CN102795627B - Multi-parameter online monitoring and optimizing control device and method of polycrystalline silicon reduction furnace - Google Patents

Abstract

The invention discloses a multi-parameter online monitoring and optimization control device and method for a polysilicon reduction furnace, which includes a multi-parameter infrared monitoring probe, an infrared image processing and visual measurement module, and a process optimization control module. Infrared images of silicon rods are collected by the data acquisition module, converted into digital images, and then analyzed and processed by the image processing module. The diameter and growth rate of silicon rods are obtained through visual measurement technology, and the temperature on the surface of silicon rods is measured by colorimetric temperature measurement. Distribution, the obtained data is connected to the display through the user interface, and the obtained measurement data is analyzed to establish an optimal control model, and the closed-loop optimal control is performed in combination with different types of polysilicon reduction furnaces. The invention can optimize and control the growth process of silicon rods, and has extremely important significance for saving energy, reducing consumption, improving production efficiency, ensuring production safety and reducing labor intensity.

Description

Multi-parameter online monitoring and optimization control device and method for polysilicon reduction furnace

technical field

The invention relates to the field of silicon reduction furnace production, in particular to a polysilicon reduction furnace multi-parameter online monitoring and optimization control device and method.

technical background

Polysilicon is the basic raw material of the electronics industry and the solar energy industry, and is widely used in semiconductor chips, high-performance sensors, optical fibers, solar panels, etc. By 2010, the global polysilicon output reached 120,000 tons, while China accounted for 50% of the total output, with an output value of nearly RMB 40 billion. It is estimated that in 2011, the global polysilicon output will reach 160,000 tons, and the proportion of China will increase to more than 60% of the total output.

The temperature distribution on the surface of silicon rods in the reduction furnace, rod diameter, growth rate and other parameters are the key links to improve the growth quality of silicon rods. During the growth of polycrystalline silicon rods, the temperature increases, the deposition rate of silicon on the silicon core increases, and the growth of silicon rods is accelerated, but the energy consumption is too much. The higher the temperature, the harsher the gas phase conditions, and the worse the uniformity of the silicon rods; the lower the temperature , the deposition rate of silicon slows down. When the temperature is lower than a certain value, the silicon rods are easy to break, which affects the continuation of growth. In addition, the change of the diameter of silicon rods over time is an important basis for temperature regulation and the distribution ratio of feed gas components. Unreasonable ratios will reduce the growth rate of silicon rods and consume a lot of energy. Therefore, there is an urgent need for a device that can monitor these parameters online and perform closed-loop optimal control in the reduction furnace.

However, at present, there are no mature instruments and equipment for production application in the theoretical research and exploratory experiments on the on-line measurement and optimal control of the multi-parameters of silicon rods in the reduction furnace at home and abroad. In the world, even Germany's Siemens and Japan's Mitsubishi Electric Corporation, which are in the leading position in polysilicon production process control, do not have mature on-line monitoring and control equipment; it is even blank in China. At present, the multi-parameters in the reduction furnace (Such as silicon rod diameter, temperature, etc.) on-line monitoring and optimal control basically stay in the theoretical research stage.

Up to now, in the actual production of more than a thousand polysilicon reduction furnaces across the country, the open-loop control mode based on the preset experience curve is adopted. The measurement is roughly estimated based on the workers' work experience and visual observation. Obviously, this method has a certain degree of subjectivity and chance.

In view of this, a device for on-line measurement of temperature distribution, rod diameter, growth rate and other parameters on the surface of silicon rods in the reduction furnace and closed-loop optimal control has been developed.

Contents of the invention

The invention solves the technical problem of multi-parameter real-time monitoring and optimal control in the growth process of polysilicon, and develops a device for accurate online measurement of silicon rod surface temperature distribution, silicon rod diameter, growth rate and other parameters and closed-loop optimal control. It is of great significance to save energy and reduce consumption, improve production efficiency, ensure production safety, and reduce labor intensity.

