for Water and Steam Properties published by IAPWS). In the model, it represents the thermal conductivity of the liquid phase (containing salt water) filling the pores. Although salt dissolution slightly affects the thermal conductivity of water, its effect is negligible at low concentrations; therefore, the thermal conductivity of pure water is used as an engineering approximation. (Intrinsic thermal conductivity of air within pores): Typical value: At standard atmospheric pressure and common operating temperatures (e.g., 25°C), the intrinsic thermal conductivity of dry air is approximately 0.026 W / (m·K). Basis for value: This value is derived from standard tables of thermophysical properties of gases. It represents the thermal conductivity of the gas phase in pores not filled with water. Since air is a poor conductor of heat, with a thermal conductivity much lower than that of water and salt crystals, the presence of air within pores is the primary cause of increased thermal resistance in dry salt deposits.

[0093] After updating the thermal property distribution of the entire deposition layer, the system immediately performs a steady-state or transient coupled heat conduction and fluid dynamics calculation within that time step. This calculation, under a given heat source power, solves for the temperature field distribution across the entire computational domain, including the heat sink and the dynamically deposited layer. Through this calculation, the system accurately obtains key temperature measurement points on the heat sink, such as the instantaneous junction temperature at the chip junction.

[0094] The system iteratively executes the above steps of "loading the environmental spectrum - calculating saturation - updating thermal conductivity - solving the temperature field" throughout the set simulation duration. By recording the instantaneous junction temperature at each time step, a complete heat dissipation performance degradation trajectory that reflects the gradual decline in heat dissipation performance under complex dynamic environments is ultimately generated for subsequent scheme evaluation and selection.

[0095] Compare the heat dissipation performance degradation trajectories and select the target surface morphology parameters that minimize the performance degradation rate.

[0096] Specifically, after the simulation, the system quantitatively evaluates the heat dissipation performance degradation trajectory generated for each design scheme. The evaluation metric typically defines a lifespan endpoint condition, such as the heat sink junction temperature rising above a preset threshold (e.g., 20 degrees Celsius) compared to its initial state. The system calculates the time required for each design scheme to reach this lifespan endpoint, or calculates its long-term heat dissipation performance degradation rate at the end of a fixed service life.

[0097] It should be noted that the long-term thermal performance degradation rate is a dimensionless indicator used to quantify and compare the degree of decline in thermal performance of different design schemes at the end of a fixed service life. Its specific calculation method is as follows: First, the system generates a thermal performance degradation trajectory (i.e., a curve showing the junction temperature changing over time) for each design scheme. From the data, two key temperature values ​​were extracted: the initial junction temperature. Simulation start time ( The final junction temperature represents the baseline heat dissipation performance under clean conditions. The junction temperature at the end of a preset fixed service life (e.g., 5000 hours) represents the heat dissipation performance after long-term salt spray deposition. The system then calculates the total temperature rise due to salt spray deposition throughout the entire service life. : Finally, the long-term thermal performance degradation rate (TPDR) is defined as the total temperature rise of the design scheme relative to the worst-performing scheme among all tested schemes (i.e., the baseline scheme or the scheme with the highest total temperature rise). Percentage: Physical Meaning and Application: The physical meaning of the TPDR index lies in its normalization of the absolute temperature rise values ​​of different schemes, making it a relative and easily comparable measure of performance degradation. A lower TPDR value indicates a stronger ability of the design scheme to resist the effects of salt spray deposition and maintain heat dissipation performance during long-term service. For example, a scheme with a TPDR of 71.5% means that, under the same service life and environment, its additional temperature rise due to salt spray deposition is only 71.5% of that of the worst-case scheme. By calculating and comparing the TPDR values ​​of various schemes, the system can objectively and quantitatively screen the target surface morphology parameters that minimize the performance degradation rate, providing direct data support for the optimized design of heat sinks.

