Design method of rotary kiln incinerator digital twin platform

Abstract

This invention belongs to the field of solid waste treatment, specifically a design method for a digital twin platform for a rotary kiln incinerator. It obtains internal concentration fields and outlet concentration data of various substances in the rotary kiln incinerator through numerical simulation, and simplifies the relevant data into a simulation dataset suitable for machine learning. A prediction model is constructed based on least squares support vector machine to learn from the simulation data, enabling the prediction of the internal concentration fields of various substances in the rotary kiln incinerator and the outlet concentration of substances. The predictive digital twin platform for the rotary kiln incinerator is built using Python software programming, and a general-purpose APP is designed to intuitively and concisely display the predicted data of the rotary kiln incinerator, providing data support for the optimization and control of the rotary kiln incinerator in actual production.

Description

Technical Field

[0001] This invention belongs to the field of solid waste treatment technology, specifically a design method for a digital twin platform for a rotary kiln incinerator. Background Technology

[0002] With societal development, human production and daily life generate a large amount of solid waste. Solid waste is complex in composition and causes serious pollution, placing a heavy burden on social development and the ecological environment. Common solid waste treatment methods include landfill, composting, and incineration. Incineration, in particular, achieves its effect by pyrolyzing solid waste at high temperatures in an incinerator, rapidly processing large quantities of solid waste and completely eliminating harmful bacteria and viruses. Furthermore, the incineration process releases a significant amount of heat, which can be effectively recovered to achieve energy conservation and emission reduction. Rotary kiln incinerators have advantages such as good material adaptability, the ability to simultaneously process multiple phases of waste, low requirements on feed size, simple operation, and ease of control, thus they are widely used in the field of solid waste treatment. However, in engineering practice, rotary kiln incinerators have also revealed some intractable problems. For example, it is difficult to effectively monitor and analyze the internal state of the rotary kiln incinerator, making it challenging to specifically adjust parameters to address the problem of excessive pollutant emissions. Summary of the Invention

[0003] Digital twin technology maps physical entities to virtual entities by constructing a digital twin in the virtual space. The virtual digital twin can reflect various characteristics of the physical entity in real time, providing information and data support and guidance. Simulation technology is the core technology for creating and operating digital twins, perfectly replicating the operational state of the physical entity in the virtual space.

[0004] The purpose of this invention is to overcome the problem of difficulty in monitoring and analyzing the internal state of rotary kiln incinerators in the prior art, and to propose a design method for a digital twin platform for rotary kiln incinerators, providing data support for the optimized control of rotary kiln incinerators.

[0005] To achieve the above requirements, the technical solution adopted by the present invention is as follows:

[0006] A design method for a digital twin platform for a rotary kiln incinerator includes the following steps:

[0007] Step a: Create a 3D model of the rotary kiln incinerator and perform mesh generation;

[0008] Step b: Use the numerical simulation software COMSOL to simulate the rotary kiln incinerator under different operating conditions;

[0009] Step c: Collect and organize the simulation data of the rotary kiln incinerator, and organize the simulation data into a simulation dataset that can be used for machine learning.

[0010] Step d: Establish a least squares support vector machine prediction model and train the prediction model using a simulation dataset;

[0011] Step e: Use Python to design a digital twin platform for the rotary kiln incinerator to achieve rapid prediction of the internal concentration field and outlet concentration of the rotary kiln incinerator.

[0012] Preferably, in step a, the three-dimensional model is meshed using tetrahedral meshes based on the solid dimensions of the rotary kiln incinerator.

[0013] The physical model of the rotary kiln incinerator includes a cylindrical kiln body, an air inlet, and a feed inlet. Under all operating conditions, the air and feed inputs are stable with consistent composition, and the combustion reaction within the rotary kiln incinerator reaches a steady state at a certain moment under each operating condition. Based on the physical model of the rotary kiln incinerator and its reaction characteristics across all operating conditions, to reduce the computational load and ensure smooth numerical simulation, the three-dimensional model of the rotary kiln incinerator is simplified as follows, mainly including:

[0014] 1) Ignore the preheating device at the front end of the feed inlet and only model the rotary kiln incinerator;

[0015] 2) Assume that the composition of solid waste is stable and ignore the impact of fluctuations in solid waste composition on combustion;

[0016] 3) Assume that the initial velocity and temperature of air and material in each air duct are uniformly distributed when entering the kiln;

[0017] 4) The gas inside the rotary kiln incinerator is considered an ideal gas.

[0018] A three-dimensional model was established based on the simplified physical dimensions of the rotary kiln incinerator. The main structure includes a cylindrical kiln body, an air inlet at the kiln head, and a feed inlet. The air inlet surrounds the feed inlet. After the three-dimensional model was meshed using tetrahedral meshes, multi-physics numerical simulations were performed for various operating conditions with a throughput of 40t to 60t and an air intake velocity of 8m / s to 12m / s.

