Method and system for controlling air volume of air bellow of sintering machine

Abstract

The invention relates to the technical field of sintering machine air bellow air volume control, in particular to a sintering machine air bellow air volume control method and system. The defects existing in a traditional manual control mode are overcome, and intelligent sintering is achieved. The control method comprises the following steps: firstly, performing data acquisition: acquiring historical parameter data and real-time parameter data of temperature, pressure and air volume of each air bellow, the main flue and the main exhaust fan; analyzing the historical parameter data by using a machine learning algorithm, and constructing a correlation model for representing a mapping relationship among the bellows parameters, the flue parameters and the optimal air volume interval; inputting the real-time parameter data into the correlation model to obtain a real-time optimal air volume interval of each air bellow, and adjusting the air volume of each air bellow and the air volume and the air pressure of the main flue and the main exhaust fan according to the real-time optimal air volume interval; and finally, continuously iteratively updating the correlation model based on the real-time parameter data, the control effect and the sintering waste gas component analysis data. The control system is used for implementing the control method. And the system is high in adaptability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sintering machine wind box air volume control, in particular to a sintering machine wind box air volume control method and system. BACKGROUND

[0002] Sintering is a key process in steel production, its essence is "suction sintering", that is, through the main suction fan to form negative pressure under the wind box group of the sintering machine trolley, to assist the combustion of the mixed material, complete the physical and chemical reaction of the material, and finally form sinter. In this process, "wind" is the core element, and the reasonable distribution and control of air volume directly determines the yield, quality and energy consumption of sinter.

[0003] In traditional sintering production, the control of wind box air volume mainly depends on the experience of operators. The operator observes the temperature and negative pressure instrument display of each wind box, and manually adjusts the flap or valve opening degree on each wind box branch pipe, in order to try to balance the air volume distribution of the whole sintering section. This manual control method has significant inherent defects:

[0004] Hysteresis: It takes time for the operator to find the temperature or negative pressure anomaly and make adjustments, and it is impossible to respond immediately to changes in working conditions.

[0005] Instability: The control effect is heavily dependent on personal experience, and there are differences in judgment and operation between different operators, making it difficult to achieve stable and consistent precise control.

[0006] Local: Manual adjustment often only focuses on the parameters of a single wind box, and lacks overall and related consideration of the whole suction system (including the main flue and main suction fan), which easily leads to low system energy efficiency.

[0007] Failure risk: When the raw material ratio and working conditions change frequently, manual experience is difficult to adapt quickly, which easily leads to uneven sintering process, "over-burning" or "under-burning", and affects product quality.

[0008] Therefore, the existing technology urgently needs a technical solution that can overcome the defects of manual control, and realize automatic, precise and intelligent adjustment of sintering machine wind box air volume. SUMMARY

[0009] The purpose of the present application is to provide a sintering machine wind box air volume control method and system to solve the shortcomings of traditional manual control method, and to overcome the instability, hysteresis and failure risk of manual adjustment depending on temperature trends, and finally realize intelligent control of the sintering process.

[0010] To achieve the above purpose, the technical solution adopted by the present application is:

[0011] The first aspect of the present invention provides a method for controlling the air volume of a sintering machine air box, comprising the following steps:

[0012] S1 Data Acquisition: Acquire historical and real-time parameter data of temperature, pressure and air volume of each air box, main flue and main exhaust fan;

[0013] S2 Modeling and Analysis: The historical parameter data is analyzed using machine learning algorithms to construct an association model that characterizes the mapping relationship between the bellows parameters, flue parameters and the optimal air volume range;

[0014] S3 Real-time Control: Input the real-time parameter data into the correlation model to obtain the real-time optimal air volume range of each air box, and adjust the air volume of each air box, as well as the air volume and air pressure of the main flue and the main exhaust fan accordingly.

[0015] S4 Dynamic Optimization: Based on the real-time parameter data, control effect, and sintering exhaust gas composition analysis data, the correlation model is continuously iterated and updated.

