Data-driven rapid prediction method, system, medium and program product for thermal gravimetric properties of chemical looping oxygen carrier materials

Abstract

The present application relates to a kind of fast prediction method, system, medium, program product based on the data-driven chemical chain oxygen carrier material thermal gravimetric property, wherein the method comprises: constructing machine learning model, with the structural features of chemical chain oxygen carrier material and temperature rise rate as input, on the one hand, the prediction of material micro-thermal gravimetric curve is realized, and based on the micro-thermal gravimetric curve of material, the thermal gravimetric curve of material is obtained, and then the thermal gravimetric property is analyzed to obtain;On the other hand, directly realize the prediction of material thermal gravimetric property such as chemical chain oxygen carrier oxygen decoupling temperature and oxygen decoupling mass fraction.Compared with prior art, the present application greatly improves the screening efficiency of chemical chain oxygen carrier, has the technical advantages such as large batch, low cost and high precision, avoids the problems such as time-consuming and laborious of traditional material experiment test, and also can expand application to metal oxide oxygen decoupling temperature identification and performance prediction in other application scenarios such as thermochemical energy storage.

Description

Technical Field

[0001] This invention relates to the field of chemical chaining combustion, and in particular to a data-driven method, system, medium, and program product for rapid prediction of the thermogravimetric properties of chemical chaining oxygen carrier materials. Background Technology

[0002] Chemical looping combustion (CLC) is a novel combustion method for in-situ CO2 capture. Its principle lies in using metal oxides as oxygen carriers to absorb O2 from the air and transfer it to the fuel, thereby achieving in-situ separation and enrichment of CO2 during combustion. Compared to traditional CLC, chemical looping with oxygen uncoupling (CLOU) requires the oxygen carrier to have lattice oxygen decoupling properties. At a certain temperature, the metal oxide decouples oxygen molecules to directly participate in the combustion and conversion reactions of hydrocarbon fuels. Oxygen decoupling can effectively promote the reaction rate of fuel gasification and combustion, alleviate the problem of carbon in the oxygen carrier volume, and thus improve the overall efficiency of the CLC process. Therefore, developing oxygen-decoupling oxygen carriers is a necessary path to realizing CLOU technology, and achieving rapid identification and accurate prediction of the thermogravimetric properties of metal oxides is a key breakthrough in designing high-performance CLOU oxygen carrier materials.

[0003] Currently, research on the thermogravimetric properties of oxygen carrier materials is mainly conducted through thermogravimetric experiments (TGA), using programmed temperature rise to obtain weight loss curves and then analyzing the thermogravimetric properties. However, TGA experiments are time-consuming and costly, and oxygen carrier materials are mainly multi-element metal oxides. The vast array of elemental combinations and structural types, along with the accurate identification and directional design of high-performance oxygen decoupling oxygen carriers, remain unresolved challenges for traditional trial-and-error experimental methods. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art by providing a data-driven method, system, medium, and program product for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials. Compared with traditional thermogravimetric experimental research methods, this invention is low-cost, highly reliable, and facilitates the rapid and accurate identification of the thermogravimetric properties of chemically chained oxygen carrier materials.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] This invention provides a data-driven method for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials, comprising the following steps:

[0007] S1: Using X-ray diffraction data as the feature descriptor of the structure of the chemically chained oxygen carrier material, and using the derivative thermogravimetric curve and thermogravimetric properties as the feature descriptors of the properties of the chemically chained oxygen carrier material, the feature descriptors of the material structure and the temperature rise rate are used as inputs, and the feature descriptors of the material properties are used as outputs to construct a machine learning model. Through training, the prediction model of the derivative thermogravimetric curve and the direct prediction model of the thermogravimetric properties of the chemically chained oxygen carrier material are obtained respectively.

