A device and method for detecting liquid-solid phase change of a sample

Abstract

This invention belongs to the technical field of liquid-solid phase change detection and discloses a device and method for detecting liquid-solid phase change of a sample. The method comprises: filling a measurement chamber with a fluid sample to be tested, turning on the detection device, cooling the sample to a preset fluid crystallization point, and then cooling it down again. During this time, a laser sensor unit acquires light transmittance, a camera module acquires a morphological image, and a temperature sensor and a pressure sensor acquire temperature and pressure data, respectively. Image features are extracted, and the light transmittance, temperature, and pressure data are normalized and converted into a one-dimensional matrix, which is used as input for feature detection in a phase change detection model. The phase change detection model, constructed using a BP neural network, processes the measurement data and acquires phase change morphological features. The technical solution described in this invention designs an integrated liquid-solid phase change detection device and method, which can improve the accuracy and reliability of phase change process detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of liquid-solid phase change detection, and in particular relates to a device and method for detecting liquid-solid phase change of a sample. Background Art

[0002] As one of the fundamental processes of material state transformation, the transformation mechanism of liquid-solid phase transition (process morphology, characteristic parameters, and phase transition point) is of great research value for thermodynamics, dynamics, materials science, and renewable energy utilization. Currently, the main methods for detecting liquid-solid phase transition processes include differential scanning calorimetry, PT phase diagrams, visual benchtop methods, X-ray diffraction, and infrared spectroscopy.

[0003] While existing liquid-solid phase change detection technologies have been widely applied and promoted in various scientific research and engineering scenarios, they are limited by inherent flaws or shortcomings of the detection methods. Using a single detection technology often only identifies specific process parameters during the phase change process, failing to fully reflect the complex liquid-solid phase change process, thus hindering the objective evaluation of the phase change behavior and mechanism. Summary of the Invention

[0004] The purpose of the present invention is to provide a sample liquid-solid phase change detection device and method to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above object, the present invention provides a sample liquid-solid phase change detection device, comprising:

[0006] A measuring cavity is provided inside the detection box, and a plurality of light sources are installed on the top of the measuring cavity; a liquid inlet, a pressurization port, and a plurality of sensor installation ports connected to the measuring cavity are provided on the top of the detection box, a liquid discharge port connected to the measuring cavity is provided on the bottom of the detection box, and corresponding observation windows are provided on the front side wall of the measuring cavity and the front side wall of the detection box;

[0007] The laser sensor unit includes a laser sensor transmitting module and a laser sensor receiving module, wherein the laser sensor transmitting module and the laser sensor receiving module are symmetrically arranged on the left and right sides of the detection box;

[0008] A sensor group, comprising a temperature sensor and a pressure sensor, wherein the temperature sensor and the pressure sensor are installed in corresponding sensor installation openings;

[0009] A camera module is provided on one side of the detection box, corresponding to the position of the front side wall of the detection box;

[0010] The control unit includes a main controller and a signal excitation module, a timing module, a data acquisition module and a temperature control module electrically connected to the main controller; the camera module, the laser sensor unit and the sensor group are all electrically connected to the data acquisition module, the signal excitation module is also electrically connected to the laser sensor emission module, and the detection box is electrically connected to the temperature control module.

[0011] Optionally, a constant temperature medium layer and a heat preservation layer are sequentially provided between the outer wall of the measurement cavity and the inner wall of the detection box.

[0012] Optionally, it is characterized in that a constant temperature medium inlet is opened on the outer wall of the detection box, and the constant temperature medium inlet passes through the insulation layer and is connected to the constant temperature medium layer.

[0013] Optionally, each laser emitter in the laser sensor transmitting module is arranged in a one-to-one correspondence with each laser receiver in the laser sensor receiving module, each laser emitter is connected to the signal excitation module, and each laser receiver is connected to the data acquisition module.

[0014] A sample liquid-solid phase change detection method, applied to a sample liquid-solid phase change detection device, comprising:

[0015] Step 1: Filling the measurement cavity with a sample to be tested, wherein the sample to be tested is in a fluid state;

[0016] Step 2: Turn on the sample liquid-solid phase change detection device, measure the phase change state of the sample to be tested in combination with a preset temperature, and record the measurement data, the measurement data including light transmittance data, morphological image data, phase change temperature data, and phase change pressure data;

[0017] Step 3: Input the measurement data into a phase change detection model integrated in the main controller to perform feature detection to obtain phase change morphological feature data, wherein the phase change detection model is constructed based on a BP neural network.

