A method for analyzing alkyl phosphate pyrolysis kinetics based on machine learning

Abstract

The application discloses a method for analyzing alkyl phosphate pyrolysis kinetics based on machine learning, and utilizes a thermogravimetric analyzer to obtain TG test data of alkyl phosphate flame retardant; the TG test data is read through Python and is converted; after relevant data is substituted into a Coats-Redfern pyrolysis kinetics classical mechanism function equation model, training set model data is formed, and a training set model is created; linear regression is established for the training set model; the training set model is verified, when a linear correlation coefficient calculated by the model is low and an activation energy is obviously inconsistent with logarithmic pre-exponential factor and another two method calculation values, the model is reselected in the training set, until an optimal pyrolysis kinetics mechanism function equation of alkyl phosphate is found; data calculated under the optimal pyrolysis kinetics mechanism function equation is exported through Python and is visualized in the form of pictures. The method is simple and fast in calculation, and provides guiding significance for screening and optimizing alkyl phosphate of different modified substrates.

Description

Technical Field

[0001] This invention relates to the field of flame retardant technology, and more specifically to a method for analyzing the pyrolysis kinetics of alkyl phosphates based on machine learning. Background Technology

[0002] Polymer materials are widely used in packaging, electronics, construction, and other fields, such as polyester, polyamide, and polyurethane. However, most polymer materials are organic compounds and are flammable. To address this flammability issue, researchers have been gradually developing flame-retardant technologies.

[0003] There are two main methods to improve the flame retardant properties of materials: in-situ copolymerization and blending. In-situ copolymerization involves incorporating flame retardants into the polymer chain through a chemical reaction, resulting in a long-lasting flame retardant effect. However, this method often requires participation in the prepolymerization process, affecting the degree of polymerization, and thus its application in flame retardant applications is currently limited. Blending, on the other hand, adds flame retardants to the polymer matrix through blending. This method is simple to operate, inexpensive, and widely favored by the market.

[0004] Currently popular flame retardants on the market mainly include halogenated flame retardants, phosphorus-based flame retardants, nitrogen-based flame retardants, phosphorus-nitrogen intumescent flame retardants, metal hydroxide flame retardants, and silicon-based flame retardants. Among them, phosphorus-based flame retardants are widely used in the flame retardant modification of various polymer materials due to their green and environmentally friendly properties, high flame retardancy, and small dosage.

[0005] The flame retardant mechanism of phosphorus-based flame retardants mainly includes two types: solid-phase and gas-phase. Solid-phase flame retardancy involves the flame retardant reacting with the polymer material during heating to form a carbon layer on the matrix surface to block heat transfer. Gas-phase flame retardancy involves the flame retardant releasing phosphorus-containing free radical quenchers during pyrolysis, which react with H· and OH· free radicals generated during the combustion of the polymer material, reducing the free radical density in the flame retardant and reducing chain reactions during combustion.

[0006] Alkyl phosphates are a class of phosphorus-based flame retardants characterized by high phosphorus content, high thermal stability, and high flame retardant efficiency. When heated, alkyl phosphates release phosphorus-containing free radicals to suppress flames and form a char layer on the matrix surface to reduce subsequent combustion. In the past decade or so, the industry has seen a surge in research and products related to alkyl phosphates, such as Clariant's Exolit OP series.

[0007] Although alkyl phosphate products have been widely used, research on the pyrolysis kinetics of alkyl phosphates is currently limited. A deeper understanding of the reaction mechanisms of flame retardants is crucial for optimizing them, screening suitable modification matrices, and designing modified formulations.

[0008] Thermogravimetric analysis (TGA) is a testing method that records the mass changes of the tested material with temperature and time in real time. This method allows for a simple and intuitive observation of the material's thermal stability and decomposition temperature. When coupled with Fourier transform infrared spectroscopy (FTIR), it can also be used to conduct real-time analysis and detection of gaseous products during thermal decomposition. The Coats-Redfern, Flynn-Wall-Ozawa, Kissinger, and Friedman methods are generally used to analyze the pyrolysis reactions of flame-retardant materials. The Coats-Redfern method can be used to infer the kinetic function of the reaction mechanism, while the other three methods are mainly used to infer and calculate pyrolysis performance parameters. Furthermore, the pyrolysis kinetic mechanism equation obtained solely by the Coats-Redfern method may not always match experimental results; it is necessary to compare it with the activation energy and pre-logarithmic factor obtained by the Kissinger and Flynn-Wall-Ozawa methods to verify the accuracy of the calculation and find the optimal pyrolysis kinetic mechanism equation for the tested material.