[0008] S2: Based on the characteristics of the target material structure and the rate of temperature rise, on the one hand, the material's derivative thermogravimetric curve is obtained using the trained derivative thermogravimetric curve prediction model, and based on the predicted material derivative thermogravimetric curve, the derivative thermogravimetric curve is converted into thermogravimetric curve through integration. The derivative thermogravimetric curve and the thermogravimetric curve are analyzed to realize the analysis and prediction of the thermogravimetric properties of the chemically chained oxygen carrier material. On the other hand, the thermogravimetric properties of the chemically chained oxygen carrier material are directly predicted using the trained direct prediction model of thermogravimetric properties.

[0009] Further, in S1, the X-ray diffraction data and the derivative thermogravimetric curve include:

[0010] For X-ray diffraction data, the diffraction angle range is unified to 10–85° with an angle step of 0.05°, and the diffraction intensity data range is unified to 0–100, totaling 1500 sets of data points with one-to-one correspondence between angle and intensity, thus achieving structured and unified X-ray diffraction data.

[0011] For the thermogravimetric curves of micro-businesses, the temperature range is uniformly set to 700-1000℃, with a total of 30 sets of data points corresponding one-to-one with temperature and mass change rate.

[0012] Furthermore, in S1, the thermogravimetric property includes:

[0013] Oxygen decoupling temperature and oxygen decoupling mass fraction, wherein the oxygen decoupling temperature includes the initial oxygen decoupling temperature and the optimal oxygen decoupling temperature.

[0014] Further, in S1, the construction of the machine learning model includes:

[0015] Machine learning models were constructed to predict the derivative thermogravimetric curves and the thermogravimetric properties of chemically chained oxygen carrier materials. The models were used to mine the mapping relationship between X-ray diffraction data, temperature rise rate and derivative thermogravimetric curves and thermogravimetric properties of oxygen carrier materials. The machine learning models included one or more of convolutional neural networks, artificial neural networks and Transformers.

[0016] Furthermore, the construction of the Transformer model includes:

[0017] The X-ray diffraction data was divided into 30 equal parts according to the diffraction angle, with each part containing 50 data points. The corresponding diffraction intensity information was converted into a 30×50 sequence format, with 30 as the time step and 50 as the feature dimension, which served as the input format for the Transformer model. After the X-ray diffraction data was embedded through a linear layer, position encoding based on sine and cosine functions was added to increase the position information of the sequence data. This was done using two Transformer Encoder layers as encoders. After encoding, the temperature rise rate information was added as input, and then a fully connected layer was used as a decoder for decoding. Finally, the data points were mapped to the differential thermogravimetric curve in a multi-output format and to the thermogravimetric properties in a single-output format.

[0018] Furthermore, in S2, the analysis and prediction of thermogravimetric properties based on differential thermogravimetric analysis includes:

[0019] The data point where the rate of mass change of the thermogravimetric curve of the micro-merchant is significantly increased is defined as the initial oxygen decoupling point, and its corresponding temperature is the initial oxygen decoupling temperature; the highest point of the thermogravimetric curve of the micro-merchant is defined as the optimal oxygen decoupling point, and its corresponding temperature is the optimal oxygen decoupling temperature.

[0020] The thermogravimetric curve is obtained by integrating the thermogravimetric curve of the micro-entrant. The difference between the mass percentage corresponding to the initial oxygen decoupling temperature and the final temperature is recorded as the oxygen decoupling mass fraction.

[0021] A second aspect of this invention provides a data-driven system for rapidly predicting the thermogravimetric properties of chemically chained oxygen carrier materials, comprising:

[0022] The model building module uses X-ray diffraction data as the feature descriptor of the material structure, the derivative thermogravimetric curve and thermogravimetric properties as the feature descriptors of the material properties, and the structural features and temperature rise rate as inputs and the performance features as outputs to build a machine learning model, which can respectively achieve the prediction of the derivative thermogravimetric curve and the direct prediction of the thermogravimetric properties of chemically chained oxygen carrier materials.

