A method for predicting the flame retardant ability of flame retardants

Abstract

The invention discloses a method for predicting the flame retardant ability of a flame retardant, and relates to the field of chemical safety technology. It fills the gap in the prior art that there is no method for directly predicting the flame retardant ability of a flame retardant. The prediction method mainly uses a statistical modeling method to establish a prediction model for the flame retardant ability of a flame retardant, and uses the quantitative relationship between the molecular structure of a flame retardant and a combustible gas mixture and the change value of the explosion limit range of the mixed system to predict the flame retardant ability of the flame retardant. It includes: using a combustible gas explosion limit test method to obtain the data sample required for the establishment of the prediction model; obtaining the change value of the explosion limit range of the combustible gas mixture before and after the addition of the flame retardant as the prediction target value; constructing a stable molecular structure of each component; calculating the molecular structure descriptor of the mixture; screening the optimal mixture molecular descriptor and the like. The prediction method of the invention can quickly obtain the flame retardant ability of the flame retardant through theoretical calculation.

Description

Technical Field

[0001] The invention relates to the technical field of chemical production safety, and in particular to a method for predicting the flame retardant ability of a flame retardant. Background Art

[0002] Flame retardants are widely used in all aspects of production and life. For example, perhalogenated hydrocarbon flame retardants are added to flammable refrigerants to suppress the risk of explosion caused by micro-leakage in the refrigeration system. In order to study the flame retardant ability of flame retardants, the commonly used method is to use the national standard "GB / T 21844-2008 Standard Test Method for Flammability Concentration Limits of Compounds (Vapor and Gases)" to determine the explosion limit of the flammable gas after adding the flame retardant, and to determine the influence of the flame retardant by analyzing the reduction in the explosion limit range (upper explosion limit - lower explosion limit) of the flammable gas after the flame retardant is added. This method is cumbersome, time-consuming, labor-intensive and has certain risks.

[0003] The existing technology mainly includes the following test methods for the flame retardant ability of flame retardants:

[0004] CN102608284B discloses a method for determining the explosion limit of a multi-component mixed gas, which uses conventional physical and chemical parameters and mixing ratios of a single gas to establish corresponding explosion limit prediction models for different types of multi-component mixed gases, and uses the established model to predict the explosion limit of an unknown mixed gas. The method includes the following steps: 1. Collection of multi-component mixed gas modeling samples and their explosion limit data; 2. Classification and processing of multi-component mixed gas modeling samples; 3. Determination of physical and chemical parameters and collection of data; 4. Establishment of a prediction model; 5. Verification and correction of the model; 6. Application of the prediction model.

[0005] CN102608285B discloses a method for predicting the combustion and explosion characteristics of an organic mixture based on a support vector machine. The method takes the known component content and conventional physical property experimental data of the organic mixture as input variables and the corresponding combustion and explosion characteristic experimental data as output variables, and utilizes a powerful machine learning algorithm, a support vector machine method, to effectively train and predict the inherent quantitative relationship of nonlinearity, uncertainty and complexity between the two, thereby establishing a stable and efficient support vector machine prediction model.

[0006] In the above-mentioned prior art prediction of the flame retardant ability of flame retardants, the basic principle is to use the components and physicochemical properties (or physical properties) of the gas mixture, and the prediction object is the explosion limit of the mixture. In its implementation steps, it is necessary to first use experimental methods or data collection methods to obtain the physicochemical parameters or physical property parameters of each component as independent variables, which is cumbersome to operate, and the physicochemical parameters obtained by the experimental method will inevitably bring the influence of experimental errors into the independent variables in the process of predicting the explosion limit. Summary of the invention

[0007] The purpose of the present invention is to provide a method for predicting the flame retardant ability of a flame retardant, which can quickly obtain the flame retardant ability of the flame retardant through theoretical calculation and is easy to operate.

[0008] To achieve the above object, the present invention adopts the following technical solutions:

[0009] A method for predicting the flame retardant ability of a flame retardant, the method using a statistical modeling method to establish a prediction model for the flame retardant ability of a flame retardant, the prediction model using a quantitative relationship between the molecular structure of a mixture of a flame retardant and a combustible gas and the variation value of the explosion limit range of the mixed system to predict the flame retardant ability of the flame retardant.

