Numerical simulation method for multiple physical phenomena of rotary kiln based on artificial intelligence reasoning

Abstract

The invention discloses a rotary kiln multi-physical phenomenon numerical simulation method based on artificial intelligence reasoning, and relates to the technical field of rotary kiln multi-physical phenomenon simulation. On the basis of an Euler multiphase flow equation, a heat transfer model, a chemical kinetic reaction model, a particle growth model, a chemical reaction model and the like are added; establishing a three-dimensional numerical model capable of accurately simulating multiple physical and chemical phenomena such as gas-solid two-phase flow movement, combustion, NOx generation, clinker calcination and particle growth in the rotary kiln; a long-short-term memory neural network is introduced to build an AI model, the AI model is used for predicting the working condition of the second half section of the model, the simulation speed is increased, and long-time simulation of the rotary kiln is achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of rotary kiln multi-physical phenomenon simulation, and in particular is a rotary kiln multi-physical phenomenon numerical simulation method based on artificial intelligence reasoning. Background Art

[0002] The rotary kiln is the core equipment of the new dry cement sintering system. The flame temperature, flame length, air supply, liquid phase volume of the material, etc. in the rotary kiln will have a great impact on the quality of the clinker. Since the temperature in the rotary kiln is extremely high, the highest temperature can reach about 1800℃, and the gas contains a lot of dust, there is currently no good experimental monitoring method to monitor the internal material movement, chemical reactions during clinker calcination, and particle growth during the operation of the rotary kiln; that is to say, the inside of the rotary kiln in the industry is still a "black box" for scientific research and design personnel;

[0003] In order to understand the temperature distribution, material movement and chemical reactions inside the rotary kiln, numerical simulation is the most widely used and economical method currently used by scientific researchers.

[0004] The patent with publication number CN117875097A discloses a simulation method for aluminum ash treatment in a rotary kiln using COMSOL. The method establishes a two-dimensional axisymmetric geometric model. Secondly, the geometric model area is divided according to the components of the rotary kiln. Then, the material parameters, boundary values, initial parameters are set, and meshing is performed. Finally, the calculation and solution are performed to obtain the simulation results of aluminum ash treatment in the rotary kiln.

[0005] The patent with publication number CN112380738A discloses a cement rotary kiln combustion field reconstruction error compensation and optimization method, storage medium and system. The method first establishes a cement rotary kiln numerical simulation model, extracts numerical simulation data and uses a neural network to obtain the relationship between combustion efficiency and boundary conditions, extracts characteristic parameters of combustion efficiency, uses the extracted characteristic parameters as error compensation modeling, and uses an online optimization algorithm to perform error compensation reconstruction on the cement rotary kiln site to achieve online real-time optimization.

[0006] The patent with announcement number CN106570244B discloses a one-dimensional mathematical model for establishing a cement rotary kiln, which includes a heat transfer model, a material bed movement model, and a clinker mineral chemical reaction model. The on-site production data is collected for the one-dimensional model boundary, and the changes of the flue gas temperature, material temperature, material bed height, wall temperature, material chemical composition, clinker mineral content, etc. in the rotary kiln with the length of the rotary kiln are calculated.

[0007] At present, the existing technology mainly focuses on the simulation of a single phenomenon, and cannot effectively simulate the interaction of the three-dimensional multi-physical field and chemical phenomena of the rotary kiln in actual operation; based on this, a solution is provided. Summary of the invention

[0008] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a numerical simulation method of multi-physical phenomena of a rotary kiln based on artificial intelligence reasoning.

[0009] A numerical simulation method for multi-physical phenomena of a rotary kiln based on artificial intelligence reasoning, the method specifically comprises the following steps:

[0010] Step 1: Use CAD software to draw the 3D geometry file of the rotary kiln, perform meshing, collect data on the kiln head temperature, kiln head pressure, kiln tail temperature, kiln tail pressure, coal feed, primary wind speed and temperature, secondary wind speed and temperature, and the content of each component in the kiln tail gas, and define boundary conditions;

[0011] Step 2: Build a numerical simulation model. The numerical simulation model is based on the Euler two-phase flow basic model. The remaining models will be built on the Euler two-phase flow model. The Euler two-phase flow basic model includes three basic equations: continuity equation, momentum equation, and energy equation.

