Belt roaster thermal system control method based on digital twinning

Abstract

This invention discloses a control method for the thermal system of a belt sintering mill based on digital twins, comprising: collecting pellet production process data; correcting and fitting the collected pellet production process data to create a mechanism model for the physicochemical changes during pellet sintering; reducing the mechanism model to a low-order mathematical model, and then, combined with the pellet sintering heat transfer calculation formula, using pellet sintering quality indicators and minimum energy consumption during sintering as constraints, solving for the values ​​of controlled variables at different time points in each process stage of pellet sintering; based on the Urgench equation, solving for the values ​​of controllable variables using the controlled variable values, and sending the controllable variable values ​​to the programmable logic controller (PLC) of the belt sintering mill thermal system, enabling the PLC to automatically control the belt sintering mill thermal system to achieve minimum energy consumption sintering. This invention ensures both pellet production quality and achieves the lowest overall production energy consumption.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pellet belt induration machine control. More particularly, the present application relates to a belt induration machine thermal system control method based on digital twinning. BACKGROUND

[0002] Oxidized pellet is one of the important methods of fine ore agglomeration. First, the concentrate powder is dried in a drying kiln, then rolled to the appropriate strength, and then added with appropriate moisture and binder in the balling machine to form green balls with uniform viscosity and sufficient strength. The green balls are preheated and baked in an oxidizing atmosphere to form pellets. The oxidized pellet process is particularly suitable for processing fine concentrate powder. The produced pellets have good cold strength, reducibility and particle size composition. In the steel industry, pellets are important blast furnace burden and can be used together with sintered ore to form a good burden structure, which is widely used in major steel enterprises.

[0003] There are three processes for producing oxidized pellets, namely shaft furnace, chain-back-ring and belt induration machine. The belt induration machine process has occupied the main market in the world due to its small footprint, low comprehensive energy consumption and high product yield. Whether the belt induration machine runs stably and the energy consumption is optimal in production is directly related to its thermal system. A good thermal system of the belt induration machine can make the belt induration machine play a better role in energy saving and consumption reduction. However, the existing belt induration machine still uses a PID control system. With the expansion of the scale and the increase of the complexity of the industrial production process, the fluctuation range of the pellet product quality and the process control variable is becoming more and more strict. The conventional PID control system cannot meet the requirements of intelligent industrial production of pellets. The PID control system adopts point-to-point control mode and cannot comprehensively consider the key operating parameters of pellet production. It is necessary to manually monitor the key parameters of pellet production in real time and make reasonable adjustments, resulting in serious energy loss and waste of human resources in actual production. SUMMARY

[0004] The purpose of the present application is to provide a belt induration machine thermal system control method based on digital twinning, which can ensure the quality of pellet production and achieve the lowest comprehensive energy consumption in production.

[0005] In order to achieve these objects and other advantages in accordance with the present application, a belt induration machine thermal system control method based on digital twinning is provided, comprising:

[0006] collecting pellet production process data;

[0007] Based on the pellet sintering process theory, the collected pellet production process data is fitted, analyzed and corrected to create a mechanism model for the physical and chemical changes in the pellet sintering process;

[0008] The mechanistic model is reduced to a low-order mathematical model using CFD reduction. Digital twin technology is used to visualize the current data of the controlled variables in the sintering space of each process segment in pellet sintering. Then, combined with the pellet sintering heat transfer calculation formula, the values ​​of the controlled variables at different time points in each process segment of pellet sintering are solved under the constraints of pellet sintering quality index and minimum energy consumption during sintering. The controlled variables refer to the physical parameters in the low-order mathematical model and the pellet sintering heat transfer calculation formula that directly affect sintering quality and sintering energy consumption.

[0009] Based on the Ergen equation, the values ​​of the controllable variables are solved by combining the values ​​of the controlled variables. The values ​​of the controllable variables are then sent to the programmable logic controller (PLC) of the belt sintering machine thermal system. This enables the PLC to automatically control the belt sintering machine thermal system to achieve sintering with the lowest energy consumption. The controllable variables refer to the physical parameters that are controlled and adjusted by the PLC of the belt sintering machine thermal system.

[0010] Preferably, the pellet production process data includes flue gas temperature, gas pressure, wind speed at different time points in each process segment during pellet sintering, as well as the temperature at different material layer depths in the pellets.