[0010] As a preferred embodiment of the present invention, the construction of the above-mentioned prediction model includes the following steps:

[0011] S1. Use the combustible gas explosion limit test method to obtain the data samples required for establishing the prediction model;

[0012] S2. Obtain the change value of the explosion limit range of the combustible gas mixture before and after the flame retardant is added as the prediction target value;

[0013] S3, construct stable molecular structures of each component;

[0014] S4, calculating the molecular structure descriptor of the mixture;

[0015] S5, screening the optimal mixture molecular descriptor;

[0016] S6. Use computational statistical methods such as multivariate linear regression MLR, neural network, support vector machine SVM, etc. to perform training modeling on the training set. The above-screened molecular descriptors are used as input variables, and the predicted target values ​​are used as output values. Through training, a mathematical model that associates the molecular structure of the mixture with the variation value of the explosion limit range is established, which is the prediction model.

[0017] As another preferred embodiment of the present invention, in step S1, the method for obtaining the data sample is: experimentally determining the values ​​of the upper and lower explosion limits of each group of combustible gases before the flame retardant is added, and calculating the difference between the upper and lower explosion limits as the explosion limit range of each group of combustible gases before the flame retardant is added.

[0018] Furthermore, in step S2, the method for obtaining the explosion limit range change value is: using a combustible gas explosion limit test method to determine the values ​​of the upper and lower explosion limits of each group of mixtures after adding the flame retardant, and calculating the difference between the upper and lower explosion limits.

[0019] Further preferably, the method for constructing a stable molecular structure of each component in step S3 is: using the molecular simulation method in HyperChem and ChemDraw software to construct the molecular structure of each component in the mixture, and using molecular conformation analysis, molecular mechanics methods, and quantum mechanics semi-empirical methods to optimize the molecular structure to obtain a stable molecular structure.

[0020] Further preferably, the specific calculation steps of the molecular structure descriptor of the mixture in step S4 are: using the topological method or quantum mechanics method in the DRAGON and CODESSA molecular descriptor calculation software to calculate the molecular structure descriptor of each component in the mixture, and according to the distribution ratio of each component in the mixture, using the weighted average and other mixture molecular descriptor calculation methods to calculate the molecular structure descriptor of the mixture.

[0021] Furthermore, the screening step of the optimal mixture molecular descriptor in step S5 is: randomly selecting part of the mixture as a training set and the remaining part as a test set, and using a genetic algorithm feature variable screening method for the training set to screen the molecular descriptor of the mixture with the strongest correlation with the predicted target value.

[0022] Furthermore, after the prediction model is established, the following steps are also included:

[0023] Further verifying the prediction model, and constructing the molecular structure of the flame retardant and the combustible gas;

[0024] Calculate molecular structure descriptors of flame retardants and flammable gases;

[0025] Selecting a corresponding molecular descriptor from the mixture molecular structure descriptors as a model input value;

[0026] The model input value is input into the prediction model, and the change value of the explosion limit range of the mixture is obtained by calculation through the prediction model. The change value is the parameter that characterizes the flame retardant ability of the flame retardant.

[0027] Furthermore, the verification step of the prediction model is: using a test set to verify the fitting ability, stability and prediction ability of the prediction model; if the verified prediction model meets the conditions, the prediction model is used to predict the flame retardant ability of the fuel; if the verified prediction model does not meet the conditions, S6 is repeated until the prediction model meets the conditions.

[0028] Furthermore, the prediction model is used to predict the flame retardant ability of flame retardants with unknown performance. For flame retardants that are outside the data sample, have not been experimentally measured, and have unknown flame retardant ability, the molecular simulation method in HyperChem and ChemDraw software is used to construct the molecular structure of the flame retardant and combustible gas.

[0029] Furthermore, the molecular structure descriptors of the flame retardant and the combustible gas are calculated using the topological method or quantum mechanics method in the DRAGON and CODESSA molecular descriptor calculation software. According to the ratio of the flame retardant and the combustible gas, the molecular structure descriptor of the mixture is calculated using the weighted average method.

[0030] Furthermore, the prediction model should meet the following conditions: the average relative error between the predicted value and the experimental value of the validation model sample is within a certain range.

[0031] Within the above certain range, for example, the average relative error is less than 15%.

[0032] The weighted average method mentioned above is:

[0033] The value of the molecular structure descriptor of each component in the mixture is multiplied by the corresponding ratio, and then the total is summed to obtain the overall value, which is then divided by the sum of the ratios of each group to obtain the molecular structure descriptor of the mixture.