[0023] The performance prediction module, on the one hand, takes X-ray diffraction data of the material as input to obtain the prediction of the derivative thermogravimetric curve of the material. Based on the derivative thermogravimetric curve, it analyzes and obtains the oxygen decoupling temperature and oxygen decoupling mass fraction of the material, realizing the rapid analysis and prediction of the thermogravimetric properties of chemically chained oxygen carrier materials. On the other hand, it takes structural data of the material as input to directly obtain the prediction of thermogravimetric property data.

[0024] Furthermore, the data-driven rapid prediction method and system for the thermogravimetric properties of chemically chained oxygen carrier materials provided by this invention can be extended to the identification and performance prediction of metal oxide oxygen decoupling related application scenarios.

[0025] The oxygen decoupling process of metal oxides can be applied to various low-carbon utilization technology scenarios, such as Chemical Looping with Oxygen Uncoupling (CLOU), Thermochemical Energy Storage (TCES), and Thermochemical Watersplitting (TCWS). Different application scenarios have varying requirements for the thermogravimetric properties of metal oxide materials. Based on the constructed predictive model of the thermogravimetric properties of chemically looping oxygen carrier materials, X-ray diffraction data of metal oxides are input, and the oxygen decoupling temperature is output. The applicable application scenarios for metal oxide materials are identified based on the classification of the initial oxygen decoupling temperature.

[0026] A third aspect of the present invention provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the above-described data-driven method for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials.

[0027] A fourth aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the above-described data-driven method for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials.

[0028] Compared with the prior art, the present invention has the following technical advantages:

[0029] 1) By constructing a machine learning framework to mine the mapping relationship between the structural data of chemically chained oxygen carrier materials and their derivative thermogravimetric curves and thermogravimetric properties, the identification capability of chemically chained oxygen carrier materials is significantly improved. Compared with traditional experimental testing, the method of this invention is more objective and accurate, providing researchers in the field of materials science with a powerful tool.

[0030] 2) This invention enables rapid identification of oxygen carrier materials with oxygen decoupling properties, allowing researchers to quickly predict and analyze the thermogravimetric properties of these materials. This method not only saves valuable time and human resources but also reduces material and energy consumption, aligning with the principles of green chemistry and sustainable development. It provides a crucial scientific foundation and technical support for advancing energy conversion and storage technologies. Attached Figure Description

[0031] Figure 1 A flowchart illustrating the workflow of a data-driven rapid prediction method for the thermogravimetric properties of chemically chained oxygen carrier materials according to an embodiment of the present invention.

[0032] Figure 2This is a schematic diagram of the structure of the Transformer model provided in an embodiment of the present invention;

[0033] Figure 3 A comparison chart of the predicted and actual values ​​of the derivative thermogravimetric curves of a portion of the materials in the training set provided by an embodiment of the present invention;

[0034] Figure 4 This is a comparison chart of the predicted and actual values ​​of the derivative thermogravimetric curves of some materials in the test set provided according to an embodiment of the present invention.

[0035] Figure 5 A comparison chart of the actual and predicted values ​​of the material's initial oxygen decoupling temperature obtained from derivative thermogravimetric curve analysis according to an embodiment of the present invention;

[0036] Figure 6 This is a comparison chart of the actual and predicted values ​​of the optimal oxygen decoupling temperature of the material obtained from the derivative thermogravimetric curve analysis according to an embodiment of the present invention.

[0037] Figure 7 This is a comparison chart of the actual and predicted values ​​of thermogravimetric curves of some materials obtained based on the thermogravimetric curve analysis of the derivative quotient according to an embodiment of the present invention.

[0038] Figure 8 This is a comparison chart of the actual and predicted values ​​of the oxygen decoupling mass fraction of the material obtained from thermogravimetric analysis according to an embodiment of the present invention.