[0011] Preferably, the fitting analysis method includes polynomial fitting, least squares fitting, and interpolation fitting.

[0012] Preferably, the order reduction methods for the mechanistic model include linear static order reduction, nonlinear static order reduction, and nonlinear dynamic order reduction.

[0013] Preferably, the mechanism model includes:

[0014]

[0015] Where R = 8.314 kJ / mol·K, k refers to the reaction rate constant at temperature T, A refers to the Arrhenius constant, E refers to the activation energy, T refers to the absolute temperature, and R refers to the gas constant.

[0016] Preferably, the formula for calculating heat transfer during pellet sintering includes:

[0017] Q = H × (T) f -T w )×A

[0018] Where Q refers to the heat absorbed by the material, H refers to the convective heat transfer coefficient, and T refers to the heat transfer coefficient. f Fluid temperature, T w The temperature of the material is referred to as A, and the convective heat transfer area is referred to as A.

[0019] Preferably, the controllable variables include at least the trolley speed, the fan speed, and the roasting temperature, and the controlled variables include at least the furnace temperature, the furnace pressure, the wind box temperature, and the wind box pressure.

[0020] The present application at least includes the following beneficial effects: the present application establishes a mechanism model according to the pellet production process data, realizes the accurate simulation of the pellet production process, and then reduces the mechanism model to a low-order mathematical model, improves the system operation efficiency while maintaining the simulation accuracy of the model, and achieves the second-level response speed adapted to the production environment, with the pellet sintering quality index and the lowest energy consumption in the sintering process as the constraint conditions, which ensures the pellet production quality and realizes the lowest comprehensive energy consumption in production. Through the application of the belt-type induration machine thermal system control method based on digital twinning, a breakthrough of 2% reduction in actual production energy consumption is realized on the basis of the original, which reduces the production cost and saves energy consumption. In addition, since the mechanism model CFD is reduced for system simulation, it is equivalent to installing a sensor at any point in the induration machine, and the controlled variable data of the coordinate point of each process section sintering space can be used for system simulation reduction. Through the establishment of digital twinning, the running system of the induration machine can be automatically controlled online, remotely monitored, the life of the components can be predicted, and the engineers can remotely diagnose and simulate the equipment failure and take remedial measures.

[0021] Other advantages, objects, and features of the present application will be apparent from the following description, and will be understood by those skilled in the art. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 Flow chart of the belt-type induration machine thermal system control method based on digital twinning of the present application;

[0023] Figure 2 Instantaneous temperature distribution curve diagram of the belt-type induration machine thermal system control method of the present application in actual application;

[0024] Figure 3 Periodic temperature change curve diagram of the belt-type induration machine thermal system control method of the present application in actual application;

[0025] Figure 4 Temperature field distribution diagram obtained by three-dimensional simulation of the belt-type induration machine thermal system control method of the present application in actual application using digital twinning technology. DETAILED DESCRIPTION

[0026] The present application will be further described in detail below with reference to the accompanying drawings, so that those skilled in the art can implement it according to the description.

[0027] It should be noted that in the description of the present application, the terms "transverse", "longitudinal", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the indicated device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0028] As shown in Figure 1 The present application provides a kind of based on digital twinning's hot system control method of belt roaster, comprising:

[0029] S1, collecting pellet production process data;

[0030] Specifically, the pellet production process data includes the flue gas temperature, air pressure, wind speed of each process section at different time points in the pellet sintering process and the temperature of different material layer depths of the pellet;

[0031] The flue gas temperature, air pressure and wind speed of each process section at different time points in the pellet sintering process can be obtained by the instrument arranged on the belt roaster, and the temperature of different material layer depths of the pellet can be obtained by inserting high-temperature measuring instruments of different depths in the pellet material layer. For example, high-temperature measuring instruments are placed at positions with material layer depths of 100mm, 200mm and 300mm, and a cycle of about 40 minutes (temperature gradient: room temperature-150 degrees-300 degrees-700 degrees-1250 degrees-850 degrees-400 degrees-150 degrees) is required for work, wherein the high-temperature 1250 degrees works for about 15-20 minutes, the temperature fluctuation is + / - 20 degrees, and the high-temperature measuring instrument can record the temperature change curve at the material layer depth within a cycle.

