WO2024103871A1 - 一种多晶硅监控方法、装置及相关设备 - Google Patents

一种多晶硅监控方法、装置及相关设备 Download PDF

Info

Publication number
WO2024103871A1
WO2024103871A1 PCT/CN2023/113706 CN2023113706W WO2024103871A1 WO 2024103871 A1 WO2024103871 A1 WO 2024103871A1 CN 2023113706 W CN2023113706 W CN 2023113706W WO 2024103871 A1 WO2024103871 A1 WO 2024103871A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
pixel
reactor
silicon rod
parameter
Prior art date
Application number
PCT/CN2023/113706
Other languages
English (en)
French (fr)
Inventor
银波
刘兴平
侯雨
刘丹丹
宋亚博
Original Assignee
新特能源股份有限公司
新特硅基新材料有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 新特能源股份有限公司, 新特硅基新材料有限公司 filed Critical 新特能源股份有限公司
Publication of WO2024103871A1 publication Critical patent/WO2024103871A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/94—Hardware or software architectures specially adapted for image or video understanding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/18—Status alarms
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
    • G08B21/18—Status alarms
    • G08B21/24—Reminder alarms, e.g. anti-loss alarms
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Definitions

  • the present application relates to the technical field of polysilicon production, and in particular to a polysilicon monitoring method, device and related equipment.
  • the polysilicon reduction process is a key link in the production process.
  • the reduction workshop includes multiple reactors. In order to timely monitor the reaction conditions in each of the multiple reactors, on-site personnel are required to conduct regular inspections of the reactors to monitor whether there are abnormal problems with silicon rods such as atomization, black spots, and bright spots in the reactors.
  • the purpose of the embodiments of the present application is to provide a polysilicon monitoring method, device and related equipment to solve the problem of poor monitoring effect in the process of monitoring abnormal conditions of silicon rods in related technologies.
  • an embodiment of the present application provides a polysilicon monitoring method, the method comprising:
  • the alarm condition includes at least one of the following:
  • the number parameter of the black spot areas in the monitoring image is greater than or equal to a first threshold
  • the area parameter of the black spot area in the monitoring image is greater than or equal to a second threshold
  • the first similarity parameter corresponding to the monitoring image is less than or equal to a third threshold value, the first similarity parameter being used to characterize the image similarity between the monitoring image and a reference image, the reference image being an image captured in the reactor in a non-atomized state;
  • the monitoring image includes highlighted pixels located within a preset interval, and the preset interval is used to characterize a pixel interval corresponding to a first position in the monitoring image, wherein the first position is a position in the reactor for placing silicon rods, and the highlighted pixels are pixels in the monitoring image whose corresponding pixel values are greater than a pixel threshold.
  • performing image analysis on the surveillance image to obtain an analysis result includes:
  • a number parameter of black spot regions in the monitoring image and an area parameter of black spot regions in the monitoring image are determined.
  • the method further includes:
  • the method further includes:
  • the input amount of the input raw materials of the reaction furnace is adjusted.
  • the method further includes:
  • a fogging alarm is generated.
  • the atomization alarm is used to indicate the position of the reactor.
  • the method before detecting the reference light signal emitted by the reference device in the reaction furnace at the current moment and obtaining the detection signal data, the method further includes:
  • the reference signal data is updated according to the test signal data.
  • the method before detecting the reference light signal emitted by the reference device in the reaction furnace at the current moment and obtaining the detection signal data, the method further includes:
  • a detection instruction is sent to the reference device, so that the reference device emits the reference light signal based on the detection instruction.
  • the method further includes:
  • the first similarity parameter is less than or equal to a third threshold value
  • at least two second similarity parameters are obtained according to the monitoring image and at least two fog image groups, wherein the at least two second similarity parameters correspond to the at least two fog image groups one by one, wherein the at least two fog image groups respectively indicate different degrees of fogging of the reactor, and the second similarity parameters are used to characterize the image similarity between the corresponding fog image group and the monitoring image;
  • the second similarity parameter corresponding to the target fogged image group is a similarity parameter with the largest value among the at least two second similarity parameters
  • the method further includes:
  • the reaction parameters include at least one of a current value, a ratio of a hydrogen input amount to a trichlorosilane input amount.
  • a pixel spacing is determined, wherein the pixel spacing is used to characterize the corresponding the shortest distance between a pixel region of a first silicon rod and a corresponding pixel region of a second silicon rod, wherein the first silicon rod and the second silicon rod are two silicon rods disposed adjacent to each other in the reaction furnace;
  • Correcting the pixel spacing according to a correction parameter to obtain a first spacing wherein the correction parameter is used to characterize a physical length corresponding to a unit pixel block in the monitoring image, and the first spacing is used to characterize a spacing between the first silicon rod and the second silicon rod at a current moment;
  • a target parameter is determined according to the silicon rod spacing and the initial spacing, wherein the initial spacing is used to characterize the spacing between the first silicon rod and the second silicon rod at an initial moment, the initial moment is the start-up moment of the reactor, and the target parameter is used to characterize the diameter of the silicon rod in the reactor at the current moment.
  • determining the pixel spacing according to the binary image includes:
  • first pixel point set is a set of pixel points corresponding to an edge portion of the first silicon rod close to the second silicon rod
  • second pixel point set is a set of pixel points corresponding to an edge portion of the second silicon rod close to the first silicon rod
  • a pixel spacing is determined according to the first pixel point set and the second pixel point set.
  • determining the pixel spacing according to the first pixel point set and the second pixel point set includes:
  • the average of the point spacings corresponding to the plurality of pixel heights is determined as the pixel spacing.
  • the method before correcting the pixel spacing according to the correction parameter to obtain the first spacing, the method further includes:
  • the correction parameter is determined according to the silicon core reference spacing and the silicon core target spacing, wherein the silicon core reference spacing is used to characterize the physical distance between the first silicon rod silicon core and the second silicon rod silicon core. distance.