[0009] However, existing methods for searching the optimal pyrolysis kinetic mechanism function equations are cumbersome, requiring extensive data export and import analysis, which is time-consuming. Therefore, there is an urgent need for a simple and fast calculation method to improve the analysis speed. Summary of the Invention

[0010] To address certain technical problems existing in the prior art, this application aims to provide a method for analyzing the pyrolysis kinetics of alkyl phosphates based on machine learning. This method is simple and fast to calculate, and provides guidance for the screening and optimization of alkyl phosphates for different modified matrices.

[0011] To solve the aforementioned technical problems, this application adopts the following technical solution:

[0012] 1. A method for analyzing the pyrolysis kinetics of alkyl phosphates based on machine learning, characterized in that the method includes the following steps;

[0013] S1, TG test data acquisition: Thermogravimetric analysis was used to obtain the TG curves of alkyl phosphate flame retardants from room temperature to 800℃ in β mode with different heating rates under nitrogen atmosphere, so as to obtain the TG test data of the remaining mass percentage w of flame retardant as a function of real-time temperature T1.

[0014] S2. Read and convert TG test data: Read TG test data using Python and convert the percentage of remaining material mass w in the data into the weight loss rate α, and convert the real-time temperature T1 of the test data into the absolute temperature T;

[0015] S3. Create a training set model: Substitute the relevant data into the classical mechanism function equation model of Coats-Redfern pyrolysis kinetics to form training set model data and create a training set model.

[0016] S4. Establish Linear Regression: Establish linear regression on the training set and train each model using linear regression. Utilize the training set models... The linear relationship between 1000 / T yields the activation energy E and the logarithmic pre-exponential factor lnA, and the linear correlation coefficient R under each functional equation is calculated. 2 ;

[0017] S5. Training Set Validation: The validation steps include:

[0018] A1. Compare the calculated values ​​of activation energy E and logarithmic pre-exponential factor lnA obtained from each training set model with the calculated values ​​obtained by the Kissinger method and the Flynn-Wall-Ozawa method.

[0019] A2. When the linear correlation coefficient R calculated by the model... 2 If the activation energy E is low and its value is significantly inconsistent with the logarithmic pre-exponential factor lnA and the values ​​calculated by the other two methods, the model will be reselected in the training set until the model has a high linear correlation coefficient and the activation energy E and logarithmic pre-exponential factor lnA calculated by the model are close to the values ​​calculated by the Kissinger method and the Flynn-Wall-Ozawa method. This verifies that the optimal mechanism equation for the pyrolysis kinetics of alkyl phosphate flame retardants has been selected.

[0020] S6. Data Visualization: Export the data calculated from the optimal mechanism equation using Python and visualize it in the form of images.

[0021] Preferably, the heating rate β in step S1 is 10-25℃ / min.

[0022] Preferably, in step S2, Python is used to read the TG test data. The conversion relationship between the real-time temperature T1 and the absolute temperature T is T = 273.15 + T1, and the conversion relationship between the remaining mass percentage w of the flame retardant and the weight loss rate α is α = 100% - w.

[0023] Preferably, the weight loss rate α in step S2 is selected from data of 5%-70% as the calculation data for Python.

[0024] Preferably, the data substitution process in step S3 is entirely completed using Python.

[0025] Preferably, the classical mechanism function equation model of Coats-Rediern pyrolysis kinetics in step S3 is a solid-phase thermal decomposition reaction equation used for mechanism analysis of the pyrolysis kinetic reaction.

[0026] The solid-phase thermal decomposition reaction equation utilizes the Arrhenius formula. Substitution Same as above

[0027] In programmed heating mode, the heating rate Equation 1 can be obtained:

[0028] Where α is the weight loss rate (%), E is the activation energy (kJ / mol), R is the gas constant (8.314 J / mol·K), A is the pre-exponential factor, and taking its logarithm yields the logarithmic pre-exponential factor lnA, f(α) is the mechanism function, and T is the absolute temperature (K); after separating the variables, Equation 2 is obtained:

[0029] Let f(α) = (1-α) n After integrating Equation 2, we obtain the Coats-Redfern equation: Where G(α) is the classical mechanism function equation of pyrolysis kinetics.

