Methods, apparatus, equipment and storage media for reverse measurement of thermophysical parameters

Abstract

This invention relates to the field of thermophysical parameter measurement technology for quartz glass, and particularly to a method, apparatus, device, and storage medium for reverse measurement of thermophysical parameters. The method first obtains an initial set of thermophysical parameters and a measured temperature, then calls a pre-trained temperature field model to predict the temperature of the initial set of thermophysical parameters, obtaining an initial predicted temperature. Next, a fitness function is constructed based on the measured temperature and the initial predicted temperature. Then, a beetle whisker algorithm is used to iteratively search and update the initial set of thermophysical parameters based on the fitness function. Finally, when a preset iteration stopping condition is met, the updated set of thermophysical parameters is input into the temperature field model for temperature prediction, outputting the target predicted temperature. This solves the problem of existing technologies being unable to obtain high-precision three-dimensional temperature fields in real time.

Description

Technical Field

[0001] This invention relates to the field of thermophysical parameter measurement technology for quartz glass, and particularly to a method, apparatus, equipment, and storage medium for reverse measurement of thermophysical parameters. Background Technology

[0002] Quartz glass, with its excellent optical transparency, high hardness, and chemical stability, has become a core material for key components such as the printhead microperforation plate and optical alignment module in inkjet-printed OLED (organic light-emitting diode) manufacturing. However, its high brittleness and hardness make traditional ultra-precision grinding prone to surface and subsurface damage, resulting in low processing efficiency and making it difficult to meet the mass production requirements of large-size, complex microstructures. Laser-assisted diamond turning, as an advanced processing technology, reduces material brittleness through local laser preheating, enabling high surface integrity and geometric precision processing, providing support for efficient OLED equipment manufacturing. However, it still faces key technical bottlenecks in practical applications.

[0003] The thermophysical properties of quartz glass exhibit significant temperature dependence, while the temperature field changes drastically during laser-assisted processing, making it difficult for traditional thermal models to accurately predict temperature distribution. Current methods for measuring thermophysical properties primarily rely on experimental measurements, which not only require expensive specialized equipment but also present significant challenges under complex conditions such as high temperatures and small samples. Furthermore, these methods are susceptible to interference from various factors, resulting in poor data repeatability and hindering the construction of high-precision thermal analysis models. Regarding temperature field acquisition, thermocouples can only provide limited point-based information, while infrared thermal imaging is limited by spatial resolution and environmental interference, making it difficult to capture temperature changes in key microscopic regions, thus restricting the application of thermal error modeling and compensation. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, the present invention aims to provide a method, device, equipment and storage medium for reverse measurement of thermophysical parameters, in order to solve the problem that the prior art cannot obtain high-precision three-dimensional temperature fields in real time.

[0005] The first aspect of this invention provides a method for inverse measurement of thermal property parameters, comprising: acquiring an initial set of thermal property parameters and a measured temperature; calling a pre-trained temperature field model to predict the temperature of the initial set of thermal property parameters to obtain an initial predicted temperature; constructing a fitness function based on the measured temperature and the initial predicted temperature; using a beetle whisker algorithm to iteratively search and update the initial set of thermal property parameters based on the fitness function; and when a preset iteration stopping condition is met, inputting the updated set of thermal property parameters into the temperature field model to predict the temperature, thereby outputting a target predicted temperature.

[0006] Optionally, in a first implementation of the first aspect of the present invention, before obtaining the initial set of thermal property parameters and the measured temperature, the method further includes: obtaining a preset search range for thermal property parameters; and using a beetle whisker algorithm to perform random sampling based on the search range for thermal property parameters to obtain the initial set of thermal property parameters.

[0007] Optionally, in a second implementation of the first aspect of the present invention, before calling the pre-trained temperature field model to predict the temperature of the initial thermophysical parameter set and obtain the initial predicted temperature, the method further includes: constructing an initial temperature field model using the COMSOL tool and performing format conversion processing on the initial temperature field model to obtain a MATLAB-adapted model; performing function reconstruction processing on the MATLAB-adapted model using the MATLAB (Matrix Laboratory) tool to obtain a MATLAB function model; constructing a training dataset using the MATLAB tool, and performing multiple rounds of iterative training on the MATLAB function model based on the training dataset to obtain the pre-trained temperature field model.

[0008] Optionally, in a third implementation of the first aspect of the present invention, the step of constructing a fitness function based on the measured temperature and the initial predicted temperature includes: obtaining a preset amplification factor, a preset time range, and a preset power range; calculating a relative error between the measured temperature and the initial predicted temperature based on the time range and the power range to obtain a relative error; calculating the mean error based on the relative error to obtain an average relative error; performing correlation calculations on the measured temperature and the initial predicted temperature to obtain a coefficient of determination; and constructing the fitness function based on the amplification factor, the relative error, the average relative error, and the coefficient of determination.

[0009] Optionally, in the fourth implementation of the first aspect of the present invention, the step of employing the beetle whisker algorithm to iteratively search and update the initial thermal property parameter set based on the fitness function includes: obtaining the beetle's initial position, initial step size, and initial search distance based on the initial thermal property parameter set; determining the left and right antenna parameter vectors based on the beetle's initial position and the initial search distance; calculating antenna comfort based on the fitness function, the left and right antenna parameter vectors, to obtain the left and right antenna fitness; calculating the fitness difference based on the left and right antenna fitness to obtain the fitness difference; determining the parameter update direction based on the fitness difference; and iteratively searching and updating the initial thermal property parameter set based on the beetle's initial position, initial step size, initial search distance, and parameter update direction.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, determining the left and right antenna parameter vectors based on the initial position of the longhorn beetle and the initial search distance includes: obtaining a unit random vector; performing antenna normalization position calculation based on the unit random vector, the initial position of the longhorn beetle, and the initial search distance to obtain the normalized position of the left and right antennas; and performing inverse normalization calculation based on the normalized position of the left and right antennas to obtain the left and right antenna parameter vectors.

