CN113564348B - A sintering production method based on machine vision and data-driven - Google Patents

Abstract

The invention discloses a sintering production method based on machine vision and data driving, which comprises the following steps: sampling various sintering raw materials, performing quality detection, recording the proportion of the sintering raw materials of different types, and uploading the proportion to a sintering expert database in real time; installing an image acquisition device at the tail of the sintering machine, detecting the acquired image by an infrared thermal imager to obtain the temperature of the material layer section at the tail of the sintering machine, the thickness of a scarlet layer and the uniformity index of the scarlet layer, and uploading the indexes in real time; installing a hyperspectral imager at the tail of the sintering machine, detecting the chemical components of finished sintered ores at the tail of the sintering machine and the uniformity of the components of the material layer section at the tail, and uploading the chemical components and the uniformity in real time; installing a high-definition camera at the outlet of the sintering cooler, detecting the granularity change of the finished sintered ore, and uploading the granularity change in real time; and the sintering expert database performs data mining and big data analysis based on various uploaded data, constructs an intelligent control system for the sintering production process, and regulates and controls the proportion of sintering raw materials so as to realize closed-loop control of the sintering process.

Description

A sintering production method based on machine vision and data-driven

technical field

The invention relates to the technical field of iron and steel metallurgy, in particular to a sintering production method based on machine vision and data driving.

Background technique

In 2019, my country's blast furnace pig iron output was 770 million tons, accounting for 62.2% of the world's pig iron output (1.24 billion tons). High basicity sinter with acidic pellets is the current charge structure in my country. In 2019, sinter accounted for 78% of the national blast furnace charge structure, and pellets accounted for about 13% (120 million tons). The production of high-quality sinter is the basis for high-quality and high-yield blast furnace smelting. However, the sintering quality parameters are complex and diverse, so their control often depends on the observation of the sintering machine tail and cross-section, and corresponding control operations are carried out according to experience. Today, with the vigorous development of artificial intelligence and computer technology, the detection technology with image processing as the core provides an advanced and accurate detection method for the inspection of the sintering machine tail section.

At present, most factories are equipped with industrial TVs to monitor the cross-section status, and in recent years, image analysis technology has been used to quantify them. According to the survey, Japan Kawasaki Steel Company uses an image analyzer to divide the image into 5 parts according to the width of the trolley, and calculates the area of various red and high temperature zones of the sintered material layer once per second. After using the image analyzer, the output and quality of sintered ore have been significantly improved. my country is also researching and improving this technology. Recently, Laiwu Iron and Steel Group has carried out equipment transformation for the 400m2 sintering machine tail thermal imager because the existing equipment cannot meet the functions of providing data support and judgment and prediction for sintering operators. The sintering machine tail thermal imager equipment can realize the online prediction of FeO in sintering ore, and at the same time detect the vertical sintering speed, end point analysis and temperature field distribution of the sintering machine tail section, and use the data to intuitively reflect the changes of sintering ore parameters. It not only improves the overall production quality and stability of sinter, but also provides more data analysis for technicians. Although many scientific research institutions and iron and steel enterprises have done a lot of work and applications on the sintering machine tail section, there are still many problems: (1) the FeO content in the sinter is not accurately predicted; (2) the identification accuracy of the section system is low, and it is difficult to make To the self-learning system, the algorithm needs to be further improved.

Hyperspectral imaging technology is an image data technology based on a lot of narrow waves developed in the past two decades. Its definition is that on the basis of multispectral imaging, in the spectral range from ultraviolet to infrared, the imaging spectrometer is used to continuously image the target object in dozens or hundreds of spectral bands within the spectral coverage. While obtaining the spatial characteristic imaging of the object, the spectral information of the measured object is also obtained. Because the reflection spectrum of an object has a "fingerprint" effect, different objects have different spectra, and the principle that the same object must have the same spectrum can distinguish different material information. At present, hyperspectral imaging technology is mainly used in agricultural food, medicine, aerospace, geological exploration and other fields, but it is rarely used in the field of metallurgy. This technology can be used with a sintering machine tail thermal imager, which can not only observe the vertical sintering speed, end point analysis, temperature field distribution and other indicators of the machine tail section in real time, but also online detect the chemical composition of the sintering section and the degree of sintering uniformity, reflecting more The change of sinter parameter index enables sinter technicians to better control the sintering process.