[0030] Preferably, in step S3, 14 classical mechanistic function equation models of pyrolysis kinetics commonly used in the Coats-Redfern equation from the literature are selected, and the data is substituted into them to create a training model.

[0031] Preferably, in step S4, the model in the training set is used. The slope and intercept parameters obtained from establishing a linear regression with the 1000 / T data can be used to calculate the activation energy E and the logarithmic pre-exponential factor lnA in the pyrolysis reaction process. The score of the training set after the linear regression is the linear correlation coefficient R. 2 .

[0032] Preferably, the classical mechanistic equation model and reaction mechanism of the pyrolysis kinetics specifically include:

[0033]

[0034]

[0035] Preferably, in step S5, the Kissinger method calculates the activation energy using a differential method, and the formula is as follows: Where T maxThe absolute temperature corresponding to the maximum weight loss rate of the alkyl phosphate is given. The weight loss rate as a function of absolute temperature is calculated using Python by performing a first-order differential calculation. The peak temperature at which the weight loss rate of the alkyl phosphate is maximized at each heating rate is then identified and substituted into the Kissinger method formula to obtain the slope and intercept, thereby determining the activation energy and the pre-logarithmic factor.

[0036] Preferably, in step S5, Flynn-Wall-Ozawa calculates the activation energy of alkyl phosphates using an integral method, with the following formula: By using Python to substitute data at different heating rates and the same weight loss rate into the Flynn-Wall-Ozawa formula, the slope obtained by linear solution can be used to calculate the activation energy of alkyl phosphate flame retardants at various weight loss rates.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] First, thermogravimetric analysis (TGA) data of alkyl phosphate flame retardants were acquired using a thermogravimetric analyzer under multiple heating rate modes. Then, Python was used to read the experimental data, and following machine learning steps, the relevant data were transformed into functional expressions in the classic Coats-Redfern model. A training set model was created, linear regression was established, and the correlation coefficients of each training set were analyzed and evaluated. These coefficients were compared with the activation energy and logarithmic exponent factors calculated by the Kissinger and Flynn-Wall-Ozawa methods to search for and select the optimal pyrolysis kinetic mechanism function equation for alkyl phosphate pyrolysis. This allows the calculation method, combined with TGA, to qualitatively and quantitatively characterize pyrolysis parameters and mechanisms. This approach can reveal the pyrolysis kinetic behavior of alkyl phosphate flame retardants at high temperatures from a microscopic perspective, and to reveal the flame-retardant mechanism of alkyl phosphate flame retardants during heating from a certain dimension.

[0039] Compared to conventional methods that calculate pyrolysis kinetics at a single heating rate, this invention uses multiple heating rates to verify the reliability of the pyrolysis kinetic mechanism function equation. Furthermore, it utilizes Python tools to quickly calculate and match the optimal pyrolysis kinetic mechanism function equation for the material based on a machine learning model. This significantly reduces computation time and improves computational accuracy, providing a computational approach for the subsequent analysis of flame retardant pyrolysis kinetics. Attached Figure Description

[0040] Figure 1 This is a flowchart of the steps in this invention to analyze the pyrolysis kinetics of alkyl phosphates based on machine learning;

[0041] Figure 2This is an example diagram showing the results calculated using the R2 mechanism model for alkyl phosphates in this invention, with a heating rate of 10℃ / min as an example. Detailed Implementation

[0042] The present application will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0043] In the description of this application, it should be understood that the terms "upper", "lower", "left", "right", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0044] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0045] A method for analyzing the pyrolysis kinetics of alkyl phosphates based on machine learning, the method comprising the following steps;

[0046] S1, TG test data acquisition: Thermogravimetric analysis was used to obtain the TG curves of alkyl phosphate flame retardants from room temperature to 800℃ in β mode with different heating rates under nitrogen atmosphere, so as to obtain the TG test data of the remaining mass percentage w of flame retardant as a function of real-time temperature T1.