[0032] S2, based on the theory of pellet sintering process, fitting analysis and correction are carried out on the collected pellet production process data, and a mechanism model for the physical and chemical changes in the pellet sintering process is created;

[0033] Since pellet sintering refers to the process of preheating green balls and sintering them in an oxidizing atmosphere to make pellet ore through solid phase solidification reaction, the main processes include the oxidation of magnetite.

[0034] Based on the oxidation process of magnetite, fitting analysis and correction are carried out on the collected pellet production process data. The fitting analysis method described herein includes polynomial fitting, least squares fitting and interpolation fitting, and one or a combination of several fitting methods can be used to obtain the mechanism model for the physical and chemical changes in the pellet sintering process.

[0035] For the hyperparameters in the mechanism model, the collected balling process data is iteratively corrected, specifically, back propagation or forward propagation can be used, and then the gradient descent method is used to obtain the optimal hyperparameter value, the gradient descent method formula is: w i k+1 = w i k - [η x Δf(w i )], wherein η is the learning rate, the value is 0-1, used to control the convergence step, Δf w (w i ) is the gradient of the loss function, w i is the i-th hyperparameter in the model, and k is the iteration number. The mechanism model with the optimal hyperparameter value is finally judged for maturity according to generalization ability, plasticity, solvability and robustness, and the judgment method can be chi-square test.

[0036] Specifically, the mechanism model comprises:

[0037]

[0038] wherein R = 8.314 kJ / mol K, k is the reaction rate constant at temperature T, A is the Arrhenius constant, E is the activation energy, T is the absolute temperature, and R is the gas constant; in a specific case, after correction, A = 2 and E = 60 kJ / mol.

[0039] S3, the mechanism model is reduced to a low-order mathematical model by CFD reduction method, the current data of the controlled variables of the sintering space in each process section in the balling and sintering process are visualized by digital twinning technology, and then combined with the balling and sintering heat calculation formula, the value data of the controlled variables of each process section at different time points in the balling and sintering process are solved, the controlled variables are physical parameters directly affecting the sintering quality and sintering energy consumption in the low-order mathematical model and the balling and sintering heat calculation formula;

[0040] Since the mechanism model is a high-order model, although it is relatively simple and accurate in describing the balling and sintering process, it is not conducive to subsequent calculation process, therefore, it needs to be reduced. Specifically, the reduction methods of the mechanism model include linear static reduction, nonlinear static reduction and nonlinear dynamic reduction.

[0041] In digital twin technology, one-dimensional and three-dimensional coupled systems can map the entire life cycle of physical systems in virtual space, including three-dimensional simulation of key parts and one-dimensional system simulation of other parts. One-dimensional system simulation mainly includes Flowmaster, Twin builder, Modelica, and other multi-physical field and multi-interface system simulation software. To achieve rapid synchronization between three-dimensional simulation and one-dimensional system, the time of three-dimensional simulation must be greatly reduced.

[0042] Three-dimensional simulation mainly has the following three forms: 1) CFD simulation; 2) response surface analysis of finite experiments; and 3) CFD reduced order model. However, CFD simulation is time-consuming and cannot map the physical equipment in a timely manner; in response surface analysis, it cannot be guaranteed that the response surface passes through all sample points, so there is a certain error; and the reduced order model is a simplified version of a high-fidelity static or dynamic model that can retain necessary behaviors and dominant effects to reduce the solution time or storage capacity required for more complex models.

[0043] In the reduction process, steady-state simulation uses singular value decomposition (SVD) to compress two-dimensional / three-dimensional solutions, and combines interpolation methods to continuously reconstruct within the parameter range to build a reduced order model. Linear time-invariant (LTI) reduction of transient simulation is a non-intrusive method that learns from previous transient simulation learning data through deep learning (unsupervised reinforcement learning mode). Its essence is to calculate convolution according to the impulse response and input, and then directly calculate the output. Therefore, an LTI system is completely characterized by an impulse response. The derivative of the step response with respect to time is the impulse response, and the step response is calculated by a three-dimensional CFD model. Therefore, based on the three-dimensional reduced order model, coupling with one-dimensional system simulation can realize the digital twin design of the entire life cycle of the physical equipment.

[0044] Linear reduction uses LTI linear time-invariant technology to import mechanism models into the reduction simulation environment, trains using the same boundary input conditions, simulates output results in several scenarios, and generates a reduced order model. Nonlinear static model reduction outputs point data, i.e., response surface reduction and interpolation methods as the basic path. After generating the reduced order model, a working condition is randomly selected within the working range, and CFD simulation and CFD reduced order model are used for calculation, with a maximum deviation of 1.2%, proving that the CFD reduced order model is approaching to be usable.