The technical scheme adopted in the present invention is:

The polysilicon reduction furnace multi-parameter online monitoring and optimization control device is characterized in that it includes a multi-parameter infrared monitoring probe, an infrared image processing and visual measurement module, and a process optimization control module. The multi-parameter infrared monitoring probe includes a near-infrared optical system, image acquisition module, the multi-parameter infrared monitoring probe is equipped with a water-cooled protective cover outside, and the front end of the water-cooled protective cover is provided with a water-cooled sealed pressure-resistant observation window, and the multi-parameter infrared monitoring probe goes deep into the polysilicon reduction furnace; the infrared image The processing and visual measurement module includes an infrared image processing module and a visual measurement module. The infrared image processing module is sequentially connected to a data acquisition module and an image processing module. The visual measurement module includes a colorimetric temperature measurement module, a diameter measurement module, and an image processing module. The output terminals are respectively connected with the colorimetric temperature measurement module and the diameter measurement module. The output terminals of the colorimetric temperature measurement module and the diameter measurement module are connected to the data output interface and the user interface. The data output interface is externally connected to the process optimization control module, and the user interface is externally connected to The monitor and the process optimization control module are connected with the original control system through a data interface.

The process optimization control module is a closed-loop optimization control model.

The image acquisition module is an infrared CCD camera.

The method for multi-parameter online monitoring and optimal control of a polysilicon reduction furnace is characterized in that it includes the following specific steps:

1) The multi-parameter infrared monitoring probe obtains the image signal of silicon rod growth in the furnace through the near-infrared optical system and infrared CCD camera by selecting the optimal infrared band and high-sensitivity infrared CCD camera, and performs A/D conversion through the data acquisition module , to obtain an infrared digital image of a polysilicon rod;

2) Send the infrared digital image obtained above into the image processing module, and perform image preprocessing and segmentation through the image processing module. Combined with the specific image, through the image denoising algorithm and smoothing filter algorithm, the maximum protection of the image surface temperature Distribution features; Find the closed edge by contour tracking, and perform sub-pixel edge detection and straight line fitting to obtain higher precision image edges;

3) Send the above-mentioned processed image to the diameter measurement module of the visual measurement module. Firstly, the infrared CCD camera is calibrated to obtain the internal and external parameters of the camera. According to the Hough line fitting algorithm, the edge of the initial target image is selected, and then the image is processed. Tracking of the edge of a small area; matching corresponding points under the guidance of the basic matrix of robust estimation; finally, according to the above algorithm, the measured pixel value is converted to the actual unit scale to obtain the real-time diameter data of the silicon rod;

4) Send the image processed in step 2 to the colorimetric temperature measurement module, and use the image processing system in the colorimetric temperature measurement module to correct the bad points of the image, and perform segmentation, smoothing, layering, and grayscale conversion on the image , then call the database, fit the gray curve, calibrate the temperature, and complete the temperature measurement;

5) Send the silicon rod diameter and surface temperature data obtained in steps 3 and 4 to the process optimization control module through the data output interface, and establish an optimal control model in combination with the working condition information, and use feedback to guide the silicon rods in the most reasonable process conditions grow under

6) Output the silicon rod diameter and surface temperature data obtained above to the terminal to obtain the real-time diameter and growth rate of the silicon rod, as well as the real-time temperature of any point on the surface of the silicon rod, and display them on the monitor.

The working principle of the present invention is:

The invention obtains the infrared image of the silicon rod in the furnace through the multi-parameter infrared monitoring probe (NICCD), collects and converts it into a digital image through the data acquisition module, and then analyzes and processes it by the image processing module, and obtains the diameter and growth of the silicon rod through the visual measurement technology. The temperature distribution on the surface of the silicon rod is measured by the colorimetric temperature measurement method, and the obtained data is connected to the display through the user interface. After analyzing the measured data, an optimal control model is established, and the closed-loop optimal control is carried out in combination with different types of polysilicon reduction furnaces.

