Digital twin system of cement decomposition furnace based on POD and deep learning

Abstract

The present invention discloses a digital twin system of a cement decomposition furnace based on POD and deep learning, comprising a data acquisition and processing module, a database retrieval module, a model reduction prediction module, a comprehensive display module and a remote control module connected in sequence. The present invention combines computational fluid dynamics with POD and deep learning to achieve rapid calculation of the entire temperature field and flow field in the entire cement decomposition furnace based on a finite number of discrete operating points, and displays them in a variety of ways. The present invention enables cement decomposition furnace production and operation personnel to more intuitively and quickly obtain the real-time changes in the temperature field and flow field in the cement decomposition furnace, so as to timely regulate the coal powder combustion and raw material decomposition in the furnace according to the changes.

Description

Technical Field

[0001] The present invention relates to cement production, and in particular to a digital twin system of a cement decomposition furnace based on POD and deep learning. Background Art

[0002] In the cement production process, the calcination and decomposition of cement raw materials in the cement decomposition furnace is a very important link in the entire cement production process, and its calcination quality is a key factor affecting the final cement quality. Therefore, in cement production, production operators will pay special attention to the temperature, flow rate and gas composition in the cement decomposition furnace to ensure that the cement raw materials can be fully decomposed. However, due to the limitations of observation technology, the current front-line production operators can only obtain data on temperature, flow rate and gas composition at limited locations, which brings great difficulties to the production operators in regulating the combustion of coal powder in the furnace and the decomposition of raw materials.

[0003] With the gradual development and improvement of CFD computational fluid dynamics technology, people can now calculate the complete operating information such as the temperature field and velocity field in the cement decomposition furnace by establishing models and setting boundary conditions. However, the time consumed by conventional methods is generally long, and it is impossible to quickly calculate the complete operating information using a set of characteristic physical quantities in the furnace. Therefore, conventional calculation methods are of limited help to the real-time regulation of production operators. Therefore, in order to enable production operators to obtain the coal powder combustion and raw material decomposition in the cement decomposition furnace in real time, and to further regulate the coal powder combustion and cement raw material decomposition, it is very necessary to propose and design a digital twin system that can accurately and quickly reflect the combustion and decomposition conditions in the cement decomposition furnace. Summary of the invention

[0004] Purpose of the invention: The purpose of the invention is to provide a digital twin system that can accurately and quickly reflect the combustion dynamics field and raw material decomposition conditions in a cement decomposition furnace.

[0005] Technical solution: The digital twin system of a cement decomposition furnace based on POD and deep learning described in the present invention comprises a data acquisition and processing module, a database retrieval module, a model reduction prediction module, a comprehensive display module and a remote control module connected in sequence;

[0006] The data acquisition and processing module can collect data from multiple specific locations in the cement decomposition furnace in real time, including temperature, gas flow rate and gas composition, and process the collected data to form a set of characteristic physical quantities and transmit them to the database retrieval module;

[0007] The database retrieval module can receive a set of characteristic physical quantities from the data acquisition and processing module, and compare the set of characteristic physical quantities with the corresponding data of the existing complete working condition at the specific position in the database; if there is a working condition that is successfully compared, the complete working condition data is directly transmitted to the comprehensive display module; if the comparison is not successful, the set of characteristic physical quantities is passed to the model reduction prediction module;

[0008] The model reduction prediction module can receive information transmitted from the database retrieval module, and based on the information, use the POD (proper orthogonal decomposition) method to reduce the order of the established furnace reaction model, wherein the coefficients of the basis vector of the POD are corrected by deep learning; after obtaining the reduced-order model, the relevant data in the current furnace, including the temperature field, the velocity field and the smoke composition information in the furnace, are quickly calculated based on the previously established grid, and then the calculation results are transmitted to the database retrieval module on the one hand, and to the comprehensive display module on the other hand;

[0009] The comprehensive display module can receive information transmitted by the database retrieval module and the model reduction prediction module, and restore and display it on the computer terminal; the information includes temperature field, velocity field and smoke composition information in the furnace;

[0010] The remote control module can receive instructions from production operators to control the amount of pulverized coal, the amount of cement raw materials, the primary air volume, the secondary air volume, and the tertiary air volume.

[0011] The data acquisition and processing module uses multiple sensors arranged in the cement decomposition furnace to collect different physical quantities at different positions in the cement decomposition furnace, including temperature, gas flow rate, and gas composition, and integrates them into a set of characteristic physical quantities and transmits them to the database query module for analysis and query.