[0045] The independent variables and dependent variables in the reduced low-order mathematical model remain unchanged, and combined with the pellet sintering heat calculation formula, the values of the controlled variables at different time points in the pellet sintering process are solved with the pellet sintering quality index and the minimum energy consumption in the sintering process as the constraint conditions.

[0046] Specifically, the pellet sintering heat calculation formula includes:

[0047] Q = H x (T f -T w ) x A

[0048] Wherein, Q refers to the material absorbs heat, H refers to the heat transfer coefficient, T f refers to the fluid temperature, T w refers to the material temperature, A refers to the heat transfer area.

[0049] And t = (Q / C p ) + t0, wherein t is the temperature of the pellet after heating, t0 is the initial temperature, C p is the specific heat capacity of the pellet, since the mechanism model is also related to the material temperature, therefore, the above reduced low-order mathematical model and the pellet sintering heat calculation formula are jointly solved with the pellet sintering quality index and the lowest energy consumption in the sintering process as the constraint condition, the value data of the controlled variables at different time points in each process section can be obtained respectively.

[0050] S4, based on the Ergun equation, the value data of the controllable variables is solved combined with the value data of the controlled variables, and the value data of the controllable variables is sent to the programmable logic controller of the belt type roaster thermal system, so that the programmable logic controller automatically controls the belt type roaster thermal system to realize the lowest energy consumption sintering, the controllable variable refers to the physical parameter controlled and adjusted by the programmable logic controller of the belt type roaster thermal system.

[0051] Since the Ergun equation describes the pressure drop of the fluid through the bed with a certain porosity and height composed of particles with a single screen size, and the above value data of the controlled variables is actually the coupling result of the temperature fluid field. Therefore, the value data of the controlled variables of the vector field source of each process section can be solved by using the Ergun equation, and the value data of the controlled variables of the vector field source of each process section is actually directly related to the physical parameter adjusted by the programmable logic controller, so the value data of the controllable variables can be solved.

[0052] Specifically, the controllable variables at least include pallet speed, fan speed, and roasting temperature. The controlled variables at least include hearth temperature, hearth air pressure, air box temperature, and air box air pressure.

[0053] It can be seen from the above examples that the mechanism model is established according to the data of the pellet production process, the accurate simulation of the pellet production process is realized, the mechanism model is reduced to a low-order mathematical model, the simulation accuracy of the model is maintained, the system operation efficiency is improved, the response speed of seconds is achieved to adapt to the production environment, the sintering quality index of the pellet and the minimum energy consumption in the sintering process are taken as the constraint conditions, the production quality of the pellet is ensured, and the minimum energy consumption in the production is realized. Through application of the belt-type indurating machine thermal system control method based on digital twinning, a breakthrough of 2% reduction in actual production energy consumption is realized on the basis of the original, the production cost is reduced, and the energy consumption is saved. In addition, since the mechanism model CFD is reduced for system simulation, it is equivalent to that a sensor is installed at any point in the indurating machine, and the controlled variable data of the coordinate point of each process section sintering space can be used for system simulation reduction. Through the establishment of digital twinning, the running system of the indurating machine can be automatically controlled online, remotely monitored, the life of the component is predicted, the engineer can remotely diagnose and simulate the fault of the equipment, and remedial measures are taken.

[0054] The following is a curve diagram of automatic adjustment of the temperature of the wind box in the actual production process:

[0055] Figure 2 It is the instantaneous temperature distribution curve of 1# wind box to 18# wind box at a moment;

[0056] Figure 3 It is the temperature change curve of 20# wind box to 24# wind box in a continuous period;

[0057] Figure 4 It is the temperature field distribution diagram obtained by using the digital twinning technology to simulate the sintering space in the pellet sintering process.

[0058] Although the embodiments of the present application have been disclosed as above, it is not limited to the application listed in the specification and the embodiments, and can be fully applied to various fields suitable for the present application, and other modifications can be easily realized by those skilled in the art, therefore, the present application is not limited to specific details and the figures shown and described herein, without departing from the general concept defined by the claims and the equivalent scope.