[0018] Optionally, the process of acquiring the measurement data specifically includes:

[0019] The temperature control module is controlled to cool the sample to be tested until the sample to be tested is cooled to a preset fluid crystallization point threshold, and then the temperature is lowered; during the cooling process of the sample to be tested, the signal excitation module is used to excite the laser sensing unit to obtain light transmittance data; the camera module is used to obtain morphological image data of the sample to be tested in the measurement cavity, the temperature sensor is used to obtain phase change temperature data, and the pressure sensor is used to obtain phase change pressure data.

[0020] Optionally, before inputting the measurement data into the phase change detection model, the method further includes:

[0021] performing feature extraction on the morphological image data to obtain image feature data;

[0022] The light transmittance data, phase change temperature data, and phase change pressure data are normalized based on a time series to obtain normalized data. The normalized data and the image feature data are converted into one-dimensional matrix data to obtain preprocessed measurement data. The preprocessed measurement data is used as input of the phase change detection model for feature detection.

[0023] Optionally, the training process of the phase change detection model specifically includes:

[0024] Acquiring training data, wherein the training data includes measurement training data and corresponding phase change morphology characteristic data;

[0025] An initial phase change detection model is constructed, the training data is input into the initial phase change detection model, the characteristics of the morphological image data are quantitatively analyzed using the light transmittance data, phase change temperature data and phase change pressure data to obtain initial phase change morphological feature data, and training is performed with the goal of minimizing the loss between the initial phase change morphological feature data and the phase change morphological feature data corresponding to the measured training data to obtain a trained phase change detection model.

[0026] The technical effects of the present invention are:

[0027] The present invention is based on the theory of multi-source information fusion, and adopts UG and CFD simulation technology to integrate the visual experimental platform method (microscope observation method and laser transmission method), temperature detection method, pressure detection method, etc., and use them together for liquid-solid transformation process detection (including morphology, light transmittance, phase change temperature and pressure). The characteristics of different detection methods are used to carry out multi-dimensional angle cross-measurement, which can avoid the defects and shortcomings of single detection methods. In addition, for the information of different magnitudes, forms and units such as images, light transmittance, phase change temperature, phase change pressure, etc. obtained by detection, convolutional neural networks and normalization algorithms can be used to convert them to the same magnitude and interval through information classification, extraction of characteristic variables, dimensionless standardization, etc., and then use the BP neural network algorithm to obtain a measurement model of a new liquid-solid phase change detection device. The present invention combines multi-dimensional detection with a neural network algorithm to design a set of integrated liquid-solid phase change detection devices and methods, which can improve the accuracy and reliability of phase change process detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0029] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0030] Figure 1 Schematic diagram of the structure of a novel liquid-solid phase change detection device in an embodiment of the present invention;

[0031] Figure 2 This is an analytical diagram of the preprocessing and fusion process of multi-source information in an embodiment of the present invention;

[0032] Figure 3 This is a flow chart of extracting morphological image feature vectors based on a convolutional neural network in an embodiment of the present invention;

[0033] Figure 4 Schematic diagram of the structure of the phase change morphological feature measurement model based on BP neural network in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but rather as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0035] It should be understood that the terms described herein are intended only to describe particular embodiments and are not intended to limit the present invention. In addition, for numerical ranges herein, it should be understood that each intermediate value between the upper and lower limits of the range is also specifically disclosed. Each smaller range between any intermediate value within a stated value or stated range and any other stated value or intermediate value within the stated range is also encompassed by the present invention. The upper and lower limits of these smaller ranges may be independently included or excluded within the scope.

[0036] Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art. Although only preferred methods are described herein, any method similar or equivalent to that described herein may also be used in the practice or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods associated with the documents. In the event of any conflict with any incorporated document, the contents of this specification shall prevail.

[0037] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present invention without departing from the scope or spirit of the invention. Other embodiments will be apparent to those skilled in the art from the present invention. The present description and examples are intended to be illustrative only.

[0038] The words “include,” “including,” “have,” “contain,” etc. used in this article are open-ended terms, meaning including but not limited to.