[0011] Optionally, in a sixth implementation of the first aspect of the present invention, the iterative search and update of the initial thermal property parameter set based on the initial position of the longhorn beetle, the initial step size, the initial search distance, and the parameter update direction includes: obtaining a preset step size decay coefficient and a preset step size ratio coefficient; updating the initial step size based on the step size decay coefficient to obtain the current step size; updating the initial search distance based on the step size ratio coefficient and the current step size to obtain the current search distance; and iteratively searching and updating the initial thermal property parameter set based on the initial position of the longhorn beetle, the current step size, the current search distance, and the parameter update direction.

[0012] A second aspect of the present invention provides a reverse measurement device for thermal property parameters, comprising: a parameter acquisition module for acquiring an initial set of thermal property parameters and a measured temperature; and a first temperature prediction module for calling a pre-trained temperature field model to predict the temperature of the initial set of thermal property parameters and obtain an initial predicted temperature.

[0013] The function construction module is used to construct a fitness function based on the measured temperature and the initial predicted temperature; the data update module is used to iteratively search and update the initial thermal property parameter set based on the fitness function using the beetle whisker algorithm; the second temperature prediction module is used to input the updated thermal property parameter set into the temperature field model to predict the temperature when the preset iteration stopping condition is met, so as to output the target predicted temperature.

[0014] A third aspect of the present invention provides a reverse measurement device for thermal properties, the reverse measurement device comprising: a memory and at least one processor, the memory storing instructions; the at least one processor calling the instructions in the memory to cause the reverse measurement device for thermal properties to perform the various steps of the reverse measurement method for thermal properties described in any of the preceding claims.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the reverse measurement method for thermal property parameters described in any of the preceding claims.

[0016] In the technical solution of this invention, an initial set of thermal property parameters and a measured temperature are first obtained. A pre-trained temperature field model is then used to predict the temperature of the initial set of thermal property parameters to obtain the initial predicted temperature. A fitness function is then constructed based on the measured temperature and the initial predicted temperature. Next, the beetle whisker algorithm is used to iteratively search and update the initial set of thermal property parameters based on the fitness function. Finally, when a preset iteration stopping condition is met, the updated set of thermal property parameters is input into the temperature field model for temperature prediction to output the target predicted temperature. This solves the problem that existing technologies cannot obtain high-precision three-dimensional temperature fields in real time, improves the prediction accuracy and working condition adaptability of the temperature field model in the manufacturing of key components for inkjet printing, and provides key technical support for intelligent thermal management, microcrack suppression, and processing quality optimization in the ultra-precision machining process of quartz glass. It also helps to ensure the accuracy of the printing nozzle orifice shape, alignment stability, and long-term working life. Attached Figure Description

[0017] Figure 1 A logic flowchart of the reverse measurement method for thermophysical parameters provided in an embodiment of the present invention;

[0018] Figure 2 This is a schematic diagram comparing simulation experimental data provided in the embodiments of the present invention;

[0019] Figure 3 A schematic diagram of the structure of the reverse measurement device for thermal property parameters provided in an embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram of the structure of the reverse measurement device for thermal property parameters provided in an embodiment of the present invention. Detailed Implementation

[0021] This invention provides a method, apparatus, device, and storage medium for reverse measurement of thermophysical parameters. In this invention, the terms "first," "second," "third," "fourth," etc. (if present)," in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0022] To achieve high-precision temperature field simulation for laser-assisted diamond turning of quartz glass and support low-damage processing of key OLED components for inkjet printing, existing temperature field analysis methods typically employ forward modeling. The forward modeling process begins by experimentally determining the thermal properties of the material (e.g., quartz glass), including thermal conductivity, specific heat capacity, absorptivity, and emissivity, which serve as the core input for the simulation model. The measured parameters are then imported into the COMSOL Multiphysics (simulation) platform to construct a three-dimensional thermal simulation model, simulating the temperature field distribution of the workpiece during laser-assisted processing and presenting it as a temperature cloud map. Next, relying on an experimental platform integrating the spindle, workpiece, and laser-assisted system, contact temperature measurement is performed at specific locations on the workpiece surface using devices such as adhesive thermocouples to obtain temperature data from the actual processing. Finally, the simulated temperature curves are compared with the measured temperature data to evaluate the accuracy and effectiveness of the thermal simulation model.

[0023] However, its application has significant limitations. This method is highly dependent on the precise input of thermophysical parameters, while the thermophysical parameters of quartz glass are strongly dependent on temperature and laser wavelength. The dynamic changes under high-temperature processing conditions are difficult to capture accurately through conventional experiments. Existing measurement methods cannot match parameter fluctuations during processing in real time, resulting in a natural deviation between the parameters input to the model and the actual working conditions. At the same time, experimental measurements not only require expensive professional equipment, but are also susceptible to interference from factors such as contact thermal resistance, sample preparation differences, and operational errors under complex conditions such as high temperature and small processing areas. The accuracy and repeatability of the measurement results are difficult to guarantee, and the measurement process is time-consuming and complex, making it difficult to adapt to the high-efficiency requirements of the processing site. More importantly, small errors in thermophysical parameters can be amplified through the coupling effect of heat conduction and laser energy absorption, leading to a significant deviation between the temperature field simulation results and the actual temperature distribution during processing, which greatly weakens the reliability of the forward model.

