Industrial silicon ingot breaking device and method based on impurity agglomerate ultrasonic identification

Abstract

An industrial silicon ingot crushing device and method based on impurity cluster ultrasonic recognition belong to the technical field of industrial silicon block processing equipment. First, the scene image and depth data of the operation area are collected and converted into three-dimensional space coordinates. Second, the echo signal is obtained by scanning the target silicon block with a ring-shaped ultrasonic array probe, and the impurity type is analyzed. Third, the ultrasonic three-dimensional space coordinates of the impurity points obtained by scanning the ring-shaped ultrasonic array probe are solved. Fourth, the size and content of the impurity clusters inside the target silicon block are analyzed, the sound-light prompt type is determined, and the impurity virtual projection is displayed in the AR display interaction device. Finally, after the operation is completed, the AR display interaction device generates a crushing report, and the power module enters a low-power mode. The industrial silicon ingot crushing device is realized. The present application can effectively improve the operation efficiency and product quality, and reduce the phenomenon of incomplete impurity removal caused by differences in manual experience and visual fatigue.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of industrial silicon block processing equipment, and relates to an industrial silicon ingot breaking device and method based on impurity agglomerate ultrasonic identification, in particular to a handheld coaxial industrial silicon ingot breaking device and method based on impurity agglomerate ultrasonic identification. BACKGROUND

[0002] In the production and processing process of industrial silicon, 3-5mm crushed silicon is often laid on the inside of the casting mold to avoid the adhesion of molten silicon to the mold, so as to facilitate demolding. However, in the casting process, part of the slag will be carried into the silicon ingot after the cooling of the molten silicon, forming a slag inclusion product. In addition, the raw material quartz sand for industrial silicon production contains mineral impurities such as hematite, mica and calcite. These oxides will reduce carbon and compound with silicon at high temperature to form elemental compounds (FeSi2, Al-Si-Fe, CaSi2, etc.), which are dispersedly distributed in the molten silicon. With the solidification of the molten silicon, these impurities will be continuously pushed to the uncooled liquid phase area by the growing solid-liquid interface due to the influence of the segregation effect (lower solubility in solid components). The natural cooling of the silicon ingot is a process from outside to inside, and finally affected by the cooling and diffusion rate of impurities, part of the impurity enrichment area and agglomerated block-shaped objects are formed in the silicon ingot. Therefore, the subsequent process of industrial silicon is to accurately break the large-size industrial silicon ingot into a block with a side length of 150-200mm, and to remove the impurities adhered to the edge and inside of the silicon ingot. The size is also the conventional breaking specification in the industry, which meets the feeding requirements of the subsequent links.

[0003] However, the current industrial silicon processing process has low automation and limited efficiency in impurity removal: most scenes still use the traditional mode of visual screening and manual crushing with a hammer, completely relying on the experience of the operator to judge the size of the silicon block, using the difference in fracture toughness between the impurity and the silicon matrix, and the principle of stress concentration cracking at the interface, by increasing the number of strikes as much as possible, refining the size of the silicon block, and distinguishing by naked eye to avoid the residual of internal impurity agglomerates, the work efficiency is difficult to meet the needs of enterprise scale production. The large fixed crushing equipment on the market, such as the double-power mobile jaw crusher station and its control method provided by Chinese invention patent (application number 202511066969.3), can realize batch crushing of silicon ingots, but due to its large size and poor mobility, it is difficult to adapt to the flexible operation needs of production enterprises, and it does not have impurity removal capability, still requiring manual picking. The existing handheld crushing device, such as the handheld rock crusher for ore mining provided by Chinese utility model patent (application number 202121758319), is more portable, but it only has basic crushing function and no impurity detection capability, which is easy to cause over-crushing, leading to increased dust and waste of silicon material. Or due to insufficient crushing, the impurities are not completely separated and remain in the silicon block, which is difficult to meet the high-purity requirements of product quality.

[0004] In summary, the production of industrial silicon blocks is faced with problems such as difficulty in impurity detection and separation, low precision of silicon ingot crushing, and poor adaptability of equipment. Therefore, it is urgent to develop a semi-automatic handheld industrial silicon ingot crushing device that can identify the size of the silicon block, detect the distribution of internal impurities, and be flexible and portable, to improve the overall efficiency and product quality of industrial silicon processing. SUMMARY

[0005] To solve the problems of the prior art, the present application provides an industrial silicon ingot crushing device and method based on ultrasonic recognition of impurity agglomerates, specifically a handheld coaxial industrial silicon ingot crushing device and method based on ultrasonic recognition of impurity agglomerates, which automatically recognizes and analyzes scene data through a binocular stereo vision algorithm, obtains the position, size and content information of impurity agglomerates in the silicon ingot / silicon block through ultrasonic detection technology, and uses sound and light warnings and visual projection to make the silicon block impurity agglomerate detection process virtual and real, more intuitive, which can effectively improve the work efficiency and product quality, and reduce the phenomenon of incomplete impurity removal caused by differences in human experience and visual fatigue.

[0006] To achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0007] An industrial silicon ingot crushing method based on ultrasonic recognition of impurity agglomerates, the industrial silicon ingot crushing method comprising the following steps:

[0008] The first step, the operator wears an AR display interaction device, carries a handheld crushing device into the work area, collects the scene image and depth data of the work area, and converts them into three-dimensional space coordinates. Specifically:

[0009] Step 1.1, the binocular camera and depth sensor of the AR display interaction device scan the work area to form scene images and pixel point cloud data , wherein is the pixel coordinate of a single pixel point in the pixel point cloud with respect to the imaging of the binocular camera, is the distance from the real position of the pixel point to the optical center of the binocular camera, i.e. the depth data.