[0012] Add sub-models to the numerical simulation model. The sub-models include:

[0013] Pulverized coal combustion simulation model, used to simulate the entire process of pulverized coal combustion;

[0014] Multiphase flow transmission and heat transfer simulation model, used to simulate the movement law and heat exchange law of burning material particles inside the rotary kiln;

[0015] Clinker calcination simulation model is used to simulate the evolution of mineral components inside the burning material under material transmission, heat transfer and mass transfer behavior;

[0016] The particle growth simulation model is used to simulate the situation in which particles grow larger during the calcination process, affecting the movement and heat exchange laws of the particles.

[0017] Pollutant generation simulation model, used to analyze the pollutants generated during the entire process;

[0018] Complete the construction of the 3D numerical simulation model of the rotary kiln.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] This application is based on the Euler multiphase flow equation, and adds heat transfer model, chemical kinetic reaction model, particle growth model, chemical reaction model, etc. to establish a three-dimensional numerical model that can accurately simulate multiple physical and chemical phenomena such as gas-solid two-phase flow movement, combustion, NOx generation, clinker calcination, and particle growth in the rotary kiln; and introduces a long short-term memory neural network to build an AI model, and uses the AI ​​model to predict the working conditions in the second half of the model, speeding up the simulation speed, and realizing long-term simulation of the rotary kiln. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is the technical roadmap of rotary kiln multi-phenomenon simulation of the present invention;

[0022] Figure 2 It is a schematic diagram of the intrinsic orthogonal decomposition of sample data of the present invention;

[0023] Figure 3 This is a detailed structure diagram of the LSTM of the present invention;

[0024] Figure 4 This is the overall neural network diagram of the AI ​​model of the present invention. DETAILED DESCRIPTION

[0025] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0026] See also Figure 1-Figure 4 The present application provides a numerical simulation method for multi-physical phenomena of a rotary kiln based on artificial intelligence reasoning, which specifically includes the following steps:

[0027] S1: Preparation;

[0028] Use CAD software to draw the 3D geometry file of the rotary kiln, perform mesh division, collect data on the kiln head temperature, kiln head pressure, kiln tail temperature, kiln tail pressure, coal feed, primary wind speed and temperature, secondary wind speed and temperature, and the content of each component in the kiln tail gas, and define boundary conditions;

[0029] S2: Numerical model construction;

[0030] (1) Basic equations: The numerical simulation model of this patent is based on the Euler two-phase flow basic model. The other models will be built on the Euler two-phase flow model. The Euler two-phase flow basic model includes three basic equations: continuity equation, momentum equation, and energy equation.

[0031] The basic Euler two-fluid model in this patent has three phases:

[0032] The first phase is the gas phase, which is also the main phase and contains various gas components such as:

[0033] There are 11 gas components in total: O2, CO2, N2, NO, CO, H2O(g), HCN, CH4.C2H4, C2H6, H2;

[0034] The second phase is coal powder solid particles, including four components: fixed carbon, volatile matter, ash, and H2O(l);

[0035] The third phase is the burning material particles, including six components: CaCO3, CAO, C2S, C3S, C3A, and C4AF;

[0036] The continuity equation for Euler-Euler two-phase flow is:

[0037]

[0038] In the formula, α q is the volume fraction of the q phase, is the q-phase velocity, is the mass flow rate transferred from phase p to phase q, is the mass flow rate transferred from phase q to phase p, S q It is the source phase.

[0039] The momentum control equation is:

[0040]

[0041] in,

[0042]

[0043] In the above formula, μ q and λ q The q phase is the shear viscosity and the volume viscosity, F q is the external force, F lift,q It is lift. is the turbulent dissipation force, Virtual mass force, Rpq is the interaction force between phases, It is the speed of meeting;

[0044] The energy governing equation is:

[0045]

[0046] In the formula, k eff,q is the thermal conductivity, S q is the source term, Q pq is the heat exchange between phase p and phase q, h q is the enthalpy of phase q, p op is the reference atmospheric pressure, p is the gauge pressure;

[0047] Furthermore, based on the Euler two-fluid model, the simulation content of the clinker calcination process inside the rotary kiln needs to add the following five sub-models:

[0048] (1) Pulverized coal combustion simulation model;

[0049] (2) Multiphase flow transmission and heat transfer simulation model;