[0024] Based on the technical problems existing in the above-mentioned forward modeling, this invention proposes a method, apparatus, device and storage medium for reverse measurement of thermophysical parameters.

[0025] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the reverse measurement method for thermophysical parameters in this invention includes:

[0026] 101. Obtain the initial set of thermophysical parameters and the measured temperature;

[0027] In this embodiment, the initial set of thermophysical parameters is randomly generated using a beetle whisker algorithm. This initial set includes four thermophysical parameters: thermal conductivity, specific heat capacity, absorptivity, and emissivity. The process of obtaining the measured temperature requires a combination of workpiece pretreatment, thermocouple placement, and continuous data acquisition. The fused silica workpiece used has a diameter of 25.4 mm and a thickness of 4 mm. To ensure stable thermocouple fixation, a 1 mm wide groove is precisely milled into its surface. The thermocouple is then adhered to the groove using thermally conductive adhesive, maintaining a 3 mm distance between the thermocouple sensing end and the center of a 1064 nm wavelength laser beam with a focused spot diameter of approximately 80 μm. This laser beam is irradiated onto the workpiece surface fixed to the machine tool spindle via a transparent diamond tool. During the experiment, the workpiece and laser beam are kept stationary, and the laser irradiation time is set to 5 minutes. Simultaneously, a data acquisition card is used to record the thermocouple temperature changes in real time from the laser activation until the workpiece naturally cools down, thus completing the acquisition of the measured temperature. The groove design on the workpiece surface ensures the stability of the thermocouple installation, avoids displacement interference during processing, and enhances the thermal contact effect through the use of thermally conductive adhesive, thereby improving the response speed and accuracy of temperature sensing. Furthermore, in addition to using thermocouples for temperature measurement, multi-source data fusion methods such as microstructured optical fibers, thermal imagers, and infrared sensors can also be used for temperature measurement; this application does not limit the scope of the application.

[0028] 102. Call the pre-trained temperature field model to predict the temperature of the initial thermophysical parameter set, and obtain the initial predicted temperature;

[0029] In this embodiment, the pre-trained temperature field model is derived from the COMSOL temperature field model through format conversion and function reconstruction. It can be directly called within the MATLAB environment. Its input parameters include a preset laser power and four parameters to be inverted: thermal conductivity, specific heat capacity, absorptivity, and emissivity, encompassed by the initial set of thermophysical parameters. The joint architecture of MATLAB and COMSOL retains the high precision of COMSOL finite element calculation while offering the convenience of MATLAB. It can quickly respond to the input of the initial set of thermophysical parameters and output stable temperature prediction results, laying a reliable foundation for subsequent parameter iteration updates. The spatial location for temperature prediction is set to 3mm radially from the workpiece center, consistent with the actual thermocouple installation location. This consistency between the model output location and the experimental temperature measurement location allows for direct comparison and analysis between the initial predicted temperature and the measured temperature, providing accurate temperature data support for the subsequent construction of the fitness function. Simultaneously, the laser interaction time in the model is set to 5 minutes according to experimental conditions, ensuring a complete match between the predicted scenario and the actual working conditions. By substituting the initial set of thermophysical parameters into the co-simulation model, and after the model's built-in finite element calculation logic, the initial predicted temperature under the corresponding parameter combination can be output.

[0030] 103. Construct a fitness function based on the measured temperature and the initial predicted temperature;

[0031] In this embodiment, the measured temperature originates from the dynamic temperature data of a specific location on the workpiece recorded by thermocouples during the experiment, while the initial predicted temperature is the temperature data of the same scene and location output by the co-simulation model based on the initial set of thermal property parameters. The two have consistency and comparability in spatiotemporal dimensions, providing a reliable data foundation for the construction of the fitness function. The fitness function, by integrating temperature deviation characteristics and correlation information, transforms the prediction effect corresponding to the initial set of thermal property parameters into a quantifiable numerical index, achieving an objective quantitative evaluation of the temperature prediction effect, avoiding bias caused by subjective judgment, and ensuring the accuracy of the parameter optimization direction.

[0032] 104. Using the longhorn beetle whisker algorithm, the initial set of thermal property parameters is iteratively searched and updated based on the fitness function;

[0033] In this embodiment, the beetle whisker algorithm is used to iteratively search and update the initial set of thermal property parameters based on a fitness function. Essentially, it simulates the antennae-sensing mechanism of a longhorn beetle while foraging, using the quantitative evaluation results of the fitness function as a guide to achieve efficient optimization and dynamic updating of thermal property parameters. First, the initial set of thermal property parameters is mapped to the initial position of the longhorn beetle in the parameter search space. Key parameters such as the initial step size, search distance, and step size decay coefficient are simultaneously set to construct a complete optimization framework. During the iteration process, the beetle whisker algorithm first generates a unit random vector to determine the search direction. Then, it calculates the normalized parameter positions corresponding to the left and right antennae of the longhorn beetle based on the current search distance. After inverse normalization, two different antennae parameter vectors are obtained. These are substituted into the fitness function to calculate the corresponding fitness values. By comparing the fitness differences between the left and right antennae, the direction of parameter optimization is determined. The smaller the fitness value, the better the parameter combination. The algorithm adjusts the longhorn beetle's movement direction accordingly. Subsequently, the current step size is updated based on a preset step size decay coefficient, and the search distance is adjusted synchronously to move the beetle along the optimal direction and update its own position. This process is repeated until the fitness function value meets the preset accuracy requirement (set according to actual needs) or the maximum number of iterations is reached, completing the iterative search and update of parameters. At this point, the parameter combination corresponding to the beetle's position is the iterative set of thermal property parameters. The beetle beetle algorithm achieves direction determination only through the antennae perception of a single individual, without the need for group cooperation. The parameter settings are simple, effectively reducing the computational complexity of the optimization process, and enabling rapid response to changes in the coupling relationship between thermal property parameters and the temperature field. The dynamic decay mechanism of step size and search distance allows the algorithm to perform a large-scale coarse search in the parameter space in the early stages, and gradually focus on the optimal region for a fine search in the later stages, balancing the globality and local accuracy of the optimization and avoiding getting trapped in local optima. Meanwhile, the fitness function’s sensitive feedback to temperature deviation ensures that each iteration of the algorithm aims to reduce prediction error, guiding the parameter combination to continuously converge toward the optimal direction. The resulting parameter set can accurately match the actual working conditions of laser-assisted processing, providing reliable support for high-precision modeling of the temperature field and effectively overcoming the limitations of traditional experimental measurement methods under complex working conditions.