[0034] Compared with the prior art, the present invention brings the following beneficial technical effects:

[0035] (1) The present invention provides a method for predicting the flame retardant ability of a flame retardant, which uses a statistical modeling method to establish a prediction model for the flame retardant ability of a flame retardant. Once a model characterizing the relationship between the molecular structure of a mixture of a flame retardant and a combustible gas and the variation value of the explosion limit range is established, the flame retardant ability of the flame retardant in the mixture system can be quickly obtained by only inputting the molecular structure of the mixture.

[0036] (2) The present invention adopts molecular structure and component volume ratio to predict the change value of the explosion limit range to characterize the flame retardant ability of the flame retardant, which can avoid the influence of experimental errors in the independent variables.

[0037] (3) The method of the present invention is simpler and more convenient. Based on the established prediction model, no experimental test is required. The flame retardant ability of the flame retardant can be quickly predicted based on the molecular structure information and the component ratio. Compared with the experimental test method, it is safer and simpler. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below in conjunction with the accompanying drawings:

[0039] Figure 1 A flow chart for constructing and applying the model of the present invention. DETAILED DESCRIPTION

[0040] The present invention proposes a method for predicting the flame retardant ability of a flame retardant. In order to make the advantages and technical solutions of the present invention clearer and more specific, the present invention is described in detail below in conjunction with specific embodiments.

[0041] The present invention uses a statistical modeling method to establish a prediction model for the flame retardant ability of a flame retardant. The principle of the model is to predict the flame retardant ability of the flame retardant by using the quantitative relationship between the molecular structure of the flame retardant and combustible gas mixture and the variation value of the explosion limit range of the mixed system.

[0042] like Figure 1 As shown, the present invention provides a method for predicting the flame retardant ability of a flame retardant, comprising:

[0043] S1. Use the combustible gas explosion limit test method to obtain the data samples required for establishing the prediction model;

[0044] S2. Obtain the change value of the explosion limit range of the combustible gas mixture before and after the flame retardant is added as the prediction target value;

[0045] S3, construct stable molecular structures of each component;

[0046] S4, calculating the molecular structure descriptor of the mixture;

[0047] S5, screening the optimal mixture molecular descriptor;

[0048] S6. Using computational statistical methods such as multivariate linear regression MLR, neural network, support vector machine SVM, etc., training modeling is performed on the training set, the above-screened molecular descriptors are used as input variables, and the predicted target values ​​are used as output values. Through training, a mathematical model associating the molecular structure of the mixture with the variation value of the explosion limit range is established, which is the prediction model;

[0049] S7. verifying the prediction model;

[0050] S8, constructing the molecular structure of the flame retardant and the combustible gas;

[0051] S9, calculating molecular structure descriptors of the mixture;

[0052] S10, selecting a corresponding molecular descriptor from the molecular structure descriptors of the mixture as a model input value;

[0053] S11. Inputting the model input value into the prediction model, and obtaining the change value of the explosion limit range of the mixture through the prediction model calculation, wherein the change value is a parameter characterizing the flame retardant ability of the flame retardant.

[0054] The following is a detailed description with reference to specific embodiments.

[0055] Embodiment 1:

[0056] The specific steps are as follows:

[0057] Step 1: Use the combustible gas explosion limit test method to obtain the data sample required for the establishment of the prediction model, experimentally determine the values ​​of the upper and lower explosion limits of each group of combustible gases before the flame retardant is added, and calculate the difference between the upper and lower explosion limits as the explosion limit range of each group of combustible gases before the flame retardant is added;

[0058] Step 2: using the combustible gas explosion limit test method to determine the value of the upper and lower explosion limits of each group of mixtures after adding the flame retardant, and calculating the difference between the upper and lower explosion limits as the explosion limit range of each group of mixtures after adding the flame retardant. At the same time, calculating the change value of the explosion limit range before and after adding the flame retardant as the predicted target value;

[0059] Step 3: Use molecular simulation methods in software such as HyperChem and ChemDraw to construct the molecular structure of each component in the mixture, and use molecular conformation analysis, molecular mechanics methods, quantum mechanics semi-empirical methods, etc. to optimize the molecular structure to obtain a stable molecular structure;

[0060] Step 4: Based on the above stable molecular structure, the molecular structure descriptors of each component in the mixture are calculated using the topological method or quantum mechanics method in the molecular descriptor calculation software such as DRAGON and CODESSA, and the molecular structure descriptors of the mixture are calculated using the weighted average molecular descriptor calculation method according to the distribution ratio of each component in the mixture;

[0061] Step 5: Randomly select 80% of the mixture as a training set and the rest as a test set. For the training set, use a genetic algorithm or other feature variable screening method to screen the molecular descriptor with the strongest correlation between the molecular structure descriptor of the mixture and the predicted target value;

