Gelling system performance prediction method based on raw material oxide
By constructing a gelling system database and establishing an oxide performance prediction model using machine learning methods, the problem of low prediction accuracy caused by incomplete raw material parameters in the existing technology is solved, and accurate prediction and efficient research and development of gelling material performance are achieved.
Patent Information
- Application Number
- CN202510256041.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-05
- Publication Date
- 2025-06-06
AI Technical Summary
The existing gelling material performance prediction methods are difficult to adapt to the diverse raw material combinations, and most of them rely on a large amount of experimental data and complex parameter systems. Especially when the raw material parameters are incomplete, the prediction accuracy is low, resulting in low R&D efficiency.
By constructing a gelling system database, analyzing the composition content and molding conditions of raw material oxides, and combining machine learning methods to establish a predictive model of the performance of each phase oxide to achieve accurate prediction of the performance of gelling materials.
This method can achieve accurate prediction of gelled material performance under limited raw material parameters, improve R&D efficiency, support the efficient use of different types of raw materials, and promote the green transformation and sustainable development of the building materials industry.
Smart Images
Figure CN120108596A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of cement materials, and in particular to a method for predicting the performance of a cementitious system based on raw material oxides, and is particularly suitable for evaluating the performance of a cementitious system under the condition of incomplete raw material parameters. Background Art
[0002] As society pursues sustainable development, greening and low-carbonization of building materials have become an important trend. Traditional cementitious materials such as cement have high energy consumption and large carbon emissions during the production process, so it is of great significance to develop new cementitious systems based on low-carbon raw materials. The performance optimization of traditional cementitious materials mainly relies on the trial-and-error cycle of "design-preparation-testing". The development cycle of a single cementitious system usually takes 6-8 months, resulting in low R&D efficiency.
[0003] In recent years, machine learning technology has shown great potential in predicting material properties, but existing methods still have many shortcomings. Existing models can usually only predict specific types of cement-based composites, and are difficult to adapt to a variety of raw material combinations. Most of them rely on a large amount of experimental data and complex parameter systems, such as detailed hydration kinetic parameters, mineral phase composition, and microstructural characteristics. Especially when different types of raw materials are used in practical applications, they often face problems of different types or incomplete parameters. In addition, the input parameters of existing models are usually limited to the proportions of specific materials, ignoring the diversity of material components and unable to effectively handle different components of the same type of materials.
[0004] Therefore, developing a simplified prediction method based on the oxide composition of raw materials, which can accurately predict the performance under limited parameters, has important theoretical significance and practical application value. This method will help overcome the limitations of existing methods, provide support for the efficient use of different types of raw materials in cementitious materials, and promote the green transformation and sustainable development of the building materials industry. Summary of the invention
[0005] In view of the shortcomings of the prior art, the present invention provides a method for predicting the performance of a cementitious system based on raw material oxides. By analyzing the composition and content of the raw material oxides and combining them with the molding conditions, accurate prediction of the performance of cementitious materials can be achieved, effectively solving the accuracy problem caused by insufficient raw material parameters in traditional predictions, providing the cement industry with a more scientific and sustainable cementitious material optimization solution, and helping to promote the industry to develop in a green and efficient direction.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:
[0007] The present invention discloses a method for predicting the performance of a gelling system based on raw material oxides, the steps of which are as follows:
[0008] (1) Construct a gelling system database by combining the phase composition, component data of each phase, mineral phase content, molding conditions and performance parameters in different gelling systems;
[0009] (2) Based on the oxide data and mineral phase content of different raw materials, the relationship between oxide content and mineral phase is obtained;
[0010] (3) Use machine learning methods to establish a prediction model for the performance of each phase of oxides.
[0011] Preferably, in step (1), the phase composition includes a clinker phase, a calcium phase, a silicon-aluminum phase and other auxiliary cementitious material phases, wherein the clinker phase mainly includes cement and clinker of different types; the calcium phase mainly includes CaCO 3 The silicon-aluminum phase mainly contains raw materials with metakaolin as the main active ingredient; the other auxiliary cementitious material phase is raw materials with hydraulic or volcanic ash activity, mainly including fly ash, silica fume and slag, and the active ingredients are mainly active CaO, SiO 2 and Al 2 O 3 .
[0012] Preferably, in step (1), the component data of each phase includes the raw material ratio of each phase and the CaO, SiO contained in the material. 2 、Al 2 O 3 components and normalize them.
[0013] Preferably, the mineral phase is mainly divided into an active phase and an inert phase.
[0014] Preferably, in step (1), the molding conditions mainly include water-cement ratio, water reducing agent, and curing age; and the performance parameters mainly include compressive flexural strength, tensile flexural strength, flexural strength and durability.
[0015] Preferably, in step (2), the different raw materials are raw materials mainly composed of silicon-aluminum phase.
[0016] Preferably, in step (3), the machine learning method includes a single regression model and an integrated regression model.
[0017] Preferably, the single regression model includes linear regression, decision tree, support vector machine, Gaussian process regression, random forest and artificial neural network; the integrated regression model is obtained by integrating two or more single regression models.