[0034] 105. When the preset iteration stop condition is met, the updated set of thermal property parameters is input into the temperature field model to predict the temperature and output the target predicted temperature.

[0035] In this embodiment, when the iteration process meets the preset stopping condition, i.e., the fitness function value reaches the preset accuracy threshold or the number of iterations reaches the maximum set limit, it means that the thermal property parameter set optimized by the beetle whisker algorithm in multiple rounds has converged to the optimal state. At this time, the updated thermal property parameter set is input into the constructed temperature field model for temperature prediction, and the target predicted temperature is finally output. The updated thermal property parameter set has fully adapted to the variation characteristics of the measured temperature through multiple rounds of iterative search, and its parameters such as thermal conductivity, specific heat capacity, absorptivity, and emissivity are highly consistent with the actual working conditions of laser-assisted processing. The temperature field model outputs the target predicted temperature based on the optimized parameter set, providing direct temperature data reference for the optimization of processing parameters, thermal error compensation, and microcrack suppression in laser-assisted diamond turning, effectively supporting the realization of the high-precision, low-damage processing requirements of quartz glass components.

[0036] In this embodiment of the invention, before obtaining the initial set of thermal property parameters and the measured temperature, the method further includes: obtaining a preset search range for thermal property parameters; and using a beetle whisker algorithm to perform random sampling based on the search range for thermal property parameters to obtain the initial set of thermal property parameters.

[0037] In this embodiment, the search range for thermophysical parameters is determined based on the material physical properties of quartz glass, existing literature data, and engineering experience. Reasonable value ranges for each parameter to be inverted, such as thermal conductivity, specific heat capacity, absorptivity, and emissivity, are clearly defined. This range comprehensively covers the possible dynamic changes of parameters under actual laser-assisted processing conditions while strictly excluding extreme values ​​that do not conform to the laws of material physics, providing clear and scientific constraints for subsequent sampling processes. Based on this, the beetle whisker algorithm is used for random sampling of the initial parameter set. The algorithm generates reasonably distributed random sampling points in the multidimensional parameter space, with a preset search range as a constraint. Each sampling point corresponds to a complete set of thermophysical parameters. After screening, the final initial thermophysical parameter set is formed. The sampling process ensures the physical validity of the parameter combination and, through the algorithm's inherent spatial search logic, enables the sampling points to initially cover key areas of the parameter space, avoiding over-concentration or deviation from the effective range. Random sampling based on the beetle whisker algorithm, compared with traditional random sampling methods, can better balance the randomness and specificity of sampling. The distribution of sampling points in the parameter space is more reasonable, which can provide a higher quality initial starting point for the subsequent iterative search of the beetle whisker algorithm. This helps the algorithm to quickly identify the optimal parameter region and shorten the convergence time.

[0038] In this embodiment of the invention, before calling the pre-trained temperature field model to predict the temperature of the initial thermophysical parameter set and obtain the initial predicted temperature, the method further includes: constructing an initial temperature field model using COMSOL tools and performing format conversion processing on the initial temperature field model to obtain a MATLAB-adapted model; performing function reconstruction processing on the MATLAB-adapted model using MATLAB tools to obtain a MATLAB function model; constructing a training dataset using MATLAB tools, and performing multiple rounds of iterative training on the MATLAB function model based on the training dataset to obtain the pre-trained temperature field model.

[0039] In this embodiment, an initial temperature field model is first constructed using COMSOL. This model accurately replicates the processing environment, workpiece geometry, and heat conduction and energy absorption patterns during laser heating in the laser-assisted processing of quartz glass. Next, the initial temperature field model is converted to a MATLAB-compatible file format, resulting in a MATLAB-compatible model, thus achieving cross-platform integration between COMSOL and MATLAB. Based on the MATLAB-compatible model, MATLAB is used to reconstruct its functions, forming a MATLAB function model. The model has five input parameters: laser power and four thermophysical properties to be inverted: thermal conductivity, specific heat capacity, absorptivity, and emissivity. The predicted output temperature is monitored at a location 3mm radially from the workpiece center, consistent with the thermocouple installation position. The laser interaction time within the model is synchronously set to 5 minutes to ensure consistency between simulation and experimental conditions. Subsequently, a training dataset is constructed using MATLAB, and the MATLAB function model is iteratively trained multiple times based on this dataset. This allows the model to gradually learn and solidify the nonlinear mapping relationship between the input parameters and the temperature output, ultimately yielding a pre-trained temperature field model. The combination of format conversion and function reconstruction not only preserves the high-precision advantage of COMSOL tools in temperature field simulation, but also achieves efficient compatibility between simulation logic and the MATLAB environment through the construction of MATLAB adaptation models and MATLAB function models, thus improving the ease of model calling.