[0062] Step 6: Use computational statistical methods such as multivariate linear regression MLR, neural network, support vector machine SVM, etc. to perform training modeling on the training set, use the above-screened molecular descriptors as input variables, and use the predicted target values ​​as output values ​​to establish a mathematical model that associates the molecular structure of the mixture with the variation value of the explosion limit range through training;

[0063] Step 7: Use the test set to verify the model's fitting ability, stability and prediction ability. If the verified model meets the conditions, the model is used to predict the flame retardant ability of the fuel. If the verified model does not meet the conditions, repeat step 6 until the model meets the application requirements.

[0064] Step 8: Use the prediction model to predict the flame retardancy of flame retardants with unknown performance. For flame retardants that are not in the data sample, have not been experimentally measured, and have unknown flame retardancy, use the molecular simulation method in software such as HyperChem and ChemDraw to construct the molecular structure of the flame retardant and the combustible gas;

[0065] Step 9: Using the topological method or quantum mechanics method in molecular descriptor calculation software such as DRAGON and CODESSA to calculate the molecular structure descriptors of the flame retardant and the combustible gas, and using weighted average and other methods to calculate the molecular structure descriptors of the mixture according to the ratio of the flame retardant and the combustible gas;

[0066] Step 10: Corresponding to the optimal molecular descriptor selected in step 5, a corresponding molecular descriptor is selected from the molecular structure descriptors of the mixture obtained by the above calculation as a model input value;

[0067] Step 11: Input the input value into the prediction model established in step 7, and obtain the change value of the explosion limit range of the mixture through model calculation. The result is the parameter that characterizes the flame retardant ability of the flame retardant.

[0068] Embodiment 2:

[0069] The specific steps are as follows:

[0070] Step 1: Select perhalogenated hydrocarbons as flame retardant components and hydrocarbon combustible gas as combustible components, and add flame retardant components to hydrocarbon combustible gas in different proportions to form a mixed gas. Use GB / T 21844-2008 Standard Test Method for Flammability Concentration Limits of Compounds (Vapor and Gases) to test the change in the explosion limit range of the gas before and after adding the flame retardant components. A total of 31 groups of values ​​are used as data samples for modeling;

[0071] Step 2: Use HyperChem software to draw the 3D molecular structure of all flame retardant components and combustible components, and use molecular mechanics method MM+ and quantum mechanics semi-empirical method AM1 to optimize the molecular structure, and finally obtain a stable 3D molecular structure;

[0072] Step 3: Use Dragon molecular simulation software to calculate the molecular structure descriptors of the flame retardant component and the combustible component respectively, and use the weighted average method to calculate the molecular structure descriptors of each group of mixtures according to the proportion of each mixture component;

[0073] Step 4: randomly select 80% of the mixture as a training set and the rest as a test set. For the training set, a genetic algorithm is used to screen the molecular descriptor combination with the strongest correlation with the change value of the explosion limit range in the molecular descriptor of the mixture as the optimal molecular descriptor group;

[0074] Step 5: Using the support vector machine algorithm, for the training set, the optimal molecular descriptor group is used as the input variable, and the change value of the explosion limit range is used as the output value to establish a prediction model;

[0075] Step 6: Use the test set to verify the model and calculate the model's coefficient of determination R 2The cross-validation coefficient Q of the leave-one-out method is 0.924. loo 2 is 0.922, and the average relative error of the model is 0.08. 2 and Q loo 2 When the value is greater than 0.6, the model is considered to have good fitting ability and stability; when the value of the mean relative error is less than 0.15, the model is considered to have good prediction ability, therefore, the verification model meets the application requirements;

[0076] Step 7: Apply the prediction model to predict the flame retardant CF 3 I Flame retardancy in propane, CF 3 The volume ratio of I to propane is 1. HyperChem was used to draw CF 3 The 3D molecular structure of I and propane was obtained by using the molecular mechanics method MM+ and the quantum mechanics semi-empirical method AM1 to optimize the molecular structure and obtain a stable 3D molecular structure;

[0077] Step 8: Use Dragon molecular simulation software to calculate CF 3 I and propane molecular descriptors, according to the group distribution ratio, the molecular descriptors of the mixture are calculated by weighted average method; according to the optimal molecular descriptor group determined in step 4, the corresponding molecular descriptors are selected from the molecular descriptors of the mixture as input values;

[0078] Step 9: Substitute the input value into the model verified in step 6 to calculate the flame retardant CF under this ratio. 3 After I, the explosion limit range of propane decreased by 4.8%, reflecting the CF predicted by the model established in this example. 3 I Flame retardancy against propane at this ratio.