[0010] Further, the handheld crushing device is provided with four high-reflective feature marker points which are not easily blocked, facilitating the recognition and data collection of the AR display interaction device. It needs to be additionally explained that the pixel point cloud data includes the data information of these high-reflective feature marker points.

[0011] Step 1.2, extract the pixel point cloud data of the work scene and convert it into three-dimensional space coordinates in a unified main field of view coordinate system , as shown in formula (1); the main field of view coordinate system takes the optical center of the binocular camera as the origin, the x-axis is to the right along the horizontal direction of the lens of the AR monitoring and interaction device 1, the y-axis is vertically upward along the lens, and the z-axis is outward perpendicular to the lens;

[0012] (1)

[0013] In the formula: is the three-dimensional space coordinate of the target silicon block in the main field of view coordinate system; is the coordinate of the image center; , are the horizontal and vertical focal lengths; is a rotation matrix, is a translation vector, both of which are built-in parameters of the binocular camera; the superscript T represents transposition.

[0014] The second step, the ring-shaped ultrasonic array probe is used to scan the target silicon block to obtain echo signals, which are pre-amplified, envelope-detected, and then analyzed for impurity types. Specifically:

[0015] Step 2.1, start the handheld crushing device, and use the ring-shaped ultrasonic array probe to emit ultrasonic waves to scan the target silicon block. When the sound wave propagates inside the target silicon block, it encounters different material interfaces, and due to the difference in medium acoustic impedance, it will produce reflection and form echo pulses. The ring-shaped ultrasonic array probe is used to synchronously receive the echo signals.

[0016] The returned echo signal is pre-amplified by an amplifier and then envelope processed. First, the signal is converted into an analytic form using Hilbert transform as shown in equation (2) to obtain the envelope amplitude. Then, the envelope detection as shown in equation (3) and the smoothing filter processing as shown in equation (4) are used to remove high-frequency oscillation to obtain the reflection intensity curve.

[0017] (2)

[0018] (3)

[0019] (4)

[0020] In the formula, is the received channel signal after pre-amplification; is the Hilbert transformed signal; is the analytic signal; is the envelope signal; is the smoothed envelope signal; j is the imaginary unit; L is the smoothing window length, which is 1-5 ms; t is the time variable, which represents the time of ultrasonic signal propagation in the target silicon block; is the integral variable.

[0021] Step 2.2, for the processed envelope signal, find its local maximum value, and the corresponds to a reflection interface, and its intensity is . As shown in equation (5):

[0022] (5)

[0023] In the formula, represents the set of local maximum values of the envelope signal; represents the time position corresponding to the i th peak value, and represents the round-trip time of ultrasonic wave propagation to the reflection interface; represents the envelope amplitude of the i th peak value, and represents the reflection intensity of the reflection interface; , respectively represent the envelope amplitudes of the adjacent peak values before and after the i th peak value.

[0024] The envelope amplitude of the peak value is normalized to estimate the reflection intensity as shown in equation (6), and impurity judgment is performed:

[0025] (6)

[0026] In the formula, is the normalized reflection intensity coefficient of the reflection interface at the i th peak value, and if If the reflection interface exists suspicious impurities, a preset threshold is used to determine the existence of the suspicious impurities, and the preset threshold ranges from 0.6 to 0.8. is the envelope amplitude of the ith peak value; is the reference interface standard amplitude.

[0027] Further, the impurity type of the suspicious impurity reflection interface is determined in two ways: (1) when the reflected wave of the ultrasonic wave is in phase with the incident wave, the reflection interface is considered to be a high-resistance impurity composed of Al-Si-Fe and other impurity agglomerates; (2) when the reflected wave of the ultrasonic wave is out of phase with the incident wave, the reflection interface is considered to be a low-resistance impurity composed of gas cavities such as bubbles and cracks.

[0028] Third step, solving the ultrasonic three-dimensional space coordinates of the impurity points scanned by the annular ultrasonic array probe. Specifically:

[0029] The ultrasonic three-dimensional space coordinates are three-dimensional space coordinates in the handheld breaking device main body coordinate system.

[0030] Step 3.1, using the time of arrival of the annular ultrasonic array probe unit at the same point to solve the distance of each annular ultrasonic array probe unit to the impurity point , as shown in formula (7):

[0031] (7)

[0032] In the formula, c is the propagation speed of the ultrasonic wave in the target silicon block; n is the annular ultrasonic array probe unit number; N is the number of annular ultrasonic array probe units; is the echo time of the nth annular ultrasonic array probe unit from transmitting the ultrasonic wave to receiving the echo signal; is the starting time of transmitting the ultrasonic wave.

[0033] Step 3.2, using multilateration and least squares method for impurity point positioning, wherein the annular ultrasonic array probe unit coordinates are fixed coordinates , and the ultrasonic three-dimensional space coordinates of each impurity point are . The multilateration equation is established as shown in formula (8):

[0034] (8)

[0035] Taking one annular ultrasonic probe unit as a reference ultrasonic probe unit, a linear equation is obtained by simplifying, as shown in formula (9):

[0036] (9)

[0037] In the formula, represents the fixed coordinate of the reference ultrasonic probe unit; d1 represents the distance from the reference ultrasonic probe unit to the impurity point; the superscript T represents transposition.

[0038] The matrix of the remaining annular ultrasonic array probe unit is constructed as shown in equation (10):

[0039] (10)

[0040] (11)

[0041] (12)

[0042] wherein, is a coefficient matrix, which is specifically shown in equation (11); is the ultrasonic three-dimensional space coordinate of the impurity point; is an observation vector, which is specifically shown in equation (12); the superscript T represents transposition.