[0040] In this embodiment of the invention, the step of constructing a fitness function based on the measured temperature and the initial predicted temperature includes: obtaining a preset amplification factor, a preset time range, and a preset power range; calculating the relative error between the measured temperature and the initial predicted temperature based on the time range and the power range to obtain a relative error; calculating the mean error based on the relative error to obtain an average relative error; performing correlation calculations on the measured temperature and the initial predicted temperature to obtain a coefficient of determination; and constructing the fitness function based on the amplification factor, the relative error, the average relative error, and the coefficient of determination.

[0041] In this embodiment, before constructing the fitness function, it is necessary to first obtain various preset basic parameters. The preset amplification factor is generally set to 1000, the preset time range is taken from the entire time interval of temperature acquisition during the experiment, and the power range is seven power levels from 2W to 14W, spaced 2W apart. Based on the above time and power ranges, using the initial predicted temperature and the measured temperature as a basis, the relative error is calculated according to the corresponding calculation expression to obtain the relative error under different time and power conditions. The specific calculation expression is as follows:

[0042] ,

[0043] in, Indicates relative error. Represents a set of thermophysical parameters. It represents any moment within the range of time values. This represents any power value within the specified range. Indicates the initial predicted temperature. This indicates the measured temperature.

[0044] Then, the mean error is calculated based on the relative error to obtain the average relative error. The specific calculation expression is as follows:

[0045] ,

[0046] in, Indicates relative error. This represents the average relative error. It represents any moment within the range of time values. This represents any power value within the specified range. Represents a set of thermophysical parameters. This indicates the amount of relative error.

[0047] Next, a correlation calculation is performed based on the measured temperature and the initial predicted temperature to obtain the coefficient of determination. The specific calculation expression is as follows:

[0048] ,

[0049] in, Indicates at a given power and thermophysical property parameter group Under the given conditions, the coefficient of determination of the model prediction results on the measured temperature data is used to measure the goodness of fit of the prediction model to temperature changes. Indicates at time ,power and thermophysical parameter group Under these conditions, the initial predicted temperature is calculated by the model; Indicates at time and power The measured temperature obtained under the given conditions; Indicates at a given power Under the conditions, the measured temperature The arithmetic mean over a time series is defined as follows: , This represents the number of temperature samples taken in the time series corresponding to the same laser power.

[0050] Finally, the fitness function is constructed based on the amplification factor, the relative error, the average relative error, and the determination coefficient. The specific calculation expression is as follows:

[0051] ,

[0052] in, This represents the fitness value corresponding to the current set of thermal property parameters. Represents a set of thermophysical parameters. It represents any moment within the range of time values. This represents any power value within the specified range. This represents the amplification factor, used to amplify the error values ​​of different parameters, thereby improving the algorithm's performance. Indicates relative error. This represents the average relative error. This represents the number of temperature samples taken within a time series corresponding to the same laser power. The coefficient of determination is represented by the coefficient of determination. This indicates the effect on all laser powers. Take the maximum value.

[0053] In this embodiment, the hierarchical calculation of relative error and average relative error captures both the local deviation of a single spatiotemporal operating condition and reflects the overall deviation level under a specific power. Combined with the coefficient of determination's characterization of the fitting degree of the changing trend, it achieves multi-dimensional and accurate quantification of temperature deviation. The introduction of the amplification factor enhances the sensitivity to error differences between different parameter groups. The fitness function proposed in this embodiment can focus on the operating condition with the most significant error, effectively guiding subsequent algorithm iterations towards minimizing the error, improving the accuracy of thermophysical parameter inversion, and adapting to experimental designs with multiple power levels, making the fitness function more closely match the complex operating conditions of actual processing.

[0054] In this embodiment of the invention, the use of the longhorn beetle whisker algorithm to iteratively search and update the initial thermal property parameter set based on the fitness function includes: obtaining the initial position, initial step size, and initial search distance of the longhorn beetle based on the initial thermal property parameter set; determining the left and right antenna parameter vectors based on the initial position and initial search distance of the longhorn beetle; calculating antenna comfort based on the fitness function, the left and right antenna parameter vectors, to obtain the left and right antenna fitness; calculating the fitness difference based on the left and right antenna fitness to obtain the fitness difference; determining the parameter update direction based on the fitness difference; and iteratively searching and updating the initial thermal property parameter set based on the initial position, initial step size, initial search distance, and parameter update direction of the longhorn beetle.

[0055] In this embodiment, the initial set of thermal property parameters is directly mapped to the initial position of the longhorn beetle in the multidimensional parameter space. Each dimensional component of this position corresponds to the initial values ​​of the four parameters to be inverted: thermal conductivity, specific heat capacity, absorptivity, and emissivity. This ensures that the initial optimization position of the longhorn beetle and the initial set of parameters correspond precisely in the parameter space. At the same time, based on the search range of thermal property parameters and engineering practice experience, the initial step size and initial search distance of the algorithm are preset. The initial step size is used to control the magnitude of each parameter update, and the initial search distance defines the spatial distance between the left and right tentacles of the longhorn beetle and the initial position. The values ​​of both need to take into account both the globality and local fineness of the optimization, laying the foundation for efficient subsequent search.