[0079] Embodiment 3:

[0080] The specific steps are as follows:

[0081] Step 1: Select perhalogenated hydrocarbons as flame retardant components and hydrocarbon combustible gas as combustible components, and add flame retardant components to hydrocarbon combustible gas in different proportions to form a mixed gas. Use GB / T 21844-2008 Standard Test Method for Flammability Concentration Limits of Compounds (Vapor and Gases) to test the change in the explosion limit range of the gas before and after adding the flame retardant components. A total of 31 groups of values ​​are used as data samples for modeling;

[0082] Step 2: Use ChemDraw software to draw the molecular structures of all flame retardant components and combustible components, and use molecular mechanics method MM+ and quantum mechanics semi-empirical method AM1 to optimize the molecular structure, and finally obtain a stable molecular structure;

[0083] Step 3: using CODESSA molecular simulation software to calculate the molecular structure descriptors of the flame retardant component and the combustible component respectively, and according to the component ratio of each mixture, using the weighted average method to calculate the molecular structure descriptor of each group of mixtures;

[0084] Step 4: randomly select 80% of the mixture as a training set and the rest as a test set. For the training set, a genetic algorithm is used to screen the molecular descriptor combination with the strongest correlation with the change value of the explosion limit range in the molecular descriptor of the mixture as the optimal molecular descriptor group;

[0085] Step 5: Using the BP neural network algorithm, for the training set, the optimal molecular descriptor group is used as the input variable, and the change value of the explosion limit range is used as the output value to establish a prediction model;

[0086] Step 6: Use the test set to verify the model and calculate the model's coefficient of determination R 2 The cross-validation coefficient Q of the leave-one-out method is 0.892. loo 2 is 0.890, and the average relative error of the model is 0.12. 2 and Q loo 2 When the value is greater than 0.6, the model is considered to have good fitting ability and stability; when the value of the mean relative error is less than 0.15, the model is considered to have good prediction ability, therefore, the verification model meets the application requirements;

[0087] Step 7: Apply the prediction model to predict the flame retardant CF 3 I Flame retardancy in propane, CF 3 The volume ratio of I to propane is 1. Use ChemDraw to draw CF 3 The 3D molecular structure of I and propane was obtained, and the molecular mechanics method MM+ and the quantum mechanics semi-empirical method AM1 were used to optimize the molecular structure and obtain a stable molecular structure;

[0088] Step 8: Calculate CF using CODESSA molecular simulation software 3 I and propane molecular descriptors, according to the group distribution ratio, the molecular descriptors of the mixture are calculated by weighted average method; according to the optimal molecular descriptor group determined in step 4, the corresponding molecular descriptors are selected from the molecular descriptors of the mixture as input values;

[0089] Step 9: Substitute the input value into the model verified in step 6 to calculate the ratio of the flame retardant CF added in this embodiment. 3 After I, the explosion limit range of propane decreased by 4.5%, reflecting the CF predicted by the model established in this example.3 I Flame retardancy against propane at this ratio.

[0090] The preferred embodiments of the present invention are described in detail above with reference to the accompanying drawings; however, the present invention is not limited thereto.

[0091] The parts not described in the present invention can be realized by referring to the prior art.

[0092] It should be noted that any equivalent manner or obvious variation manner made by those skilled in the art under the guidance of this specification should be within the protection scope of the present invention.

Claims

1. A method for predicting the flame retardant ability of a flame retardant, characterized in that: The prediction method is to use a statistical modeling method to establish a prediction model for the flame retardant ability of the flame retardant, and the prediction model is to use the quantitative relationship between the molecular structure of the flame retardant and the combustible gas mixture and the change value of the explosion limit range of the mixed system to predict the flame retardant ability of the flame retardant; The construction of the prediction model includes the following steps: S1. Use the combustible gas explosion limit test method to obtain the data samples required for establishing the prediction model; S2. Obtain the change value of the explosion limit range of the combustible gas mixture before and after the flame retardant is added as the prediction target value; S3, construct stable molecular structures of each component; S4, calculating the molecular structure descriptor of the mixture; S5, screening the optimal mixture molecular structure descriptor; S6. Using multiple linear regression MLR, neural network, support vector machine SVM calculation statistical methods, training modeling is performed on the training set, the molecular structure descriptors screened out above are used as input variables, and the predicted target value is used as the output value. A mathematical model associating the molecular structure of the mixture with the variation value of the explosion limit range is established through training, which is the prediction model; In step S1, the method for obtaining the data sample is: experimentally determining the values ​​of the upper and lower explosion limits of each group of combustible gases before adding the flame retardant, and calculating the difference between the upper and lower explosion limits as the explosion limit range of each group of combustible gases before adding the flame retardant; In step S2, the method for obtaining the change value of the explosion limit range is: using the combustible gas explosion limit test method to determine the values ​​of the upper and lower explosion limits of each group of mixtures after the flame retardant is added, and calculating the difference between the upper and lower explosion limits as the explosion limit range of each group of mixtures after the flame retardant is added. At the same time, the change value of the explosion limit range before and after the flame retardant is added is calculated as the predicted target value.