[0098] Finally, the system ranks and compares the evaluation results of all design schemes. Through screening, it determines the design scheme that minimizes the long-term heat dissipation performance degradation rate or maximizes the time to reach the end of its lifespan, and outputs the corresponding heat sink surface morphology parameters as target surface morphology parameters. This output provides clear and quantitative design guidance for subsequent heat sink product development and manufacturing, and can be directly used to guide the parameter setting of surface treatment processes or micromachining processes.

[0099] In a specific embodiment of the present invention, a continuous evolution process is also included: during long-term operation of the system, the steps of acquiring electrical impedance spectrum signals to update the multiphysics coupling model are repeatedly executed, key control parameters are continuously iterated, and historical data are recorded.

[0100] Establish a rule-based association model between historical data of key control parameters and environmental statistical characteristics within the corresponding time period.

[0101] When faced with a new work environment, its environmental statistical characteristics are obtained, and the initial values ​​of key control parameters applicable to the new environment are directly deduced through association rule model.

[0102] Specifically, the purpose of this step is to endow the entire prediction and correction system with the ability to learn and self-evolve over a long period of time. The system will no longer be fixed after a single calibration, but will be able to continuously learn from new environmental data during continuous operation, optimize the underlying logic of its own model, and thus achieve faster and more accurate "prior" predictions of different future environmental conditions.

[0103] To achieve this, the system periodically repeats the core closed-loop calibration process—from acquiring online monitoring signals to predicting heat dissipation performance—over a longer timescale, such as weeks or months. With each complete closed-loop calibration, the system uses the latest acquired actual change curve to iteratively adjust and optimize the control parameters among the deposition seed characteristic parameters, namely the adhesion efficiency coefficient and the crystal growth rate coefficient. The system establishes a historical database specifically for recording historical data of the control parameters that have stabilized after each iterative adjustment, and stores this data in association with the long-term statistical characteristics of the salt spray environment within that calibration cycle. These long-term statistical characteristics include the average salt spray concentration for that cycle, the statistical variance of temperature and humidity fluctuations, and the frequency of strong wind events.

[0104] As historical database data pairs accumulate, the system initiates an association rule mining algorithm in the background, such as a decision tree-based algorithm. The goal of this algorithm is to learn and build association rules from this data. These rules aim to reveal the potential nonlinear mapping relationship between the long-term statistical characteristics of the salt spray environment and the optimal control parameters. Its mathematical essence is to construct an empirical model: ,in, It is a parameter vector that includes the adhesion efficiency coefficient and the crystal growth rate coefficient. It is a vector that contains the aforementioned environmental statistical characteristics. It is a function model that the algorithm fits by learning from historical data.

[0105] It should be noted that the function model This is the core of the "continuous evolution step" in this invention, aiming to establish a rapid derivation channel from macroscopic environmental statistical characteristics to the optimal parameters of a microscopic sedimentary model. This model, through machine learning algorithms, learns and solidifies the complex nonlinear relationship between the environment and sedimentary physical processes from long-term accumulated historical data, thereby achieving "prior" predictions of new environments. Model construction and mathematical implementation: Function model. In engineering, various machine learning regression algorithms can be used to construct the model. In this embodiment, a gradient boosting decision tree model or a deep neural network model is preferred. Input vector E: The input to the model is an environmental statistical feature vector, the dimension of which is determined by the selected features. For example, it can be a vector containing 5 elements. Output vector P: The model's output is the key control parameter vector of the sedimentation kinetics model, i.e. Model Structure: If a gradient boosting decision tree model is used, model G consists of hundreds of sequentially trained decision trees, each dedicated to fitting the residual of the previous tree. If a deep neural network model is used, model G is a fully connected network with multiple hidden layers. By introducing nonlinearity through activation functions (such as ReLU), it can fit extremely complex functional relationships.