[0056] Based on this, by generating normalized random direction vectors and combining them with the initial position and initial search distance of the longhorn beetle, left and right antenna parameter vectors are constructed. The dimension of the random direction vectors is consistent with the number of parameters to be inverted, and their randomness ensures that the algorithm can conduct unbiased exploration in the parameter space. Subsequently, the left and right antenna parameter vectors are substituted into the constructed fitness function to complete the quantitative calculation of antenna comfort. The results are the left and right antenna fitness. The output value of the fitness function directly characterizes the quality of the corresponding parameter vectors; the lower the value, the higher the degree of fit between the temperature prediction result corresponding to the parameter combination and the measured temperature. Therefore, the difference in fitness between the left and right antennas directly reflects the distribution of quality in the parameter spaces on both sides.

[0057] In this embodiment, the fitness difference is calculated based on the fitness of the left and right antennae. The sign and absolute value of this difference determine the parameter update direction. If the difference is positive, it indicates that the parameter region where the left antenna is located is better, and the update direction points to the left antenna; if the difference is negative, the region where the right antenna is located is better, and the update direction points to the right antenna. Finally, by combining the initial position of the longhorn beetle, the initial step size, the initial search distance, and the determined parameter update direction, the iterative search and update of the initial thermal property parameter set is completed. Guided by the update direction, the movement amplitude is controlled by the initial step size, and the initial position of the longhorn beetle is moved along the optimal direction to a new parameter space position. The parameter combination corresponding to this new position is the iteratively updated thermal property parameter set. The whole process retains the reasonable value basis of the initial parameter set and achieves directional optimization of the parameter combination through antenna perception and direction judgment.

[0058] In this embodiment of the invention, determining the left and right antenna parameter vectors based on the initial position of the longhorn beetle and the initial search distance includes: obtaining a unit random vector; performing antenna normalization position calculation based on the unit random vector, the initial position of the longhorn beetle, and the initial search distance to obtain the normalized position of the left and right antennas; and performing inverse normalization calculation based on the normalized position of the left and right antennas to obtain the left and right antenna parameter vectors.

[0059] In this embodiment, a unit random vector is first obtained. This unit random vector is initially obtained through random sampling, and then its magnitude is normalized to ensure that the vector magnitude is always 1. This ensures that the search direction is uniformly and unbiasedly distributed in the parameter space, preventing the algorithm from getting stuck in a local search due to directional bias. The expression for calculating the unit random vector is:

[0060] ,

[0061] in, Represents a unit random vector. The function can generate random numbers. Indicates the parameter dimension. Indicates to Calculate its Euclidean norm. The parameters to be inverted here are thermal conductivity, specific heat capacity, absorptivity, and emissivity, totaling four terms. Therefore, the parameter dimension is 4.

[0062] Then, based on the unit random vector, the initial position of the longhorn beetle, and the initial search distance, the normalized position of the tentacles is calculated to obtain the normalized positions of the left and right tentacles. The specific calculation expression is as follows:

[0063] ,

[0064] ,

[0065] in, Indicates the normalized position of the right antenna. This indicates the normalized position of the left antenna. The normalized positions of the right antenna and the left antenna are both normalized values, meaning that the values ​​of all components are within [0,1], thus eliminating the influence on parameters of different orders of magnitude. express The beetle's location at all times. express The beetle's search distance at any given time.

[0066] In order to normalize the position of the right antenna and the normalized position of the left antenna Mapping back to the actual solution space requires inverse normalization, the specific calculation expression of which is:

[0067] ,

[0068] in, Indicates the first denomination of the longhorn beetle after denormalization. The position of the dimension; For the normalized longhorn beetle The position of the dimension; , These are the normalized longhorn beetle numbers. The maximum and minimum values ​​of the dimension's position. Based on the above formula, the normalized position of the right tentacle is calculated. and the normalized position of the left antenna Perform inverse normalization to obtain the left antenna parameter vector. and right antenna parameter vector .

[0069] In this embodiment of the invention, the iterative search and update of the initial thermal property parameter set based on the initial position of the longhorn beetle, the initial step size, the initial search distance, and the parameter update direction includes: obtaining a preset step size decay coefficient and a preset step size ratio coefficient; updating the initial step size based on the step size decay coefficient to obtain the current step size; updating the initial search distance based on the step size ratio coefficient and the current step size to obtain the current search distance; and iteratively searching and updating the initial thermal property parameter set based on the initial position of the longhorn beetle, the current step size, the current search distance, and the parameter update direction.

[0070] In this embodiment, a preset step size decay coefficient and a step size ratio coefficient are first obtained. The step size decay coefficient is a constant less than 1, used to control the step size to gradually shrink during the iteration process, and is generally set to 0.95. The step size ratio coefficient is a preset correlation coefficient, used to establish a quantitative correspondence between the step size and the search distance, that is, the ratio coefficient between the step size and the search distance, and is generally set to 0.95.

[0071] Then, using the initial step size as a baseline, it is scaled and updated in each iteration using a step size decay coefficient. That is, the current step size is equal to the initial step size (or the current step size in the previous iteration) multiplied by the step size decay coefficient, so that the step size gradually decreases as the number of iterations increases, realizing the transition of the optimization process from large-scale exploration to local fine-tuning. The specific calculation expression is as follows:

[0072] ,

[0073] in, express Step size of time, express The step size of time.

[0074] Subsequently, based on the step length scaling factor and the updated current step length, the initial search distance is updated to ensure that the search distance changes synchronously with the step length, thus ensuring that the antennae's sensing range matches the adjustment range of the step length. The specific calculation expression is as follows:

[0075] ,

[0076] in, express Search distance at any given moment express Step size of time, This represents the step size scaling factor.