  • the initial spacing is determined based on a first reference image acquired at an initial moment, and a memory space occupied by the first reference image and a memory space occupied by the monitoring image are both greater than or equal to a memory threshold.
  • the method further includes:
  • a connected domain analysis is performed on the silicon rod image included in the monitoring image to determine N gap blocks, where the gap blocks are used to indicate gaps between multiple silicon particles on the surface of the silicon rod, and N is a positive integer;
  • the quality information of the silicon rods in the reactor is determined according to the block areas of the M gap blocks, wherein the M gap blocks are the first M gap blocks arranged in descending order of block areas among the N gap blocks.
  • the acquiring of the monitoring image of the reactor includes:
  • a monitoring image of the reactor is acquired.
  • the method further includes:
  • the pixel values of the highlighted pixel point are respectively compared with the pixel values of a plurality of reference points for similarity to obtain a plurality of third similarity parameters, wherein the plurality of third similarity parameters correspond to the plurality of reference points one by one, and each of the plurality of reference points corresponds to a silicon rod bright spot in a temperature interval;
  • the bright spot information of the target point the bright spot information of the highlighted pixel point located in the preset interval is determined, wherein the bright spot information includes the temperature information of the corresponding silicon rod bright spot and the processing flow information of the corresponding silicon rod bright spot.
  • the method further includes:
  • the highlighted The pixel is stored as a new reference point.
  • the method further includes:
  • gas leakage information is output.
  • the method further includes:
  • the operating status information includes the start and stop data of the reactor
  • the first inspection route includes P inspection points arranged in order, where P is an integer greater than 1, and each inspection point corresponds to a reactor in a startup state;
  • the inspection robot is controlled to stop in sequence between the P points to be inspected.
  • the method further includes:
  • the first inspection route overlaps with the obstacle area, determining a second inspection route according to the obstacle area, the initial inspection route and the running status information, wherein the second inspection route includes at least part of the P points to be inspected, and the second inspection route does not include the obstacle area, wherein the obstacle area is an area corresponding to an obstacle that hinders the movement of the inspection robot;
  • the inspection robot is controlled to continue inspection.
  • the obstacle area is determined according to the moving speed of the obstacle and/or the volume of the obstacle.
  • an embodiment of the present application further provides a polysilicon monitoring device, the device comprising:
  • An image acquisition module used for acquiring monitoring images of the reactor
  • An image analysis module used to perform image analysis on the monitoring image to obtain analysis results
  • An alarm module is used to output an alarm message when the analysis result meets the alarm condition
  • the alarm condition includes at least one of the following:
  • the number parameter of the black spot areas in the monitoring image is greater than or equal to a first threshold
  • the area parameter of the black spot area in the monitoring image is greater than or equal to a second threshold
  • the first similarity parameter corresponding to the monitoring image is less than or equal to a third threshold value, the first similarity parameter being used to characterize the image similarity between the monitoring image and a reference image, the reference image being an image captured in the reactor in a non-atomized state;
  • the monitoring image includes highlighted pixels located within a preset interval, and the preset interval is used to characterize a pixel interval corresponding to a first position in the monitoring image, wherein the first position is a position in the reactor for placing silicon rods, and the highlighted pixels are pixels in the monitoring image whose corresponding pixel values are greater than a pixel threshold.
  • an embodiment of the present application further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the above-mentioned polysilicon monitoring method when executed by the processor.
  • an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned polysilicon monitoring method are implemented.
  • the black spots on silicon rods, reactor atomization, and bright spots on silicon rods that lead to unqualified silicon rods in the silicon rod production process are quantitatively monitored, so that when the analysis results meet the alarm conditions, alarm information is output in time to remind the inspection personnel to deal with the abnormal production conditions of the reactor, reduce the output rate of unqualified silicon rods, and improve the monitoring effect of silicon rods.
  • FIG1 is a schematic flow chart of a polysilicon monitoring method provided in an embodiment of the present application.
  • FIG2 is a schematic diagram of a plurality of silicon rods in a reactor provided in an embodiment of the present application
  • FIG3 is a schematic diagram of a first silicon rod and a second silicon rod provided in an embodiment of the present application
  • FIG4 is a schematic diagram of a monitoring image after binarization provided in an embodiment of the present application.
  • FIG5 is a schematic structural diagram of a polysilicon monitoring device provided in an embodiment of the present application.
  • FIG. 6 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application.
  • FIG. 1 is a flow chart of a polysilicon monitoring method provided by the embodiment of the present application. As shown in FIG. 1 , the method includes the following steps:
  • Step 101 Acquire a monitoring image of a reactor.
  • the monitoring image includes an image of a first position, where the first position is a position in the reactor for placing silicon rods.
  • the monitoring image may be an image of the interior of the furnace (referring to a color image) captured by an image acquisition device through the sight glass of the silicon rod reactor.
  • the image acquisition device may be a fixed camera arranged facing the sight glass of the silicon rod reactor, or may be a camera mounted on an inspection robot (the inspection robot periodically inspects multiple reactors in the reaction workshop according to a preset program).
  • the monitoring image acquisition frequency may be once per second, once per five seconds, or once per minute, etc.
  • the user may adaptively select the monitoring image acquisition frequency according to actual needs, and the embodiment of the present application does not limit this.
  • the reactor is provided with a plurality of grooves for inserting silicon rods and silicon cores.
  • the first position is used to indicate the position of the groove, that is, the lower end position of the silicon rod extending out of the groove after the silicon rod and silicon core are inserted into the groove.
  • Step 102 Perform image analysis on the monitoring image to obtain analysis results.
  • the analysis result includes at least a number parameter of black spot areas in the monitoring image, an area parameter of black spot areas in the monitoring image, a first similarity parameter corresponding to the monitoring image, and data such as highlight pixels included in the monitoring image.
  • the first similarity parameter is used to characterize the image similarity between the monitoring image and the reference image
  • the reference image is an image captured in the reactor in a non-atomized state
  • the highlighted pixel point is a pixel point in the monitoring image whose corresponding pixel value is greater than a pixel threshold.