[0039] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0040] Example 1

[0041] like Figure 1 - Figure 4 As shown, this embodiment provides a sample liquid-solid phase change detection device, including:

[0042] A measuring cavity is provided inside the detection box, and a plurality of light sources are installed on the top of the measuring cavity; a liquid inlet, a pressurization port, and a plurality of sensor installation ports connected to the measuring cavity are provided on the top of the detection box, a liquid discharge port connected to the measuring cavity is provided on the bottom of the detection box, and corresponding observation windows are provided on the front side wall of the measuring cavity and the front side wall of the detection box;

[0043] The laser sensor unit includes a laser sensor transmitting module and a laser sensor receiving module, wherein the laser sensor transmitting module and the laser sensor receiving module are symmetrically arranged on the left and right sides of the detection box;

[0044] A sensor group, comprising a temperature sensor and a pressure sensor, wherein the temperature sensor and the pressure sensor are installed in corresponding sensor installation openings;

[0045] A camera module is provided on one side of the detection box, corresponding to the position of the front side wall of the detection box;

[0046] The control unit includes a main controller and a signal excitation module, a timing module, a data acquisition module and a temperature control module electrically connected to the main controller; the camera module, the laser sensor unit and the sensor group are all electrically connected to the data acquisition module, the signal excitation module is also electrically connected to the laser sensor emission module, and the detection box is electrically connected to the temperature control module.

[0047] A sample liquid-solid phase change detection method, applied to a sample liquid-solid phase change detection device, comprising:

[0048] Step 1: Filling the measurement cavity with a sample to be tested, wherein the sample to be tested is in a fluid state;

[0049] Step 2: Turn on the sample liquid-solid phase change detection device, measure the phase change state of the sample to be tested in combination with a preset temperature, and record the measurement data, the measurement data including light transmittance data, morphological image data, phase change temperature data, and phase change pressure data;

[0050] Step 3: Input the measurement data into a phase change detection model integrated in the main controller to perform feature detection to obtain phase change morphological feature data, wherein the phase change detection model is constructed based on a BP neural network.

[0051] This embodiment is based on the theory of multi-source information fusion, and adopts UG and CFD simulation technology to integrate the visual test bench method (microscope observation method and laser transmission method), temperature detection method, pressure detection method, etc., and use them together for liquid-solid transformation process detection (including morphology, light transmittance, phase change temperature and pressure). The characteristics of different detection methods are used to carry out multi-dimensional angle cross-measurement, which can avoid the defects and shortcomings of single detection methods. In addition, for the information of different magnitudes, forms and units such as images, light transmittance, phase change temperature, phase change pressure, etc. obtained by detection, convolutional neural networks and normalization algorithms can be used to convert them to the same magnitude and interval through information classification, extraction of characteristic variables, dimensionless standardization, etc., and then use the BP neural network algorithm to obtain a measurement model of a new liquid-solid phase change detection device. The present invention combines multi-dimensional detection with a neural network algorithm to design a set of integrated liquid-solid phase change detection devices and methods, which can improve the accuracy and reliability of phase change process detection.

[0052] Limited by the inherent defects or deficiencies of the measurement method, when a single detection technology is used for liquid-solid phase change process detection, it is impossible to conduct a comprehensive and accurate scientific evaluation of the phase change behavior and internal mechanism. In view of this, the present embodiment aims to make full use of the respective advantages of the visual laboratory bench method (microscope observation method, laser transmission method), pressure detection method, and temperature detection method, and simultaneously use them for liquid-solid phase change process detection of the same material system. Different methods can cross-complement and mutually verify, thereby providing more comprehensive liquid-solid phase change information. On this basis, the BP neural network algorithm can be used to integrate the optical transmittance, phase change temperature, phase change pressure and phase change morphological characteristics to construct a measurement model for a new liquid-solid phase change detection device. Further, it can provide strong support for a comprehensive understanding of the liquid-solid transformation mechanism and behavioral characteristics.

[0053] This embodiment makes full use of the visual laboratory bench method (microscope observation method and laser transmission method), phase change pressure detection, phase change temperature detection and other methods for the liquid-solid transformation process detection of the same material system. Specifically, the microscope observation method can provide clearly visible material morphological characteristics, and the laser transmission method, pressure detection method, and thermocouple temperature measurement method are used to provide accurate phase change process parameters and boundary adjustment. On this basis, by taking advantage of the neural network algorithm, the morphological characteristics of the material are integrated with the laser transmittance, phase change temperature, phase change pressure, etc. to construct an accurate and reliable liquid-solid phase change measurement model.

[0054] Based on the theory of multi-source information fusion, this embodiment employs microscopic observation, laser transmission, pressure detection, and temperature detection methods to simultaneously detect the characteristics of the liquid-solid phase change process. This method obtains relatively more information about the phase change process, improving the accuracy and reliability of the measurement results. In device development, UG mapping technology was used to construct the basic structure of the liquid-solid phase change measurement cavity, sensor installation location, and visual observation window. CFD fluid simulation technology was then used to further optimize the measurement device structure and sensor parameters to ensure compatibility and stability between the various measurement modules.