[0018] The present invention has the following beneficial effects:
[0019] 1. The present invention proposes a method for predicting the performance of a cementitious system based on raw material oxides. Only the component content and molding conditions of the raw material oxides are required as input parameters, which is applicable to various types of cement and cementitious materials and has strong versatility.
[0020] 2. The present invention can predict the performance of the content and oxide composition of any phase, solving the problem of prediction failure caused by insufficient raw material parameters in existing prediction methods. Through this method, the designer can reasonably adjust the proportion of each phase of the gelling system according to the oxide content of the raw materials, thus realizing accurate prediction and control of the performance of the gelling system.
[0021] 3. The present invention establishes a performance prediction model based on raw material oxides through the correlation between raw material oxides and mineral phases, expands the operability of performance prediction, and provides a data-driven scientific method for the design and optimization of the cementitious system. Through the machine learning model, it is possible to accurately predict the impact of different oxide compositions on the performance of the cementitious system, thereby guiding the optimization and proportioning design of raw materials, reducing the cost of experimental trial and error, improving the efficiency of cementitious material research and development, and providing theoretical support and practical basis for the efficient use of other auxiliary cementitious materials and the development of green cementitious materials. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a flow chart of the performance prediction of the gelling system of the present invention;
[0023] Figure 2 This is the correlation analysis diagram between oxide content and mineral phase;
[0024] Figure 3 It is the advantage map of two models: support vector regression and Gaussian process regression;
[0025] Figure 4 This is a graph of the prediction accuracy of the model constructed in Example 1. DETAILED DESCRIPTION
[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0027] Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.
[0028] refer to Figure 1 As shown, the present invention discloses a method for predicting the performance of a gelling system based on raw material oxides, the steps are as follows:
[0029] (1) Obtain the specific component data of clinker phase, calcium phase, silicon-aluminum phase and other auxiliary cementitious material phases, mineral phase content and water-cement ratio, water reducer, curing age and other auxiliary forming conditions and their corresponding performance data in different cementitious systems to build a cementitious system database;
[0030] Among them, the main distinguishing features of clinker phase, calcium phase, silicon-aluminum phase and other auxiliary cementitious material phases are as follows: clinker phase mainly refers to different types of cement and clinker, etc.; calcium phase mainly includes CaCO 3 The raw materials with quartz as the main component are as inert fillers, which partially dissolve in the cementitious system and release calcium ions to participate in the hydration reaction; the silicon-aluminum phase mainly contains raw materials with metakaolin as the main active component, which provides silicate activity during the hydration process and contains inert mineral phases such as quartz; other auxiliary cementitious material phases mainly include fly ash, silica fume and slag and other auxiliary cementitious materials, whose active components are mainly active CaO, SiO 2 and Al 2 O 3 , with potential hydration activity or pozzolanic effect.
[0031] The specific composition data of clinker phase, calcium phase, silicon-aluminum phase and other auxiliary cementitious material phases include the raw material ratio of each phase and the CaO, SiO 2 、Al 2 O 3 components and normalize them.
[0032] The mineral phase is mainly divided into an active phase mainly composed of metakaolin and an inert phase represented by quartz. The performance parameters include but are not limited to compressive flexural strength, tensile flexural strength, flexural strength and durability.
[0033] (2) The oxide data and mineral phase content of different raw materials are obtained through the database, and correlation analysis is performed to obtain the relationship between the oxide content and the mineral phase; among them, the different raw materials are mainly raw materials with silicon-aluminum phase as the main component.
[0034] (3) Based on the database, a machine learning method is used to establish a prediction model for the performance of each phase of oxides, and the prediction and verification are carried out based on the existing raw materials.
[0035] Among them, the machine learning methods include single regression models and integrated regression models. The single regression models include linear regression, decision tree, support vector machine, Gaussian process regression, random forest and artificial neural network; the integrated regression model is obtained by integrating two or more single regression models.
[0036] The present invention will be further described below in conjunction with specific embodiments.
[0037] Example 1
[0038] This embodiment is mainly aimed at a cementitious system with silicate cement as clinker, limestone as calcium phase, calcined clay as silicon-aluminum phase and slag and coal gangue as other auxiliary cementitious material phases, and the target performance is 28-day compressive strength.
[0039] Take the existing oxide composition prediction of P 42.5 cement, limestone, calcined clay and fly ash as an example:
[0040] (1) Through literature retrieval, the specific component data of clinker phase, calcium phase, silicon-aluminum phase and other auxiliary cementitious material phases in different cementitious systems, the mineral phase content and water-cement ratio, water reducer, age and other auxiliary forming conditions and their corresponding performance data were obtained to form a cementitious system database; among them, the clinker phase mainly refers to different types of cement and clinker, and the calcium phase mainly includes CaCO 3 The raw materials are the main components, the silicon-aluminum phase mainly contains raw materials with kaolin as the main active ingredient, and other auxiliary cementitious material phases mainly include other auxiliary cementitious materials such as slag and fly ash. The mineral phases are mainly inert phases and active phases. The age is 28 days, and the corresponding performance data is compressive strength.