[0016] Furthermore, in step S1, the historical parameter data covers different raw material ratios and operating conditions, and the data collection time span is not less than 12 months.

[0017] Further, in step S2, the step of constructing the association model includes:

[0018] The standardized temperature, pressure, and air volume of all bellows at a given moment are integrated with the main flue pressure and the main exhaust fan air volume into an input feature vector.

[0019] The historical parameter data is used to train a machine learning model to establish a mapping relationship between the input feature vector and the optimal airflow range of each bellows.

[0020] Furthermore, the input feature vector is represented as:

[0021]

[0022] Where Tn, Pn, and Qn represent the standardized temperature, pressure, and air volume of the nth air box, respectively, Py represents the main flue pressure, Qy represents the main exhaust fan air volume, and n is the total number of air boxes.

[0023] Furthermore, the optimal airflow range for each bellows output in the correlation model is defined by the following formula:

[0024]

[0025] Where i represents the bellows number. μ represents the standardized minimum / maximum value of the model output. Qi This represents the average airflow of bellows i under historical normal operating conditions. This represents the standard deviation of the historical air volume of bellows i.

[0026] Furthermore, the machine learning model is a hybrid model, which includes an LSTM neural network unit for handling time series dependencies and a random forest model for performing regression analysis.

[0027] Further, in step S4, the dynamic optimization step includes updating the control parameters by solving for the objective function using an optimization algorithm, wherein the objective function is:

[0028] minΣ|Qi-Qi^target|+λ*ΔPm

[0029] Where Qi is the actual air volume of the i-th air box, Qi^target is its target air volume, ΔPm is the main flue pressure change, and λ is the weighting coefficient.

[0030] A second aspect of the present invention provides a sintering machine air box airflow control system for implementing the sintering machine air box airflow control method described above, comprising:

[0031] Data acquisition module: configured in each air box, main flue and main exhaust fan, configured to collect temperature, pressure and air volume parameters;

[0032] The dynamic modeling module is communicatively connected to the data acquisition module and is configured to perform machine learning algorithm training to build the association model based on historical data.

[0033] The optimization decision module is connected to the dynamic modeling module and is configured to solve the objective function and generate control commands based on the correlation model and real-time data using an optimization algorithm.

[0034] The execution control module is connected to the optimization decision module and is configured to drive the regulating valves of each air box and the frequency converter of the main exhaust fan according to the control command.

[0035] Feedback calibration module: connected to the dynamic modeling module, configured to receive sintering exhaust gas composition analysis data and correct the model weights of the associated model accordingly.

[0036] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art:

[0037] This invention discloses a method and system for controlling the air volume of a sintering machine's air box. By using a machine learning model, a mapping relationship between air volume distribution and system state is constructed, thereby achieving precise and intelligent adjustment of the air volume of the air box.

[0038] By introducing a feedback calibration mechanism based on the composition of sintering exhaust gas, the system is equipped with online learning and self-adaptation capabilities. The system can adapt to different raw materials and operating conditions, solving the problems of lag and failure under manual control.

[0039] By independently generating a dynamic optimal airflow range for each bellows and using optimization algorithms to seek the optimal control commands, over-burning or under-burning is effectively prevented, thus improving the quality of sintered ore.

[0040] This invention promotes the increase in the output and quality of sintered ore by achieving "homogeneous sintering". At the same time, it reduces the energy consumption of the main exhaust fan by optimizing the air volume distribution, reduces maintenance costs, and brings significant economic benefits. Attached Figure Description

[0041] The following sections will describe some specific embodiments of the invention in detail by way of example and not limitation, with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar parts or portions. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings:

[0042] Figure 1 This is a flowchart of the sintering machine air box air volume control method provided in this invention;

[0043] Figure 2 This is a structural diagram of the sintering machine air box air volume control system provided in this invention. Detailed Implementation

[0044] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] This example discloses a method and system for controlling the air volume of a sintering machine's air box, aiming to solve the instability, lag, and failure of traditional manual temperature and trend control and adjustment in the sintering process, and to achieve intelligent sintering.