[0050] (3) Clinker calcination simulation model;

[0051] (4) Particle growth simulation model;

[0052] (5) Pollutant generation simulation model, the pollutant is mainly NOx;

[0053] For the study of rotary kiln, the first thing is the simulation of pulverized coal combustion, which determines the overall temperature field of the rotary kiln;

[0054] Furthermore, the multiphase flow transmission and heat transfer simulation studied the movement law and heat exchange law of the burning material particles inside the rotary kiln;

[0055] Furthermore, the simulation model of calcination of calcined materials is used to study the evolution of the mineral composition inside the calcined materials under the material transmission / heat transfer / mass transfer behavior;

[0056] Furthermore, during the calcination process, the particles will gradually grow, which in turn affects the movement law and heat exchange law of the particles of the calcination material;

[0057] Finally, considering that pollutants will be generated in the whole process, and the generation of pollutants is related to the gas temperature field and gas composition, a pollutant generation simulation model is used; the above five parts are interrelated and influence each other. The simulation of a certain problem alone is inaccurate. Only when all the above five parts are considered in the same model, the simulation results are true and reliable, and have the value of guiding construction.

[0058] (2) The specific simulation model of pulverized coal combustion is as follows:

[0059] The pulverized coal combustion model involves the gas combustion model and the solid combustion model. The gas combustion model is that the volatile matter of the pulverized coal will emit combustible gas. This part uses the fast chemical reaction model:

[0060]

[0061] Among them, R i,r is the reaction rate (kmol / s), v″ i,r and v′ i,r is the stoichiometric coefficient, A and B are empirical constants, which are 4.0 and 0.5 respectively, YR is the reactant mass fraction, Mw,r is the mole fraction of the reactant, YP is the reactant mass fraction, Mw,p is the mole fraction of the reactant, ε is the turbulent dissipation rate, and k is the turbulence intensity.

[0062] The solid combustion model uses a reduced core model:

[0063]

[0064] Where P o2 is the partial pressure of oxygen around the particle, k fm is the film resistance, k am is the ash layer resistance, k rm is the inner core surface reaction resistance coefficient;

[0065] In the pulverized coal combustion model, gas combustion is added to the main phase, and solid combustion is added to the second phase, where pulverized coal combustion involves the transfer of the fixed carbon component C of the second phase to the component CO2 of the first phase, which is added using the heterogeneous chemical reaction module;

[0066] Among them, the multiphase flow transmission and heat transfer simulation model is as follows:

[0067] The interphase (gas-solid) transmission model uses the Huilin & Gidaspow drag model, and the specific theory is:

[0068] The force on the particle in the air is:

[0069]

[0070] Among them C d is the drag coefficient, ρ f is the fluid density (kg / m3), u is the fluid velocity (m / s), v p is the particle velocity (m / s); there are many ways to calculate the drag coefficient, and different calculation methods represent different drag models;

[0071]

[0072] in is the Ergun drag coefficient, is the Wen-Yu drag coefficient, ψ is the weight coefficient, and its value is (0,1);

[0073] The heat transfer model uses the GUNN model, specifically:

[0074]

[0075] Among them, the clinker calcination model is:

[0076] The clinker calcination model is added to the third phase, with a total of five chemical models. The reaction type and chemical kinetic parameters of each chemical reaction model have been obtained through experiments, as shown in Table 1:

[0077] Table 1 Reaction types and chemical kinetic parameters

[0078]

[0079] The particle growth model is as follows:

[0080] The particle enlargement model uses the PBM model, which can simulate particle growth relatively and quickly. PBM ignores particle breakage and agglomeration and only considers particle growth. At the same time, the discrete method is used to solve the PBM model, and the PBM model is coupled with the CFD model to simulate the clinker calcination process. The particle enlargement model is added to the third phase.