[0041] (2) The oxide data and mineral phase content of different raw materials are obtained through the database, and the correlation analysis is performed to obtain the relationship between the oxide content and the mineral phase; among them, the different raw materials are mainly raw materials with silicon and aluminum phase as the main components. The relationship between the oxide content and the mineral phase is as follows: Figure 2 As shown in the figure, the contents of active phase and inert phase are respectively related to Al 2 O 3 、SiO 2 The content of CaO is highly correlated; CaO is directly related to clinker and calcium phase raw materials, and directly affects the performance of cementitious materials. 2 、Al 2 O 3 It is reasonable to use the content to represent the activity of the raw materials.
[0042] (3) Based on the database, a machine learning method is used to establish a prediction model for the performance of each phase of oxide; wherein the machine learning method includes all applicable single regression models and integrated regression models; the single regression model includes linear regression, decision tree, support vector machine, Gaussian process regression, random forest and artificial neural network; the integrated regression model is obtained by integrating two or more single regression models. This embodiment combines the advantages of support vector regression and Gaussian process regression models, such as Figure 3 As shown in Figure 2, an integrated performance prediction model is constructed with a prediction accuracy of more than 0.98. Figure 4 shown.
[0043] (4) Given a water-cement ratio of 0.4, a water-reducing agent content of 0, and an age of 28 days, the contents of the clinker phase, calcium phase, silicon-aluminum phase, and other auxiliary cementitious material phases of P 42.5 cement, limestone, calcined clay, and fly ash, as well as their CaO, SiO 2 、Al 2 O 3 The performance prediction is shown in Table 1.
[0044] Table 1 Composition and content of each phase
[0045]
[0046] (5) The experiment was carried out according to the ratio in Table 1, and the actual compressive strength was 31.9 MPa with an error of 3.13%, which verified the feasibility of the prediction model.
[0047] Example 2
[0048] The same as Example 1, except that the content of each phase is different, and the performance prediction is carried out. The composition and content of each phase are shown in Table 2.
[0049] Table 2 Composition and content of each phase
[0050]
[0051] The experiment was carried out according to the ratio in Table 2, and the actual compressive strength was 36.8 MPa with an error of 1.87%, which verified the feasibility of the prediction model.
[0052] According to the above embodiments, the performance prediction of the content and oxide composition of any phase can be carried out, which solves the problem of prediction failure caused by insufficient raw material parameters in existing prediction methods. Through this method, the designer can reasonably adjust the proportion of each phase of the gel system according to the oxide content that is relatively easy to obtain in the raw materials, thereby realizing accurate prediction and control of the performance of the gel system.
[0053] The embodiments described above are only descriptions of the preferred modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.
Claims
1. A method for predicting the performance of a gelling system based on raw material oxides, characterized in that: Here are the steps: (1) Construct a gelling system database by combining the phase composition, component data of each phase, mineral phase content, molding conditions and performance parameters in different gelling systems; (2) Based on the oxide data and mineral phase content of different raw materials, the relationship between oxide content and mineral phase is obtained; (3) Use machine learning methods to establish a prediction model for the performance of each phase of oxides.
2. The method for predicting the performance of a gelling system based on raw material oxides according to claim 1, characterized in that: In step (1), the phase composition includes clinker phase, calcium phase, silicon-aluminum phase and other auxiliary cementitious material phases, wherein the clinker phase mainly includes cement and clinker of different types; the calcium phase mainly includes raw materials with CaCO3 as the main component; the silicon-aluminum phase mainly includes raw materials with kaolin as the main active ingredient; the other auxiliary cementitious material phases are raw materials with hydraulic or volcanic ash activity, mainly including fly ash, silica ash and slag, and the active ingredients are mainly active CaO, SiO2 and Al2O3.
3. A method for predicting the performance of a gelling system based on raw material oxides according to claim 1 or 2, characterized in that: In step (1), the component data of each phase include the raw material ratio of each phase and the CaO, SiO2, and Al2O3 components contained in the material, and they are normalized.
4. The method for predicting the performance of a gelling system based on raw material oxides according to claim 1, characterized in that: The mineral phase is mainly divided into an active phase and an inert phase.
5. The method for predicting the performance of a gelling system based on raw material oxides according to claim 1, characterized in that: In step (1), the molding conditions mainly include water-cement ratio, water reducing agent, and curing age; the performance parameters mainly include compressive flexural strength, tensile flexural strength, flexural strength and durability.
6. The method for predicting the performance of a gelling system based on raw material oxides according to claim 2, characterized in that: In step (2), the different raw materials are mainly raw materials mainly composed of silicon-aluminum phase.
7. The method for predicting the performance of a gelling system based on raw material oxides according to claim 1, characterized in that: In step (3), the machine learning method includes a single regression model and an integrated regression model.
8. The method for predicting the performance of a gelling system based on raw material oxides according to claim 1, characterized in that: The single regression model includes linear regression, decision tree, support vector machine, Gaussian process regression, random forest and artificial neural network; the integrated regression model is obtained by integrating two or more single regression models.