  • the monitoring image can be analyzed to distinguish the background image blocks, the normal silicon rod image blocks, and the silicon rod black spot image blocks included in the monitoring image, and the number parameter and area parameter corresponding to the silicon rod black spot image blocks are determined based on the distinction result, wherein the number parameter and area parameter corresponding to the silicon rod black spot image blocks are
  • the number i.e., the parameter of the number of black spot areas in the monitoring image
  • the area parameter corresponding to the silicon rod black spot block i.e., the area parameter of the black spot area in the monitoring image
  • the block area of each silicon rod black spot block in the multiple silicon rod black spot blocks included in the monitoring image can be understood as the block area of each silicon rod black spot block in the multiple silicon rod black spot blocks included in the monitoring image.
  • the image analysis of the surveillance image may be completed based on a connected domain algorithm
  • the connected domain algorithm may be a two-pass scanning method or a seed-filling method.
  • the process of acquiring the highlight pixel point may be: performing grayscale processing on the monitoring image to obtain a grayscale image; and then performing binarization processing on the grayscale image to obtain the highlight pixel point.
  • the maximum value method can be used for image grayscale processing
  • the average value method can be used for image grayscale processing
  • the weighted average value method can be used for image grayscale processing.
  • the embodiment of the present application does not limit the specific algorithm used for image grayscale processing.
  • a global threshold method can be applied (that is, the pixel value of each pixel in the grayscale image is compared with the aforementioned pixel threshold.
  • the pixel threshold compared with each pixel in the grayscale image is the same), or a local adaptive dynamic threshold algorithm can be applied (that is, the pixel threshold to be compared for each pixel is dynamically calculated.
  • the pixel threshold compared with each pixel in the grayscale image can be the same or different).
  • the embodiments of the present application do not limit this.
  • Step 103 When the analysis result meets the alarm condition, output alarm information.
  • the alarm condition includes at least one of the following:
  • the number parameter of the black spot areas in the monitoring image is greater than or equal to a first threshold
  • the area parameter of the black spot area in the monitoring image is greater than or equal to a second threshold
  • the first similarity parameter corresponding to the monitoring image is less than or equal to a third threshold
  • the highlighted pixel points included in the monitoring image are located in a preset interval, and the preset interval is used to represent a pixel interval corresponding to a first position in the monitoring image, and the first position is a position in the reactor for placing silicon rods.
  • the abnormal growth of silicon rods caused by excessive spread of black spots in the silicon rods can be quantified.
  • the silicon rods with abnormal growth and/or the silicon rods with too large black spots are identified as the silicon rods with abnormal growth, and an alarm message is generated accordingly to remind the staff in the reduction workshop to deal with the silicon rods with abnormal growth in time (such as increasing the amount of hydrogen in the reactor, or reducing the temperature in the reactor, etc.) to avoid further deterioration of the black spot problem in the reactor.
  • the above-mentioned area parameter can be understood as the block area of the black spot area with the largest block area among the multiple black spot areas included in the monitoring image, and can also be understood as the sum of the block areas of the multiple black spot areas included in the monitoring image.
  • the user can determine the reference meaning of the above-mentioned area parameter according to actual needs, and the embodiments of the present application do not limit this.
  • the aforementioned area parameter may be 25 or 45.
  • the image similarity between the monitoring image and the reference image can be compared, that is, a first similarity parameter is determined, and the first similarity parameter is compared with a third threshold value to determine whether atomization occurs in the reactor.
  • the highlighted pixel points within the preset interval can be called abnormal points.
  • the abnormal points are marked on the monitoring image to facilitate users (such as inspection personnel in the workshop where the reactor is located) to locate the abnormal points and speed up the handling efficiency of the abnormal points (i.e., silicon rod bright spots).
  • the bright spot of the silicon rod can be understood as an abnormally high temperature spot caused by production failure at the lower end of the silicon rod (that is, the insertion position of the silicon rod in the reactor).
  • This abnormally high temperature spot will appear in the field of vision as a bright light spot. If the abnormally high temperature spot is not identified and dealt with in time, the lower end of the silicon rod will experience high-temperature melting, which will cause the silicon rod to fall and bring about more serious economic losses.
  • multiple highlighted pixel points in addition to applying a preset interval to screen multiple highlighted pixel points, multiple highlighted pixel points can also be screened by applying a quantity threshold, that is, among multiple highlighted pixel points, multiple adjacent highlighted pixel points are determined as a candidate point group, and the candidate point group in which the number of highlighted pixel points is less than or equal to the quantity threshold is filtered out to further improve the accuracy of the determined abnormal points.
  • a quantity threshold that is, among multiple highlighted pixel points, multiple adjacent highlighted pixel points are determined as a candidate point group, and the candidate point group in which the number of highlighted pixel points is less than or equal to the quantity threshold is filtered out to further improve the accuracy of the determined abnormal points.
  • a plurality of highlighted pixel points may be preliminarily screened based on a quantity threshold, and then a preset interval may be applied to further screen the plurality of highlighted pixel points after the preliminarily screening to finally determine the abnormal points, which can significantly improve the accuracy of the determined abnormalities.
  • the inspection robot can return to the reactor sight glass indicated by the aforementioned alarm message to re-inspect the silicon rods in the reactor. If the re-inspection passes, the periodic inspection mode is restored. If the re-inspection fails, the alarm message is generated again.
  • the output alarm information includes at least one of a black spot problem, a fogging problem and a silicon rod bright spot problem.
  • the black spots on silicon rods, reactor atomization, and bright spots on silicon rods that lead to unqualified silicon rods in the silicon rod production process are quantitatively monitored, so that when the analysis results meet the alarm conditions, alarm information is output in time to remind the inspection personnel to deal with the abnormal production conditions of the reactor, reduce the output rate of unqualified silicon rods, and improve the monitoring effect of silicon rods.