[0106] Training Process: The fitting (i.e., training) process of the function model G is based on data pairs (environmental features E, optimal parameters P) accumulated in the historical database. The system periodically executes the following training process in the background: Data Preparation: Extracting data pairs from all records in the historical database to form a training set. Model Training: Inputting the training set into the selected machine learning algorithm. The algorithm iteratively adjusts the internal parameters of model G (such as the structure and split points of a decision tree, or the weights and biases of a neural network) by minimizing a loss function (such as mean squared error MSE). The loss function is defined as: in It is the number of samples in the training set. It is the first The true optimal parameters for each sample (obtained through closed-loop calibration). The model is for the first Predicted parameters for each environmental feature. Model validation and update: After training, the generalization ability of the model is evaluated using methods such as cross-validation. If the performance of the new model is better than the old model, the system updates it to the currently effective function model G. Engineering applications: Once the function model G is successfully trained, it becomes an efficient "experience surrogate model". When the system is deployed to a new environment, it no longer needs to undergo a lengthy, zero-based closed-loop calibration process. The system only needs to run in the new environment for a short period of time (e.g., one week) to collect enough environmental data to calculate the statistical feature vector of the new environment. Then, the function model G is directly called to perform a forward computation: Received It is a high-quality initial parameter derived from historical experience and applicable to the new environment. The system utilizes this... This is used to initialize the deposition dynamics model, thereby enabling the rapid construction of a prediction model that is highly matched to the new environment even without measured decay data, which greatly improves the model's generalization ability and deployment efficiency.

[0107] Once the association rules are established and verified to be effective, the system possesses evolutionary capabilities. When the cooling system is deployed to a completely new application scenario, the system can first monitor the new environment for a period of time to obtain its long-term statistical characteristics of the salt spray environment. Then, instead of waiting for the lengthy online monitoring and closed-loop calibration process, it can be directly... Input to function In the process, a highly reliable initial parameter vector is calculated through deduction. The system utilizes this This allows for the direct initialization of sedimentation seed feature parameters, enabling the rapid construction of a prediction model that highly matches the new environment even without measured attenuation data, thus greatly improving the model's generalization ability and deployment efficiency.

[0108] Reference Figure 2 The second aspect of the present invention provides a heat dissipation prediction and correction system based on a salt spray deposition model, comprising: an environment and deposition sensing module, a multiphysics model modeling module, a multiphysics heat transfer simulation module, an online state monitoring and analysis module, a model calibration and evolution module, and a prediction and design optimization module.

[0109] The environment and sedimentation sensing module is connected to the multiphysics modeling module. The multiphysics heat transfer simulation module and the online condition monitoring and analysis module are both connected to the model calibration and evolution module. The multiphysics modeling module and the model calibration and evolution module are both connected to the prediction and design optimization module.

[0110] The environment and sedimentation sensing module collects fluid dynamics parameters and droplet physical property parameters in the salt spray environment, and establishes a sedimentation dynamics model based on the fluid dynamics parameters and droplet physical property parameters to generate sedimentation seed characteristic parameters.

[0111] The multiphysics modeling module constructs a multiphysics coupled model that combines thermal conduction, moisture migration, and electrical conductivity effects based on the characteristic parameters of the sediment seed.

[0112] The multiphysics heat transfer simulation module drives the multiphysics coupled model to perform heat transfer simulation and generates an initial prediction curve of heat dissipation performance.

[0113] The online status monitoring and analysis module acquires the electrical impedance spectrum signal of the surface deposited layer of the radiator when it is running in a salt spray environment, and analyzes a curve showing the actual change in heat dissipation performance based on the electrical impedance spectrum signal.

[0114] The model calibration and evolution module adjusts the key control parameters in the sedimentation kinetics model in reverse based on the deviation between the initial predicted curve and the actual change curve, and recalculates based on the adjusted key control parameters to obtain the corrected sedimentation seed characteristic parameters.

[0115] The prediction and design optimization module updates the multiphysics coupling model using the corrected deposition seed characteristic parameters, and outputs the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model.