[0043] The ultrasonic three-dimensional space coordinate of the impurity point is solved by using the least square method as shown in equation (13) :

[0044] (13)

[0045] wherein, the superscript T represents transposition.

[0046] Fourthly, the size and content of the impurity cluster in the target silicon block are analyzed, and the acousto-optic prompt type is determined. Specifically:

[0047] Step 4.1, on the basis of the impurities determined in step 3, the impurities are clustered in space to form independent impurity cluster regions , and the size and content of the impurity cluster are analyzed. The size is approximately considered as the minimum circumscribed cuboid size of the impurity cluster, and the calculation idea is as follows: first, the covariance matrix of the impurity points is calculated as shown in equation (14); second, the covariance matrix is decomposed as shown in equation (15); finally, the coordinate difference of the vertex of the minimum circumscribed cuboid model in the axial direction is calculated as shown in equation (16), and the actual size of the impurity cluster is output. The impurity content in the target silicon block is approximately obtained by dividing the total minimum circumscribed cuboid volume of all impurity clusters in the scanning region by the volume of the target silicon block in the scanning region.

[0048] (14)

[0049] (15)

[0050] (16)

[0051] wherein: is the covariance matrix; N is the number of impurity points contained in the impurity cluster; is the set of all impurity points contained in the kth impurity cluster; is the centroid of the impurity cluster; is the principal eigenvector; is the eigenvalue; is the minimum circumscribed cuboid size of the impurity cluster, wherein , the corresponding is the length, width, and height of the minimum circumscribed cuboid, respectively.

[0052] Step 4.2, according to the size and content difference of the impurity cluster, specific determination threshold and sound and light prompt type are divided, as follows:

[0053] If there is an impurity cluster size 10mm, it is determined that there is a larger impurity cluster in the region, and the warning system flashes yellow sound and light warning, reminding the operator to crush and remove impurities;

[0054] If the maximum impurity cluster size 10mm, but the impurity content ratio is higher than 2%, the warning system flashes red sound and light warning, prompting the operator to dispose of the target silicon block as waste;

[0055] The rest are determined to be qualified materials, and the warning system maintains green light source without alarm.

[0056] Step 5, display the impurity virtual projection in the AR display interactive device. Specifically:

[0057] Step 5.1, the coordinates of each impurity point obtained in step 3.2 are rigidly registered and corrected by rotation and translation. The rotation matrix R and the translation vector t are calculated as follows: first, the pre-set fixed coordinate set of the high-reflective feature identification point in step 1.1 under the handheld crushing device system is , The observation coordinate set measured by the AR display interactive device according to step 1.2 is The perspective-n-point positioning method (PnP algorithm) is used to solve the rotation matrix R and the translation vector t, with the minimum overall re-projection error as the optimization design target, as shown in formula (17):

[0058] (17)

[0059] Step 5.1.1, the specific calculation method is as follows: the center of the pre-set fixed coordinate set of the high-reflective feature identification point is calculated as shown in formula (18) ​ ; calculate the high-reflection feature landmark point observation coordinate set of the center ; get the preset fixed coordinate set decentralized as shown in formula (20) .

[0060] (18)

[0061] (19)

[0062] (20)

[0063] (21)

[0064] wherein, , the corresponding four high-reflection feature landmarks.

[0065] Step 5.1.2, calculate the decentralized covariance matrix H as shown in formula (22), and solve the left singular vector matrix U, singular value diagonal matrix and right singular vector matrix by singular value decomposition SVD as shown in formula (23).

[0066] (22)

[0067] (23)

[0068] Step 5.1.3, finally get the rotation matrix R as shown in formula (24), and get the translation vector t as shown in formula (25).

[0069] (24)

[0070] (25)

[0071] wherein, is the determinant operator; R is the rotation matrix; t is the translation vector.

[0072] Step 5.1.4, thus the impurity ultrasonic three-dimensional space coordinates obtained in step 3.2 can be converted into three-dimensional space coordinates under the AR display interactive device main field of view coordinate system, as shown in formula (26):

[0073] (26)

[0074] Step 5.2, the impurity three-dimensional space coordinates calculated in step 5.1 are displayed in the AR display interaction device The projection display is performed in combination with the sound-light prompt realized in step 4.2, and the virtual-real combined display of the impurity distribution in the target silicon ingot can effectively realize efficient screening and crushing of the industrial silicon ingot by the operator.

[0075] Further, since the personnel operation visual angle is subjectively and dynamically adjusted, the rotation matrix R and the translation vector t in step 5.1 should be dynamically updated in real time at a frequency of 30 Hz, and the display projection refresh in the AR display interaction device is also completed at a refresh frequency of 30 Hz, and the update delay is controlled within 50 ms.

[0076] In the sixth step, after the operation is completed, the AR display interaction device generates a crushing report, and the power module enters a low-power mode. Specifically:

[0077] After the operator manually issues the operation completion instruction through the reserved button, the system extracts the key data from the process log and outputs the crushing report in a preset format, and the core information includes basic operation information and quality judgment information, such as operation time, average impurity content, product qualification rate, and abnormal alarm information. After the operator views the output report on the AR display interaction device, the operator can manually close, and then the report is stored in the system history record, and the AR display interaction device enters a low-power mode.