  • performing image analysis on the surveillance image to obtain an analysis result includes:
  • a number parameter of black spot regions in the monitoring image and an area parameter of black spot regions in the monitoring image are determined.
  • grayscale processing of the monitoring image is performed, on the one hand, to facilitate the subsequent connected domain analysis process and ensure the accuracy of the connected domain analysis; on the other hand, it is to filter out the background noise in the monitoring image.
  • the process of performing connected domain analysis on the silicon rod image included in the grayscale image to determine at least one black spot block included in the silicon rod image may be:
  • At least one black spot block included in the silicon rod image is determined.
  • the silicon rod image included in the grayscale image is first determined to avoid interference caused by background blocks and improve the extraction efficiency and accuracy of black spot blocks.
  • the process of performing inverse binarization processing on the silicon rod image included in the grayscale image may be to compare the pixel value of each pixel block in the silicon rod image included in the grayscale image with a pixel threshold, and set the pixel value of the pixel block greater than the pixel threshold to 0 (i.e., black), and set the pixel value of the pixel block less than or equal to the pixel threshold to 255 (i.e., white).
  • the number parameter of the black spot area in the monitoring image can be obtained accordingly, and then the area parameter of the black spot area in the monitoring image can be determined by counting the number of pixel blocks included in each black spot block.
  • the method further includes:
  • reaction furnaces there are multiple reaction furnaces in the reduction workshop, that is, there are multiple silicon rod reaction furnaces.
  • at least one reference device is arranged in each reaction furnace.
  • the reference device may be an infrared emitter disposed in the reaction furnace.
  • At least one inspection robot is arranged in the reduction workshop, and the inspection robot inspects the multiple reactors in the reduction workshop in turn according to the preset inspection route.
  • the inspection robot moves to the target inspection position (i.e. the position where the signal receiving component carried by the inspection robot is relative to the sight glass of the reactor)
  • the inspection robot will receive the reference light signal emitted by the reference device in the reactor at the current moment, and form detection signal data based on the actually received light signal, and the detection signal data at least includes the light energy of the light signal received by the inspection robot.
  • a central control server is also provided in the restoration workshop.
  • the central control server is connected to the inspection robot in communication and constitutes an inspection system. People will upload multiple detection signal data detected during the inspection process to the central control server for storage and data processing.
  • the reference signal data at least includes the light energy when the reference light signal is emitted.
  • the light energy of the reference light signal will gradually attenuate due to the interference of the silicon powder in the reactor, and as the dispersion degree of the silicon powder in the reactor increases, the light energy attenuation amplitude of the reference light signal will also increase. Therefore, by utilizing the characteristic that the light energy attenuation amplitude of the reference light signal is positively correlated with the dispersion degree of the silicon powder in the reactor, the dispersion degree of the silicon powder in the reactor can be indirectly determined by calculating the light attenuation parameters.
  • the above-mentioned light attenuation parameter determination operation can be performed by the inspection robot or by the central control server.
  • the central control server can be used to calculate the light attenuation parameter to reduce the energy consumption of the inspection robot and extend the service life of the inspection robot.
  • the light attenuation parameter When the light attenuation parameter is greater than or equal to the light attenuation threshold, it indicates that the silicon powder dispersion level in the reactor exceeds the critical standard. At this time, it can be determined that there is atomization in the reactor, that is, atomization information is output to guide inspection personnel to deal with the reactor with atomization problems in a timely manner.
  • the method further includes:
  • the input amount of the input raw materials of the reaction furnace is adjusted.
  • the conversion coefficient should be understood as the correlation coefficient between the light energy attenuation amplitude of the reference light signal and the decrease in the input amount of silicon material.
  • the conversion coefficient can be obtained by fitting the correlation between the decrease in the input amount of silicon material and the light energy attenuation amplitude under different atomization conditions; the aforementioned silicon material can be understood as trichlorosilane.
  • the process of adjusting the input amount of the input raw material of the reactor may be: based on the decrease in the input amount of the silicon material indicated by the adjustment parameter, the input amount of the silicon material of the reactor is reduced, for example, the input amount of trichlorosilane in the reactor is reduced by 200 kg/h.
  • the above adjustment operation can be performed by the central control server sending a first atomization control instruction to the inspection robot, which is executed by the inspection robot; the central control server can also send a second atomization control instruction to the inspection robot.
  • the chemical control instructions are given to the control component of the input raw materials of the reactor, and the execution is completed through the control component.
  • the light attenuation parameter in the reactor at the current moment can be determined.
  • the light attenuation parameter is greater than or equal to the light attenuation threshold, it can be determined that there is an atomization condition in the reactor.
  • the obtained light attenuation parameter is processed by applying the conversion coefficient to determine the adjustment parameter for adjusting the input amount of the input raw material of the reactor, thereby suppressing the atomization condition in the reactor.
  • the method further includes:
  • a fogging alarm is generated, and the fogging alarm is used to indicate the position of the reaction furnace.
  • the inspection robot after determining that there is an atomization condition in the reactor and disposing of the atomization device in the reactor based on the preset logic, that is, after adjusting the input amount of the input raw materials of the reactor according to the adjustment parameters, in order to observe whether the control means effectively solve the atomization condition in the reactor, the inspection robot will stay at the inspection position of the reactor for a preset time (for example: 1 minute, 3 minutes, etc.). During the stay, the inspection robot will obtain the reference light signal emitted by the reference device at the first detection moment and form first signal data, as well as the reference light signal emitted by the reference device at the second detection moment and form second signal data.
  • a preset time for example: 1 minute, 3 minutes, etc.
  • the aforementioned first optical attenuation value that is, the optical energy attenuation amplitude of the reference optical signal emitted by the reference device at the first detection moment
  • the aforementioned second optical attenuation value that is, the optical energy attenuation amplitude of the reference optical signal emitted by the reference device at the second detection moment
  • first detection time and the second detection time are any two different times within the preset time.