In addition, physical properties such as strength and particle size of sinter are also important indicators in quality inspection. The early detection of sintered products was mostly manual sampling. This method is not only not suitable for the current sintering production rhythm, the sampling cycle is long, which affects the timeliness of inspection, and the manual sampling environment is harsh, dusty, noisy, labor-intensive and safe. The hidden danger is great and affects the health of workers. With the large-scale and modernization of domestic and foreign iron and steel enterprises, the production process has become more and more high-speed and automated, and higher requirements have been placed on the accuracy, scientificity and timeliness of sinter quality inspection. It is imperative to replace manual quality inspection.

Although a lot of work has been done on sintering machine tail section imaging, particle size detection, chemical composition detection, sintering big data, etc., the selection of corresponding technical algorithms, the matching between technologies, and the closed-loop control system for the sintering production process have not yet been formed. Therefore, it is necessary to study a method based on the above equipment technology combined with big data analysis drive to realize intelligent control of sintering production process, provide sintering workers with important sintering data parameters and results, and improve the stability of sintering production and the quality of sintered minerals.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a sintering production method based on machine vision and data driving. The main problem to be solved is to develop intelligent and unmanned sintering, and to obtain more basic parameters affecting sintering behavior in the sintering production process in real time, especially It is a parameter that cannot be measured in real time by manual operation. It can realize online detection of finished ore quality and real-time optimization and adjustment of sintering process parameters, reduce the influence of human operation on the sintering process during sintering, and improve the quality of sintered ore and also improve the quality of sintering staff. working environment.

In order to solve the above-mentioned technical problems, the embodiments of the present invention provide the following solutions:

A sintering production method based on machine vision and data-driven, comprising the following steps:

Step 1: Sampling various sintering raw materials for quality inspection, recording the proportion of different sintering raw materials, and uploading the result data to the sintering expert database in real time;

Step 2: Install an image acquisition device and an infrared thermal imager at the tail of the sintering machine, and the collected images are detected by the infrared thermal imager to obtain the temperature of the material layer section at the tail of the sintering machine, the thickness of the red layer, and the uniformity index of the red layer. Upload the result data to the sintering expert database in real time;

Step 3. Install a hyperspectral imager at the tail of the sintering machine to detect the chemical composition of the finished sintered ore at the tail of the sintering machine and the uniformity of the cross-sectional composition of the material layer at the tail, and upload the result data to the sintering expert database in real time;

Step 4. Install a high-definition camera at the outlet of the sintering cooler to detect the particle size change of the finished sinter, and upload the result data to the sintering expert database in real time;

Step 5: The sintering expert database conducts data mining and big data analysis based on the received result data, constructs an intelligent control system for the sintering production process, and adjusts the proportion of sintering raw materials to realize closed-loop control of the sintering process.

Preferably, the step one specifically includes:

Manually check and match the sintering raw materials, and manually input the data into the sintering expert database;

An automatic sampling device is installed under the distributing roller of the sintering machine to sample the sintering mixture from time to time, check the quality, particle size and proportion data, and modify the raw material data in the sintering expert database in real time.

Preferably, the image acquisition device used in the second step is an infrared detector.

Preferably, in the second step, a water-cooling protection device is arranged around the infrared thermal imager, and the infrared thermal imager converts the electronic video signal measured by the infrared detector into a two-dimensional plane imaging infrared system. The infrared thermal image is displayed on the monitor for real-time manual viewing, and intelligently identifies the temperature of the material layer section at the tail of the sintering machine, the thickness of the red layer, and the uniformity index of the red layer, and uploads the result data to the sintering expert database in real time.

Preferably, in the third step, the hyperspectral imager is installed around the infrared thermal imager, a water cooling protection device is arranged around the hyperspectral imager, and the hyperspectral imager is real-time based on a machine vision system Reconnaissance the spectral signal of the material layer section at the tail of the sintering machine, converts the spectral signal into chemical composition according to the spectral library, and uploads the chemical composition and distribution data of the material layer section at the rear of the sintering machine to the sintering expert database in real time.

Preferably, in the step 4, detecting the particle size change of the finished sintered ore specifically includes:

A high-definition camera is installed at the outlet of the sintering cooler, and the captured images are sent to the sintering control room in real time. At the same time, a group of high-definition photos are taken every 10 seconds, and the particle size composition of the sintered ore is calculated based on the machine vision system, and the result data is uploaded to the sintering ore in real time. Expert database.

Preferably, in the step 5, the working process of the constructed sintering production process intelligent control system includes:

Collect various result data obtained in real time, carry out data mining and big data analysis, calculate the solution in real time when there is a problem, and upload it to the sintering control room, and modify the type and ratio of sintering raw materials based on the solution.

Preferably, the way of modification includes manual modification and automatic modification by the system.