[0046] Specifically, see Figure 1 The construction of this control method includes the following steps:

[0047] S1 Data Acquisition: Acquire historical and real-time parameter data of temperature, pressure and air volume of each air box, main flue and main exhaust fan during continuous production process;

[0048] S2 Modeling and Analysis: Machine learning algorithms are used to analyze the historical parameter data to construct an association model that represents the mapping relationship between the bellows parameters, flue parameters and the optimal air volume range;

[0049] S3 Real-time Control: Input the above real-time parameter data into the correlation model to obtain the real-time optimal air volume range of each air box. Based on this range, the system automatically adjusts the opening of the regulating valve of each air box to match its precise air volume requirements, and simultaneously and collaboratively optimizes the air volume and air pressure of the main flue and the main exhaust fan, thereby achieving global stability and balanced distribution of the air volume of the entire exhaust system.

[0050] S4 Dynamic Optimization: Based on real-time parameter data, control effects, and sintering exhaust gas composition analysis data, the correlation model is continuously iterated and updated.

[0051] In step S1, the collected historical parameter data must cover a variety of typical operating conditions under different raw material ratios, and the data collection time span should be no less than 12 months in order to fully capture the impact of the production cycle.

[0052] In step S2, the steps for constructing the correlation model include: integrating the standardized temperature, pressure, and air volume of all wind boxes at a certain moment with the main flue pressure and the main exhaust fan air volume into an input feature vector; and then using historical parameter data to train a machine learning model to establish a mapping relationship between the input feature vector and the optimal air volume range of each wind box.

[0053] The above input feature vector is represented as:

[0054] Where Tn, Pn, and Qn represent the standardized temperature, pressure, and air volume of the nth air box, respectively, Py represents the main flue pressure, Qy represents the main exhaust fan air volume, and n is the total number of air boxes.

[0055] The optimal airflow range for each bellows output in the above correlation model is defined by the following formula:

[0056] Where i represents the bellows number. This represents the standardized minimum / maximum value of the model output. This represents the average airflow of bellows i under historical normal operating conditions. This represents the standard deviation of the historical air volume of bellows i.

[0057] In this example, the machine learning model is a hybrid model, which includes LSTM neural network units for handling time series dependencies and a random forest model for performing regression analysis. The hybrid architecture of the model balances the accuracy of time series prediction with the stability of static regression.

[0058] In step S4, the dynamic optimization step includes updating the control parameters by solving the objective function using an optimization algorithm. The objective function is:

[0059] minΣ|Qi-Qi^target|+λ*ΔPm

[0060] Where Qi is the actual air volume of the i-th air box, Qi^target is its target air volume, ΔPm is the main flue pressure change, and λ is the weighting coefficient.

[0061] This example also provides a sintering machine air box airflow control system to implement the above-mentioned sintering machine air box airflow control method. (See also...) Figure 2 The system includes:

[0062] Data acquisition module: configured in each air box, main flue and main exhaust fan, configured to collect temperature, pressure and air volume parameters;

[0063] The dynamic modeling module is communicatively connected to the data acquisition module and is configured to perform machine learning algorithm training to build the association model based on historical data.

[0064] The optimization decision module is connected to the dynamic modeling module and is configured to solve the objective function and generate control commands based on the correlation model and real-time data using an optimization algorithm.

[0065] The execution control module is connected to the optimization decision module and is configured to drive the regulating valves of each air box and the frequency converter of the main exhaust fan according to the control command.

[0066] Feedback calibration module: connected to the dynamic modeling module, configured to receive sintering exhaust gas composition analysis data and correct the model weights of the associated model accordingly.

[0067] In summary, the beneficial effects of the present invention are as follows:

[0068] (1) By using data-driven machine learning models, the reliance on human experience is replaced, which solves the lag and instability of traditional control methods and responds to changes in working conditions in real time.

[0069] (2) By dynamically generating an independent optimal air volume range for each wind box and performing global collaborative optimization, the fine and uniform distribution of air volume in the sintering section is realized, which effectively improves the yield and quality consistency of sintered ore.