[0012] The database retrieval module has two data input ports, one port receives a group of characteristic physical quantities transmitted from the data acquisition and processing module, and the other port receives the complete operating condition data in the furnace calculated by the model reduction prediction module for a certain group of characteristic physical quantities; the database retrieval module has three data output ports, two ports are connected to the model reduction prediction module, one port transmits a group of characteristic physical quantities to it to calculate the complete operating condition data in the furnace, the other port transmits all the complete operating condition data contained in the database to it for calculating the reduced-order model, and the third port is connected to the comprehensive display module in the furnace, and transmits the existing complete operating condition data in the database to it for display.

[0013] The model order reduction prediction module has two input ports, one input port receives a set of characteristic physical quantities transmitted from the database retrieval module for calculation, and the other input port receives multiple complete operating condition data transmitted from the database retrieval module for calculating the reduced order model; the model order reduction prediction module has two output ports, one output port is connected to the database retrieval module, and the complete operating condition data obtained by this calculation is transmitted to it for storage, and the other output port is connected to the comprehensive display module in the furnace, and the complete operating condition data obtained by this calculation is transmitted to it for display.

[0014] The comprehensive display module has two input ports, one input port is connected to the database retrieval module, and the other input port is connected to the model reduction prediction module. After receiving the complete operating condition data transmitted from the above two ports, the comprehensive display module in the furnace processes and displays the data.

[0015] The data display forms of the comprehensive display module include tables, line graphs, bar graphs, scatter plots, contour plots, profile graphs, 3D stereograms, dynamic pictures and videos.

[0016] The adjustable operating parameters of the remote control module include coal input amount, raw material amount, primary air volume, secondary air volume, tertiary air volume, secondary air temperature, and tertiary air temperature.

[0017] The specific operation steps of the system are:

[0018] (1) A three-dimensional model of the cement decomposition furnace is established according to the specific structure and actual size of the cement decomposition furnace, and the model is meshed; an appropriate mathematical model is selected, and the operating condition information of the cement decomposition furnace under multiple conditions is collected as boundary conditions, and CFD numerical simulations are performed on the temperature field, flow field, coal powder combustion, and cement raw material decomposition in the cement decomposition furnace under these operating conditions; the numerical simulation results under multiple operating conditions are stored as complete operating condition data in the database retrieval module;

[0019] (2) Using the SNAPSHOT algorithm in POD, a set of optimal canonical orthogonal bases is extracted based on multiple complete working conditions in the existing database, and then a set of optimal orthogonal basis coefficients is obtained using the deep learning method; based on this set of orthogonal bases and this set of orthogonal basis coefficients, a similar low-order model is obtained and stored in the model reduction prediction module;

[0020] (3) When it is necessary to reconstruct and predict the current combustion conditions in the furnace and the decomposition conditions of the cement raw materials, the data acquisition and processing module will first use multiple sensors arranged in the furnace to obtain a set of characteristic physical quantities, and then this set of characteristic physical quantities will be transmitted to the database retrieval module; if the database contains a complete working condition corresponding to this set of characteristic physical quantities, the complete working condition data will be directly transmitted to the comprehensive display module for display; if the database does not contain a complete working condition corresponding to this set of characteristic physical quantities, the set of characteristic physical quantities will be transmitted to the model reduction prediction module, and based on the previously established low-order model, the complete working condition data corresponding to this set of characteristic physical quantities will be quickly calculated; this set of complete working condition data will be transmitted to the comprehensive display module in the furnace for display, and on the other hand, it will be transmitted to the database retrieval module for storage for the next retrieval;

[0021] (4) Whenever the number of complete working conditions stored in the database retrieval module changes, the database retrieval module will pass all the complete working conditions it contains to the model reduction prediction module, and re-extract the optimal orthogonal basis and corresponding coefficients based on these complete working conditions to obtain a new reduced-order model to iteratively process the previous reduced-order model;

[0022] (5) When the complete working conditions are displayed on the screen, the production and operation personnel can interact with the displayed content by using a mouse or touching the screen. In response to the problems exposed by the current working conditions, the production and operation personnel can change the corresponding parameters in the remote control module to complete the control of the cement decomposition furnace operating parameters.

[0023] A computer storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned digital twin system of a cement decomposition furnace based on POD and deep learning.