The beneficial effects brought by the technical solutions provided by the embodiments of the present invention include at least:

In the embodiment of the present invention, based on the image recognition technology and big data mining technology, the thermal imaging data of the sintering machine tail section is obtained in real time through the infrared thermal imager of the sintering machine tail, and the indicators such as temperature, thickness of the red layer, and uniformity of the red layer are obtained; The hyperspectral online detection technology obtains the chemical composition of the sinter in real time; the particle size change of the finished sinter is obtained in real time based on the high-definition camera; finally, the detection results and the quality of the sintering raw materials are uploaded to the sintering expert database, and based on the big data mining and analysis, the sintering production process is constructed. The intelligent control system realizes the online detection of sinter particle size and chemical composition and the real-time optimization and adjustment of process parameters, thereby improving the quality of sintered ore, reducing the influence of manual operation in the sintering process, and finally realizing the unmanned and intelligent sintering process.

In addition, the method of the present invention greatly reduces the labor intensity of sintering workers, improves the working environment, and the modification of sintering ore blending and the detection of sintering quality become more accurate and objective due to the reduction of human factors, and the detection data can be queried in real time, which is more conducive to Direct production operations.

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present invention more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained from these drawings without creative effort.

1 is a flowchart of a machine vision-based and data-driven sintering production method provided by an embodiment of the present invention;

FIG. 2 is a schematic diagram of the installation position of the equipment in the sintering process provided by the embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

An embodiment of the present invention provides a machine vision-based and data-driven sintering production method, as shown in FIG. 1 , the method includes the following steps:

Step 1: Take samples of various sintering raw materials for quality inspection, record the proportion of different types of sintering raw materials, and upload the result data to the sintering expert database in real time.

This step specifically includes: manually detecting and proportioning the sintering raw materials, and manually inputting the data into the sintering expert database; installing an automatic sampling device under the distributing roller of the sintering machine, sampling the sintering mixture from time to time, and testing the quality, particle size, Proportion data, modify the raw material data in the sintering expert database in real time.

Step 2: Install an image acquisition device and an infrared thermal imager at the tail of the sintering machine, and the collected images are detected by the infrared thermal imager to obtain the temperature of the material layer section at the tail of the sintering machine, the thickness of the red layer, and the uniformity index of the red layer. Upload the result data to the sintering expert database in real time.

In this step, the image acquisition device used is an infrared detector, and a water cooling protection device is arranged around the infrared thermal imager to reduce the influence of high temperature on the instrument. The infrared thermal imager converts the electronic video signal measured by the infrared detector into an infrared thermal image through a two-dimensional plane imaging infrared system, which is displayed on a display for manual real-time viewing, and intelligently identifies the temperature and temperature of the material layer section at the tail of the sintering machine. The thickness of the red layer, the uniformity of the red layer and other indicators, the result data will be uploaded to the sintering expert database in real time.

Step 3: Install a hyperspectral imager at the tail of the sintering machine to detect the chemical composition of the finished sintered ore at the tail of the sintering machine and the uniformity of the cross-sectional composition of the material layer at the tail, and upload the result data to the sintering expert database in real time.

In this step, the hyperspectral imager is installed around the infrared thermal imager, and there is also a water cooling protection device around the hyperspectral imager. The hyperspectral imager is based on the machine vision system to detect the spectral signal of the section of the material layer at the tail of the sintering machine in real time. The spectral library converts the spectral signal into chemical composition, and uploads the chemical composition and distribution data of the material layer section at the tail of the sintering machine to the sintering expert database in real time.

Step 4: Install a high-definition camera at the outlet of the sintering cooler to detect the particle size change of the finished sinter, and upload the result data to the sintering expert database in real time.

Among them, the detection of particle size changes of finished sintered ore specifically includes:

A high-definition camera is installed at the outlet of the sintering cooler, and the captured images are sent to the sintering control room in real time. At the same time, a group of high-definition photos are taken every 10 seconds, and the particle size composition of the sintered ore is calculated based on the machine vision system, and the result data is uploaded to the sintering ore in real time. Expert database.

Step 5: The sintering expert database conducts data mining and big data analysis based on the received result data, constructs an intelligent control system for the sintering production process, and adjusts the proportion of sintering raw materials to realize closed-loop control of the sintering process.

In this step, the working process of the constructed sintering production process intelligent control system includes:

Collect various result data obtained in real time, carry out data mining and big data analysis, calculate the solution in real time when there is a problem, and upload it to the sintering control room, and modify the type and ratio of sintering raw materials based on the solution. The modification methods include manual modification and automatic modification by the system.