[0055] During the measurement process, a multi-channel signal source is used to simultaneously stimulate laser sensors, temperature sensors, pressure sensors, and electron microscopes. The measured light transmittance, phase transition temperature, phase transition pressure, and phase transition morphological images are collected in real time by a single acquisition module to ensure the synchronization and consistency of all measurement results. The liquid-solid phase transition morphological images are first converted into a one-dimensional array using a convolutional neural network as the phase transition morphological feature data. The feature data is then normalized and the morphological information is classified by extracting the feature data. The transmittance, phase transition temperature, and phase transition pressure are dimensionlessly normalized using a normalization algorithm. Based on this, a BP neural network algorithm is used with light transmittance, phase transition temperature, and phase transition pressure as input variables and phase transition morphological feature data as output variables to establish a mathematical model linking the phase transition morphological features with the process parameters, including phase transition temperature, phase transition pressure, and light transmittance.

[0056] This embodiment provides a liquid-solid phase change process detection method based on multi-source information fusion, which can improve the reliability and accuracy of measurement results and facilitate a more comprehensive scientific evaluation of the behavioral characteristics and internal mechanisms of the phase change process.

[0057] In this example, a newly developed liquid-solid phase transition detection device integrates a visualization platform method (microscope observation and laser transmission), pressure detection, and temperature detection to obtain comprehensive information about the liquid-solid transition process. Furthermore, all measurement results are pre-processed using a convolutional neural network and a normalization algorithm, and a BP neural network is then used to construct a measurement model for the new liquid-solid phase transition detection device.

[0058] The specific technical solution is to first maintain the substance under test in a fluid state, then use negative pressure technology to fill the entire measurement chamber with the fluid under test. At the beginning of the measurement, the fluid under test is cooled by an external temperature control device, inducing the fluid to transform from liquid to solid. Simultaneously, the PC control unit triggers the laser sensor, electron microscope, temperature sensor, and pressure sensor to synchronize their operation at the same time frequency, and the signal acquisition unit simultaneously collects the various parameters obtained from the detection. The electron microscope probe is used to capture the morphological changes of the fluid under test during the transformation from liquid to solid, namely microscopic changes such as molecular aggregation, arrangement, and crystallization. The laser sensor (transmitter and receiver) is used to detect changes in light transmittance during the material's morphological transformation, providing a quantitative reference for visualizing the morphological transformation process. Temperature and pressure sensors installed in the measurement chamber detect subtle changes in temperature and pressure during the transformation from liquid to solid, which is used to analyze the behavioral characteristics and internal mechanisms of the liquid-solid phase transition. At the end of the measurement, the captured image and the measurement results of each sensor are extracted, and the information is classified and preprocessed using a convolutional neural network and dimensionless normalization. Furthermore, a BP neural network algorithm is used to construct a mathematical model between morphological characteristics and light transmittance, phase transition temperature, and phase transition pressure, which is used to comprehensively and accurately analyze the liquid-solid phase transition process. This allows for liquid-solid phase transition process detection based on multi-source information fusion.

[0059] Liquid-Solid Phase Change Detection Chamber Module: In this embodiment, the liquid-solid phase change detection chamber is designed as a cube. The inner wall surface should have a surface finish of at least 0.8μm to ensure the tightness of each module within the measurement chamber. A boss is designed at the top of the measurement chamber to measure the inflow of liquid into the chamber and the increase in test pressure. Mounting holes for temperature and pressure sensors are located on either side of the boss and are assembled using NPT threads. Thirteen perfectly symmetrical holes are designed on the left and right end faces of the measurement chamber to accommodate the transmitting and receiving laser sensor probes, respectively. A sapphire glass observation window is designed on the front of the measurement chamber for mounting an electron microscope probe. A low-heat, multi-point light source is located at the top of the measurement chamber to provide bright light for the electron microscope probe. A drain port is designed at the bottom of the measurement chamber to remove waste liquid after the measurement chamber experiment. Four height-adjustable support feet are also installed at the bottom to adjust the height and horizontal position of the measurement chamber. In addition, the outer wall of the measurement chamber is designed with a temperature-control layer to regulate the temperature of the fluid being measured. Its inlet and outlet are located at opposite corners of the measurement chamber. Furthermore, an insulation layer is designed outside the temperature-control layer, tightly wrapped with multiple layers of thermal insulation to minimize heat exchange between the measurement chamber and the outside world.