[0047] S2. Read and convert TG test data: Read TG test data using Python and convert the percentage of remaining material mass w in the data into the weight loss rate α, and convert the real-time temperature T1 of the test data into the absolute temperature T;

[0048] S3. Create a training set model: Substitute the relevant data into the classical mechanism function equation model of Coats-Redfern pyrolysis kinetics to form training set model data and create a training set model.

[0049] S4. Establish Linear Regression: Establish linear regression on the training set and train each model using linear regression. Utilize the training set models... The linear relationship between 1000 / T yields the activation energy E and the logarithmic pre-exponential factor lnA, and the linear correlation coefficient R under each functional equation is calculated. 2 ;

[0050] S5. Training Set Validation: The validation steps include:

[0051] A1. Compare the calculated values ​​of activation energy E and logarithmic pre-exponential factor lnA obtained from each training set model with the calculated values ​​obtained by the Kissinger method and the Flynn-Wall-Ozawa method;

[0052] A2. When the linear correlation coefficient R calculated by the model... 2 If the activation energy E is low and its value is significantly inconsistent with the logarithmic pre-exponential factor lnA and the values ​​calculated by the other two methods, the model will be reselected in the training set until the model has a high linear correlation coefficient and the activation energy E and logarithmic pre-exponential factor lnA calculated by the model are close to the values ​​calculated by the Kissinger method and the Flynn-Wall-Ozawa method. This verifies that the optimal pyrolysis kinetic mechanism function equation for alkyl phosphate flame retardants has been selected.

[0053] S6. Data Visualization: Export the data calculated under the pyrolysis kinetic mechanism function equation using Python and visualize it in the form of images.

[0054] The flowchart of the steps for analyzing the pyrolysis kinetics of alkyl phosphates based on the above machine learning is as follows: Figure 1 As shown.

[0055] In step S1, a thermogravimetric analyzer was used to obtain the TG curves of the alkyl phosphate flame retardant under different heating rates β and a nitrogen atmosphere using a multiple heating rate mode, thereby obtaining experimental test data. In this experiment, the main focus was on studying the combustion changes of the alkyl phosphate flame retardant under a nitrogen atmosphere as it was heated from room temperature to 800℃ at different heating rates (10-25℃ / min). Specifically, under a heating rate β of 10-25℃ / min and a nitrogen atmosphere, the temperature was increased from room temperature to 800℃, and the mass of 10 mg of alkyl phosphate flame retardant was recorded as a function of temperature, yielding the percentage of remaining material mass w at the real-time temperature T1.

[0056] In step S2, the TG test data is read using Python, and the remaining mass percentage w of the flame retardant material is converted into a weight loss rate α, where α = 100% - w. The real-time temperature T1 of the test data is converted into an absolute temperature T, where T = 273.15 + T1. The weight loss rate α is calculated using data ranging from 5% to 70% in Python.

[0057] In step S3, the relevant data is substituted into 14 commonly used classical mechanism function equation models of Coats-Redfern pyrolysis kinetics using Python to form training set model data, and the training set model is created.

[0058] Step S4 involves establishing linear regression: Linear regression is established on the training set, and each model is trained using linear regression. The training set models are then used... The linear relationship between 1000 / T yields the activation energy E and the logarithmic pre-exponential factor lnA, and the linear correlation coefficient R under each functional equation is calculated. 2 .

[0059] In step S5, the training set model is validated by comparing and analyzing the calculated data using Python. When the model calculates the linear correlation coefficient R... 2 Models with low activation energy E and significantly inconsistent with the values ​​calculated by the logarithmic pre-exponential factor lnA and the other two methods will be reselected in the training set until the linear correlation coefficient of the training set models is high and the activation energy E and logarithmic pre-exponential factor lnA calculated by the models are close to the values ​​calculated by the Kissinger method and the Flynn-Wall-Ozawa method. This verifies that the optimal mechanism equation for the pyrolysis kinetics of alkyl phosphate flame retardants has been selected.

[0060] In step S6, the data calculated under the pyrolysis kinetic mechanism function equation is exported using Python and visualized as an image.