[0116] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A heat dissipation prediction and correction method based on a salt spray deposition model, characterized in that, include: S1. Collect fluid dynamics parameters and droplet physical property parameters in the salt spray environment, and establish a deposition dynamics model based on the fluid dynamics parameters and droplet physical property parameters to generate deposition seed characteristic parameters; S2. Based on the characteristic parameters of the sediment seed, construct a multi-physics coupled model that combines thermal conduction, moisture migration and electrical conductivity effects; S3. Drive the multiphysics coupling model to perform heat transfer simulation and generate an initial prediction curve of heat dissipation performance; S4. Obtain the electrical impedance spectrum signal of the surface deposited layer of the radiator when it is running in a salt spray environment, and analyze the actual change curve of heat dissipation performance based on the electrical impedance spectrum signal. S5. Based on the deviation between the initial predicted curve and the actual change curve, the key control parameters in the sedimentation dynamics model are adjusted in reverse, and the corrected sedimentation seed characteristic parameters are obtained by recalculating based on the adjusted key control parameters. S6. Update the multiphysics coupling model using the corrected sedimentation seed characteristic parameters, and output the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model.

2. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 1, characterized in that, The deposition dynamics model established based on fluid dynamics parameters and droplet physical property parameters generates deposition seed characteristic parameters, including: Within the observation channel connected to the radiator environment, airflow velocity field data and temperature and humidity data are acquired to form fluid dynamic parameters; Within the observation channel, data on the droplet size distribution and charge characteristics of salt spray droplets are acquired to form droplet physical property parameters; The fluid dynamics parameters and droplet physical properties parameters are input into the Lagrange model used to simulate the transport and attachment of particles in the flow field. The spatial distribution of attachment efficiency and the initial pore structure characteristics are calculated, and the spatial distribution of attachment efficiency and the initial pore structure characteristics are used together as the deposition seed characteristic parameters.

3. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 2, characterized in that, The process involves constructing a multiphysics coupled model based on the characteristic parameters of the deposited seeds, which integrates thermal conduction, moisture migration, and electrical conductivity effects. Based on the initial pore structure characteristics in the sediment seed feature parameters, the equivalent porous media network parameters of the sediment layer are defined; A capillary adsorption model for describing the behavior of liquid films in pores is established based on the equivalent porous media network parameters, and parameters for moisture storage and migration are generated. By combining moisture storage and migration parameters with ion ionization model, the equivalent conductivity distribution parameters of the deposition layer are calculated. The equivalent porous media network parameters, moisture storage and migration parameters, and equivalent conductivity distribution parameters are integrated with the geometric model of the heat sink to form a multiphysics coupling model.

4. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 1, characterized in that, The driving multiphysics coupling model performs heat transfer simulation and generates an initial prediction curve for heat dissipation performance, including: A discrete time step sequence is defined, and within each time step, the instantaneous physical state of the deposition layer, including the instantaneous thickness field and the instantaneous equivalent thermal conductivity distribution, is calculated using a multiphysics coupling model. The instantaneous physical state is applied as a dynamic thermal resistance boundary condition to the surface mesh of the heat sink, and the transient heat conduction equation is solved to obtain the predicted temperature value of the critical node of the heat sink at the current time step. By concatenating the predicted temperature values ​​for all time steps in chronological order, an initial prediction curve of the heat dissipation performance as it evolves over time is generated.

5. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 1, characterized in that, The process of acquiring the electrical impedance spectrum signal of the surface deposited layer of the heat sink during operation in a salt spray environment, and analyzing a curve showing the actual change in heat dissipation performance based on the electrical impedance spectrum signal, includes: At least one pair of sensing electrodes are deployed in the predetermined monitoring area of ​​the radiator, and a sweep frequency excitation signal is applied through the sensing electrodes; The response signal on the sensing electrode is acquired synchronously, and the complex impedance data sequence that varies with frequency is calculated. The complex impedance data sequence is used as the impedance spectrum signal. The impedance spectrum signal was fitted by an equivalent circuit model to extract characteristic impedance parameters that are strongly correlated with the accumulation state of the deposition layer. Establish a functional model that maps characteristic impedance parameters to the core temperature of the heat sink; The characteristic impedance parameter that changes over time is input into the functional relationship model to calculate the time series of the radiator core temperature, and the time series is used as the actual change curve of the heat dissipation performance.

6. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 1, characterized in that, The process involves adjusting the key control parameters in the sedimentation kinetics model in reverse based on the deviation between the initial predicted curve and the actual change curve, and recalculating based on the adjusted key control parameters to obtain the corrected sedimentation seed characteristic parameters, including: Calculate the difference between the initial prediction curve and the actual change curve at corresponding time points to generate a deviation sequence; The deviation sequence is input into a parameter adaptive adjustment algorithm that aims to minimize the statistical error of the deviation sequence; The adhesion efficiency coefficient and crystal growth rate coefficient in the deposition kinetics model are iteratively optimized using a parameter adaptive adjustment algorithm until the statistical error meets the convergence condition, and the corrected deposition seed characteristic parameters are output.

7. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 1, characterized in that, After outputting the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model, the following are also included: Substitute a set of different radiator surface morphology parameters into the updated multiphysics coupling model; For each set of radiator surface morphology parameters, the updated multiphysics coupling model is driven to perform long-term service simulation to generate the corresponding heat dissipation performance degradation trajectory. Compare the heat dissipation performance degradation trajectories and select the target surface morphology parameters that minimize the performance degradation rate.

8. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 7, characterized in that, The updated multiphysics coupling model is used for long-term service simulation to generate corresponding heat dissipation performance degradation trajectories, including: Load an accelerated environment spectrum containing periodic temperature and humidity changes as simulation input; At each time step of the simulation process, the dynamic moisture absorption of the deposition layer is calculated using a multiphysics coupling model based on the current environmental conditions in the acceleration environment spectrum. The equivalent thermal conductivity of the deposited layer is updated in real time based on the dynamic moisture absorption, and the instantaneous core temperature of the radiator is calculated based on the updated equivalent thermal conductivity, thus forming a heat dissipation performance degradation trajectory.

9. The heat dissipation prediction and correction method based on the salt spray deposition model according to claim 1, characterized in that, It also includes a continuous evolutionary process: During long-term system operation, the steps of acquiring electrical impedance spectrum signals and updating the multiphysics coupling model are repeatedly executed, key control parameters are continuously iterated, and historical data are recorded. Establish a rule model linking historical data of key control parameters with environmental statistical characteristics within the corresponding time period; When faced with a new work environment, its environmental statistical characteristics are obtained, and the initial values ​​of key control parameters applicable to the new environment are directly deduced through association rule model.

10. A heat dissipation prediction and correction system based on a salt spray deposition model, applied to the heat dissipation prediction and correction method based on a salt spray deposition model as described in any one of claims 1-9, characterized in that, include: The environment and sedimentation sensing module collects fluid dynamics parameters and droplet physical property parameters in the salt spray environment, and establishes a sedimentation dynamics model based on the fluid dynamics parameters and droplet physical property parameters to generate sedimentation seed characteristic parameters; The multiphysics modeling module constructs a multiphysics coupled model that combines thermal conduction, moisture migration, and electrical conductivity effects based on the characteristic parameters of the sediment seed. The multiphysics heat transfer simulation module drives the multiphysics coupled model to perform heat transfer simulation and generates an initial prediction curve of heat dissipation performance. The online status monitoring and analysis module acquires the electrical impedance spectrum signal of the surface deposited layer of the heat sink when it is running in a salt spray environment, and analyzes a curve showing the actual change in heat dissipation performance based on the electrical impedance spectrum signal. The model calibration and evolution module adjusts the key control parameters in the sedimentation kinetics model in reverse based on the deviation between the initial prediction curve and the actual change curve, and recalculates based on the adjusted key control parameters to obtain the corrected sedimentation seed characteristic parameters. The prediction and design optimization module updates the multiphysics coupling model using the corrected deposition seed characteristic parameters, and outputs the corrected heat dissipation performance prediction results based on the updated multiphysics coupling model.

Images