[0060] Multi-sensor detection module: In this embodiment, the multi-sensor detection module includes a laser sensor, an electron microscope probe, a temperature sensor, and a pressure sensor. Among them, the laser sensor includes a transmitting probe and a receiving probe (a total of 13 pairs), which are symmetrically installed on the left and right end faces of the measurement cavity and are evenly distributed in an array. The laser is emitted from the transmitter to the receiver. As the substance to be measured changes from liquid to solid, the transmittance of the laser changes accordingly. Based on the light transmittance, the phase change process can be quantitatively analyzed. During the liquid-solid phase change process, the electron microscope probe is used to observe the aggregation, crystallization and arrangement of the molecules of the substance to be measured, and to capture the image information of the phase change morphological characteristics in real time. In addition, the temperature measurement during the phase change process uses a thermocouple thermometer with high sensitivity and good stability to synchronously detect the temperature variable during the liquid-solid phase change process. The pressure measurement uses a capacitive pressure sensor with high sensitivity to synchronously detect the temperature variable during the liquid-solid phase change process. The measurement results of light transmittance, phase change temperature, phase change pressure and morphological characteristics are used together to construct a liquid-solid phase change measurement model using a BP neural network algorithm after data classification and preprocessing.

[0061] Signal Control and Acquisition Module: This module includes a multi-channel signal excitation unit, a high-precision electronic timer, a multi-channel signal acquisition unit, and a PC centralized control unit. In this embodiment, the multi-channel signal excitation unit provides the required operating power to the temperature sensor, pressure sensor, laser sensor, and electron microscope probe, triggering them to begin operating simultaneously. The high-precision electronic timer provides a unified time reference for reading all measurement signals, laying the foundation for the fusion of multiple signals. The multi-channel signal acquisition unit is primarily used for high-frequency recording of measured data and real-time transmission to the PC unit for storage. Furthermore, the PC unit uses the single-chip microcomputer unit to achieve real-time communication and centralized control with various measurement devices.

[0062] Processing and fusion of measurement data:

[0063] In this embodiment, to achieve consistency across different measurement units, magnitudes, and intervals, all measurement results were unified based on real-time sampling time. Furthermore, a convolutional neural network was used to convert the liquid-solid phase transition image information into a one-dimensional array. A normalization algorithm was used to perform dimensionless preprocessing on the transmittance, temperature, and pressure, converting them to the same interval and magnitude. Morphological information was then classified using the morphological feature data. Furthermore, a BP neural network algorithm was used to cross-fuse all characteristic parameters, further establishing a liquid-solid phase transition measurement model based on morphological features, transmittance, phase transition temperature, and phase transition pressure.

[0064] Working Principle of the New Liquid-Solid Phase Change Detection Device: In this embodiment, the substance to be measured is first ensured to be in a liquid, flowable state. A vacuum negative pressure technique is then used to fill the entire measurement chamber with the fluid to be measured. Before measurement begins, the temperature within the measurement chamber is ensured to be at least 10°C above the crystallization point of the fluid to be measured, and the background pressure within the measurement chamber is maintained at 0.1 MPa. A level calibrator is used to adjust the measurement chamber to a horizontal position. At the start of the measurement, the PC unit activates the temperature control unit, electronic timer, multi-channel excitation source, sensor unit, and signal acquisition unit, all of which begin operating simultaneously.

[0065] The internal temperature of the measurement cavity is regulated by a refrigeration cycle fluid, inducing the fluid to be measured to cool down until the phase change process is completed. When approaching the crystallization point, the cooling rate must be strictly controlled to ensure that the phase change process is accurately detected. The changes in the temperature and pressure of the substance to be measured are synchronously detected by temperature sensors and pressure sensors. In particular, the changes in temperature and pressure in the measurement cavity during liquid-solid transformation must be accurately captured. At the same time, PID technology is used to achieve linkage with the temperature control unit to adjust the output power of the temperature control unit. An electron microscope probe is used to detect in real time the changes in the aggregation, arrangement, and crystallization state of the molecules of the substance to be measured. An array laser sensor is used to synchronously detect the changes in transmittance during the liquid-solid transformation process. The morphological characteristics of the phase change are analyzed based on characteristic variables such as phase change temperature, phase change pressure, and light transmittance.

[0066] During the detection process, the data acquisition unit stores the detection information such as phase change temperature, phase change pressure, light transmittance, morphological image, etc. synchronously in the PC unit with the same time reference. Then, through technical means such as convolutional neural network, normalization algorithm, and BP neural network, a measurement model of the new liquid-solid phase change detection device can be constructed.