[0024] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the digital twin system of a cement decomposition furnace based on POD and deep learning is implemented.

[0025] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0026] 1. The present invention can calculate the temperature field and flow field data in the entire cement decomposition furnace according to a set of characteristic physical quantities, so that the production and operation personnel can fully understand the temperature field, three-dimensional flow field, coal powder combustion and cement raw material decomposition data in the furnace, and thus regulate the raw material decomposition process according to the data;

[0027] 2. The present invention uses a reduced-order model to quickly calculate the production and operation data of the entire cement decomposition furnace, and realizes the real-time display of temperature field, flow field and other data in the cement decomposition furnace, which is convenient for production and operation personnel to make rapid and accurate adjustments based on the real-time simulation data in the furnace. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a system schematic diagram of the present invention. DETAILED DESCRIPTION

[0029] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0030] refer to Figure 1 The digital twin system of cement decomposition furnace based on POD and deep learning described in the present invention includes a data acquisition and processing module 1, a database retrieval module 2, a model reduction prediction module 3, a comprehensive display module 4 and a remote control module 5. The data acquisition module 1 receives instructions from production operation personnel and is connected to the database retrieval module 2. The database retrieval module 2 is connected to the data acquisition and processing module 1 on the one hand, and to the model reduction prediction module 3 and the comprehensive display module 4 on the other hand. The model reduction prediction module 3 is connected to the database retrieval module 2 on the one hand, and is also connected to the comprehensive display module 4 on the other hand. The comprehensive display module is connected to the database retrieval module 2 and the model reduction prediction module 3 on the one hand, and is connected to the remote control module 5 on the other hand.

[0031] The data acquisition and processing module 1 can receive instructions from production operators, and use multiple sensors arranged in the cement decomposition furnace to collect different physical quantities such as temperature, gas flow rate, gas composition, etc. at different positions in the decomposition furnace, and integrate them into a set of characteristic physical quantities and pass them to the database query module for analysis and query.

[0032] The database retrieval module 2 can receive a set of characteristic physical quantities transmitted from the data acquisition module 1, and search the database for matching comprehensive operating condition information. If there is matching comprehensive operating condition information, the comprehensive operating condition information will be transmitted to the comprehensive display module 4 for display; if there is no matching comprehensive operating condition information, the received set of characteristic physical quantities will be transmitted to the model reduction prediction module 3 for calculation. The database retrieval module 2 has two data input ports, one port can receive a set of characteristic physical quantities transmitted from the data acquisition processing module 1; the other port can receive the complete operating condition information in the furnace calculated by the model reduction prediction module 3 for a certain set of characteristic physical quantities. The database retrieval module has three data output ports, two ports are connected to the model reduction prediction module, one port transmits a set of characteristic physical quantities to it to calculate the complete operating condition data in the furnace, and the other port transmits all the complete operating condition data contained in the database to it for calculating the reduced order model; the third port is connected to the comprehensive display module in the furnace, and transmits the existing complete operating condition data in the database to it for display.

[0033] Based on the existing complete operating condition data in the database, the model reduction prediction module 3 extracts a set of optimal canonical orthogonal bases according to the SNAPSHOT algorithm of POD, and uses the deep learning method to find the optimal set of coefficient matrices. Based on this set of orthogonal bases and this set of coefficient matrices, a similar low-order model is obtained. Based on the existing reduced-order model, the model reduction prediction module 3 uses a set of characteristic physical quantities transmitted from the database retrieval module 2 as boundary conditions, and can quickly calculate the corresponding comprehensive operating condition information. The model reduction prediction module 3 has two input ports, one input port can receive a set of characteristic physical quantities transmitted from the database retrieval module 2 for calculation, and the other input port can receive the complete operating condition information transmitted from the database retrieval module 2 to calculate the reduced-order model. The model reduction prediction module 3 has two output ports, one output port is connected to the database retrieval module 2, and the complete operating condition information obtained by this calculation is transmitted to it for storage; the other output port is connected to the comprehensive display module 4 in the furnace, and the complete operating condition information obtained by this calculation is transmitted to it for display.

[0034] The integrated display module 4 can receive the complete working condition information transmitted, and process the complete working condition information according to the requirements of the production operation personnel and display it on the display screen. The integrated display module 4 has two input ports, one input port is connected to the database retrieval module 2, and the other input port is connected to the model reduction prediction module 3 to receive the complete working condition information transmitted from the above two ports. The data display forms of the integrated display module 4 include tables, line graphs, bar graphs, scatter plots, contour maps, profiles, 3D stereograms, dynamic pictures and videos. The interactive methods of the integrated display module 4 include mouse commands, keyboard commands, touch feedback, and voice feedback.