2. The method for predicting the flame retardant ability of a flame retardant according to claim 1, characterized in that: The method for constructing a stable molecular structure of each component in step S3 is: using the molecular simulation method in HyperChem and ChemDraw software to construct the molecular structure of each component in the mixture, and using molecular conformation analysis, molecular mechanics method, and quantum mechanics semi-empirical method to optimize the molecular structure to obtain a stable molecular structure.

3. The method for predicting the flame retardant ability of a flame retardant according to claim 1, characterized in that: The specific calculation steps of the molecular structure descriptor of the mixture in step S4 are: using the topological method or quantum mechanics method in the DRAGON and CODESSA molecular descriptor calculation software to calculate the molecular structure descriptor of each component in the mixture, and according to the distribution ratio of each component in the mixture, using the weighted average mixture molecular descriptor calculation method to calculate the molecular structure descriptor of the mixture.

4. The method for predicting the flame retardant ability of a flame retardant according to claim 1, characterized in that: In step S5, the steps for screening the optimal mixture molecular structure descriptor are as follows: randomly selecting part of the mixture as a training set and the remaining part as a test set, and using a genetic algorithm feature variable screening method for the training set to screen the molecular structure descriptor of the mixture with the strongest correlation with the predicted target value.

5. The method for predicting the flame retardant ability of a flame retardant according to claim 1, characterized in that: After establishing the prediction model, the following steps are also included: Further verifying the prediction model, and constructing the molecular structure of the flame retardant and the combustible gas; Calculate molecular structure descriptors of mixtures of flame retardants and flammable gases; Selecting a corresponding molecular structure descriptor from the mixture molecular structure descriptors as a model input value; The model input value is input into the prediction model, and the change value of the explosion limit range of the mixture is obtained by calculation through the prediction model. The change value is the parameter that characterizes the flame retardant ability of the flame retardant.

6. The method for predicting the flame retardant ability of a flame retardant according to claim 5, characterized in that: The verification steps of the prediction model are as follows: using the test set to verify the fitting ability, stability and prediction ability of the prediction model. If the average relative error value is less than 0.15, the model is considered to have good prediction ability; if the verification prediction model meets the conditions, the prediction model is used to predict the flame retardant ability of the fuel; if the verification prediction model does not meet the conditions, S6 is repeated until the prediction model meets the conditions.

7. The method for predicting the flame retardant ability of a flame retardant according to claim 6, characterized in that: The prediction model is used to predict the flame retardancy of flame retardants with unknown performance. For flame retardants with unknown flame retardancy that are outside the data sample and have not been experimentally measured, the molecular simulation method in HyperChem and ChemDraw software is used to construct the molecular structure of the flame retardant and combustible gas.

8. The method for predicting the flame retardant ability of a flame retardant according to claim 6, characterized in that: The molecular structure descriptors of the flame retardant and the combustible gas are calculated using the topological method or quantum mechanics method in the DRAGON and CODESSA molecular descriptor calculation software. According to the ratio of the flame retardant and the combustible gas, the molecular structure descriptor of the mixture is calculated using the weighted average method.

9. The method for predicting the flame retardant ability of a flame retardant according to claim 6, characterized in that: The conditions that the prediction model should meet are: the average relative error between the predicted value and the experimental value of the verification model sample is within a certain range.

10. The method for predicting the flame retardant ability of a flame retardant according to claim 8, characterized in that: The weighted average method is as follows: the value of the molecular structure descriptor of each component in the mixture is multiplied by the corresponding ratio, and then the total sum is obtained to obtain the overall value, and then divided by the sum of the ratios of each group to obtain the molecular structure descriptor of the mixture.

Images