[0078] An industrial silicon ingot crushing device based on impurity cluster ultrasonic identification, which realizes the above-mentioned industrial silicon ingot crushing method through the industrial silicon ingot crushing device, the industrial silicon ingot crushing device includes two core components, an AR display interaction device and a handheld crushing device, and the two can realize bidirectional data transmission through high anti-interference data lines or Bluetooth pairing. Among them, a coaxial integrated annular ultrasonic array probe and a detachable tungsten alloy hammer are designed at the gun head of the handheld crushing device to realize identification detection and crushing of the impurity-containing target silicon block. Specifically:

[0079] The AR display interaction device comprises an image perception module, a control module, a power module, a display module and a communication module. The image perception module mainly comprises a binocular camera and a depth sensor, which are used for scanning and recording working images in a field of view, and identifying position and posture information of a handheld crushing device; the control module comprises an image processor and a power supply management unit, which are used for receiving internal impurity signals obtained by scanning a target silicon block by the annular ultrasonic array probe, identifying and analyzing the signal data, and completing three-dimensional space coordinate conversion of impurity position display projection. Meanwhile, images of the scanned and crushed qualified target silicon block are recorded and stored in historical data of this work, which is used for subsequent statistics and report generation; the display module mainly comprises a semi-transparent and semi-reflective silicon-based micro organic light emitting diode (OLED) display screen, which is used for displaying the internal impurity distribution of the scanned target silicon block in the field of view of the wearer in real time; the power module provides stable power supply for the AR display interaction device, and supports entering a low-power mode after the whole operation is completed; the communication module supports two transmission modes, including Bluetooth pairing and data line transmission, and the two modes can be flexibly switched.

[0080] The handheld crushing device comprises an ultrasonic detection module, a warning module, a communication module, a crushing execution module and a holding damping module. The ultrasonic detection module is used for scanning a target silicon block and collecting internal acoustic echo signals thereof; the warning module is used for real-time processing and analyzing acoustic data monitored by the ultrasonic detection module, judging whether there is an impurity agglomerate distribution in the target silicon block, and having an audible and visual alarm function; the communication module is used for transmitting the analyzed data and related parameters to the AR display interaction device; the crushing execution module mainly comprises a servo driver and a detachable tungsten alloy impact crushing hammer, the servo driver drives the hammer head to perform crushing operation according to adjustable working parameters, and the type and size of the hammer head are replaced and adjusted according to actual construction requirements; the holding damping module wraps a non-slip rubber sleeve at the handle, and a damping spring is arranged in the device to absorb high-frequency vibration generated in the crushing process, thereby improving the use comfort and stability.

[0081] Compared with the prior art, the present application has the following advantages:

[0082] (1) The high-frequency ultrasonic array probe is integrated at the impact hammer head of the handheld crushing device, coaxial arrangement is realized, the detection, warning and crushing process are integrated and automated, and the operation efficiency and product quality are greatly improved.

[0083] (2) The real-time ultrasonic three-dimensional space positioning of the internal impurities of the target silicon block is realized according to the scanning data of the annular ultrasonic array probe and the delay and sum algorithm, the detection accuracy is high, the response speed is fast, the ultrasonic three-dimensional space coordinates are matched with the main field of view coordinate system, and the real-time visualization of the impurities can be realized.

[0084] (3) The timely acousto-optic early warning of the handheld crushing device and the field impurity projection superposition of the AR display interaction device make the detection process virtual and real fusion, more intuitive, and effectively reduce the impurity removal phenomenon caused by artificial experience difference and visual fatigue.

[0085] (4) The device connection mode of wired and wireless dual-link complementation and the replaceable impact hammer head design make the application have good adaptability in the crushing of industrial silicon ingots under different production lines and material conditions. BRIEF DESCRIPTION OF DRAWINGS

[0086] Figure 1 is a flowchart of the use of the application;

[0087] Figure 2 is a schematic diagram of the circuit module composition;

[0088] Figure 3 is a schematic diagram of the structure of the application;

[0089] Figure 4 is a schematic diagram of the identification of impurities in silicon blocks exceeding the standard;

[0090] In the figure: 1 AR display interaction device; 2 handheld crushing device; 3 high anti-interference data line; 4 annular ultrasonic array probe; 5 hammer head. DETAILED DESCRIPTION

[0091] The application will be further described below in combination with specific implementation cases.

[0092] An industrial silicon ingot crushing method based on impurity agglomerate ultrasonic identification, in particular a handheld coaxial industrial silicon ingot crushing method based on impurity agglomerate ultrasonic identification, the handheld coaxial industrial silicon ingot crushing method comprising the following steps:

[0093] First step, the operator wears the AR display interaction device 1, carries the handheld crushing device 2 into the work area, collects the scene image and depth data of the work area, and converts them into three-dimensional space coordinates. Specifically:

[0094] Step 1.1, the binocular camera and depth sensor of the AR display interaction device 1 scan the work area to form scene images and pixel point cloud data , wherein is the pixel coordinate of a single pixel point in the pixel point cloud about the imaging of the binocular camera, is the distance from the real position of the pixel point to the optical center of the binocular camera, i.e. the depth data.

[0095] Further, the handheld crushing device 2 is provided with 4 high-reflectivity feature marker points that are not easily blocked, facilitating the identification and data collection of the AR display interaction device 1. It needs to be additionally explained that the pixel point cloud data The data information covering these high reflection feature points.

[0096] Step 1.2, extracting the pixel point cloud data of the work scene , converting into three-dimensional space coordinates in a unified subject field of view coordinate system , as shown in formula (1); the subject field of view coordinate system takes the binocular camera optical center as the origin, the x-axis is right along the horizontal of the AR monitoring and interaction device 1 lens, the y-axis is up along the vertical of the lens, and the z-axis is outward perpendicular to the lens.