[0081]

[0082] The Hillert model is used to calculate and predict the clinker calcination model, specifically:

[0083]

[0084] Where d0 is the initial particle size, A is the pre-exponential factor of particle growth, E p is the particle growth activation energy, R is the gas constant, T is the particle temperature, and time is the calcination time of the particles after entering the kiln;

[0085] Among them, the pollutant generation simulation model is specifically as follows:

[0086] The pollutant model uses a finite chemical reaction rate model, where the chemical reaction rate is:

[0087]

[0088] ”'

[0089] Among them, C j,r is the molar concentration (kmol / s), η i,r With η i,r are reactants and products"

[0090] The equivalent number, Γ represents the effect of the reversible reaction, v i,r With v i,r is the stoichiometric number;

[0091] Among them, the chemical reaction rate constant often uses the Arrhenius formula:

[0092]

[0093] Among them, A r is the pre-exponential factor, T is the temperature, β is the temperature coefficient, E r is the activation energy, R is the gas constant;

[0094] In addition, there is a catalytic reduction reaction. This reaction model is based on the finite chemical rate model by adding a catalyst mass phase, namely:

[0095]

[0096] The pollutant model components are all gases, so they are added to the first phase;

[0097] At this point, the 3D numerical simulation of the rotary kiln has been completed.

[0098] S3: numerical simulation calculation correction;

[0099] Calculate the numerical simulation model built by S2 and verify the accuracy of the built model with the field data collected by S1. If there are any differences, adjust the parameters;

[0100] S4: Long short-term memory neural network (LSTC) for long-term computation;

[0101] The model built by S2 basically realizes the simulation of all behaviors in the clinker calcination process in the rotary kiln. However, the chemical reaction size of the numerical model built by S2 is very different. The fastest chemical reaction rate (combustion) is 1e9 times the slowest reaction rate (particle enlargement), and the stiffness of the chemical reaction is too large.

[0102] Although the use of implicit algorithms can reduce chemical stiffness to a certain extent, it takes at least thirty minutes for the material to enter the kiln and exit the kiln. The amount of simulation calculations required to complete more than thirty minutes using numerical simulation is huge. If the rotary kiln must be recalculated every time the rotary kiln parameters are adjusted, the calculation efficiency is too slow. This is the biggest difficulty in fully simulating the rotary kiln.

[0103] The total calculation time of the rotary kiln is 1800s. This patent uses the numerical model built by S2 to calculate the first 500s, and derives the velocity field, temperature field, and three-phase gas-solid component field data of the rotary kiln. The flow field feature analysis method based on the long short-term memory neural network (LSTC) is used, and the unsupervised training method is used to fully explore the implicit features in the time-series signal, and the low-dimensional characterization and feature analysis of the complex time series features of the flow field in the rotary kiln are performed. The rotary kiln operation AI model is obtained, and the AI ​​model is used to calculate the subsequent 1300s, which can greatly save calculation time.

[0104] (1) Intrinsic orthogonal decomposition of sample data

[0105] Intrinsic orthogonal decomposition or principal component analysis is a method of vector data statistical analysis that can reduce the order of high-dimensional flow field data and map it to a low-dimensional orthogonal basis mode space, thereby analyzing the main characteristics of the flow field and its corresponding basis mode coefficients. In essence, it maximizes the sample variance in each dimension after reducing the flow field to a low dimension.

[0106] First, the sample (flow field data) calculated by simulation needs to be standardized. Let the original data be x i , where i = 1, 2, 3, ... r, sample data x i is an n-dimensional vector and r is the number of samples.

[0107]

[0108] The covariance matrix of the standardized data can be further obtained:

[0109]

[0110] By solving the eigenvalues ​​of the nXn-order covariance matrix, the first m-order eigenvalues ​​can be recorded as λ 1 , 2 , …, λ m , the corresponding fundamental mode eigenvector can be recorded as ξ 1 , 2 ,…,ξ m The selection of m value is determined according to the proportion of variance values ​​of different fundamental modes to the total variance value, ensuring that the characteristic components contained in the fundamental mode account for more than 95% of the entire sample space.

[0111] Then the original sample data can be approximately expressed as:

[0112] X=U λ ξ

[0113] In this way, the fundamental mode coefficient matrix and a small number of POD fundamental modes can be used to represent most of the information of the original sample.

[0114] (2) Convolutional layer

[0115] In order to effectively extract the time series features of different positions in the flow field, one-dimensional convolution is combined to extract the features of the time series signal. The size of all convolution kernels in the model is 9, which is a relatively large size. The number of convolution kernels is thirty-two, and the activation function uses the RELU function to obtain a larger range of local time series information, which is conducive to the convergence of the model.