[0061] The classical mechanistic function equation model of Coats-Redfern pyrolysis kinetics used in this embodiment is an equation for mechanistic analysis of the pyrolysis kinetic reaction. It is a solid-phase thermal decomposition reaction equation, and the specific equations are as follows:

[0062] For solid-phase thermal decomposition reactions, the Arrhenius equation is used. Substitution In programmed heating mode, the heating rate

[0063] Equation 1 can be obtained:

[0064] Where α is the weight loss rate (%), E is the activation energy (kJ / mol), R is the gas constant (8.314 J / mol·K), A is the pre-exponential factor, and taking its logarithm yields the logarithmic pre-exponential factor lnA, f(α) is the mechanism function, and T is the absolute temperature (K); after separating the variables, Equation 2 is obtained:

[0065] Let f(α) = (1-α) nAfter integrating Equation 2, we obtain the Coats-Redfern equation: Where G(α) is the classical mechanism function equation of pyrolysis kinetics.

[0066] Preferably, 14 classical functional equation models G(α) commonly used in the Coats-Redfern equation from the literature are selected, based on the training set models. The slope and intercept parameters obtained from establishing a linear regression with the 1000 / T data can be used to calculate the activation energy E and the logarithmic pre-exponential factor lnA in the pyrolysis reaction process. The score of the training set after the linear regression is the linear correlation coefficient R. 2 .

[0067] The classical mechanistic function equation models and reaction mechanisms of the 14 described pyrolysis kinetics are shown in Table 1.

[0068] Table 1. Classical mechanistic functional equation models and reaction mechanisms of the 14 described pyrolysis kinetics

[0069]

[0070]

[0071]

[0072] In this example, the Kissinger method calculates the activation energy using the differential method, and the formula is as follows: Where T max The absolute temperature corresponding to the maximum weight loss rate of the alkyl phosphate is given. The weight loss rate as a function of absolute temperature is calculated using Python by performing a first-order differential calculation. The peak temperature at which the weight loss rate of the alkyl phosphate is maximized at each heating rate is then identified and substituted into the Kissinger method to obtain the slope and intercept, thereby determining the activation energy and the pre-logarithmic factor.

[0073] In this example, Flynn-Wall-Ozawa calculated the activation energy of alkyl phosphates using an integral method, with the following formula: By using Python to substitute data at different heating rates and the same weight loss rate into the Flynn-Wall-Ozawa formula, the slope obtained by linear solution can be used to calculate the activation energy of alkyl phosphate flame retardants at various weight loss rates.

[0074] Python was used to read and convert the TG test data of the flame retardant at a heating rate of 10℃ / min. The relevant data were then substituted into 14 commonly used classical pyrolysis kinetic function equation models of Coats-Redfem to create a training set model. The specific correlation coefficient calculation results are shown in Table 2. The results indicate that, according to the functional equation of the pyrolysis kinetic mechanism model R2... The calculated activation energy and logarithmic pre-exponential factor are close to those calculated by the Kissinger method (see Table 3) and the Flynn-Wall-Ozawa method (see Table 4), and have high self-linear correlation coefficients.

[0075] Then, Python was used to read and convert the TG test data of the flame retardant at different heating rates, and the relevant data were substituted into the functional equation of the pyrolysis kinetic mechanism model R2. The calculated activation energy and logarithmic pre-exponential factor are shown in Table 5. These are similar to those calculated using the Kissinger method and the Flynn-Wall-Ozawa method, and exhibit a high linear correlation coefficient. This indicates that the pyrolysis kinetic mechanism model for this alkyl phosphate flame retardant is R², and the pyrolysis kinetic mechanism function equation is: The solid-phase reaction mechanism is a phase boundary controlled reaction (contracting area).

[0076] Table 2 shows the calculated correlation coefficients of alkyl phosphate flame retardants under 14 pyrolysis kinetic equations at a heating rate of 10 °C / min.

[0077]

[0078] Table 3. Activation energies and pre-logarithmic factors of the pyrolysis process of alkyl phosphate flame retardants calculated by the Kissinger method at different heating rates.

[0079]

[0080] Table 4. Activation energies of alkyl phosphate flame retardants during pyrolysis calculated by the Flynn-Wall-Ozawa method at different weight loss rates.

[0081] Weight loss rate (α) E(kJ / mol) 5％ 190.453 10％ 215.081 15％ 224.491 20％ 229.331 30％ 232.229 40％ 232.222 50％ 231.304 60％ 228.357 70％ 206.264

[0082] Table 5. Alkyl phosphate flame retardants at different heating rates Calculation results of correlation coefficients under the functional equation of pyrolysis kinetic mechanism

[0083]

[0084] The results of calculations using the R2 mechanism model for alkyl phosphates at a heating rate of 10 °C / min were exported using Python and visualized as an image, as shown in the attached figure. Figure 2 .