[0067] After the measurement is completed, the temperature control unit slowly adjusts the temperature of the substance to be measured to above the crystallization point. The sensor unit then synchronously detects characteristic variables such as temperature, pressure, light transmittance, and morphology within the measurement chamber, thereby enabling reverse process detection and analysis of the liquid-to-solid phase transition. In this embodiment, the liquid inlet at the top of the measurement chamber is designed to also serve as a boost port for the background pressure of the fluid to be measured. By increasing the background pressure of the test using an external booster device and repeating the above measurement process, the liquid-to-solid phase transition process under high-pressure conditions can be tested and studied.

[0068] Figure 1 This is a schematic diagram of the structure of a new liquid-solid phase change detection device. It mainly detects characteristic parameters such as morphology, transmittance, phase change temperature, and phase change pressure during the liquid-solid transformation process to obtain quantitative phase change process information of the substance to be tested, supporting the analysis and research of phase change behavior characteristics and internal mechanisms. The device mainly includes unit modules such as experimental measurement cavity, temperature and pressure sensors, laser sensor, electron microscope probe, etc. Example of the structural features of the detection device:

[0069] In this embodiment, the liquid-solid phase change measurement chamber is made of food-grade 316L stainless steel and has a cubic structure with an internal single-side length of 93.0 mm ± 0.2 mm. The top boss is a hollow cylinder with a height of 45.0 mm ± 0.2 mm and an inner diameter of 8.0 mm ± 0.1 mm. The single-side wall thickness of the measurement chamber and boss is 3.5 mm ± 0.1 mm, with a maximum pressure tolerance of 100 MPa. A temperature control medium circulation layer is uniformly wrapped around the outer wall of the measurement chamber, with a thickness of ≥ 20.0 mm ± 0.2 mm. The temperature control medium inlet and outlet are circular pipes with an inner diameter of 25.0 mm ± 0.2 mm, located in the upper left and lower right corners of the measurement chamber, respectively. The observation window on the front of the measurement chamber is circular with a radius of 40.0 mm ± 0.1 mm and is made of high-pressure-resistant sapphire glass. In addition, the left and right end faces of the measuring cavity are respectively designed with 13 array-type symmetrical holes for installing laser sensor probes, and the inner diameter of the holes is designed to be 9.0mm±0.1mm. On both sides of the top boss of the measuring cavity, there is a hole with an inner diameter of 10.0mm±0.1mm, respectively, for installing the temperature sensor and pressure sensor. At the center of the bottom of the measuring cavity, there is a circular drain port with an inner diameter of 30.0mm±0.2mm. Figure 1 illustrate.

[0070] Before the measurement begins, the substance to be measured is filled into the entire measurement chamber in a fluid state using vacuum negative pressure technology to maintain a stable temperature. The measurement chamber is adjusted to an absolute horizontal angle using a level calibrator and support feet to ensure the stability of the device during the test process. At the same time, the laser sensor emission angle and the electron microscope probe's observation position are precisely adjusted to estimate the crystallization point temperature of the fluid to be measured, and the target temperature of the temperature control module is set to 3 to 5°C higher than the crystallization point of the fluid to be measured. At the beginning of the measurement, the PC control unit, through the single-chip microcomputer, simultaneously turns on the temperature control device, multi-channel signal excitation source, multi-sensor unit, data acquisition unit, electronic timer unit, etc., and starts working simultaneously. During the measurement process, the temperature control module first outputs a constant cooling power. When the temperature is cooled to a target temperature of 3 to 5°C above the crystallization point of the fluid to be measured, the cooling output power of the temperature control module is reduced, causing the temperature of the measurement chamber to slowly decrease. In the sensor module, a high-resolution electron microscope probe captures image information such as the aggregation, arrangement, and crystallization of fluid molecules. A uniformly distributed array of 13 laser sensors comprehensively measures the transmittance of the fluid under test, quantitatively reflecting the fluid's morphological changes. High-sensitivity temperature and pressure sensors precisely detect temperature and pressure changes during phase transitions. All sensor measurement results are preprocessed using a convolutional neural network and normalization algorithm before they can be used in the research and construction of the liquid-solid phase transition measurement model.

[0071] After determining the expected crystallization temperature of the fluid to be tested, the external temperature control system of the measurement chamber is activated, the refrigeration temperature slowly decreases, and the PC controller sends experimental test instructions to each unit. The multi-channel signal excitation source triggers the electronic timer, electron microscope probe, laser sensor, temperature sensor, pressure sensor, etc. to start working synchronously, and the liquid-solid transformation process is detected in real time at the same time test frequency to obtain characteristic information such as images, transmittance, phase change temperature, and phase change pressure during the phase change process, which is used to study and analyze the mechanism of the phase change process. After the test is completed, the temperature is slowly increased using the temperature control system to achieve reverse detection of the liquid-solid phase change process. Before the experiment begins, the test background pressure in the measurement chamber is changed to conduct testing and research on the liquid-solid transformation process under different pressure conditions.