[0077] Finally, starting from the beetle's initial position in the parameter space (corresponding to the initial set of thermal property parameters), and following the previously determined parameter update direction, the movement amplitude is controlled by the current step size. Simultaneously, the current search distance constrains the antenna's sensing range. The beetle's position is then moved along the optimal direction to a new parameter space coordinate system. The parameter combination corresponding to this coordinate system is the iteratively updated set of thermal property parameters. The specific calculation expression is as follows:

[0078] ,

[0079] in, for The location of the longhorn beetle at all times; for The location of the longhorn beetle at all times. express Step size of time, Represents a unit random vector. This represents the fitness difference calculated based on the fitness of the right antenna and the left antenna, used to determine the direction of parameter updates. The symbolic function, whose function value is expressed as:

[0080] ,

[0081] Among them, when the independent variable When the value of is greater than 0, the function value is 1; when is greater than 0, the function value is When the value of is 0, the function value is 0; when , the function value is 0. When the value of is less than 0, the function value is -1. When the function value is 1, it means that the parameter region where the left antenna is located is better, and the update direction is towards the left antenna. When the function value is -1, the region where the right antenna is located is better, and the update direction is towards the right antenna. When the function value is 0, it means that the fitness of the parameter regions corresponding to the left and right antennas is equal, and there is no need to update the parameters in a specific direction.

[0082] Furthermore, the iterative search of the longhorn beetle whisker algorithm relies on a clearly defined parameter space boundary. Without search range constraints, the parameter space will expand excessively due to unlimited dimensions and values, significantly increasing computational resource consumption and potentially leading to meaningless, large-scale random exploration, reducing optimization efficiency, prolonging convergence time, and even failing to obtain an effective solution due to parameters deviating from reasonable ranges. Therefore, this application embodiment sets search ranges for thermal property parameters. For example, the search range for thermal conductivity is set between 1 W / (m·K) and 3 W / (m·K), the search range for specific heat capacity is set between 600 J / (kg·K) and 1600 J / (kg·K), the absorptivity is set between 0.03 and 0.1, and the emissivity is set between 0.4 and 0.9.

[0083] Furthermore, to verify the validity of the measured thermal parameter set, temperature-dependent thermophysical parameters were introduced into the simulation model, and the temperature changes under test laser power conditions (5W, 10W, and 15W) were calculated, such as... Figure 2 As shown, the black solid line (5W·measured), red solid line (15W·measured), and blue solid line (10W·measured) correspond to the measured temperatures at different power levels, respectively. The black dashed line (5W·before optimization), red dashed line (15W·before optimization), and blue dashed line (10W·before optimization) correspond to the temperatures measured using the traditional method at different power levels, respectively. The black dotted line (5W·after optimization), red dotted line (15W·after optimization), and blue dotted line (10W·after optimization) correspond to the temperatures measured using the reverse measurement method of thermal property parameters provided in this application at different power levels, respectively. The results show that the temperature curve measured using the reverse measurement method of thermal property parameters provided in this application has a higher degree of agreement with the measured temperature curve, indicating that the identified thermal property parameters have good rationality.

[0084] The above describes the method for reverse measurement of thermophysical parameters in embodiments of the present invention. The following describes the apparatus for reverse measurement of thermophysical parameters in embodiments of the present invention. Please refer to [link to relevant documentation]. Figure 3 One embodiment of the reverse measurement device for thermophysical parameters in this invention includes:

[0085] Parameter acquisition module 301: used to acquire the initial set of thermophysical parameters and the measured temperature;

[0086] First temperature prediction module 302: used to call a pre-trained temperature field model to predict the temperature of the initial thermophysical parameter set and obtain the initial predicted temperature;

[0087] Function construction module 303: used to construct a fitness function based on the measured temperature and the initial predicted temperature;

[0088] Data update module 304: used to iteratively search and update the initial thermal property parameter set based on the fitness function using the beetle whisker algorithm;

[0089] The second temperature prediction module 305 is used to input the updated set of thermal property parameters into the temperature field model to predict the temperature when the preset iteration stop condition is met, so as to output the target predicted temperature.

[0090] Based on the same ideas as the methods in the above embodiments, the apparatus provided in this application can implement the methods in the above embodiments.

[0091] above Figure 3The reverse measurement device for thermal property parameters in this embodiment of the invention is described in detail from the perspective of modular functional entities. The reverse measurement device for thermal property parameters in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0092] Figure 4 This is a schematic diagram of the structure of a reverse measurement device for thermal properties provided in an embodiment of the present invention. The reverse measurement device 400 can vary considerably due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 410 (e.g., one or more processors) and a memory 420, and one or more storage media 430 (e.g., one or more mass storage devices) storing application programs 433 or data 432. The memory 420 and storage media 430 can be temporary or persistent storage. The program stored in the storage media 430 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the reverse measurement device 400 for thermal properties. Furthermore, the processor 410 may be configured to communicate with the storage media 430 and execute the series of instruction operations in the storage media 430 on the reverse measurement device 400 for thermal properties to implement the steps of the reverse measurement method for thermal properties provided in the above-described method embodiments.