[0039] Figure 9 This is a comparison chart of the actual and predicted values ​​of the material initial oxygen decoupling temperature according to an embodiment of the present invention;

[0040] Figure 10 This is a comparison chart of the actual and predicted values ​​of the oxygen decoupling mass fraction of the material according to an embodiment of the present invention. Detailed Implementation

[0041] Overall, this invention provides a data-driven method, system, and medium for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials. Specifically, the method uses X-ray diffraction data as a feature descriptor of the oxygen carrier material's structure and the derivative thermogravimetric curve as a feature descriptor of the material's properties. Structural features and temperature rise rate are used as inputs, and performance features are used as outputs to construct a machine learning model to predict the derivative thermogravimetric curve. Based on the material's derivative thermogravimetric curve, further analysis yields the oxygen decoupling temperature and oxygen decoupling mass fraction data, enabling the analysis and prediction of the thermogravimetric properties of the chemically chained oxygen carrier material. This invention constructs a mapping relationship between the structural data of the oxygen carrier material and the derivative thermogravimetric curve through a machine learning model. It is low-cost, highly reliable, and enables rapid identification of oxygen carrier materials with oxygen decoupling characteristics, avoiding the time-consuming and labor-intensive problems of traditional material testing experiments.

[0042] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Component models, material names, connection structures, control methods, algorithms, and other features not explicitly described in this technical solution are considered common technical features disclosed in the prior art.

[0043] Example 1

[0044] This embodiment discloses a data-driven method for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials, referencing... Figure 1 As shown, the specific steps include the following:

[0045] S1: Using X-ray diffraction data as the feature descriptor of the structure of the chemically chained oxygen carrier material, and using the derivative thermogravimetric curve and thermogravimetric properties as the feature descriptors of the properties of the chemically chained oxygen carrier material, the feature descriptors of the material structure and the temperature rise rate are used as inputs, and the feature descriptors of the material properties are used as outputs to construct a machine learning model. Through training, the prediction model of the derivative thermogravimetric curve and the direct prediction model of the thermogravimetric properties of the chemically chained oxygen carrier material are obtained respectively.

[0046] S2: Based on the characteristics of the target material structure and the temperature rise rate, on the one hand, the material's derivative thermogravimetric curve is obtained using the trained derivative thermogravimetric curve prediction model. Based on the predicted material derivative thermogravimetric curve, the derivative thermogravimetric curve is converted into thermogravimetric curve through integration. The derivative thermogravimetric curve and the thermogravimetric curve are analyzed to achieve the analysis and prediction of the thermogravimetric properties of the chemically chained oxygen carrier material. On the other hand, the thermogravimetric properties of the chemically chained oxygen carrier material are directly predicted using the trained direct prediction model for thermogravimetric properties. Specifically, in step S1, the structuring of the X-ray diffraction data and the acquisition of the oxygen decoupling temperature include:

[0047] For X-ray diffraction data, the diffraction angle range is unified to 10–85° with an angle step of 0.05°, and the diffraction intensity data range is unified to 0–100, totaling 1500 sets of data points with one-to-one correspondence between angle and intensity, thus achieving structured and unified X-ray diffraction data.

[0048] For the thermogravimetric curves of micro-businesses, the temperature range is uniformly set to 700-1000℃, with a total of 30 sets of data points corresponding one-to-one with temperature and mass change rate.

[0049] In specific implementation, step S1, the construction of the machine learning model, includes:

[0050] A machine learning model was constructed to predict the derivative thermogravimetric curves of chemically chained oxygen carrier materials. The model is used to mine the mapping relationship between X-ray diffraction data, temperature rise rate, and the derivative thermogravimetric curves of the oxygen carrier materials. The machine learning model includes one or more of the following: convolutional neural network, artificial neural network, random forest, and Transformer. Specifically, a schematic diagram of the Transformer model is shown below. Figure 2 As shown.

[0051] In specific implementation, this step uses X-ray diffraction data and temperature rise rate as model inputs, and derivative thermogravimetric curve data as model outputs to construct a Transformer-based derivative thermogravimetric curve prediction model. A total of 28 sets of data were used, divided into training and testing sets in an 8:2 ratio. The training set was used to train the model, and the testing set was used to verify the model's prediction performance. After model training, the model was analyzed using R... 2 Alternatively, the mean absolute error (MAE) can be used for model evaluation, calculated as follows:

[0052]

[0053] Among them, y i For the true value, For predicted values, This is the average value.