[0070] (3) A feedback calibration mechanism based on the composition of sintering exhaust gas was introduced, which enables the system to optimize the control model in reverse according to the final production results (product quality), forming an intelligent closed loop and improving the system's adaptability.

[0071] (4) While ensuring sintering quality, the power consumption of the main exhaust fan was reduced by optimizing the air volume distribution globally, thus achieving a green and low-carbon production process.

[0072] The above embodiments are only for illustrating the technical concept and features of the present invention. Their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be used to limit the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method of controlling the volume of air in a sinter machine wind box, characterized in that, Comprising the following steps: S1 data acquisition: obtain the historical parameter data and real-time parameter data of the temperature, pressure and air volume of each air box, main flue and main exhaust fan; S2 modeling analysis: analyze the historical parameter data using a machine learning algorithm to construct a correlation model representing the mapping relationship between air box parameters, flue parameters and the optimal air volume interval; S3 real-time control: input the real-time parameter data into the correlation model to obtain the real-time optimal air volume interval of each air box, and adjust the air volume of each air box, as well as the air volume and air pressure of the main flue and main exhaust fan accordingly; S4 dynamic optimization: based on the real-time parameter data, control effect and sintering waste gas composition analysis data, continuously iteratively update the correlation model.

2. The method of controlling the volume of air in the wind boxes of a sintering machine as claimed in claim 1, characterized in that, In step S1, the historical parameter data covers different raw material ratio conditions, and the data collection time span is not less than 12 months.

3. The method of controlling the volume of air in the wind boxes of a sintering machine as claimed in claim 1, characterized in that, In step S2, the steps of constructing the correlation model include: Integrate the standardized temperature, pressure and air volume of all air boxes at a certain moment, and the main flue pressure and main exhaust fan air volume into an input feature vector; Train a machine learning model using the historical parameter data to establish the mapping relationship between the input feature vector and the optimal air volume interval of each air box.

4. The method of controlling the volume of air in the wind boxes of a sintering machine as claimed in claim 3, characterized in that, The input feature vector is represented as: Where Tn, Pn and Qn represent the standardized temperature, pressure and air volume of the nth air box, Py represents the main flue pressure, Qy represents the main exhaust fan air volume, and n is the total number of air boxes.

5. The method according to claim 3 or 4, characterized in that The optimal air volume interval output by the correlation model for each air box is defined by the following formula: where i denotes the windbox number, denotes the normalized minimum / maximum of the model output, denotes the average air flow of windbox i in historical normal operation, denotes the standard deviation of the historical air flow of windbox i.

6. The method of controlling the volume of air in the wind boxes of a sintering machine as claimed in claim 3, characterized in that, The machine learning model is a hybrid model that includes an LSTM neural network unit for handling time series dependencies and a random forest model for regression analysis.

7. The method of controlling the volume of air in the wind boxes of a sintering machine as claimed in claim 1, characterized in that, In step S4, the dynamic optimization step includes updating the control parameters by solving the objective function using an optimization algorithm, where the objective function is: minΣ|Qi-Qi^target|+λ*ΔPm Where Qi is the actual air volume of the ith air box, Qi^target is its target air volume, ΔPm is the main flue pressure change, and λ is the weight coefficient.

8. A sintering machine plenum air volume control system for implementing the sintering machine plenum air volume control method according to any one of claims 1 to 7, characterized by Comprising: Data acquisition module: configured in each air box, main flue and main exhaust fan, configured to collect temperature, pressure and air volume parameters; Dynamic modeling module, in communication connection with the data acquisition module, configured to execute machine learning algorithm training and construct the correlation model based on historical data; Optimization decision module, connected with the dynamic modeling module, configured to solve the objective function based on the correlation model and real-time data through an optimization algorithm to generate control instructions; Execution control module, connected with the optimization decision module, configured to drive each air box regulating valve and main exhaust fan frequency converter to act according to the control instructions; Feedback calibration module: connected with the dynamic modeling module, configured to receive sintering waste gas composition analysis data and correct the model weight of the correlation model accordingly.

Images