[0019] Preferably, in step b, referring to the actual operating conditions of the rotary kiln incinerator, four interfaces are selected and added: fluid flow, fluid heat transfer, chemical reaction, and mass transfer, and multiphysics coupling is performed. Specifically, a multiphysics model of the rotary kiln incinerator is constructed based on COMSOL software, and the four physical fields of fluid flow, fluid heat transfer, mass transfer, and chemical reaction are coupled together: non-isothermal flow is used to couple the fluid flow with the fluid heat transfer interface; fluid flow affects mass transfer, so reactive flow is used to couple the fluid flow with the mass transfer interface; the temperature generated by fluid heat transfer can affect the chemical reaction rate, and the endothermic or exothermic nature of the reaction affects the total heat transferred by the fluid, so coupling is achieved by adding a chemical reaction heat source under the fluid heat transfer interface and changing the model input temperature under the chemical reaction interface; the mass transfer field and the chemical reaction influence each other, and they are coupled through concentration changes and reaction rates.

[0020] Preferably, in step c, the concentration fields of each substance and the time-varying data of the outlet concentration in the simulation data are exported, simplified, and after removing outliers, they are organized into a simulation dataset. The rotary kiln incinerator is simulated under different operating conditions (processing capacity 40t~60t, inlet gas velocity 8m / s~12m / s) using multiphysics simulation. The concentration field distribution data of the longitudinal section of the rotary kiln incinerator after stabilization under each operating condition is exported. The data is simplified and distorted to form the simulation dataset. Simulations are performed with different feed rates based on different processing capacities. After removing outliers from the simulation data, a rotary kiln incinerator concentration field dataset is established.

[0021] Preferably, in step d, the program for the least squares support vector machine prediction model is written, and the prediction model is trained using a simulation dataset. The least squares support vector machine learning algorithm is used to learn from the simulation dataset to predict the concentration field of the rotary kiln incinerator. In the prediction method of this invention, the factors affecting the concentration field include the feed rate and the air intake rate. The feed rate and air intake rate are used as inputs, and the concentration field data is used as the output. The least squares support vector machine prediction model is... Radial basis function (RBF) is σ is the bandwidth of the kernel function; y is the output data of the least squares support vector machine, i.e., the concentration field data; x is the input data of the prediction model, i.e., the feed rate and air intake rate data; n represents the total number of dimensions; K(x i α(x) is a nonlinear mapping from input data to a high-dimensional space, obtained through radial basis kernel function operations. i Let α represent the parameter in the i-th dimension, and b be an undetermined constant. The prediction model is trained using a simulated dataset to obtain the optimal α. i The values ​​of b and 'b' represent the results of the least squares support vector machine prediction model.

[0022] Based on a simulation dataset constructed using multiphysics simulation, a machine learning program written in Python was used to predict the concentration field and outlet concentration. The machine learning type employed in this invention is least squares support vector machine. The simulation dataset constructed through machine learning multiphysics simulation enables real-time prediction of the concentration field inside and at the outlet of a rotary kiln incinerator.

[0023] Preferably, in step e, a digital twin platform for the rotary kiln incinerator is designed using Python. This platform allows for the acquisition of material concentration field cloud maps and outlet concentration-time curves. Switching between different substances can be completed simply by clicking the relevant panel. The digital twin platform is built using Python: QtDesigner is used to arrange panels for switching between carbon monoxide and other substances, with relevant canvases, input boxes, and buttons arranged under each panel. The least squares support vector machine prediction model is configured to respond by clicking relevant buttons. This allows the concentration data to be obtained by reading the input box content, predicting it using the least squares support vector machine, and then displaying the concentration data as cloud maps and curves on the corresponding canvases. Finally, the digital twin platform for the rotary kiln incinerator is packaged into an app.

[0024] Machine learning code is packaged and encapsulated into a digital twin platform app using Python software. An intuitive user interface is designed to present data to users in the form of cloud maps and curves. Users of the digital twin platform can perform various operations through this intuitive interface without needing to understand the underlying principles, and it can be used on any computer.

[0025] The beneficial effects of this invention are:

[0026] This invention utilizes simulation technology and digital twin technology, combined with machine learning, to construct a digital twin of a rotary kiln incinerator, enabling online prediction of the concentration field within the rotary kiln incinerator and providing data support for pollutant emission control in rotary kiln incinerators. Attached Figure Description

[0027] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0028] Figure 1 This is a flowchart illustrating the design of a digital twin system for a rotary kiln incinerator according to an embodiment of the present invention.