[0054] The results show that the trained Transformer model has the following prediction performance for the differential thermal curve: the average R0 of the training set is... 2 The R-value is 0.97, and the average R-value on the test set is [value missing]. 2 The value is 0.60. The comparison charts of the differential thermogravimetric curves of some materials predicted by the model on the training and test sets with the actual values ​​are shown below. Figure 3 , Figure 4As shown, the listed materials include: CuO, CuO modified with different MgO and Al2O3 ratios (Cu90MgAl-1000, Cu70MgAl-1000, Cu90Al-1000), 60% cerium-doped CuO (Cu60Ce), and CuO with ZrO2 support (CuO-ZrO2).

[0055] In specific implementation, the oxygen decoupling temperature is predicted based on the analysis of the derivative thermogravimetric curves. The model predicts the corresponding derivative thermogravimetric curves, and further analysis of each curve yields the oxygen decoupling temperature of the corresponding material, including the initial oxygen decoupling temperature and the optimal oxygen decoupling temperature. A comparison between the actual and predicted values ​​of the initial oxygen decoupling temperature is provided. Figure 5 As shown, the comparison between the actual and predicted values ​​of the optimal oxygen decoupling temperature is as follows: Figure 6 As shown, the mean absolute error between the true and predicted values ​​of the initial oxygen decoupling temperature is 7.73℃ on the training set and 15.00℃ on the test set; the mean absolute error between the true and predicted values ​​of the optimal oxygen decoupling temperature is 2.27℃ on the training set and 8.33℃ on the test set.

[0056] In specific implementation, the thermogravimetric curve of the micro-material is transformed into a thermogravimetric curve through integration, and the oxygen decoupling mass fraction is further predicted. For some materials, the actual and predicted thermogravimetric curve values ​​are compared... Figure 7 As shown. The model predicts the corresponding derivative thermogravimetric curves, and further integrates each derivative thermogravimetric curve to convert it into a thermogravimetric curve, thus obtaining the oxygen decoupling mass fraction data of the corresponding material. A comparison between the actual and predicted values ​​of the oxygen decoupling mass fraction is shown below. Figure 8 As shown, the mean absolute error between the true and predicted values ​​of oxygen decoupling mass fraction is 0.24 on the training set and 0.8 on the test set.

[0057] In this embodiment, a machine learning model is constructed to mine the mapping relationship between the material structure data, temperature rise rate, and derivative thermogravimetric curve of the chemically chained oxygen carrier. Based on this derivative thermogravimetric curve, the model enables the analysis and prediction of the thermogravimetric properties of the chemically chained oxygen carrier material. This model significantly improves the ability to identify the thermogravimetric properties of oxygen carrier materials, avoiding the time-consuming and labor-intensive problems of traditional material testing experiments.

[0058] This invention also discloses a data-driven rapid prediction system for the thermogravimetric properties of chemically chained oxygen carrier materials, specifically including:

[0059] The model building module uses X-ray diffraction data as the feature descriptor of the material structure and the derivative thermogravimetric curve as the feature descriptor of the material properties. It uses structural features and temperature rise rate as inputs and performance features as outputs to build a machine learning model to achieve direct prediction of the derivative thermogravimetric curve of chemically chained oxygen carrier materials.

[0060] The performance prediction module, based on the analysis of the derivative thermogravimetric curve of the material, obtains the oxygen decoupling temperature and oxygen decoupling mass fraction data of the material, enabling rapid analysis and prediction of the thermogravimetric properties of chemically bonded oxygen carrier materials.

[0061] The modules described above as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0062] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional units.

[0063] This embodiment also proposes a computer-readable storage medium storing computer instructions for instructing a computer to execute the aforementioned data-driven method for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials. The storage medium can be an electronic medium, magnetic medium, optical medium, electromagnetic medium, infrared medium, or a semiconductor system or propagation medium. The storage medium may also include semiconductor or solid-state memory, magnetic tape, removable computer disk, random access memory (RAM), read-only memory (ROM), hard disk, and optical disc. Optical discs may include optical disc-read-only memory (CD-ROM), optical disc-read-write (CD-RW), and DVD.