[0035] The remote control module 5 can adjust the operating parameters of the cement decomposition furnace based on the comprehensive display module 4 to control the production process in the decomposition furnace. The operating parameters that can be adjusted by the remote control module include coal intake, raw material quantity, primary air volume, secondary air volume, tertiary air volume, secondary air temperature, and tertiary air temperature.

[0036] When building a digital twin system for a cement decomposition furnace, the following steps are included:

[0037] (1) The production and operation personnel first establish a three-dimensional model of the cement decomposition furnace according to its specific structure and actual size, and divide the model into grids. They select a suitable mathematical model, and collect the operating information of the cement decomposition furnace under multiple normal operating conditions as boundary conditions. They perform CFD numerical simulations on the coal powder combustion, cement raw material decomposition, and gas flow in the cement decomposition furnace under these operating conditions, and store the numerical simulation results under multiple operating conditions in the database retrieval module.

[0038] (2) The model reduction prediction module will then read all the complete operating condition information stored in the database, and use the SNAPSHOT algorithm in the POD to extract a set of optimal canonical orthogonal bases based on these existing multiple complete operating conditions. Then, a deep learning method is used to obtain a set of optimal orthogonal basis coefficients. Based on this set of orthogonal bases and this set of orthogonal basis coefficients, a similar low-order model is obtained and stored in the model reduction prediction module.

[0039] (3) Whenever the number of complete working conditions stored in the database retrieval module changes, the database retrieval module will pass all the complete working conditions it contains to the model reduction prediction module, and re-extract the optimal orthogonal basis and corresponding coefficients based on these complete working conditions to obtain a new reduced-order model for iterative processing of the previously established model.

[0040] (4) After obtaining the reduced-order model, the temperature field and flow field in the cement decomposition furnace can be quickly calculated. First, the database retrieval module will receive a set of characteristic physical quantities transmitted from the data acquisition and processing module, and then this set of characteristic physical quantities will be retrieved in the database. If the database contains complete temperature field and flow field data corresponding to this set of characteristic physical quantities, the data will be directly transmitted to the comprehensive display module for display; if the database does not contain complete operating condition data corresponding to this set of characteristic physical quantities, then this set of characteristic physical quantities will be passed to the model reduction prediction module, and based on the previously established low-order model, the complete operating condition data corresponding to this set of characteristic physical quantities will be quickly calculated. This set of complete operating condition data is transmitted to the comprehensive display module for display on the one hand, and to the database retrieval module on the other hand for storage for the next retrieval.

[0041] In this paper, a cement decomposition furnace with a designed output of 4000t / d is taken as the object, and the digital twin system of the cement decomposition furnace is constructed according to the temperature field and flow field data under different actual outputs.

[0042] Firstly, a three-dimensional model of cement decomposition furnace is established according to its specific structure and actual size, and the model is meshed, and the mesh is a hexagonal structured mesh. According to the drawn computational grid model, the software is used to numerically simulate the coal powder combustion and raw material decomposition process in the cement decomposition furnace. The mathematical models used in this process include: the turbulence model is the standard k-ε model, the particle phase injection is the injection model, the particle phase is the DPM model, the effect of turbulence on the particle phase is the random orbit model, the release of coal powder volatiles adopts the single rate model, the coke combustion adopts the dynamic / diffusion controlled rate reaction model and the P-1 radiation model. According to the actual operation of the cement decomposition furnace, boundary conditions are set, such as parameters such as coal quantity, raw material quantity, primary air volume, secondary air volume, tertiary air volume, secondary air temperature and tertiary air temperature. CFD numerical simulations are performed on the coal powder combustion and raw material decomposition process in the cement decomposition furnace under five operating conditions of 3800t / d, 3900t / d, 4000t / d, 4100t / d and 4200t / d respectively. The complete operating condition data in the cement decomposition furnace under the five operating conditions are stored in the database retrieval module 2.