[0097] Second step, the echo signal is obtained by scanning the target silicon block through the annular ultrasonic array probe 4, and the impurity type analysis is carried out after pre-amplification and envelope detection. Specifically:

[0098] Step 2.1, starting the handheld crushing device 2, and using the annular ultrasonic array probe 4 to emit ultrasonic waves to scan the target silicon block. When the sound wave propagates in the target silicon block, it will produce reflection and form echo pulse at the interface of different materials due to the difference of medium acoustic impedance. The annular ultrasonic array probe synchronously receives the echo signal.

[0099] The returned echo signal is pre-amplified by the amplifier and then envelope processed. First, the signal is converted into an analytical form using Hilbert transform as shown in formula (2) to obtain the envelope amplitude; then, the envelope detection as shown in formula (3) and the smoothing filter processing as shown in formula (4) are used to remove high-frequency oscillation and obtain the reflection intensity curve.

[0100] Further, L in formula (4) is the smoothing window length, which is 1-5ms, and 3ms is taken in this embodiment.

[0101] Step 2.2, for the processed envelope signal, the local maximum value is obtained, and the of each peak corresponds to a reflection interface, and the intensity is , as shown in formula (5).

[0102] The envelope amplitude of the peak is normalized to estimate the reflection intensity as shown in formula (6), and the impurity is judged:

[0103] Further, the in formula (6) is the normalized reflection intensity coefficient of the reflection interface at the i th peak, and the preset threshold range is 0.6-0.8, and 0.7 is taken in this embodiment. is the reference interface standard amplitude, and 2.0 is taken in this embodiment.

[0104] Further, the suspicious impurity existing reflection interface is determined by two ways: (1) when the reflected wave of the ultrasonic wave is in phase with the incident wave, the reflection interface is considered to be a high-resistance impurity composed of Al-Si-Fe and other impurity agglomerates; (2) when the reflected wave of the ultrasonic wave is out of phase with the incident wave, the reflection interface is considered to be a low-resistance impurity composed of bubbles, cracks and other gas cavities.

[0105] Third step, solving the ultrasonic three-dimensional space coordinates of the impurity points scanned by the ring-shaped ultrasonic array probe 4. Specifically:

[0106] The ultrasonic three-dimensional space coordinates are three-dimensional space coordinates in the handheld breaking device 2 main body coordinate system.

[0107] Step 3.1, using the time of arrival of the ring-shaped ultrasonic array probe unit to the same point to solve the distance of each ring-shaped ultrasonic array probe unit to the impurity point , as shown in formula (7).

[0108] Further, in formula (7), c is the propagation speed of ultrasonic wave in the target silicon block, and the specific embodiment takes 8430 m / s; N is the number of ring-shaped ultrasonic array probe units, and the specific embodiment takes 12.

[0109] Step 3.2, using multi-lateral ranging and least square method for impurity point positioning, wherein the ring-shaped ultrasonic array probe unit coordinates are fixed coordinates , and the ultrasonic three-dimensional space coordinates of each impurity point are . The multi-lateral ranging equation is shown in formula (8).

[0110] Taking one ring-shaped ultrasonic probe unit as a reference ultrasonic probe unit, a linear equation is obtained, as shown in formula (9).

[0111] Further, a matrix is constructed for the remaining ring-shaped ultrasonic array probe units, as shown in formula (10).

[0112] Further, the ultrasonic three-dimensional space coordinates of the impurity points are solved by using the least square method, as shown in formula (13). .

[0113] Fourth step, analyzing the size and content of the impurity agglomerate in the target silicon block, and performing acousto-optic prompt type determination. Specifically:

[0114] Step 4.1, on the basis of the impurities determined in the third step, forming independent impurity agglomerate regions according to spatial clustering ​, and the size and content of the impurity agglomerates are analyzed. The size is approximately considered as the minimum circumscribed cuboid size of the impurity agglomerates, and the calculation idea is as follows: first, the covariance matrix of the impurity points is calculated as shown in formula (14); second, the covariance matrix is decomposed as shown in formula (15); and finally, the difference of the vertex coordinates of the minimum circumscribed cuboid model in the axial direction is calculated as shown in formula (16) to output the actual size of the impurity agglomerates. The impurity content in the target silicon block is approximately obtained by dividing the total minimum circumscribed cuboid volume of the impurity agglomerates in the scanning area by the volume of the target silicon block in the scanning area.

[0115] Step 4.2, according to the size and content difference of the impurity agglomerates, the specific determination threshold and the sound and light prompt type are divided, as follows:

[0116] If the size of the impurity agglomerates is greater than 10 mm, 10 mm, it is determined that there is a large impurity agglomerate in the region, and the warning system flashes yellow sound and light warning to remind the operator to perform crushing treatment and remove the impurities;

[0117] If the size of the largest impurity agglomerate is less than 10 mm, 10 mm, but the impurity content ratio is higher than 2%, the warning system flashes red sound and light warning to prompt the operator to dispose the target silicon block as waste;

[0118] The rest are determined as qualified materials, and the warning system maintains green light source without alarm.

[0119] Step 5, display the impurity virtual projection in the AR display interactive device 1. Specifically:

[0120] Step 5.1, the coordinates of each impurity point obtained in step 3.2 are rigidly registered and corrected by rotation and translation. The rotation matrix R and the translation vector t are calculated as follows: first, the pre-set fixed coordinate set of the high-reflective feature identification point in the handheld crushing device 2 system in step 1.1 is , The observation coordinate set measured by the AR display interactive device 1 in step 1.2 is , and the perspective-n-point positioning method (PnP algorithm) is used to solve the rotation matrix R and the translation vector t, with the minimum overall re-projection error as the optimization design target, as shown in formula (17).