[0064] This embodiment also provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned data-driven method for rapidly predicting the thermogravimetric properties of chemically chained oxygen carrier materials. When this computer program product runs on various compatible processors, it can quickly and accurately call upon the necessary computing resources and process the input data related to the chemically chained oxygen carrier material, such as X-ray diffraction data and temperature rise rate, according to a predetermined process. By executing the constructed machine learning model, whether it is a convolutional neural network, an artificial neural network, or a Transformer model, it can achieve the mapping calculation from material structural features to performance features, thereby accurately predicting the derivative thermogravimetric curve and thermogravimetric properties of the chemically chained oxygen carrier material. Furthermore, this computer program product also has good compatibility and can be adapted to various operating systems and hardware platforms.

[0065] Example 2

[0066] This embodiment discloses a data-driven method for rapid prediction of the oxygen decoupling temperature of chemically chained oxygen carrier materials, which specifically includes the following steps:

[0067] Using X-ray diffraction data as the feature descriptor of the material structure and the initial oxygen decoupling temperature as the feature descriptor of the material properties, with structural features and temperature rise rate as inputs and initial oxygen decoupling temperature as output, a Transformer model is constructed using machine learning to directly predict the initial oxygen decoupling temperature of chemically chained oxygen carrier materials.

[0068] Eighty sets of X-ray diffraction data and initial oxygen decoupling temperature data for chemically chained oxygen carrier materials were collected and divided into training and testing sets at a 9:1 ratio. The training set was used to train the model, and the testing set was used to verify the model's performance. After model training, R... 2 Conduct model evaluation.

[0069] The results show that the trained Transformer model has the following prediction performance for the initial oxygen decoupling temperature: (Training set R) 2 The value is 0.767, and the test set R is... 2 The value is 0.636. The comparison between the model's predicted and actual values ​​for the initial oxygen decoupling temperature of the material on the training and test sets is shown in the figure below. Figure 9 As shown.

[0070] Example 3

[0071] This embodiment discloses a data-driven method for rapid prediction of oxygen decoupling mass fraction in chemically chained oxygen carrier materials, specifically including:

[0072] Using X-ray diffraction data as a feature descriptor for material structure and oxygen decoupling mass fraction as a feature descriptor for material properties, with structural features and temperature rise rate as inputs and oxygen decoupling mass fraction as output, a Transformer model is constructed using machine learning to directly predict the oxygen decoupling mass fraction of chemically chained oxygen carrier materials.

[0073] A total of 130 sets of X-ray diffraction data and oxygen decoupling mass fraction data of chemically chained oxygen carrier materials were collected and divided into training and testing sets at a 9:1 ratio. The training set was used to train the model, and the testing set was used to verify the model's performance. After model training, based on R... 2 Conduct model evaluation.

[0074] The results show that the trained Transformer model has the following prediction performance for oxygen decoupling quality fraction: Training set R 2 The value is 0.988, and the test set R is... 2 The value is 0.890. The comparison between the model's predicted and actual values ​​for the oxygen decoupling mass fraction of the material on the training and test sets is shown in the figure below. Figure 10 As shown.

[0075] Example 4

[0076] This embodiment discloses a classification and identification method based on the initial oxygen decoupling temperature, which enables the identification of application scenarios related to the oxygen decoupling process of metal oxides. The method specifically includes the following steps:

[0077] The oxygen decoupling temperature was obtained by analyzing the differential thermogravimetric curve prediction model for chemically chained oxygen carrier materials constructed in Example 1, or by directly obtaining the initial oxygen decoupling temperature of the material using the thermogravimetric property prediction model constructed in Example 2, further identifying the application scenarios of the oxygen decoupling process of metal oxide materials. A total of 164 sets of structural and performance data of different metal oxide materials studied under the CLOU, TCES, and TCWS scenarios were used. Their X-ray diffraction data and temperature rise rates were input to obtain the oxygen decoupling temperature analysis. Based on the classification of initial oxygen decoupling temperature, less than 750℃, 750–950℃, and greater than 950℃ were respectively assigned to CLOU, TCES, and TCWS. The final identification results of the material application scenarios are shown in Table 1. The identification accuracy rate for the CLOU scenario was 96.55%, for the TCES scenario it was 67.57%, and for the TCWS scenario it was 96.39%.