  • the difference between the first light attenuation value and the second light attenuation value is determined as the light attenuation change parameter. If the light attenuation change parameter is less than or equal to the change threshold, it indicates that the atomization condition of the reactor is not well improved after adjustment, so it is necessary to generate an atomization alarm to instruct the inspection personnel to go to the reactor for manual treatment. If the light attenuation change parameter is greater than the change threshold, it indicates that the atomization condition of the reactor is improved as expected after adjustment, so the inspection robot will leave the inspection position of the reactor and inspect the aforementioned multiple reactors in turn based on the preset inspection program.
  • the atomization alarm can be implemented by at least one of the following measures: buzzing of a buzzer, lighting of an indicator light, flashing of an indicator light, voice broadcast, etc., and the embodiments of the present application are not limited to this.
  • the method before detecting the reference light signal emitted by the reference device in the reaction furnace at the current moment and obtaining the detection signal data, the method further includes:
  • the reference signal data is updated according to the test signal data.
  • test signal data can be collected and used to update the reference signal data to adapt to the light energy attenuation caused by blurred viewing mirrors or light energy attenuation caused by reference equipment failure in actual applications, thereby ensuring that the light attenuation parameters obtained subsequently have higher accuracy.
  • the aforementioned initial moment and the current moment are in the same time period of a silicon reduction reaction. That is to say, before a silicon reduction reaction is performed, a test signal data can be obtained, and the latest test signal data obtained can be used as the reference signal data to ensure that the light attenuation parameters calculated each time have a high accuracy.
  • the method before detecting the reference light signal emitted by the reference device in the reaction furnace at the current moment and obtaining the detection signal data, the method further includes:
  • a detection instruction is sent to the reference device, so that the reference device emits the reference light signal based on the detection instruction.
  • the reference light signal of the reference device is controlled by the inspection robot or the central control server to adapt to the situation that the inspection time of the inspection robot to each reactor is not fixed in the actual application, that is, the reference device only sends the reference light signal when the inspection robot is at the target inspection position.
  • This can not only avoid the problem of mismatch between the patrol time of the inspection robot and the emission time of the reference light signal of the reference device, but also greatly reduce the number of times the reference light signal of the reference device is emitted, thereby reducing the energy consumption of the reference device and increasing the service life of the reference device.
  • the method further includes:
  • the first similarity parameter is less than or equal to a third threshold value
  • at least two second similarity parameters are obtained according to the monitoring image and at least two fog image groups, wherein the at least two second similarity parameters correspond to the at least two fog image groups one by one, wherein the at least two fog image groups respectively indicate different degrees of fogging of the reactor, and the second similarity parameters are used to characterize the image similarity between the corresponding fog image group and the monitoring image;
  • the second similarity parameter corresponding to the target fogged image group is a similarity parameter with the largest value among the at least two second similarity parameters
  • the degree of atomization of the reactor there are differences in the degree of atomization of the reactor. For example, there are three categories: light atomization, moderate atomization, and heavy atomization.
  • the aforementioned at least two fog image groups can be formed. It should be emphasized that the number of fog images included in each fog image group can be the same or different; for any fog image group, the number of fog images included therein is greater than or equal to 1.
  • the second similarity parameter corresponding to the fog image group can be understood as the average of multiple similarity values obtained by calculating the image similarity between the multiple fog images included in the fog image and the monitoring image. For example, if a fog image group is set to include fog image No. 1, fog image No. 2, and fog image No. 3, where the image similarity value between fog image No. 1 and the monitoring image is A1, the image similarity value between fog image No. 2 and the monitoring image is A2, and the image similarity value between fog image No. 3 and the monitoring image is A3, then the similarity parameter corresponding to the fog image group is (A1+A2+A3) ⁇ 3.
  • a preset processing flow may be triggered to automatically handle the atomization condition in the reactor according to the atomization processing instruction carried in the atomization warning information.
  • the atomization warning information may also be fed back to the user, so that the user can determine the location of the reactor according to the atomization warning information and make adjustments to the reactor.
  • the atomization condition of the reactor should be dealt with in time.
  • the method further includes:
  • the reaction parameters include at least one of a current value, a ratio of a hydrogen input amount to a trichlorosilane input amount.
  • the historical control information corresponding to the atomization warning information may be determined as the control information of the reactor, or the control information of the reactor may be determined with reference to the historical control information corresponding to the atomization warning information.
  • the control information may only instruct to adjust the ratio of hydrogen input to trichlorosilane input; and when the atomization warning information indicates that the current atomization condition of the reactor is similar to the moderate to severe atomization condition, the control information will not only instruct to adjust the ratio of hydrogen input to trichlorosilane input, but also instruct to reduce the current value of the reactor (which can be achieved by stopping the current).
  • the method further includes:
  • pixel spacing is used to characterize the shortest distance between a pixel region corresponding to a first silicon rod and a pixel region corresponding to a second silicon rod, wherein the first silicon rod and the second silicon rod are two silicon rods adjacently arranged in the reaction furnace;
  • Correcting the pixel spacing according to a correction parameter to obtain a first spacing wherein the correction parameter is used to characterize a physical length corresponding to a unit pixel block in the monitoring image, and the first spacing is used to characterize a spacing between the first silicon rod and the second silicon rod at a current moment;
  • a target parameter is determined according to the silicon rod spacing and the initial spacing, wherein the initial spacing is used to characterize the spacing between the first silicon rod and the second silicon rod at an initial moment, the initial moment is the start-up moment of the reactor, and the target parameter is used to characterize the diameter of the silicon rod in the reactor at the current moment.
  • the first silicon rod and the second silicon rod can be understood as two silicon rods closest to the sight glass of the reactor among the multiple silicon rods included in the reactor.
  • the arrangement of the plurality of silicon rods in the reactor may be as shown in FIG. 2 (the shaded circle in FIG. 2 The circles are used to indicate the silicon rods in the reaction furnace), and the two silicon rods circled by the dotted boxes in FIG. 2 are the first silicon rod and the second silicon rod respectively.