[0029] Figure 2 This is a three-dimensional structural diagram of the rotary kiln incinerator according to an embodiment of the present invention;

[0030] Figure 3 This is a grid division diagram of the rotary kiln incinerator according to an embodiment of the present invention;

[0031] Figure 4 This is a diagram of the digital twin platform interface according to an embodiment of the present invention. Detailed Implementation

[0032] The following describes embodiments of the present invention in conjunction with the accompanying drawings.

[0033] According to an embodiment of the present invention, the design flowchart is as follows: Figure 1 As shown, this paper provides a technical solution and a design method for a digital twin platform for a rotary kiln incinerator. This method can accurately predict the internal concentration field of the rotary kiln incinerator, which is difficult to measure, thus providing data support for controlling pollutant emissions. According to an implementation example of this application, the digital twin technical solution for the rotary kiln incinerator includes three parts: numerical simulation, machine learning, and software development.

[0034] (1) Numerical simulation and data export.

[0035] The technical solution of this invention uses COMSOL software to perform numerical simulation of a rotary kiln incinerator to obtain concentration field data inside the rotary kiln incinerator.

[0036] First, a simplified three-dimensional physical model of the rotary kiln incinerator is established, and the simplified three-dimensional structural diagram of the rotary kiln incinerator is shown below. Figure 2 As shown, a 3D structure of the rotary kiln incinerator was created using the geometry module in COMSOL software, and simplified into a cylinder with a diameter of 3.5m and a length of 13m according to the solution requirements. A feed inlet and five air inlets were set, with the feed inlet located at the center of the circular surface and the five air inlets evenly distributed around it. After establishing the 3D model of the rotary kiln incinerator, the mesh density was determined in the mesh options according to the model dimensions and solution requirements. A standard mesh was selected to ensure computational accuracy and save computation time. Tetrahedral meshes were used to mesh the fluid domain of the rotary kiln incinerator, as shown in the mesh diagram. Figure 3 As shown. Referring to the actual operating conditions of a rotary kiln incinerator, four interfaces were selected: fluid flow, fluid heat transfer, chemical reaction, and mass transfer. The chemical reaction interface determines the internal reaction of the rotary kiln incinerator and is coupled using multiphysics. Adding non-isothermal flow couples the fluid flow with the fluid heat transfer interface; adding reactive flow couples the mass transfer with the fluid flow interface; adding a reaction heat source condition under the fluid heat transfer module couples the fluid heat transfer with the chemical reaction interface; and the mass transfer with the chemical reaction interface is automatically coupled through concentration changes. This multiphysics coupling allows the numerical simulation to approximate the actual operating conditions as closely as possible.

[0037] The gas entering the rotary kiln incinerator is air. The boundary conditions at the mass transfer interface inlet are simplified to 21% O2 and 79% N2. The feed material is a mixture of water, carbon, and non-combustible components. Ignoring fluctuations in material composition, the boundary conditions at the mass transfer interface inlet are fixed at 20% water, 53.6% carbon, and 26.4% non-combustible components. Transient iterative calculations are performed using actual production data (daily throughput of 40-60 tons), with a total time of 200 seconds to ensure stability under all operating conditions and a time step of 0.1 seconds. After solving, the concentration fields of each substance and the outlet concentration changes are exported using the results processing module. The exported simulation data is then compiled into a simulation dataset suitable for machine learning.

[0038] (2) Machine learning (constructing a least squares support vector machine prediction model).

[0039] This invention uses least squares support vector machines to learn from the simulation dataset obtained from numerical simulations, enabling the prediction of the internal concentration field and outlet concentration of various substances in a rotary kiln incinerator. The simulation dataset obtained from the numerical simulations is used as the training set for machine learning; through training, applicable parameters are obtained, thus yielding the relevant machine learning prediction model.

[0040] Least squares support vector machine uses equality constraints instead of inequality constraints in traditional support vector machine. It selects the quadratic norm, which allows for approximation errors in samples in practical applications, as the loss function. This reduces computational complexity, ensures convergence speed, and effectively avoids the "curse of dimensionality," resulting in a model with better generalization ability.

[0041] This invention employs Python software to write machine learning programs. To eliminate the influence of dimensions between data features, the sample data is normalized. After normalizing the concentration fields of CO, CO2, and H2O and the outlet concentration data in the same way, a prediction model for the concentration field and outlet concentration of the rotary kiln incinerator is established using a least-squares support vector machine.

[0042] (3) Software setup (building a digital twin software platform).