Claims

1. A belt-type calciner thermal system control method based on digital twinning, characterized by, The method comprises the following steps: Collecting pellet production process data, including flue gas temperature, air pressure, wind speed at different time points of each process section in the pellet sintering process, and temperature of different material layer depths of the pellets, wherein the temperature of different material layer depths of the pellets is obtained by inserting high-temperature measuring instruments at different depths in the pellet material layer, and the high-temperature measuring instruments record the temperature change curve at the material layer depth where they are located within a period; Based on the theory of pellet sintering process, the collected pellet production process data is fitted, analyzed, and corrected to create a mechanistic model for the physicochemical changes during pellet sintering. This mechanistic model includes the Arrhenius equation: k=A exp Where R = 8.314 kJ / mol·K, k refers to the reaction rate constant at temperature T, A refers to the Arrhenius constant, E refers to the activation energy, T refers to the absolute temperature, and R refers to the gas constant; the model accuracy is improved by iteratively optimizing the hyperparameters, and the iterative optimization uses the gradient descent method: w i k+1 =w i k -[η×Δf(w i ], where η is the learning rate, ranging from 0 to 1, used to control the convergence step size, Δf w (w) i Let w be the gradient of the loss function. i Let be the i-th hyperparameter in the model, and k be the number of iterations; The mechanism model is reduced to a low-order mathematical model by a CFD reduction method, wherein the CFD reduction method comprises: for steady-state simulation, singular value decomposition is used to compress the solution and combined with an interpolation method to reconstruct the reduced model; for transient simulation, a linear time-invariant reduction technique is used, and a step response to time derivative is convoluted as a pulse response to learn from transient simulation data through an unsupervised reinforcement learning mode; The current data of the controlled variables of the sintering space of each process section in the pellet sintering is visualized and displayed by using a digital twin technology, wherein the digital twin technology comprises coupling of a three-dimensional reduced model and a one-dimensional system simulation to realize full life cycle mapping of the physical system. In combination with a pellet sintering heat transfer calculation formula, the pellet sintering quality index and the lowest energy consumption in the sintering process are taken as constraint conditions, the value data of the controlled variables at different time points in each process section in the pellet sintering are solved, the pellet sintering heat transfer calculation formula comprises: Q=H*(T f − T w )*A, wherein Q refers to heat absorption of the material, H refers to a convective heat transfer coefficient, T f refers to fluid temperature, T w refers to material temperature, and A refers to a convective heat transfer area; and according to t=(Q / C p )+t0, wherein t is the temperature after the pellet is heated, t0 is an initial temperature, and C p is the specific heat capacity of the pellet, the low-order mathematical model after the reduction and the pellet sintering heat transfer calculation formula are jointly solved with the pellet sintering quality index and the lowest energy consumption in the sintering process as constraint conditions, the value data of the controlled variables at different time points in each process section are obtained, and the controlled variables refer to physical parameters in the low-order mathematical model and the pellet sintering heat transfer calculation formula which directly affect the sintering quality and the sintering energy consumption. Based on the Ergun equation, the value data of the controllable variables are solved combined with the value data of the controlled variables, and the value data of the controllable variables are downloaded to the programmable logic controller of the belt-type roaster thermal system, so that the programmable logic controller automatically controls the belt-type roaster thermal system to realize the lowest energy consumption sintering, wherein the controllable variables refer to physical parameters controlled and adjusted by the programmable logic controller of the belt-type roaster thermal system.

2. The digital-twin-based control method of a hot system of a belt sintering machine according to claim 1, characterized in that, The pellet production process data includes flue gas temperature, air pressure, wind speed at different time points of each process section in the pellet sintering process, and temperature of different material layer depths of the pellets.

3. The digital-twin-based control method of a hot system of a belt sintering machine according to claim 1, characterized in that, The fitting analysis method comprises polynomial fitting, least squares fitting, and interpolation fitting.

4. The digital-twin-based control method of a hot system of a belt sintering machine according to claim 1, characterized in that, The reduction mode of the mechanism model comprises linear static reduction, nonlinear static reduction, and nonlinear dynamic reduction.

5. The digital-twin-based control method of a hot system of a belt sintering machine according to claim 1, characterized in that, The controllable variables at least include pallet speed, fan speed, and sintering temperature, and the controlled variables at least include hearth temperature, hearth air pressure, wind box temperature, and wind box air pressure.

Images