  • the first spacing between the first silicon rod and the second silicon rod in the monitoring image in the current period is obtained, and the difference between the initial spacing and the first spacing is compared to determine the automated monitoring method of the target parameter, replacing the manual visual monitoring method, avoiding interference from human factors, making the monitored silicon rod diameter have higher accuracy, and improving the monitoring effect of the silicon rod diameter change rate.
  • the distance between the first silicon rod and the second silicon rod will gradually shorten, and therefore, the first distance is less than or equal to the initial distance.
  • the first spacing can be subtracted from the initial spacing to obtain the spacing difference, that is, the diameter growth of the first silicon rod or the second silicon rod.
  • the diameter growth rate of the first silicon rod or the second silicon rod can be obtained by dividing the diameter growth by the time difference (that is, the time difference between the initial period and the current period).
  • the diameter growth plus the diameter of the silicon core can be added to obtain the diameter of the first silicon rod or the second silicon rod in the current period.
  • the process of obtaining a binary image may be:
  • Grayscale processing is performed on the monitoring image to obtain a grayscale image, and the pixel value of each pixel block in the silicon rod image included in the grayscale image is compared with the pixel threshold, and the pixel value of the pixel block greater than the pixel threshold is set to 0 (i.e., black), and the pixel value of the pixel block less than or equal to the pixel threshold is set to 255 (i.e., white), so as to form the binary image.
  • a pixel area corresponding to the first silicon rod and a pixel area corresponding to the second silicon rod are determined in the aforementioned binary image, and then the pixel spacing is determined according to the pixel area of the first silicon rod and the pixel area corresponding to the second silicon rod.
  • the smallest point spacing among multiple point spacings can be determined as the pixel spacing, wherein each point spacing among the multiple point spacings is used to refer to the spacing between a first target point and a second target point, the first target point is any pixel point in a pixel area corresponding to the first silicon rod, and the second target point is any pixel point in a pixel area corresponding to the second silicon rod.
  • N ⁇ M pixel pitches can be obtained.
  • the distance between the parallel first edge line and the second edge line can be determined as the pixel pitch, wherein the first edge line is a straight line obtained by fitting a side edge of a second pixel region close to the first pixel region, the second edge line is a straight line obtained by fitting a side edge of the first pixel region close to the second pixel region, the first pixel region is a pixel region corresponding to the first silicon rod, and the second pixel region is a pixel region corresponding to the second silicon rod.
  • the aforementioned pixel spacing deviates from the physical distance between the first silicon rod and the second silicon rod in the current period. Therefore, after obtaining the pixel spacing, the pixel spacing will be corrected based on the correction parameters so that the obtained first spacing has better data accuracy.
  • determining the pixel spacing according to the binary image includes:
  • first pixel point set is a set of pixel points corresponding to an edge portion of the first silicon rod close to the second silicon rod
  • second pixel point set is a set of pixel points corresponding to an edge portion of the second silicon rod close to the first silicon rod
  • a pixel spacing is determined according to the first pixel point set and the second pixel point set.
  • the first edge line and the second edge line can be determined respectively based on the first pixel point set and the second pixel point set based on a straight line fitting method, and then the distance between the first edge line and the second edge line is determined as the pixel spacing.
  • the first center point can be obtained by performing mean calculation on multiple pixel points in the pixel point set.
  • the second center point of the second pixel point set can be obtained based on the mean calculation, and then the distance between the first center point and the second center point is determined as the pixel spacing.
  • determining the pixel spacing according to the first pixel point set and the second pixel point set includes:
  • the average of the point spacings corresponding to the plurality of pixel heights is determined as the pixel spacing.
  • the pixel interval corresponding to the multiple pixel heights is the intersection of the pixel interval corresponding to the first pixel point set and the pixel interval corresponding to the second pixel point set. For example, if the pixel interval corresponding to the first pixel point set is [50, 105] and the pixel interval corresponding to the second pixel point set is [60, 125], then the pixel interval corresponding to the multiple pixel heights is [60, 105].
  • the pixel intervals corresponding to the multiple pixel heights are first set to filter out the data interference caused by the multiple pixel points corresponding to the ends of the silicon rods (the diameter growth rate of the ends of the silicon rods is faster than the diameter growth of the entire silicon rod, and the silicon crystals at the ends of the silicon rods grow differently, with obvious concave and convex areas), and then the mean calculation method is applied to ensure that the obtained pixel spacing can accurately indicate the spacing between the first silicon rod and the second silicon rod in the monitoring image during the current time period.
  • the method before correcting the pixel spacing according to the correction parameter to obtain the first spacing, the method further includes:
  • the correction parameter is determined according to a silicon core reference spacing and the silicon core target spacing, wherein the silicon core reference spacing is used to characterize the physical distance between the first silicon rod silicon core and the second silicon rod silicon core.
  • the silicon core reference spacing can be measured before acquiring the monitoring image; thereafter, the silicon core target spacing is obtained by performing image analysis on the correction image of the initial period, that is, the pixel distance between the silicon core of the first silicon rod and the silicon core of the second silicon rod in the correction image, and the aforementioned correction parameter can be obtained by dividing the silicon core reference spacing by the silicon core target spacing.
  • the correction parameters can be set to be recalculated every time the furnace is started for production, so as to reduce the adverse effects of factors such as silicon core placement errors and light interference, and further improve the accuracy of the target parameters obtained subsequently.
  • the initial spacing is determined according to a first reference image acquired at an initial moment, and the memory space occupied by the first reference image and the memory space occupied by the monitoring image are both greater than or equal to the internal Storage threshold.
  • multiple reference images are determined from the multiple original images, and the original images with low clarity are filtered out (that is, the original images whose memory space is smaller than the memory threshold are filtered out), and then the reference image with the earliest acquisition time among the multiple reference images is determined as the first reference image (that is, the correction image) to reduce the diameter interference caused by the sedimentation of silicon crystals on the surface of the silicon core, so that the determined correction parameters have higher accuracy.