The beneficial effects of the present invention are:

1) Compared with the existing temperature measuring equipment, the present invention has the advantages of large temperature measurement range, high precision and long-term continuous measurement;

2) Compared with the current artificial experience estimation of silicon rod diameter and growth rate, the present invention has the advantage of continuous and accurate measurement;

3) Compared with the existing fixed control mode, the present invention has the advantage of providing a corresponding closed-loop optimization control model for different furnace types and working conditions.

Description of drawings

Fig. 1 is a functional block diagram of the system structure of the present invention.

The optimal control model schematic diagram of Fig. 2 the present invention,

Fig. 3 is a schematic diagram of the surface temperature monitoring results during the silicon rod growth process.

Fig. 4 is a schematic diagram of the monitoring results of the silicon rod diameter during the growth process of the silicon rod.

Detailed ways

When the present invention is described further below in conjunction with example and accompanying drawing, but should not limit protection scope of the present invention with this.

The structural composition of the present invention is shown in Figure 1.

The polysilicon reduction furnace multi-parameter online monitoring and optimization control device is characterized in that it includes a multi-parameter infrared monitoring probe 1, an infrared image processing and visual measurement module 2, and a process optimization control module 3. The multi-parameter infrared monitoring probe 1 includes a near-infrared Optical system 4, image acquisition module 5, the multi-parameter infrared monitoring probe is equipped with a water-cooled protective cover outside, and the front end of the water-cooled protective cover is equipped with a water-cooled sealed pressure-resistant observation window, and the multi-parameter infrared monitoring probe goes deep into the polysilicon reduction furnace; the infrared image The processing and visual measurement module 2 includes an infrared image processing module and a visual measurement module. The infrared image processing module is connected in turn to a data acquisition module 6 and an image processing module 7. The visual measurement module includes a colorimetric temperature measurement module 8 and a diameter measurement module 9. , the output ends of the image processing module 7 are respectively connected with the colorimetric temperature measurement module 8 and the diameter measurement module 9, and the output ends of the colorimetric temperature measurement module 8 and the diameter measurement module 9 are all connected to the data output interface 12 and the user interface 10, and the data The output interface is externally connected to a process optimization control module 3, the user interface is externally connected to a display 11, and the process optimization control module 3 is connected to the original control system through a data interface.

Process optimization control module 3 is a closed-loop optimization control model.

The image acquisition module 5 is an infrared CCD camera.

A method for multi-parameter online monitoring and optimal control of a polysilicon reduction furnace includes the following specific steps:

1) The multi-parameter infrared monitoring probe 1 obtains the image signal of silicon rod growth in the furnace through the near-infrared optical system 4 and infrared CCD camera by selecting the optimal infrared band and high-sensitivity infrared CCD camera, and performs A through the data acquisition module 6. /D conversion to obtain infrared digital images of polysilicon rods;

2) Send the infrared digital image obtained above into the image processing module 7, and carry out image preprocessing and segmentation through the image processing module 7, combined with the specific image, through the image denoising algorithm and smoothing filter algorithm, the image is protected to the greatest extent Surface temperature distribution characteristics; find the closed edge by contour tracking, and perform sub-pixel edge detection and straight line fitting to obtain higher precision image edges;

3) Send the above-mentioned processed image to the diameter measurement module 9 of the visual measurement module. Firstly, the infrared CCD camera is calibrated to obtain the internal and external parameters of the camera. The edge of the initial target image is selected according to the Hough line fitting algorithm, and then the image is Track the edge of the small area; match the corresponding points under the guidance of the basic matrix of robust estimation; finally, convert the measured pixel value and the actual unit scale according to the above algorithm to obtain the real-time diameter data of the silicon rod;