[0121] Step 5.1.1, the specific calculation method is as follows: the center of the pre-set fixed coordinate set of the high-reflective feature identification point is calculated as shown in formula (18); and the center of the observation coordinate set of the high-reflective feature identification point is calculated as shown in formula (19). ​; the decentralized preset fixed coordinate set is obtained as shown in formula (20) ; the decentralized observed coordinate set is obtained as shown in formula (21) .

[0122] Step 5.1.2, the decentralized covariance matrix H is calculated as shown in formula (22), and the left singular vector matrix U, the singular value diagonal matrix and the right singular vector matrix are solved by singular value decomposition SVD as shown in formula (23).

[0123] Step 5.1.3, finally, the rotation matrix R is obtained as shown in formula (24), and the translation vector t is obtained as shown in formula (25).

[0124] Step 5.1.4, the impurity ultrasonic three-dimensional space coordinates obtained in step 3.2 can be converted into three-dimensional space coordinates under the AR display interactive device 1 main field of view coordinate system as shown in formula (26).

[0125] Step 5.2, the impurity three-dimensional space coordinates calculated in step 5.1 in the AR display interactive device 1 are projected and displayed, and the sound and light prompts realized in step 4.2 are matched, so that the impurity distribution in the target silicon ingot is displayed in a virtual-real combination, which can effectively realize efficient screening and crushing of the industrial silicon ingot by the operator.

[0126] Further, since the personnel operation angle is subjectively and dynamically adjusted, the rotation matrix R and the translation vector t in step 5.1 should be updated in real time with the personnel at a frequency of 30Hz, and at the same time, the display projection refresh in the AR display interactive device 1 is completed, the refresh frequency is 30Hz, and the update delay is controlled within 50ms.

[0127] Step 6, after the operation is completed, the AR display interactive device 1 generates a crushing report, and the power module enters a low-power mode. Specifically:

[0128] After the operator manually issues an operation completion instruction through the reserved button, the system extracts key data from the process log and outputs a crushing report in a preset format, and the core information includes basic operation information and quality judgment information, such as operation time, average impurity content, product qualification rate, and abnormal alarm information. After the operator views the output report on the AR display interactive device 1, the operator can manually close it, and then the report is stored in the system history record, and the AR display interactive device 1 enters a low-power mode.

[0129] An industrial silicon ingot breaking device based on impurity cluster ultrasonic identification, which realizes the above-mentioned industrial silicon ingot breaking method, the industrial silicon ingot breaking device includes AR display interactive equipment 1 and handheld breaking device 2 two core components, both of which can realize two-way data transmission through high anti-interference data line 3 or Bluetooth pairing. Wherein the coaxial integrated annular ultrasonic array probe 4 and detachable tungsten steel alloy hammer head 5 are designed at the gun head of the handheld breaking device, which can realize the identification detection and breaking of the impurity-containing target silicon block. Specifically:

[0130] The AR display interactive equipment 1 includes image perception module, control module, power module, display module and communication module. Among them, the image perception module is mainly binocular camera and depth sensor, which is used for scanning and recording the working image in the field of view, identifying the position and attitude information of the handheld breaking device 2; the control module includes image processor and power supply management unit, which is used for receiving the internal impurity signal obtained by scanning the target silicon block by the annular ultrasonic array probe 4, and identifying and analyzing the signal data to complete the three-dimensional space coordinate conversion of the impurity position display projection. At the same time, the image of the qualified target silicon block that has been scanned and broken is recorded and stored into the working history data for subsequent statistics and report generation; the display module is mainly a semi-transparent and semi-reflective silicon-based micro organic light emitting diode (OLED) display screen, which is used for real-time display of the internal impurity distribution of the scanned target silicon block in the field of view of the wearer; the power module provides stable power supply for the AR display interactive equipment 1, and supports entering low-power mode after the whole operation is completed; the communication module supports two transmission modes, including Bluetooth pairing and data line transmission, which can be flexibly switched.

[0131] The handheld breaking device 2 includes ultrasonic detection module, early warning module, communication module, breaking execution module and holding damping module. Among them, the ultrasonic detection module is used for scanning the target silicon block and collecting its internal acoustic echo signal; the early warning module is used for real-time processing and analyzing the acoustic data monitored by the ultrasonic detection module, judging whether there is impurity cluster distribution in the target silicon block, and having sound and light alarm function; the communication module is used for transmitting the analyzed data and related parameters to the AR display interactive equipment 1; the breaking execution module is mainly a servo driver and a detachable tungsten steel alloy impact breaking hammer, the servo driver drives the hammer head 5 to break according to adjustable working parameters, and the type and size of the hammer head can be replaced and adjusted according to actual construction requirements; the holding damping module wraps the handle with a non-slip rubber sleeve, and the inside of the equipment is provided with damping springs to absorb high-frequency vibration generated in the breaking process, improving the comfort and stability of use.

[0132] The above described embodiments only express the implementation ways of the present application, but cannot be understood as the limitation to the scope of the present application. It should be pointed out that, for the person skilled in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. An industrial silicon ingot breaking method based on ultrasonic recognition of impurity agglomerates, characterized by, The industrial silicon ingot breaking method comprises the following steps: In the first step, an operator wears an AR display interaction device and carries a handheld breaking device into a work area to collect scene images and depth data of the work area and convert them into three-dimensional space coordinates; In the second step, echo signals are obtained by scanning the target silicon block with a ring-shaped ultrasonic array probe, and impurity type analysis is performed after pre-amplification and envelope detection; In the third step, the ultrasonic three-dimensional space coordinates of the impurity points obtained by scanning with the ring-shaped ultrasonic array probe are solved; the ultrasonic three-dimensional space coordinates are three-dimensional space coordinates in the handheld breaking device main coordinate system; In the fourth step, the size and content of the impurity agglomerates in the target silicon block are analyzed, and an acousto-optic prompt type is determined; In the fifth step, a virtual projection of the impurities is displayed in the AR display interaction device; In the sixth step, after the work is completed, the AR display interaction device generates a breaking report, and the power module enters a low-power mode.