In a specific embodiment, three kinds of mineral powders (A, B, C), limestone, and anthracite are used as sintering raw materials, and testing equipment is installed on the sintering production line of a steel enterprise to carry out demonstration sintering experiments. For example, as shown in Table 1 and Table 2, the sintering ratio scheme simulates the actual industrial sintering production ratio of the sintering plant.

The installation location of the equipment in the sintering process is shown in Figure 2. An automatic sampling device is installed under the sintering cloth roller, an infrared thermal imager and a hyperspectral imager are installed at the tail of the sintering machine, and a high-definition camera is installed at the exit of the sintering cooler. All testing instruments are connected to a computer equipped with a sintering expert database, and the computer is unified for testing and control.

Table 1 Chemical composition table of experimental raw materials for sintering

raw material name TFe SiO₂ CaO MgO Al₂O₃ A mineral powder 66.50 5.82 0.44 0.45 0.46 B mineral powder 63.63 3.78 0.07 0.05 2.06 C mineral powder 60.25 6.27 0.12 0.06 1.35 limestone 2.1 50.35 4.22

Table 2 Name and ratio of sintering raw materials

raw material name A mineral powder B mineral powder C mineral powder limestone anthracite Proportion 27.4 34.5 16.5 16.0 5.6

1. Experimental process

Before the sintering production practice, for the quality inspection of sintering raw materials, the chemical composition and particle size composition of three kinds of mineral powder, limestone and anthracite were tested to simulate the actual industrial sintering production in a sintering plant. All data are entered into the sintering expert database. In addition, the sintering production data of the sintering plant in the past three months was obtained, and the data was uploaded to the sintering expert database to build a simple sintering big data computing platform.

After starting the sintering production practice, the infrared thermal imager at the tail of the sintering machine is used to detect the temperature of the section of the tail material layer of the sintering machine, the thickness of the red layer, the uniformity of the red layer and other indicators; the hyperspectral imager is used to detect the sintering of the finished sintering machine tail. The chemical composition of the ore and the uniformity of the cross-sectional composition of the tail material layer; the high-definition camera is installed at the outlet of the sintering cooler to detect the particle size change of the finished sintered ore; the above detection data are collected and input into the sintering expert database in real time to provide the computing platform. data and calculation parameters.

When the big data computing platform calculates the best results, the proportioning data and prediction results are displayed on the computer display interface, and the sintering staff can choose to change the proportioning automatically or manually.

2. Experimental results and analysis

Since the actual sintering production in the sintering plant uses limestone and quicklime as the flux, and coke powder and anthracite as the fuel, and only limestone as the flux and anthracite as the fuel for the demonstration sintering production, the particle size composition and drum strength of the sintered ore produced by sintering are consistent with the actual sintering production. The quality of finished ore produced by sintering varies greatly.

According to the data transmitted by the high-definition camera at the tail of the sinter cooler, the proportion of sinter with a particle size of less than 10mm is 41.3%, and the particle size composition is poor. Three groups of samples of the finished ore were tested for drum strength, and the average drum strength was 70.6%. It was detected by infrared thermal imager that the uniformity of the fiery red layer at the tail section of the sintering machine was poor, the sintered ore at the top of the material layer was severely overburned, and the sintered ore at the bottom of the material layer had a tendency of light burning. It was detected by the hyperspectral imager at the tail of the sintering machine that the chemical composition and uniformity of the middle part of the sintering machine tail section were good, while the chemical composition and uniformity of the top and bottom sections of the sintering machine were poor.

After uploading the recent data to the sintering expert database, and using the self-built big data computing platform to analyze, a new sintering raw material ratio scheme is obtained. The revised scheme is shown in Table 3, which reduces the ratio of low-silicon B mineral powder. In order to increase the overall SiO2 content, at the same time increase the limestone content to correct the alkalinity, and moderately reduce the raw material ratio to reduce the effect of overburning of the sinter. The sintering expert database sends the modification plan to the display screen for the sintering staff to view, and automatically changes the sintering raw material ratio after approval, and at the same time reminds the sintering staff to adjust the batching structure.

Table 3 Name and ratio of sintering raw materials after modification

raw material name A mineral powder B mineral powder C mineral powder limestone anthracite Proportion 28.6 32.4 17.3 16.9 4.8

After the big data computing platform modified the ratio of sintering raw materials and the sintering staff adjusted the distribution rollers to change the structure of the sintering raw materials, the demonstration sintering experiment was carried out again. The sintering results are as follows: the data from the high-definition camera at the rear of the sintering cooler shows that the proportion of sintered ore with a particle size of less than 10 mm is 50.5%, and the particle size composition is significantly improved; It is 75.6%; it is detected by the infrared thermal imager that the uniformity of the fire red layer at the tail section of the sintering machine has also been significantly improved, the overburning tendency of the sintered ore at the top of the material layer is reduced, and the sintering ore at the bottom has no tendency to lightly burn; through the hyperspectral imager at the tail of the sintering machine It was detected that the chemical composition and uniformity of the middle and bottom of the sintering machine tail section were better, and the chemical composition and uniformity of the sintering top were also significantly improved.