[0085] Compared to conventional methods that calculate pyrolysis kinetics at a single heating rate, this invention uses multiple heating rates to verify the reliability of the pyrolysis kinetic function equation. Furthermore, by utilizing Python tools and machine learning, it quickly calculates and matches the optimal pyrolysis kinetic function equation for the material, significantly reducing computation time and improving accuracy. This provides a computational approach for the subsequent analysis of flame retardant pyrolysis kinetics.

[0086] The above embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of protection of this application. Any non-substantial changes and substitutions made by those skilled in the art based on this application shall fall within the scope of protection claimed by this application.

Claims

1. A method for resolving the pyrolysis kinetics of alkyl phosphates based on machine learning, characterized by, The method comprises the following steps: S1, TG test data acquisition: TG curves of the flame retardant from room temperature to 800°C were obtained by using a thermal gravimetric analyzer under different heating rates β of the alkyl phosphate flame retardant under different heating rates w TG test data with real-time temperature T1 changes S2, Reading and converting TG test data: Reading TG test data by Python and converting the remaining mass percentage of materials in the data to weight loss rate w α , reading the real-time temperature of test data T1 to absolute temperature T ;​ S3, creating a training set model: after substituting the relevant data into the Coats-Redfern pyrolysis kinetics classical mechanism function equation model, a training set model data is formed, and a training set model is created; S4, establishing linear regression: establishing linear regression on the training set, linear regression training is carried out for each model; the linear relationship between the training set model and and The activation energy E and the logarithmic pre-exponential factor lnA are obtained, and the linear correlation coefficient under each function equation is calculated R 2 ; wherein is the pyrolysis kinetics classical mechanism function equation; S5, training set verification: the verification step comprises: A1, the activation energy obtained from each training set model E and the logarithmic pre-exponential factor lnA The calculated values were compared with those obtained by the Kissinger method and the Flynn-Wall-Ozawa method; A2, the linear correlation coefficient calculated by the model R 2 low and activation energy E and logarithmic pre-exponential factor lnA If the calculated values of the two methods are obviously inconsistent, the model will be reselected in the training set until the linear correlation coefficient of the model is high and the activation energy calculated by the model is close to the activation energy calculated by the Kissinger method and the Flynn-Wall-Ozawa method E and logarithmic pre-exponential factor lnA When the calculated values obtained by the Kissinger method and the Flynn-Wall-Ozawa method are close, it is verified that the optimal pyrolysis kinetic mechanism function equation of the alkyl phosphate flame retardant is selected. S6, data visualization: the data calculated under the optimal pyrolysis kinetics mechanism function equation are exported by Python and visualized in the form of pictures.

2. The method for analyzing the pyrolysis kinetics of alkyl phosphate salts based on machine learning according to claim 1, characterized in that: The temperature increase rate in the step S1 β is 10-25°C / min.

3. The method for analyzing the pyrolysis kinetics of alkyl phosphate salts based on machine learning according to claim 1, characterized in that: The step S2 reads the TG test data by using Python, the real-time temperature T1 The conversion relationship between the absolute temperature T The conversion relationship between the residual mass percentage of the flame retardant w α The conversion relationship between the weight loss rate .​​ 4. The method for analyzing the pyrolysis kinetics of alkyl phosphate salts based on machine learning according to claim 1, characterized in that: The weight loss rate in the step S2 α Select 5-70% of the data as the calculation data of Python.

5. The method for analyzing the pyrolysis kinetics of alkyl phosphate salts based on machine learning according to claim 1, characterized in that: The data substitution process in the step S3 is completed by Python.

6. The method for analyzing the pyrolysis kinetics of alkyl phosphate salts based on machine learning according to claim 1, characterized in that: The slope and intercept parameters obtained by establishing linear regression in the step S4 with the data in the training set model can calculate the activation energy in the pyrolysis reaction process and The slope and intercept parameters obtained by establishing linear regression in the step S4 with the data in the training set model can calculate the activation energy in the pyrolysis reaction process E and the logarithmic pre-exponential factor lnA The score of the training set after linear regression is obtained, that is, the linear correlation coefficient of the model R 2 .

Images