[0078] Table 1. Identification Results of Material Application Scenarios

[0079]

[0080] The above description of the embodiments is provided to enable those skilled in the art to understand and use the invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the invention should be within the protection scope of the present invention.

Claims

1. A data-driven method for rapid prediction of the thermogravimetric properties of chemically chained oxygen carrier materials, characterized in that, Includes the following steps: S1: Using X-ray diffraction data as the feature descriptor of the structure of the chemically chained oxygen carrier material, and using the derivative thermogravimetric curve and thermogravimetric properties as the feature descriptors of the properties of the chemically chained oxygen carrier material, the feature descriptors of the material structure and the temperature rise rate are used as inputs, and the feature descriptors of the material properties are used as outputs to construct a machine learning model. Through training, the prediction model of the derivative thermogravimetric curve and the direct prediction model of the thermogravimetric properties of the chemically chained oxygen carrier material are obtained respectively. S2: Based on the characteristics of the target material structure and the temperature rise rate, on the one hand, the material's derivative thermogravimetric curve is obtained using the trained derivative thermogravimetric curve prediction model, and based on the predicted material derivative thermogravimetric curve, the derivative thermogravimetric curve is converted into thermogravimetric curve through integration. The derivative thermogravimetric curve and the thermogravimetric curve are analyzed to realize the analysis and prediction of the thermogravimetric properties of the chemical chain oxygen carrier material. On the other hand, the thermogravimetric properties of the chemical chain oxygen carrier material are directly predicted using the trained direct prediction model of thermogravimetric properties. In S1, the X-ray diffraction data has a diffraction angle range of 10~85° with an angle step of 0.05° and a diffraction intensity range of 0~100, totaling 1500 sets of data points with one-to-one correspondence between angle and intensity, thus achieving structured uniformity of X-ray diffraction data; for the differential thermogravimetric curve, the temperature range is uniformly set to 700~1000℃, totaling 30 sets of data points with one-to-one correspondence between temperature and mass change rate. In S1, the thermogravimetric property includes oxygen decoupling temperature and oxygen decoupling mass fraction, wherein the oxygen decoupling temperature includes initial decoupling temperature and optimal decoupling temperature; In S1, the machine learning model is a Transformer model, and its construction process includes: dividing the X-ray diffraction data into 30 equal parts according to the diffraction angle, each part containing 50 sets of data points, and converting the corresponding diffraction intensity information into a 30×50 sequence format, where 30 is the time step and 50 is the feature dimension, to match the input format of the Transformer model; after the X-ray diffraction data is embedded through a linear layer, position encoding based on sine and cosine functions is added to add position information to the sequence data, and two Transformer Encoder layers are used as encoders for encoding; based on the encoded data, the input of temperature rise rate information is added, and then a fully connected layer is used as a decoder for decoding, and finally the data points are mapped to the differential thermogravimetric curve in a multi-output form and mapped to the thermogravimetric properties in a single-output form; In S2, the analysis and prediction of thermogravimetric properties based on derivative thermogravimetric curves includes: defining the data point in the derivative thermogravimetric curve where the rate of mass change increases significantly as the initial oxygen decoupling point, and the corresponding temperature as the initial oxygen decoupling temperature; defining the highest point of the derivative thermogravimetric curve as the optimal oxygen decoupling point, and the corresponding temperature as the optimal oxygen decoupling temperature; performing integral calculation on the derivative thermogravimetric curve to obtain the thermogravimetric curve; and recording the difference between the mass percentage corresponding to the initial oxygen decoupling temperature and the final temperature as the oxygen decoupling mass fraction.