To sum up, based on the image recognition technology and big data mining technology, the present invention obtains the thermal imaging data of the sintering machine tail section in real time through the infrared thermal imager of the sintering machine tail, and obtains indicators such as temperature, thickness of the red layer, and uniformity of the red layer; The chemical composition of sintered ore is obtained in real time through hyperspectral online detection technology; the particle size change of finished sintered ore is obtained in real time based on high-definition cameras; finally, the detection results and the quality of sintering raw materials are uploaded to the sintering expert database, and the sintering production process is constructed based on big data mining and analysis The intelligent control system realizes the online detection of sinter particle size and chemical composition and the real-time optimization and adjustment of process parameters, thereby improving the quality of sintered ore, reducing the influence of human operation in the sintering process, and finally realizing the unmanned and intelligent sintering process.

In addition, the method of the present invention greatly reduces the labor intensity of sintering workers, improves the working environment, and the modification of sintering ore blending and the detection of sintering quality become more accurate and objective due to the reduction of human factors, and the detection data can be queried in real time, which is more conducive to Direct production operations.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection of the present invention. within the range.

Claims (1)

1. a sintering production method based on machine vision and data drive, is characterized in that, comprises the following steps: Step 1: Sampling various sintering raw materials for quality inspection, recording the proportion of different sintering raw materials, and uploading the result data to the sintering expert database in real time; The first step specifically includes: Manually check and match the sintering raw materials, and manually input the data into the sintering expert database; An automatic sampling device is installed under the distributing roller of the sintering machine to sample the sintering mixture from time to time, check the quality, particle size and proportion data, and modify the raw material data in the sintering expert database in real time; Step 2: Install an image acquisition device and an infrared thermal imager at the tail of the sintering machine, and the collected images are detected by the infrared thermal imager to obtain the temperature of the material layer section at the tail of the sintering machine, the thickness of the red layer, and the uniformity index of the red layer. Upload the result data to the sintering expert database in real time; The image acquisition device used in the second step is an infrared detector; In the second step, a water cooling protection device is arranged around the infrared thermal imager, and the infrared thermal imager converts the electronic video signal measured by the infrared detector into an infrared thermal image through a two-dimensional plane imaging infrared system. , display it on the display for manual real-time viewing, and intelligently identify the temperature of the material layer section at the tail of the sintering machine, the thickness of the red layer, and the uniformity index of the red layer, and upload the result data to the sintering expert database in real time; Step 3. Install a hyperspectral imager at the tail of the sintering machine to detect the chemical composition of the finished sintered ore at the tail of the sintering machine and the uniformity of the cross-sectional composition of the material layer at the tail, and upload the result data to the sintering expert database in real time; In the third step, the hyperspectral imager is installed around the infrared thermal imager, a water cooling protection device is arranged around the hyperspectral imager, and the hyperspectral imager is based on the machine vision system to detect the sintering machine in real time. The spectral signal of the tail section of the material layer is converted into chemical composition according to the spectral library, and the chemical composition and distribution data of the section of the tail material layer of the sintering machine are uploaded to the sintering expert database in real time; Step 4. Install a high-definition camera at the outlet of the sintering cooler to detect the particle size change of the finished sinter, and upload the result data to the sintering expert database in real time; In the step 4, detecting the particle size change of the finished sintered ore specifically includes: A high-definition camera is installed at the outlet of the sintering cooler, and the captured images are sent to the sintering control room in real time. At the same time, a group of high-definition photos are taken every 10 seconds, and the particle size composition of the sintered ore is calculated based on the machine vision system, and the result data is uploaded to the sintering ore in real time. expert database; Step 5: The sintering expert database conducts data mining and big data analysis based on the received result data, constructs an intelligent control system for the sintering production process, and regulates the proportion of sintering raw materials, so as to realize the closed-loop control of the sintering process; In the fifth step, the working process of the constructed sintering production process intelligent control system includes: Collect the result data obtained in real time, carry out data mining and big data analysis, calculate the solution in real time when there is a problem, and upload it to the sintering control room, and modify the type and ratio of sintering raw materials based on the solution; the way of modification includes manual labor Manual modification and system automatic modification.

Images