[0072] Figure 2 This paper briefly demonstrates the preprocessing and fusion analysis of multi-source information. Through preprocessing, normalization, and correlation analysis, data from different detection methods are cross-fused and used for quantitative analysis of the liquid-solid phase transition of the fluid under test. This process primarily involves data conversion of morphological image information, normalization of optical transmittance, phase transition temperature, and phase transition pressure, unified classification of multiple information, and correlation analysis between morphological features and quantitative parameters. Image information acquired by an electron microscope probe is used to visually reflect morphological characteristics such as molecular aggregation, arrangement, and crystallization during the liquid-solid phase transition. A convolutional neural network algorithm is first used to extract feature variables and convert the image information into a one-dimensional array. A normalization algorithm is then used to normalize the array elements. For transmittance, phase transition temperature, and pressure variables with different units and magnitudes, a normalization algorithm is used to find standardized parameters to convert all data to the same interval. Using the real-time time recorded by an electronic timer as a benchmark, optical transmittance, phase transition temperature, and phase transition pressure are classified and associated with the different development stages of the phase transition morphological characteristics, achieving cross-fusion of all feature parameters. On this basis, the BP neural network algorithm is further used to study and obtain the mathematical relationship between all characteristic variables, which is the measurement model of the new liquid-solid phase change detection device.

[0073] Because the units, magnitudes, intervals, and types of signals measured by different sensors vary significantly, it is necessary to convert them to the same standard using a common time reference before classifying, cross-integrating, and analyzing the information. For images that visually reflect the morphology of phase transitions, a convolutional neural network is first used to extract a one-dimensional array, which is then converted into one-dimensional feature data using a normalization algorithm. For the light transmittance, phase transition temperature, and phase transition pressure during the phase transition process, the parameters that can be used for standardization are first determined, and then dimensionless processing is performed to convert them into standard values ​​between 0 and 1. Furthermore, using the same time reference, all information is uniformly classified based on the morphological characteristics of different phase transition processes. Based on the morphological characteristics of liquid-solid phase transitions, the other three characteristic variables are used to quantitatively analyze the morphological characteristics, and the mathematical relationship between the morphological characteristics and light transmittance, phase transition temperature, and phase transition pressure is studied.

[0074] The process of extracting morphological image feature values ​​using convolutional neural networks is as follows: Figure 3 shown.

[0075] Based on the solid morphology under an electron microscope, a convolution kernel is established to identify solid-phase crystals and their arrangement. An image of the object's morphology captured within the measurement chamber is input, and the convolution kernel is used to convolve the input image with the image's digital information to generate a convolutional layer matrix. Subsequently, a pooling procedure is used to reduce the image size while retaining important information. After multiple layers of convolution and pooling, the final pooling layer is activated using the ReLU function, as shown in formula (1).

[0076] f(x)=max(0,x) (1)

[0077] The activated matrix contains the main information of the morphological image. The activated matrix is ​​flattened and converted into a one-dimensional vector, and then the phase change morphological feature data is output in the form of a one-dimensional array.

[0078] Figure 4 The liquid-solid phase transition morphological characteristic measurement model constructed using the BP neural network algorithm is presented. After determining the input and output variables, the input and output data are normalized using the Min-Max normalization method so that their normalized values ​​are dimensionless quantities within the range of 0 to 1:

[0079]

[0080] Where Z represents the input variable or output variable, Z max and Z min Represent the maximum and minimum values ​​respectively.

[0081] Using light transmittance, phase transition temperature, and phase transition pressure as input, the sensor's characteristic parameters of light transmittance, phase transition temperature, and phase transition pressure are measured and Fourier transformed to extract a true and effective target signal. These three characteristic parameters are preprocessed as input variables and converted into dimensionless variables through normalization to reduce the data gradient and accelerate model convergence. These preprocessed input variables are then fed into a constructed BP neural network to train and construct a measurement model for determining the morphological characteristics of liquid-solid phase transitions.

[0082] When using machine learning algorithms to construct a liquid-solid phase transition morphological characteristic measurement model, calibration is performed using standard materials in a stable test environment. During the calibration process, an electron microscope is used to capture the phase transition process of the substance being measured within the measurement chamber. Manual intervention is performed to determine the captured phase transition process images and the various states of the transition process. Simultaneously, the data within the measurement chamber detected by the three sensors is recorded during each capture. The phase transition morphological characteristic data from the electron microscope images is extracted using a convolutional neural network and used as the output set. The data from the three sensors recorded at the time corresponding to each frame is used as the input set and fed into the BP neural network for model training.