2. The method according to claim 1, wherein the method is characterized by, The first step specifically comprises: Step 1.1, forming scene image and pixel point cloud data by scanning the working area through the binocular camera and depth sensor of the AR display interaction device wherein is the pixel coordinate of a single pixel point in the pixel point cloud with respect to the imaging of the binocular camera, is the distance from the real position of the pixel point to the optical center of the binocular camera, i.e. the depth data; Step 1.2, extract the pixel point cloud data of the operation scene , convert to three-dimensional space coordinates in a unified subject field of view coordinate system ; the subject field of view coordinate system takes the optical center of the binocular camera as the origin, the x axis is to the right along the horizontal direction of the lens of the AR monitoring and interaction device 1, the y axis is vertically upward along the lens, and the z axis is outward perpendicular to the lens. （1） In the formula: is the three-dimensional space coordinate of the target silicon block under the main field of view coordinate system; is the coordinate of the image center; , is the horizontal focal length and the vertical focal length; is the rotation matrix, is the translation vector, both of which are binocular camera built-in parameters; The superscript T represents transposition; The handheld breaking device is provided with four high-reflection feature marker points for recognition and data collection by the AR display interaction device.

3. The method according to claim 2, wherein the method is characterized by, The second step specifically comprises: In step 2.1, the handheld breaking device is started, an ultrasonic wave is emitted to scan the target silicon block by using the ring-shaped ultrasonic array probe, and echo signals are received synchronously by using the ring-shaped ultrasonic array probe; The echo signals are pre-amplified and envelope-processed; the signal is converted into an analytical form by using Hilbert transform according to formula (2) to obtain an envelope amplitude; a reflection intensity curve is obtained by envelope detection according to formula (3) and smoothing filter processing according to formula (4); （2） （3） （4） wherein, is the pre-amplified received channel signal; is the Hilbert transformed signal; is the analytic signal; is the envelope signal; is the smoothed envelope signal; j is the imaginary unit; L is the smoothing window length; t is the time variable representing the time of the ultrasound signal propagation within the target silicon piece; is the integral variable; Step 2.2, for the processed envelope signal, find its local maximum, and the value at each peak is corresponding to a reflection interface, whose strength is ; as shown in equation (5): （5） wherein, denotes a set of local maxima of the envelope signal; denotes a time position corresponding to the i-th peak, denotes a round-trip time of the ultrasound wave propagating to the reflecting interface; denotes an envelope amplitude of the i-th peak, denotes a reflection intensity of the reflecting interface; , denote envelope amplitudes of the neighboring peaks before and after the i-th peak, respectively; The reflection intensity is estimated after the envelope amplitude of the peak value is normalized, and impurity judgment is performed: （6） In the formula, is the normalized reflection intensity coefficient of the reflection interface at the i th peak, if If the normalized reflection intensity coefficient exceeds a preset threshold, it is judged that the reflection interface has a suspicious impurity; is the envelope amplitude of the i th peak; is the reference interface standard amplitude.

4. The method according to claim 3, wherein the method is characterized by, In step 2.2, the preset threshold range is 0.6-0.8; The reflection interface with suspicious impurities has two types of impurity type discrimination methods: (1) when the reflection wave of the ultrasonic wave is in phase with the incident wave, it is considered that the reflection interface is a high-resistance impurity composed of impurity agglomerates; (2) when the reflection wave of the ultrasonic wave is out of phase with the incident wave, it is considered that the reflection interface is a low-resistance in impurity composed of a gas cavity. The third step specifically comprises:

5. The method according to claim 4, wherein the method is characterized by, Take one ring-shaped ultrasonic probe unit as a reference ultrasonic probe unit to obtain a linearization equation as shown in formula (9): Step 3.1, time of arrival of the same point by the ring ultrasonic array probe units solving the distance of each ring ultrasonic array probe unit to the impurity point ; （7） In the formula, c is the propagation speed of ultrasonic waves inside the target silicon block; n is the annular ultrasonic array probe unit number; N is the number of annular ultrasonic array probe units; is the nth annular ultrasonic array probe unit, which is the echo time from transmitting ultrasonic waves to receiving echo signals; is the start time of transmitting ultrasonic waves; Step 3.2, impurity point positioning by multilateration and least square method, wherein the annular ultrasonic array probe unit coordinates are set fixed coordinates , and the ultrasonic three-dimensional space coordinates of each impurity point are ; the multilateration equation is established as shown in formula (8): （8） A matrix is constructed for the remaining ring-shaped ultrasonic array probe units as shown in formula (10): （9） wherein denotes the fixed coordinate set of the reference ultrasound probe unit; d1 denotes the distance of the reference ultrasound probe unit to the impurity point; the superscript T denotes the transpose; The fourth step specifically comprises: （10） （11） （12） wherein is a coefficient matrix; is a three-dimensional spatial coordinate of the impurity point; is an observation vector; the superscript T denotes transposition; Solving ultrasonic three-dimensional coordinates of impurity points by least square method As shown in equation (13): （13）。 6. The method according to claim 5, wherein the method is characterized by, The covariance matrix of the impurity points is calculated, the covariance matrix is subjected to eigenvalue decomposition, the coordinate difference of the vertex of the minimum circumscribed cuboid model in the axial direction is calculated, and the actual size of the impurity agglomerate is output; Step 4.1, on the basis of the impurities determined in the third step, grouping them into separate impurity clusters according to the space , performing the size and content analysis of the impurity clusters thereof; The impurity content in the target silicon block is approximately obtained by dividing the sum of the minimum circumscribed cuboid volumes of all the impurity agglomerates in the scanning region by the volume of the target silicon block in the scanning region; In step 4.2, specific determination thresholds and acousto-optic prompt types are divided according to the size and content differences of the impurity agglomerates, and the specific conditions are as follows: （14） （15） （16） wherein: is a covariance matrix; N is the number of impurity points contained in the impurity cluster; is the set of all impurity points contained in the kth impurity cluster; is the centroid of the impurity cluster; is the principal eigenvector; is the eigenvalue; is the minimum circumscribed cuboid size of the impurity cluster, wherein is the corresponding is the length, width, and height of the minimum circumscribed cuboid, respectively; In the remaining cases, the material is determined to be qualified, the warning system maintains a green light source, and no alarm is given. If the size of the impurity cluster is 10 mm, it is determined that there is a large impurity cluster in the area, the pre-warning system issues a flashing yellow sound and light warning, the crushing process is carried out, and the impurities are removed; If the maximum impurity agglomerate size 10 mm, but the impurity content ratio is higher than 2%, the pre-warning system sends a flashing red audible and visual warning, and the target silicon block is treated as waste material. The fifth step specifically comprises:

7. The method according to claim 6, wherein the method is characterized by, Step 5.1 specifically comprises: Step 5.1, obtain the coordinates of each impurity point from step 3.2 The rigid body registration and coordinate correction are performed by rotation and translation. The rotation matrix R and the translation vector t are calculated as follows: The preset fixed coordinate set of the high-reflection feature marker point under the handheld crushing device system is , , the observation coordinate set measured by the AR display interaction device according to step 1.2 is , the rotation matrix R and the translation vector t are solved, and the overall re-projection error is minimized as the optimization design target, as shown in formula (17): （17） Step 5.2, impurity three-dimensional space coordinates in AR display interaction device The projection display is carried out, and the impurity distribution in the target silicon ingot is displayed in a virtual-real combined manner in combination with the sound-light prompt realized in step 4.2, so that the screening and crushing of the industrial silicon ingot are realized.

8. The method according to claim 7, wherein the method is characterized by, Step 5.1.3, a rotation matrix R and a translation vector t are obtained; Step 5.1.1, calculating the high-reflective feature landmark preset fixed coordinate set center of high-reflective feature landmark observation coordinate set center of , obtaining the decentralized preset fixed coordinate set and the decentralized observation coordinate set ; （18） （19） （20） （21） wherein, corresponding four high-reflection signature points; Step 5.1.2, compute the whitening covariance matrix H by singular value decomposition (SVD) to solve for the left singular vector matrix U, the singular value diagonal matrix and the right singular vector matrix ; （22） （23） ​ （24） （25） wherein is a determinant operator; R is a rotation matrix; t is a translation vector; Step 5.1.4, convert the ultrasonic three-dimensional space coordinates of the obtained in step 3.2 into three-dimensional space coordinates under the AR display interactive device main field of view coordinate system Step 5.1.4, convert the ultrasonic three-dimensional space coordinates of the obtained in step 3.2 into three-dimensional space coordinates under the AR display interactive device main field of view coordinate system As shown in formula (26): （26） In the step 5.1, the rotation matrix R and the translation vector t are dynamically updated in real time at a frequency of 30 Hz, while completing the display projection refresh in the AR display interactive device, the refresh frequency is 30 Hz, and the update delay is controlled within 50 ms.

9. The method according to claim 8, wherein the method is characterized by, The sixth step is specifically: After issuing the job completion instruction, key data is extracted from the process log, and a fragmentation report is output in a preset format, with core information including basic job information and quality judgment information; after viewing the output report on the AR display interactive device, the AR display interactive device enters a low-power mode.

10. An industrial silicon ingot breaking apparatus based on ultrasonic recognition of impurity agglomerates, characterized by, The industrial silicon ingot fragmentation method of any one of claims 1-9 is realized by an industrial silicon ingot fragmentation device, which includes an AR display interactive device and a handheld fragmentation device, both of which can perform bidirectional data transmission; a coaxial integrated annular ultrasonic array probe and a detachable hammer head are designed at the gun head of the handheld fragmentation device to realize identification and fragmentation of impurity-containing target silicon blocks; specifically: The AR display interactive device includes an image perception module, a control module, a power module, a display module, and a communication module; the image perception module is a binocular camera and a depth sensor, which is used to scan and record the working image in the field of view, and identify the position and attitude information of the handheld fragmentation device; the control module includes an image processor and a power supply management unit, which is used to receive the internal impurity signal obtained by scanning the target silicon block by the annular ultrasonic array probe, and to identify and analyze the signal data to complete the three-dimensional space coordinate conversion of the impurity position display projection; at the same time, the image of the qualified target silicon block that has been scanned and fragmented is recorded; the communication module supports Bluetooth pairing and data line transmission; The handheld fragmentation device includes an ultrasonic detection module, a warning module, a communication module, a fragmentation execution module, and a holding damping module; the ultrasonic detection module is used to scan the target silicon block and collect its internal acoustic echo signal; the warning module is used to analyze the acoustic data monitored by the ultrasonic detection module in real time, to determine whether there are impurity agglomerates in the target silicon block, and to have an audible and visual alarm function; the communication module is used to transmit the analyzed data and related parameters to the AR display interactive device; the fragmentation execution module is a servo driver and a detachable impact fragmentation hammer; the holding damping module is provided with a damping spring.

Images