4) Send the image processed in step 2 to the colorimetric temperature measurement module 8, use the image processing system in the colorimetric temperature measurement module to correct the bad points of the image, and perform segmentation, smoothing, layering, and grayscale on the image Convert, then call the database, fit the gray curve, calibrate the temperature, and complete the temperature measurement;

5) Send the silicon rod diameter and surface temperature data obtained in steps 3 and 4 to the process optimization control module 3 through the data output interface, and establish an optimization control model based on the working condition information, and use feedback to guide the silicon rod in the most reasonable process grow under the conditions;

6) Output the silicon rod diameter and surface temperature data obtained above to the terminal to obtain the real-time diameter and growth rate of the silicon rod, as well as the real-time temperature of any point on the surface of the silicon rod, and display them on the display 11 .

Invention principle of the present invention is as follows:

In the visual measurement module, the real-time growth diameter of silicon rods is measured according to the captured growth images of silicon rods.

(1) Calibration of the camera system

，其对应的二维像点齐次坐标为

，空间点

与图像点

之间的射影关系为To convert the pixel value obtained in image processing into the actual unit scale value, it is necessary to calibrate the system, that is, to determine the geometric camera model of the camera. The homogeneous coordinates of the three-dimensional point of the known space target are

, and its corresponding two-dimensional pixel homogeneous coordinates are

, the space point

with image points

The projective relationship between

Use the least squares method to solve the overdetermined linear equations, and give the external parameters; solve the internal parameters, if the camera has no lens distortion, it can be solved by an overdetermined linear equation, if there is radial distortion, it can be solved by a three-variable optimization search . Finally, the internal and external parameters, focal length f, radial distortion coefficient k, rotation matrix R and translation vector T are solved.

(2) Edge detection and Hough straight line fitting

Edge detection utilizes first and second derivative detection based on detecting discontinuities in luminance values. The gradient of a two-dimensional function f(x,y) is defined as a vector

The magnitude of this vector is

Generally simplified to

or

They have a value of zero in the constant luminance region.

Use Hough transform for line detection and linking. First, do peak detection, find the Hough transform unit that contains the maximum value, set the Hough transform unit in the neighborhood of the found maximum value point to 0, and repeat this step until you find the desired one. up to the peak number. For each peak, find the positionally correlated pixels and combine them into line segments.

(3) Image corresponding point matching

Matching the above detected feature points is to find the image points in different images corresponding to the same spatial point.

，x(x,y,1)、x′(x′,y′,1)分别表示左、右摄像机的对应点的齐次坐标。由基本矩阵的定义可知，Assuming that two pictures are taken of silicon rods at the same moment, the corresponding feature point sets in the two pictures are

, x(x,y,1), x′(x′,y′,1) represent the homogeneous coordinates of the corresponding points of the left and right cameras respectively. From the definition of the fundamental matrix, we know that

Right now

in, , satisfying the constraint ||f||＝1;

After the basic matrix and accurate stereo matching points are obtained by the robust estimation algorithm, more corresponding point matching is performed under the guidance of the basic matrix.

(4) Establishment of optimal control model

By measuring the obtained process parameters and combining with the working condition information to establish an optimal control model to guide the growth of silicon rods to achieve the goals of reducing costs, improving product quality, energy saving and emission reduction

The colorimetric temperature measurement method of the colorimetric temperature measurement system is as follows:

The radiation and distribution of a general object with a thermodynamic temperature T is described by Planck's radiation law:

In the formula, M _eλ is the spectral radiance, where λ is the wavelength, T is the thermodynamic temperature, C ₁ =3.741832×10 ^-12 wcm ² is the first radiation constant, C ₂ =1.438786×10 ⁴ μmK is the second radiation constant ; When c ₂ /λT>>1, Planck's formula can be replaced by Wien's formula, which can be simplified as:

In the region of λT<2698μm·K, the error between Wien's formula and Planck's formula is less than 1%;

If the spectral radiance M(T, λ ₁ ) and M(T, λ ₂ ) emitted by the same point of the object are measured at the same time at two wavelengths λ ₁ and λ ₂ , then the ratio of the two can be obtained The temperature of the point, the formula is:

suppose , that is, approximately treat the object as a gray body, then the formula (9) can be simplified as:

This is the calculation formula of the colorimetric temperature measurement method.