[0083] After training, 20% of the images and the corresponding data from the three sensors were selected to verify the model's accuracy. After inputting the data from the three sensors at the corresponding moment, the BP neural network determined the phase transition state and compared it with the image's feature data and content. If the accuracy rate exceeded 95%, the model was validated. The established model for determining liquid-solid phase transition states uses light transmittance, phase transition temperature, and phase transition pressure as input variables, and phase transition morphology characteristics as output variables.

[0084] The above description is merely a preferred embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A detection method using a sample liquid-solid phase change detection device, characterized in that: include: The device comprises: A measuring cavity is provided inside the detection box, and a plurality of light sources are installed on the top of the measuring cavity; a liquid inlet, a pressurization port, and a plurality of sensor installation ports connected to the measuring cavity are provided on the top of the detection box, a liquid discharge port connected to the measuring cavity is provided on the bottom of the detection box, and corresponding observation windows are provided on the front side wall of the measuring cavity and the front side wall of the detection box; The laser sensor unit includes a laser sensor transmitting module and a laser sensor receiving module, wherein the laser sensor transmitting module and the laser sensor receiving module are symmetrically arranged on the left and right sides of the detection box; A sensor group, comprising a temperature sensor and a pressure sensor, wherein the temperature sensor and the pressure sensor are installed in corresponding sensor installation openings; A camera module is provided on one side of the detection box, corresponding to the position of the front side wall of the detection box; A control unit includes a main controller and a signal excitation module, a timing module, a data acquisition module, and a temperature control module electrically connected to the main controller; the camera module, the laser sensor unit, and the sensor group are all electrically connected to the data acquisition module, the signal excitation module is also electrically connected to the laser sensor emission module, and the detection box is electrically connected to the temperature control module; A constant temperature medium layer and a heat preservation layer are sequentially arranged between the outer wall of the measuring cavity and the inner wall of the detection box; A constant temperature medium inlet is provided on the outer wall of the detection box, and the constant temperature medium inlet passes through the insulation layer and is connected to the constant temperature medium layer; Each laser transmitter in the laser sensor transmitting module is provided in a one-to-one correspondence with each laser receiver in the laser sensor receiving module, each laser transmitter is connected to the signal excitation module, and each laser receiver is connected to the data acquisition module; The detection method comprises: Step 1: Filling the measurement cavity with a sample to be tested, wherein the sample to be tested is in a fluid state; Step 2: Turn on the sample liquid-solid phase change detection device, measure the phase change state of the sample to be tested in combination with a preset temperature, and record the measurement data, the measurement data including light transmittance data, morphological image data, phase change temperature data, and phase change pressure data; Step 3: Input the measurement data into a phase change detection model integrated in the main controller to perform feature detection to obtain phase change morphological feature data, wherein the phase change detection model is constructed based on a BP neural network.

2. The method according to claim 1, characterized in that The process of obtaining the measurement data specifically includes: The temperature control module is controlled to cool the sample to be tested until the sample to be tested is cooled to a preset fluid crystallization point threshold, and then the temperature is lowered; during the cooling process of the sample to be tested, the signal excitation module is used to excite the laser sensing unit to obtain light transmittance data; the camera module is used to obtain morphological image data of the sample to be tested in the measurement cavity, the temperature sensor is used to obtain phase change temperature data, and the pressure sensor is used to obtain phase change pressure data.

3. The method according to claim 1, characterized in that Before inputting the measurement data into the phase change detection model, the method further includes: performing feature extraction on the morphological image data to obtain image feature data; The light transmittance data, phase change temperature data, and phase change pressure data are normalized based on a time series to obtain normalized data. The normalized data and the image feature data are converted into one-dimensional matrix data to obtain preprocessed measurement data. The preprocessed measurement data is used as input of the phase change detection model for feature detection.

4. The method according to claim 1, wherein The training process of the phase change detection model specifically includes: Acquiring training data, wherein the training data includes measurement training data and corresponding phase change morphology characteristic data; An initial phase change detection model is constructed, the training data is input into the initial phase change detection model, the characteristics of the morphological image data are quantitatively analyzed using the light transmittance data, phase change temperature data and phase change pressure data to obtain initial phase change morphological feature data, and training is performed with the goal of minimizing the loss between the initial phase change morphological feature data and the phase change morphological feature data corresponding to the measured training data to obtain a trained phase change detection model.

Images