[0043] This invention uses Python to build a digital twin platform and create an app that intuitively and concisely displays the predicted data of a rotary kiln incinerator. The digital twin platform interface is shown in the image below. Figure 4As shown. Based on practical application needs, this app has the function of real-time prediction and display of the concentration field inside the rotary kiln incinerator and the concentration of substances at the outlet. The app has input windows for throughput and air velocity. By inputting the relevant parameters within the specified range and clicking the plotting button, the app can obtain a cloud map of the concentration field and a curve of the outlet concentration changing over time. Switching between different substances can be completed simply by clicking the relevant panel.

[0044] Based on the above embodiments, this invention combines numerical simulation technology and machine learning to construct a digital twin platform for rotary kiln incinerators. This digital twin platform enables the prediction of the internal concentration field and outlet concentration of the rotary kiln incinerator under different throughput and intake conditions. In actual production, due to the complexity of the reactions inside the rotary kiln incinerator, changes in operating conditions cannot effectively control pollutant emissions. Using this rotary kiln incinerator digital twin platform, pollutant emissions and concentration distribution can be quickly predicted, providing data support for optimizing the control of the rotary kiln incinerator.

Claims

1. A design method for a digital twin platform for a rotary kiln incinerator, characterized in that, Includes the following steps: Step a: Create a 3D model of the rotary kiln incinerator and perform mesh generation; Step b: Use the numerical simulation software COMSOL to simulate the rotary kiln incinerator under different operating conditions; Step c: Collect and organize the simulation data of the rotary kiln incinerator, and organize the simulation data into a simulation dataset that can be used for machine learning. Step d: Establish a least squares support vector machine prediction model and train the prediction model using a simulation dataset; Step e: Use Python to design a digital twin platform for the rotary kiln incinerator to achieve rapid prediction of the internal concentration field and outlet concentration of the rotary kiln incinerator; In step b, referring to the actual operating conditions of the rotary kiln incinerator, four interfaces are selected and added: fluid flow, fluid heat transfer, chemical reaction, and mass transfer, and multiphysics coupling is performed. Non-isothermal flow is used to couple the fluid flow with the fluid heat transfer interface. Fluid flow affects mass transfer, so reactive flow is used to couple the fluid flow with the mass transfer interface. The temperature generated by fluid heat transfer affects the chemical reaction rate, and the endothermic or exothermic reaction affects the total heat of fluid heat transfer. Coupling is achieved by adding a chemical reaction heat source under the fluid heat transfer interface and changing the model input temperature under the chemical reaction interface. The mass transfer field and the chemical reaction influence each other, and the two are coupled through the concentration change relationship and the reaction rate.

2. The design method for the digital twin platform of the rotary kiln incinerator according to claim 1, characterized in that, In step a, the three-dimensional model is meshed using tetrahedral meshes based on the solid dimensions of the rotary kiln incinerator.

3. The design method for the digital twin platform of the rotary kiln incinerator according to claim 1, characterized in that, In step c, the concentration fields of each substance and the data on the change of outlet concentration over time in the simulation data are exported, simplified and corrupted, and then organized into a simulation dataset.

4. The design method for the digital twin platform of the rotary kiln incinerator according to claim 1, characterized in that, In step d, the factors affecting the concentration field include the feed rate and the air intake rate. The feed rate and air intake rate are used as inputs, and the concentration field data is used as the output. The least squares support vector machine prediction model is y= The radial basis function (RBF) is... σ is the bandwidth of the kernel function, y is the output data of the least squares support vector machine, i.e., the concentration field data, x is the input data of the prediction model, i.e., the feed rate and air intake rate data, n represents the total number of dimensions, and K(x) i α(x) is a nonlinear mapping from input data to a high-dimensional space, obtained through radial basis kernel function operations; i Let α represent the parameter in the i-th dimension, and b be an undetermined constant. The prediction model is trained using a simulated dataset to obtain the optimal α. i The values ​​of b and 'b' represent the results of the least squares support vector machine prediction model.

5. The design method for a digital twin platform for a rotary kiln incinerator according to claim 1, characterized in that, In step e, the material concentration field cloud map and the outlet concentration change curve over time are obtained through the rotary kiln incinerator digital twin platform. Switching between different materials can be completed simply by clicking the relevant panel.

6. The design method for the digital twin platform of the rotary kiln incinerator according to claim 5, characterized in that, A digital twin platform for a rotary kiln incinerator was built using Python: QtDesigner was used to arrange panels for switching between carbon monoxide and other substances, with relevant canvases, input boxes, and buttons arranged under each panel; the least squares support vector machine prediction model was set to respond by clicking the relevant buttons, so that after reading the content of the input box, the concentration data was obtained by least squares support vector machine prediction, and the concentration data was displayed on the corresponding canvas in the form of cloud maps and curves.

Images