[0116] (3) Long short-term memory neural network LSTM;

[0117] Long short-term memory neural networks can learn long-term dependent information and are suitable for processing and predicting important events with relatively long intervals and delays in time series, avoiding the problem of gradient vanishing or gradient exploding during back propagation; the LSTM network structure consists of a forget gate, an input gate, and an output gate.

[0118] Forget gate: The forget gate takes the output ht-1 of the previous layer and the sequence data xt to be input to this layer as input, and obtains the output ft through an activation function sigmorid. The output value of ft is in the interval [0,1], indicating the probability that the cell state of the previous layer is forgotten, 1 is completely retained, and 0 is completely forgotten.

[0119] f t =σ(Wf ·[h t-1 ,x t ]+b f )

[0120] Input gate: The input gate consists of two parts. The first part uses the sigmoid activation function and the output is it. The second part uses the tanh activation function and the output is is the output of this layer, it takes values ​​in the range [0,1], the degree to which information in the representation is preserved, Indicates the new information that is retained by this layer.

[0121] i t =σ(W f ·[h t-1 ,x t ]+b i )

[0122]

[0123] So far, ft is the output of the forget gate, which controls the degree to which the cell state Ct-1 in the previous layer is forgotten. The multiplication operation of the two outputs of the input gate indicates how much new information is retained. Based on this, we can update the cell state Ct of this layer with the new information.

[0124]

[0125] Output gate: The output gate is used to control how much of the cell state of this layer is filtered. First, the sigmoid activation function is used to obtain an ot with a value in the [0,1] interval. Then, the cell state Ct is processed by the tanh activation function and multiplied by ot, which is the output ht of this layer.

[0126] o t =σ(W o [h t-1 ,x t ]+b o )

[0127] h t =o t tanh(C t )

[0128] There are three main steps:

[0129] Step 1: Forward calculate the output value of each neuron. For LSTM, the calculations are performed separately according to the algorithm introduced above.

[0130] Step 2: Determine the optimization objective function. In the early stages of training, the output value will be inconsistent with the expected value, so the error term of each neuron is calculated to construct the loss function.

[0131] Step 3: Update the network weight parameters according to the gradient of the loss function. The back propagation of the LSTM error term includes two levels: one is the spatial level, which propagates the error term to the previous layer of the network. The other is the temporal level, which propagates back along time, that is, starting from the current time t, calculate the error at each moment.

[0132] Then jump to the first step and repeat steps 1, 2, and 3 until the network error is less than the given value.

[0133] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A numerical simulation method for multi-physical phenomena of rotary kiln based on artificial intelligence reasoning, characterized in that: The method specifically comprises the following steps: Step 1: Use CAD software to draw the 3D geometry file of the rotary kiln, perform meshing, collect data on the kiln head temperature, kiln head pressure, kiln tail temperature, kiln tail pressure, coal feed, primary wind speed and temperature, secondary wind speed and temperature, and the content of each component in the kiln tail gas, and define boundary conditions; Step 2: Build a numerical simulation model. The numerical simulation model is based on the Euler two-phase flow basic model. The remaining models will be built on the Euler two-phase flow model. The Euler two-phase flow basic model includes three basic equations: continuity equation, momentum equation, and energy equation. Add sub-models to the numerical simulation model. The sub-models include: Pulverized coal combustion simulation model, used to simulate the entire process of pulverized coal combustion; Multiphase flow transmission and heat transfer simulation model, used to simulate the movement law and heat exchange law of burning material particles inside the rotary kiln; Clinker calcination simulation model is used to simulate the evolution of mineral components inside the burning material under material transmission, heat transfer and mass transfer behavior; The particle growth simulation model is used to simulate the situation in which particles grow larger during the calcination process, affecting the movement and heat exchange laws of the particles. Pollutant generation simulation model, used to analyze the pollutants generated during the entire process; Complete the construction of the 3D numerical simulation model of the rotary kiln.

2. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1 is characterized in that: After completing step 2, you need to perform the following steps: Calculate the numerical simulation model built in step 2, and verify the accuracy of the built model based on the field data collected in step 1. If there are any differences, adjust the parameters.

3. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1 is characterized in that: The Euler two-phase flow model has three phases, specifically: The first phase is the gas phase, which contains various gas components such as: There are 11 gas components in total: O2, CO2, N2, NO, CO, H2O(g), HCN, CH4.C2H4, C2H6, H2; The second phase is coal powder solid particles, including four components: fixed carbon, volatile matter, ash, and H2O(l); The third phase is the burning material particles, including six components: CaCO3, CAO, C2S, C3S, C3A, and C4AF; Among them, the first phase is the main phase.

4. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 3 is characterized in that: The continuity equation for Euler-Euler two-phase flow is: In the formula, α q is the volume fraction of the q phase, is the q-phase velocity, is the mass flow rate transferred from phase p to phase q, is the mass flow rate transferred from phase q to phase p, S q It is the source phase; The momentum control equation is: in, In the formula, μ q and λ q The q phase is the shear viscosity and the volume viscosity, F q is the external force, F lift,q It is lift. is the turbulent dissipation force, Virtual mass force, Rpq is the interaction force between phases, It is the speed of meeting; The energy governing equation is: In the formula, k eff,q is the thermal conductivity, S q is the source term, Q pq is the heat exchange between phase p and phase q, h q is the enthalpy of phase q, p op is the reference atmospheric pressure and p is the gauge pressure.

5. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1 is characterized in that: in, The pulverized coal combustion simulation model includes a gas combustion model and a solid combustion model; The gas combustion model is that the volatile matter of coal powder will emit combustible gas. This part uses the fast chemical reaction model: Among them, R i,r is the reaction rate (kmol / s), v i ” ,r With v i ' ,r is the stoichiometric coefficient, A and B are empirical constants, 4.0 and 0.5 respectively, YR is the reactant mass fraction, Mw,r is the reactant mole fraction, YP is the reactant mass fraction, Mw,p is the reactant mole fraction, ε is the turbulent dissipation rate, and k is the turbulence intensity; The solid combustion model uses a reduced core model: Where P o2 is the partial pressure of oxygen around the particle, k fm is the film resistance, k am is the ash layer resistance, k rm is the inner core surface reaction resistance coefficient; In the pulverized coal combustion model, gas combustion is added to the main phase, and solid combustion is added to the second phase. The pulverized coal combustion involves the transfer of the fixed carbon component carbon of the second phase to the carbon dioxide component of the first phase. This chemical reaction involving the transfer of substances between phases uses the heterogeneous chemical reaction module. Among them, the multiphase flow transmission and heat transfer simulation model includes interphase transmission model and heat transfer model; The interphase transmission model uses the Huilin & Gidaspow drag model, and the interphase transmission model is the gas-solid transmission model; The force on the particle in the air is: In the formula, C d is the drag coefficient, ρ f is the fluid density (kg / m3), u is the fluid velocity (m / s), v p is the particle velocity (m / s); C D =ψ CDErgun +(1-ψ)C DWen-Yu In the formula, C DErgun is the Ergun drag coefficient, C DWen-Yu is the Wen-Yu drag coefficient, ψ is the weight coefficient, and its value is (0,1); The heat transfer model uses the GUNN model, specifically: Among them, the clinker calcination simulation model is added to the third phase, with a total of five chemical models. The reaction type and chemical kinetic parameters of each chemical reaction model are obtained through experiments; The particle growth model is as follows: The particle enlargement model uses the PBM model, and the discrete method is used to solve the PBM model. The PBM model is coupled with the CFD model to simulate the clinker calcination process. The particle enlargement model is added to the third phase. The Hillert model is used to calculate and predict the clinker calcination model, specifically: Where d0 is the initial particle size, A is the pre-exponential factor of particle growth, E p is the particle growth activation energy, R is the gas constant, T is the particle temperature, and time is the calcination time; Among them, the pollutant generation simulation model is specifically as follows: The pollutant model uses a finite chemical reaction rate model, where the chemical reaction rate is: Among them, C j,r is the molar concentration (kmol / s), η' i,r and η” i,r is the number of equivalents of reactants and products, Γ represents the effect of reversible reaction, v″ i,r and v′ i,r is the stoichiometric number; The chemical reaction rate constant uses the Arrhenius formula: Among them, A r is the pre-exponential factor, T is the temperature, β is the temperature coefficient, E r is the activation energy, R is the gas constant; In addition, there is a catalytic reduction reaction. This reaction model is based on the finite chemical rate model by adding a catalyst mass phase, namely: The pollutant model components are all gases, so they are added to the first phase; At this point, the 3D numerical simulation of the rotary kiln has been completed.

6. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1 is characterized in that: After calculating the first 500s of rotary kiln simulation using the numerical simulation model built in step 2, the velocity field, temperature field, and three-phase gas-solid composition field data of the rotary kiln are derived; The flow field feature analysis method based on long short-term memory neural network is used, and the implicit features in the time series signal are fully mined by unsupervised training method. The low-dimensional characterization and feature analysis of the complex time series characteristics of the flow field in the rotary kiln are performed, and the rotary kiln operation AI model is obtained. The AI ​​model is used to calculate the subsequent 1300s.

7. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1 is characterized in that: The specific process of low-dimensional characterization and feature analysis of the complex time series characteristics of the flow field in the rotary kiln is as follows: First, perform the intrinsic orthogonal decomposition of the sample data and standardize the sample calculated by simulation, that is, the flow field data; let the original data be x i , where i = 1, 2, 3, ... r, sample data x i is an n-dimensional vector, r is the number of samples; The covariance matrix of the standardized data is obtained as: By solving the eigenvalue of the nXn order covariance matrix, the first m order eigenvalues ​​can be recorded as λ 1 , 2 , …, λ m , the corresponding fundamental mode eigenvector can be recorded as ξ 1 , 2 ,…,ξ m ; The selection of m value is determined according to the proportion of variance values ​​of different fundamental modes to the total variance value, ensuring that the characteristic components contained in the fundamental mode account for more than 95% of the entire sample space; Then the original sample data can be approximately expressed as: X=U λ ξ; Secondly, the feature extraction of time-course signals is performed by combining one-dimensional convolution. The size of all convolution kernels in the model is a larger size of nine, the number of convolution kernels is thirty-two, and the activation function uses the RELU function to obtain local time-course information. With the help of long short-term memory neural network, long-term dependent information can be learned to process and predict important events with relatively long intervals and delays in time series, avoiding the problem of gradient vanishing or gradient exploding during back propagation; the long short-term memory neural network consists of a forget gate, an input gate, and an output gate.

8. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1, characterized in that: The forget gate takes the output ht-1 of the previous layer and the sequence data xt to be input to this layer as input, and obtains the output ft through an activation function sigmorid; the output value of ft is in the interval [0,1], indicating the probability that the cell state of the previous layer is forgotten, 1 is completely retained, and 0 is completely forgotten; the specific formula is: f t =σ(W f ·[h t-1 ,x t ]+b f ); Output gate: The input gate consists of two parts. The first part uses the sigmoid activation function and the output is it. The second part uses the tanh activation function and the output is is the output of this layer, it takes values ​​in the range [0,1], the degree to which information in the representation is preserved, Indicates the new information that is retained in this layer; i t =σ(W f ·[h t-1 ,x t ]+b i ) So far, ft is the output of the forget gate, which controls the degree to which the cell state Ct-1 in the previous layer is forgotten. The multiplication operation of the two outputs of the input gate indicates how much new information is retained. Based on this, we can update the cell state Ct of this layer with the new information; Output gate: The output gate is used to control the number of filters of the cell state of this layer. First, the sigmoid activation function is used to obtain an ot with a value in the [0,1] interval. Then, the cell state Ct is processed by the tanh activation function and multiplied by ot, which is the output ht of this layer. the t =σ(W o [h t-1 ,x t ]+b o ) h t =o t ·tanh(C t )。 9. The method for numerical simulation of multi-physical phenomena of rotary kiln based on artificial intelligence reasoning according to claim 1, characterized in that: The specific way to learn long-term dependent information with the help of long short-term memory neural network is: Step 1: Forward calculate the output value of each neuron; for LSTM, calculate separately according to the algorithm introduced above; Step 2: Determine the optimization objective function; in the early stages of training, the output value and the expected value will be inconsistent, so the error term value of each neuron is calculated to construct the loss function; Step 3: Update the network weight parameters according to the gradient guidance of the loss function; the back propagation of the LSTM error term includes two levels: one is the spatial level, propagating the error term to the previous layer of the network; the other is the temporal level, back propagating along time, that is, starting from the current time t, calculating the error at each moment; Then jump to the first step and repeat steps 1, 2, and 3 until the network error is less than the given value.

Images