[0043] Afterwards, the model reduction prediction module will read the five complete operating condition information stored in the database, and take the 4000t / d operating condition as the basic operating condition. Based on these existing complete operating conditions, the SNAPSHOT algorithm in the POD is used to extract a set of optimal canonical orthogonal bases, and then a set of optimal orthogonal basis coefficients is obtained using the deep learning method. Based on this set of orthogonal bases and this set of orthogonal basis coefficients, a similar low-order model is obtained and stored in the model reduction prediction module 3.

[0044] Establish as Figure 1The digital twin system of the cement decomposition furnace shown. When the production operation personnel need to obtain the temperature field and flow field conditions in the current cement decomposition furnace, they issue instructions at the data acquisition and processing module 1. After receiving the instructions sent by the production operation personnel, the data acquisition and processing module 1 uses the sensors arranged in the furnace to collect the temperature, flow rate and gas composition at different positions in the furnace. These physical quantities are then integrated into a set of characteristic physical quantities and passed to the database retrieval module 2. After receiving this set of characteristic physical quantities, the database retrieval module 2 will compare it with the complete operating condition data in its own database. If the characteristic physical quantity matches each of the operating conditions of 3800t / d, 3900t / d, 4000t / d, 4100t / d and 4200t / d in the database, then the complete operating condition data will be transmitted to the comprehensive display module 4 for display.

[0045] If the characteristic physical quantity cannot find matching complete working condition information in the database, then the set of characteristic physical quantities will be transmitted to the model reduction prediction module 3, and the complete working condition data corresponding to the set of characteristic physical quantities will be quickly calculated based on the previously established reduced order model. This set of complete working condition data is transmitted to the comprehensive display module 4 for display on the one hand, and to the database retrieval module on the other hand for storage for the next retrieval. At this time, there are 6 complete working condition information in the database.

[0046] After the integrated display module 4 receives the complete operating condition data transmitted from the database retrieval module 2 or the model reduction prediction module 3, the complete operating condition data will be parsed in the integrated display module 4 and displayed on the interactive device in various ways. After that, the production operation personnel can view the temperature field and flow field data in the current cement decomposition furnace through the interactive device, and can control the production operation of the cement decomposition furnace in the remote control module based on this data.

[0047] Whenever the number of complete working conditions stored in the database changes, for example, the number of complete working condition data increases from 5 to 6, the database retrieval module 2 will pass all the complete working conditions it contains to the model reduction prediction module 3, and re-extract the optimal canonical orthogonal basis and corresponding coefficients based on these complete working conditions to obtain a new reduced-order model for iterative processing of the previously established model.

Claims

1. A digital twin system of cement decomposition furnace based on POD and deep learning, characterized in that: It includes a data acquisition and processing module, a database retrieval module, a model reduction prediction module, a comprehensive display module and a remote control module which are connected in sequence; The data acquisition and processing module can collect data from multiple specific locations in the cement decomposition furnace in real time, including temperature, gas flow rate and gas composition, and process the collected data to form a set of characteristic physical quantities and transmit them to the database retrieval module; The database retrieval module can receive a set of characteristic physical quantities from the data acquisition and processing module, and compare the set of characteristic physical quantities with the corresponding data of the existing complete working condition at the specific position in the database; if there is a working condition that is successfully compared, the complete working condition data is directly transmitted to the comprehensive display module; if the comparison is not successful, the set of characteristic physical quantities is passed to the model reduction prediction module; The model reduction prediction module can receive information transmitted from the database retrieval module, and based on the information, use the POD (proper orthogonal decomposition) method to reduce the order of the established furnace reaction model, wherein the coefficients of the basis vector of the POD are corrected by deep learning; After obtaining the reduced-order model, the relevant data in the current furnace, including the temperature field, velocity field and flue gas composition information in the furnace, are quickly calculated based on the pre-established grid. The calculation results are then transferred to the database retrieval module on the one hand and to the comprehensive display module on the other hand. The comprehensive display module can receive information transmitted by the database retrieval module and the model reduction prediction module, and restore and display it on the computer terminal; the information includes temperature field, velocity field and smoke composition information in the furnace; The remote control module can receive instructions from production operators to control the amount of pulverized coal, the amount of cement raw materials, the primary air volume, the secondary air volume, and the tertiary air volume.

2. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The data acquisition and processing module uses multiple sensors arranged in the cement decomposition furnace to collect different physical quantities at different positions in the cement decomposition furnace, including temperature, gas flow rate, and gas composition, and integrates them into a set of characteristic physical quantities and transmits them to the database query module for analysis and query.

3. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The database retrieval module has two data input ports, one port receives a group of characteristic physical quantities transmitted from the data acquisition and processing module, and the other port receives the complete operating condition data in the furnace calculated by the model reduction prediction module for a certain group of characteristic physical quantities; the database retrieval module has three data output ports, two ports are connected to the model reduction prediction module, one port transmits a group of characteristic physical quantities to it to calculate the complete operating condition data in the furnace, the other port transmits all the complete operating condition data contained in the database to it for calculating the reduced-order model, and the third port is connected to the comprehensive display module in the furnace, and transmits the existing complete operating condition data in the database to it for display.

4. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The model order reduction prediction module has two input ports, one input port receives a set of characteristic physical quantities transmitted from the database retrieval module for calculation, and the other input port receives multiple complete operating condition data transmitted from the database retrieval module for calculating the reduced order model; the model order reduction prediction module has two output ports, one output port is connected to the database retrieval module, and the complete operating condition data obtained by this calculation is transmitted to it for storage, and the other output port is connected to the comprehensive display module in the furnace, and the complete operating condition data obtained by this calculation is transmitted to it for display.

5. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The comprehensive display module has two input ports, one input port is connected to the database retrieval module, and the other input port is connected to the model reduction prediction module. After receiving the complete operating condition data transmitted from the above two ports, the comprehensive display module in the furnace processes and displays the data.

6. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The data display forms of the comprehensive display module include tables, line graphs, bar graphs, scatter plots, contour plots, profile graphs, 3D stereograms, dynamic pictures and videos.

7. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The adjustable operating parameters of the remote control module include coal input amount, raw material amount, primary air volume, secondary air volume, tertiary air volume, secondary air temperature, and tertiary air temperature.

8. According to claim 1, a digital twin system of a cement decomposition furnace based on POD and deep learning is characterized in that: The specific operation steps of the system are: (1) A three-dimensional model of the cement decomposition furnace is established according to the specific structure and actual size of the cement decomposition furnace, and the model is meshed; an appropriate mathematical model is selected, and the operating condition information of the cement decomposition furnace under multiple conditions is collected as boundary conditions, and CFD numerical simulations are performed on the temperature field, flow field, coal powder combustion, and cement raw material decomposition in the cement decomposition furnace under these operating conditions; the numerical simulation results under multiple operating conditions are stored as complete operating condition data in the database retrieval module; (2) Using the SNAPSHOT algorithm in POD, a set of optimal canonical orthogonal bases is extracted based on multiple complete working conditions in the existing database, and then a set of optimal orthogonal basis coefficients is obtained using the deep learning method; based on this set of orthogonal bases and this set of orthogonal basis coefficients, a similar low-order model is obtained and stored in the model reduction prediction module; (3) When it is necessary to reconstruct and predict the current combustion conditions in the furnace and the decomposition conditions of the cement raw materials, the data acquisition and processing module will first use multiple sensors arranged in the furnace to obtain a set of characteristic physical quantities, and then this set of characteristic physical quantities will be transmitted to the database retrieval module; if the database contains a complete working condition corresponding to this set of characteristic physical quantities, the complete working condition data will be directly transmitted to the comprehensive display module for display; if the database does not contain a complete working condition corresponding to this set of characteristic physical quantities, then this set of characteristic physical quantities will be transmitted to the model reduction prediction module, and based on the previously established low-order model, the complete working condition data corresponding to this set of characteristic physical quantities will be quickly calculated; This set of complete working condition data is transmitted to the comprehensive display module in the furnace for display, and is transmitted to the database retrieval module for storage for the next retrieval. (4) Whenever the number of complete working conditions stored in the database retrieval module changes, the database retrieval module will pass all the complete working conditions it contains to the model reduction prediction module, and re-extract the optimal orthogonal basis and corresponding coefficients based on these complete working conditions to obtain a new reduced-order model to iteratively process the previous reduced-order model; (5) When the complete working conditions are displayed on the screen, the production and operation personnel can interact with the displayed content by using a mouse or touching the screen. In response to the problems exposed by the current working conditions, the production and operation personnel can change the corresponding parameters in the remote control module to complete the control of the cement decomposition furnace operating parameters.

9. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a digital twin system of a cement decomposition furnace based on POD and deep learning is implemented as described in any one of claims 1 to 8.

10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, a digital twin system of a cement decomposition furnace based on POD and deep learning is implemented as described in any one of claims 1-8.

Images