Claims (3)

1. A multi-parameter online monitoring and optimization control device of a polycrystalline silicon reduction furnace is characterized by comprising a multi-parameter infrared monitoring probe, an infrared image processing and vision measuring module and a process optimization control module, wherein the multi-parameter infrared monitoring probe comprises a near-infrared optical system and an image acquisition module, a water-cooling protective cover is arranged outside the multi-parameter infrared monitoring probe, a water-cooling type sealed pressure-resistant observation window is arranged at the front end of the water-cooling protective cover, and the multi-parameter infrared monitoring probe extends into the polycrystalline silicon reduction furnace; the infrared image processing and vision measuring module comprises an infrared image processing module and a vision measuring module, the infrared image processing module is sequentially connected with a data acquisition module and an image processing module, the vision measuring module comprises a colorimetric temperature measuring module and a diameter measuring module, the output end of the image processing module is respectively connected with the colorimetric temperature measuring module and the diameter measuring module, the output ends of the colorimetric temperature measuring module and the diameter measuring module are respectively connected with a data output interface and a user interface, the data output interface is externally connected with a process optimization control module, the user interface is externally connected with a display, and the process optimization control module is connected with an original control system through a data interface; the process optimization control module is a closed-loop optimization control model.

2. The polysilicon reducing furnace multi-parameter on-line monitoring and optimization control device according to claim 1, wherein the image acquisition module is an infrared CCD camera.

3. A method for multi-parameter online monitoring and optimal control of a polycrystalline silicon reduction furnace is characterized by comprising the following specific steps:

the multi-parameter infrared monitoring probe obtains an image signal of silicon rod growth in the furnace through a near-infrared optical system and an infrared CCD camera by selecting the infrared CCD camera with the optimal infrared band and high sensitivity, and obtains an infrared digital image of the polysilicon silicon rod through A/D conversion by a data acquisition module;

2) the obtained infrared digital image is sent to an image processing module, image preprocessing and segmentation are carried out through the image processing module, and the image surface temperature distribution characteristics are protected to the maximum extent through an image denoising algorithm and a smoothing filtering algorithm in combination with a specific image; finding out a closed edge by contour tracing, and performing sub-pixel level edge detection and straight line fitting to obtain a high-precision image edge;

3) the processed image is transmitted to a diameter measuring module of a vision measuring module, firstly, the infrared CCD camera is calibrated to obtain the internal and external parameters of the camera, the initial target image edge is selected according to the Hough linear fitting algorithm, and then the small area edge of the image is tracked; carrying out corresponding point matching under the guidance of a basic matrix of robustness estimation; finally, converting the measured pixel value and the actual unit system scale according to the algorithm to obtain real-time diameter data of the silicon rod;

4) conveying the image processed in the step 2 to a colorimetric temperature measurement module, correcting a dead pixel of the image by using an image processing system in the colorimetric temperature measurement module, segmenting, smoothing, layering and carrying out gray level conversion on the image, then calling a database, fitting a gray level curve, calibrating the temperature and finishing temperature measurement;

5) the diameter and surface temperature data of the silicon rod obtained in the steps 3 and 4 are transmitted to a process optimization control module through a data output interface, and an optimization control model is established by combining working condition information so as to feedback and guide the growth of the silicon rod under the most reasonable process condition;

6) and outputting the obtained diameter and surface temperature data of the silicon rod to a terminal to obtain the real-time diameter and growth rate of the silicon rod and the real-time temperature of any point on the surface of the silicon rod, and displaying the real-time diameter and growth rate and the real-time temperature on the surface of the silicon rod in a display.

Images