[0093] The thermophysical parameter reverse measurement device 400 may also include one or more power supplies 440, one or more wired or wireless network interfaces 450, one or more input / output interfaces 460, and / or one or more operating systems 431, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 4 The illustrated structure of the reverse measurement device for thermal properties does not constitute a limitation on the reverse measurement device for thermal properties. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0094] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of a method for reverse measurement of thermophysical parameters.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0096] If the integrated unit is implemented as 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, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for reverse measurement of thermophysical parameters, characterized in that, include: Obtain the initial set of thermophysical parameters and the measured temperature; The pre-trained temperature field model is invoked to predict the temperature of the initial set of thermal property parameters, and the initial predicted temperature is obtained. Obtain a preset amplification factor, a preset time range, and a preset power range; calculate the relative error between the measured temperature and the initial predicted temperature based on the time range and the power range to obtain the relative error; calculate the mean error based on the relative error to obtain the average relative error; perform correlation calculations on the measured temperature and the initial predicted temperature to obtain the coefficient of determination; construct a fitness function based on the amplification factor, the relative error, the average relative error, and the coefficient of determination. The expression for the fitness function is as follows: ； Among them, represents the amplification factor, represents the relative error, represents the thermophysical property parameter group, represents any moment within the time value range, represents any power within the power value range, represents the average relative error, represents the sampling number of the measured temperature under the time series corresponding to the same laser power, represents the coefficient of determination, represents for all laser powers take the maximum value; Based on the initial set of thermal property parameters, obtain the initial position, initial step size, and initial search distance of the longhorn beetle; determine the left and right antenna parameter vectors based on the initial position and initial search distance; calculate antenna comfort based on the fitness function, the left and right antenna parameter vectors, to obtain the left and right antenna fitness; calculate the fitness difference based on the left and right antenna fitness, to obtain the fitness difference; determine the parameter update direction based on the fitness difference; and iteratively search and update the initial set of thermal property parameters based on the initial position, initial step size, initial search distance, and parameter update direction. When the preset iteration stop condition is met, the updated set of thermal property parameters is input into the temperature field model to predict the temperature and output the target predicted temperature.

2. The method for reverse measurement of thermophysical parameters according to claim 1, characterized in that, Before obtaining the initial set of thermophysical parameters and the measured temperature, the process also includes: Obtain the preset search range for thermal property parameters; The initial set of thermal property parameters is obtained by randomly sampling based on the search range of the thermal property parameters using the beetle whisker algorithm.

3. The method for reverse measurement of thermophysical parameters according to claim 1, characterized in that, Before calling the pre-trained temperature field model to predict the temperature of the initial set of thermophysical parameters and obtaining the initial predicted temperature, the method further includes: An initial temperature field model was constructed using the COMSOL tool, and the initial temperature field model was then converted to a MATLAB-compatible model. The MATLAB-adapted model was reconstructed using MATLAB tools to obtain the MATLAB function model. A training dataset was constructed using MATLAB tools, and the MATLAB function model was iteratively trained multiple times based on the training dataset to obtain the pre-trained temperature field model.

4. The method for reverse measurement of thermophysical parameters according to claim 1, characterized in that, The determination of the left and right antenna parameter vectors based on the initial position of the longhorn beetle and the initial search distance includes: Obtain a unit random vector; Based on the unit random vector, the initial position of the longhorn beetle, and the initial search distance, the normalized position of the antennae is calculated to obtain the normalized position of the left antennae and the normalized position of the right antennae. Based on the normalized position of the left antenna and the normalized position of the right antenna, inverse normalization calculation is performed to obtain the parameter vectors of the left antenna and the right antenna.

5. The method for reverse measurement of thermophysical parameters according to claim 1, characterized in that, The iterative search and update of the initial thermal property parameter set based on the initial position of the longhorn beetle, the initial step size, the initial search distance, and the parameter update direction includes: Obtain the preset step size decay coefficient and the preset step size ratio coefficient; The initial step size is updated based on the step size decay coefficient to obtain the current step size; The initial search distance is updated based on the step size ratio coefficient and the current step size to obtain the current search distance; The initial set of thermal property parameters is iteratively searched and updated based on the initial position of the longhorn beetle, the current step size, the current search distance, and the parameter update direction.

6. A reverse measurement device for thermophysical parameters, characterized in that, include: Parameter acquisition module: used to acquire the initial set of thermophysical parameters and measured temperature; First temperature prediction module: used to call a pre-trained temperature field model to predict the temperature of the initial set of thermal property parameters and obtain the initial predicted temperature; The function construction module is used to obtain a preset amplification factor, a preset time range, and a preset power range; calculate the relative error between the measured temperature and the initial predicted temperature based on the time range and the power range to obtain the relative error; calculate the mean error based on the relative error to obtain the average relative error; perform correlation calculations on the measured temperature and the initial predicted temperature to obtain the coefficient of determination; and construct a fitness function based on the amplification factor, the relative error, the average relative error, and the coefficient of determination. The expression for the fitness function is as follows: ； in, This represents the magnification factor. This represents the relative error. Represents a set of thermophysical property parameters. It represents any moment within the range of time values. This represents any power value within the specified range. This represents the average relative error. This represents the number of temperature samples taken within a time series corresponding to the same laser power. This represents the coefficient of determination. This indicates the effect on all laser powers. Take the maximum value; The data update module is used to: obtain the initial position, initial step size, and initial search distance of the longhorn beetle based on the initial thermophysical parameter set; determine the left and right antenna parameter vectors based on the initial position and initial search distance; calculate antenna comfort based on the fitness function, the left and right antenna parameter vectors, to obtain the left and right antenna fitness; calculate the fitness difference based on the left and right antenna fitness to obtain the fitness difference; determine the parameter update direction based on the fitness difference; and iteratively search and update the initial thermophysical parameter set based on the initial position, initial step size, initial search distance, and parameter update direction. The second temperature prediction module is used to input the updated set of thermal property parameters into the temperature field model to predict the temperature when the preset iteration stop condition is met, so as to output the target predicted temperature.

7. A reverse measurement device for thermophysical parameters, characterized in that, The reverse measurement device for thermophysical parameters includes: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors invokes the instructions in the memory to cause the thermophysical parameter reverse measurement device to perform the steps of the thermophysical parameter reverse measurement method as described in any one of claims 1-5.

8. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the reverse measurement method for thermophysical parameters as described in any one of claims 1-5.

Images