  • the above-mentioned memory threshold may be 50kb.
  • the memory threshold may be adaptively adjusted according to actual needs.
  • the embodiment of the present application does not limit the specific value of the memory threshold.
  • the measured silicon core reference spacing is set as L1
  • the original image with the earliest acquisition time and an image size exceeding 50Kb is determined as the first reference image.
  • the first reference image is analyzed to obtain the silicon core target spacing a between the first silicon rod silicon core and the second silicon rod silicon core in the correction image.
  • the method further includes:
  • a connected domain analysis is performed on the silicon rod image included in the monitoring image to determine N gap blocks, where the gap blocks are used to indicate gaps between multiple silicon particles on the surface of the silicon rod, and N is a positive integer;
  • the quality information of the silicon rods in the reactor is determined according to the block areas of the M gap blocks, wherein the M gap blocks are the first M gap blocks arranged in descending order of block areas among the N gap blocks.
  • the looseness condition of the silicon rod to be tested can be determined by further analyzing the monitoring image, and a warning message can be generated accordingly to guide the staff to take disposal measures that match the looseness condition of the silicon rod to be tested.
  • the monitoring image Before performing connected domain analysis on the silicon rod image included in the monitoring image, the monitoring image may be subjected to dynamic threshold binarization processing to filter out part of the image noise included in the monitoring image, thereby improving the analysis accuracy of the connected domain analysis performed later.
  • the gap blocks in the silicon rod image included in the monitoring image can be clearly displayed.
  • the blocks shown in the dotted box in FIG4 are gap blocks, and the gap blocks shown are used to indicate the particle gaps between the particles on the surface of the loose material silicon rod.
  • the gap blocks should be white blocks, and the area parts of non-gap blocks should be black blocks.
  • the specific value of M may be 5.
  • the process of determining the quality information of the silicon rod to be inspected according to the block areas of the M gap blocks may be:
  • the ratio of the mean parameter to the benchmark parameter is determined as the quality parameter
  • the quality information is generated according to the quality parameter.
  • the benchmark parameter can be understood as the maximum gap area that can be reached between the particles on the surface of the loose material silicon rod.
  • the quality parameter is less than or equal to 35%
  • the quality information is used to indicate that the cauliflower condition of the silicon rod to be detected is in the early stage
  • the quality parameter is 35% to 65%
  • the quality information is used to indicate that the cauliflower condition of the silicon rod to be detected is in the middle stage
  • the quality parameter is 65% to 85%
  • the quality information is used to indicate that the cauliflower condition of the silicon rod to be detected is in the late stage
  • the quality parameter is greater than 85%, the quality information is used to indicate that the cauliflower condition of the silicon rod to be detected is in the final stage.
  • the acquiring of the monitoring image of the reactor includes:
  • a monitoring image of the reactor is acquired.
  • infrared images are periodically acquired through an infrared camera, and image analysis is performed on the infrared images. Only when the infrared image analysis result indicates that the difference between the temperature of the first position and the average temperature of the silicon rod is greater than or equal to the temperature difference value, the monitoring image is continuously acquired. This method can reduce the acquisition frequency of monitoring images, thereby reducing the energy consumption in the process of silicon rod bright spot monitoring, and reducing the cost of silicon rod bright spot monitoring.
  • the temperature of the first position can be obtained by counting the mean of the spectral energy of the aforementioned preset interval in the infrared image
  • the average temperature of the silicon rod can be obtained by counting the mean of the spectral energy of the target interval corresponding to the silicon rod in the infrared image.
  • the silicon powder content of the reactor can be collected in real time or periodically through a silicon powder content sensor pre-installed in the reactor.
  • the silicon powder content sensor can be set in the tail gas pipe of the reactor.
  • the silicon powder content sensor can convert the silicon powder content of the reactor into an electronic signal (such as a voltage signal/current signal, etc.). As the silicon powder content in the reactor increases, the corresponding electronic signal output by the silicon powder content sensor will also increase accordingly.
  • the silicon powder content in the reactor when there is no atomization state, the silicon powder content in the reactor will increase in an orderly manner, and with the emergence of atomization, the silicon powder content in the reactor will surge. Based on the correlation between the above-mentioned atomization condition and the silicon powder content in the reactor, the method of monitoring the increase rate of the silicon powder content in the reactor can be used to preliminarily determine whether there is an atomization condition in the reactor.
  • the single reaction time of the reactor (generally 100 hours, which can be adaptively adjusted according to actual needs) is evenly divided to obtain N monitoring cycles.
  • the increase rate of the silicon powder content of the reactor in the monitoring cycle is calculated accordingly, and the calculated increase rate is compared with the increase rate threshold.
  • the calculated increase rate is greater than or equal to the increase rate threshold, atomization warning information is generated.
  • the single reaction time of the reactor is set to 100 hours and the value of N is 3600, the length of each monitoring cycle is 1 minute; in application, as the value of N increases, the timeliness of the generated atomization warning information will also increase accordingly.
  • the above-mentioned growth rate threshold can be determined by the user based on experience, or it can be determined by counting the first growth rate of the silicon powder content when the reactor is in an atomized state and the second growth rate of the silicon powder content when the reactor is in a non-atomized state. At this time, the determined growth rate threshold is greater than the second growth rate and less than the first growth rate.
  • the method further includes:
  • the pixel values of the highlighted pixel point are respectively compared with the pixel values of a plurality of reference points for similarity to obtain a plurality of third similarity parameters, wherein the plurality of third similarity parameters correspond to the plurality of reference points one by one, and each of the plurality of reference points corresponds to a silicon rod bright spot in a temperature interval;
  • the bright spot information of the target point the bright spot information of the highlighted pixel point located in the preset interval is determined, wherein the bright spot information includes the temperature information of the corresponding silicon rod bright spot and the processing flow information of the corresponding silicon rod bright spot.