2. The method for rapid prediction of thermogravimetric properties of chemically chained oxygen carrier materials based on data-driven methods according to claim 1, characterized in that, In S1, the machine learning model also includes convolutional neural networks and artificial neural networks.

3. A data-driven rapid prediction system for the thermogravimetric properties of chemically chained oxygen carrier materials, characterized in that, include: The model building module uses X-ray diffraction data as the feature descriptor of the chemical chain oxygen carrier material structure, and the derivative thermogravimetric curve and thermogravimetric properties as the feature descriptors of the chemical chain oxygen carrier material properties. The feature descriptors of the material structure and the temperature rise rate are used as inputs, and the feature descriptors of the material properties are used as outputs to build a machine learning model. Through training, a prediction model for the derivative thermogravimetric curve and a direct prediction model for the thermogravimetric properties of the chemical chain oxygen carrier material are obtained respectively. The performance prediction module, based on the characteristics of the target material structure and the rate of temperature rise, uses the trained differential thermogravimetric curve prediction model to obtain the differential thermogravimetric curve of the material. Based on the predicted differential thermogravimetric curve, the differential thermogravimetric curve is converted into thermogravimetric curve through integration. The differential thermogravimetric curve and the thermogravimetric curve are analyzed to realize the analysis and prediction of the thermogravimetric properties of the chemical chain oxygen carrier material. On the other hand, the module uses the trained direct prediction model of thermogravimetric properties to directly obtain the prediction of the thermogravimetric properties of the chemical chain oxygen carrier material. The X-ray diffraction data includes diffraction angles ranging from 10° to 85° with an angle step of 0.05°, and diffraction intensity data ranging from 0° to 100°, totaling 1500 sets of data points with one-to-one correspondence between angle and intensity, achieving structured and unified X-ray diffraction data; for the thermogravimetric curves, the temperature range is unified from 700°C to 1000°C, totaling 30 sets of data points with one-to-one correspondence between temperature and mass change rate. The thermogravimetric properties include oxygen decoupling temperature and oxygen decoupling mass fraction, wherein the oxygen decoupling temperature includes the initial decoupling temperature and the optimal decoupling temperature; The machine learning model is a Transformer model, and its construction process includes: dividing the X-ray diffraction data into 30 equal parts according to the diffraction angle, each part containing 50 sets of data points, and converting the corresponding diffraction intensity information into a 30×50 sequence format, where 30 is the time step and 50 is the feature dimension, to match the input format of the Transformer model; after the X-ray diffraction data is embedded through a linear layer, position encoding based on sine and cosine functions is added to add position information to the sequence data, and two Transformer Encoder layers are used as encoders for encoding; based on the encoded data, the input of temperature rise rate information is added, and then a fully connected layer is used as a decoder for decoding, and finally the data points are mapped to the differential thermogravimetric curve in a multi-output form and mapped to the thermogravimetric properties in a single-output form; In S2, the analysis and prediction of thermogravimetric properties based on derivative thermogravimetric curves includes: defining the data point in the derivative thermogravimetric curve where the rate of mass change increases significantly as the initial oxygen decoupling point, and the corresponding temperature as the initial oxygen decoupling temperature; defining the highest point of the derivative thermogravimetric curve as the optimal oxygen decoupling point, and the corresponding temperature as the optimal oxygen decoupling temperature; performing integral calculation on the derivative thermogravimetric curve to obtain the thermogravimetric curve; and recording the difference between the mass percentage corresponding to the initial oxygen decoupling temperature and the final temperature as the oxygen decoupling mass fraction.

4. A storage medium containing computer-executable instructions, characterized in that, When executed by a computer processor, the storage medium of the computer-executable instructions is used to perform the data-driven rapid prediction method for the thermogravimetric properties of chemically chained oxygen carrier materials as described in any one of claims 1 to 2.

5. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the data-driven rapid prediction method for the thermogravimetric properties of chemically chained oxygen carrier materials as described in any one of claims 1 to 2.

Images