  • the bright spots of silicon rods in different temperature ranges have different corresponding pixel values in the image, and the processing procedures are also different. Therefore, by collecting the reference point information corresponding to different temperatures and the corresponding processing procedure information in the historical production period, and comparing the similarity between the abnormal points determined in the current production period and the reference points, the reference point closest to the abnormal points determined in the current production period is determined, that is, the target point is determined, and then the bright spot information of the abnormal point is determined accordingly with reference to the bright spot information of the target point.
  • the bright spot information determination process of the abnormal points determined in the current production period can be effectively simplified, the processing efficiency of the abnormal points can be improved, and the adverse effects of the bright spots of the silicon rods corresponding to the abnormal points can be further reduced.
  • the bright spot information of the target point can be directly used as the bright spot information of the abnormal point; and in another example, the bright spot information of the target point can be used as the initial template of the bright spot information of the abnormal point, so that the user can revise or add or delete the initial template according to the actual situation, so that the bright spot information of the determined abnormal point is compatible with the actual silicon rod bright spot temperature and processing flow.
  • the above-mentioned processing flow includes but is not limited to: changing the temperature value in the reactor, changing the ratio of hydrogen to trichlorosilane in the reactor, changing the hydrogen input amount in the reactor, changing the current value in the reactor, changing the trichlorosilane input amount in the reactor and stopping the reactor.
  • the grayscale image corresponding to each reference point can be compared with the grayscale image corresponding to the first reference image for image similarity, and the image similarity parameters of the grayscale image corresponding to each reference point can be obtained, and then the reference point corresponding to the image similarity parameter with the largest value among multiple image similarity parameters can be determined as the target point.
  • a highlight image located in the preset interval is determined.
  • the method further includes:
  • the highlighted pixel point is stored as a new reference point.
  • the abnormal point can be stored in the database as a new reference point to iteratively enrich the reference point data stored in the database, so that the data accuracy of the bright spot information of the abnormal point determined subsequently can be further improved.
  • the method further includes:
  • gas leakage information is output.
  • the inspection robot includes a walking mechanism and a mechanical arm
  • the gas sensor for monitoring the concentration of hydrogen and hydrogen chloride can be arranged on the walking mechanism or on the mechanical arm.
  • being installed on a robotic arm can better detect leakage of the reactor, because the robotic arm needs to extend to the sight glass on the reactor, and when the gas sensor is installed on the robotic arm, the gas sensor can be close to the sight glass, and the sight glass itself is the key part for detecting whether gas leakage occurs in the reactor.
  • gas leakage information is output to indicate the location of the reactor where the leakage occurs and the leakage status, thereby ensuring safe operation of the reduction workshop.
  • the method further includes:
  • the operating status information includes the start and stop data of the reactor
  • the first inspection route includes P inspection points arranged in order, where P is an integer greater than 1, and each inspection point corresponds to a reactor in a startup state;
  • the inspection robot is controlled to stop in sequence between the P points to be inspected.
  • the number of inspection points on the initial inspection route can be consistent with the number of reactors in the reduction workshop, and the inspection points and reactors are set in a one-to-one correspondence.
  • the first inspection route is determined based on the initial inspection route and the operating status information by obtaining the operating status information of the reactor corresponding to each inspection point on the initial inspection route, for example, whether the reactor corresponding to each inspection point is in the start-up state or in the stop state, retaining the inspection points where the corresponding reactors are in the start-up state, and then planning these inspection points where the corresponding reactors are in the start-up state to obtain the first inspection route.
  • the first stop monitoring sequence corresponding to the first inspection route indicates a stop detection sequence between the points to be inspected on the first inspection route.
  • the inspection robot scheduling platform obtains the initial inspection route and the operating status information of the reactor in the reduction workshop to determine the first inspection route, and the machine scheduling platform controls the inspection robot to conduct inspections in the reduction workshop according to the first inspection route; in this way, the planned route with a monitoring sequence is used for monitoring, the reactor that needs to be monitored is monitored, and the reactor that needs priority monitoring is monitored first, thereby realizing timely monitoring of the operating status of products in multiple reactors in the reduction workshop.
  • the method further includes:
  • the first inspection route overlaps with the obstacle area, determining a second inspection route according to the obstacle area, the initial inspection route and the running status information, wherein the second inspection route includes at least part of the P points to be inspected, and the second inspection route does not include the obstacle area, wherein the obstacle area is an area corresponding to an obstacle that hinders the movement of the inspection robot;
  • the inspection robot is controlled to continue inspection.
  • the second inspection route is determined based on the initial inspection route and operating status information on the premise of excluding the obstacle area, so that the inspection robot can avoid obstacles during the inspection process, thereby improving work efficiency and safety during the inspection.
  • the obstacle area is determined according to the moving speed of the obstacle and/or the size of the obstacle. Volume determined.
  • the size of the obstacle area can be determined based on the volume of the obstacle; when the obstacle is a moving object, it is necessary to jointly determine the obstacle area corresponding to the obstacle based on the moving speed of the obstacle and the volume of the obstacle, so as to reduce the probability of collision between the inspection robot and the obstacle and ensure the safe use of the inspection robot.
  • the moving speed of the obstacle can be obtained based on an infrared sensor installed on the inspection robot body, or based on a radar sensor installed on the inspection robot body, or based on a camera installed on the inspection robot body.
  • the infrared sensor When the infrared sensor is installed on the inspection robot body, there may be two or more infrared sensors. When the number of the infrared sensors is two, the two infrared sensors are arranged at a certain distance.
  • the inspection robot When a camera is installed on the inspection robot body, the inspection robot obtains image information on its route, and then uses the inter-frame difference method to calculate the moving speed of obstacles on its route.
  • FIG. 5 is a schematic diagram of the structure of a polysilicon monitoring device 500 provided in an embodiment of the present application.
  • the polysilicon monitoring device 500 includes:
  • An image acquisition module 501 is used to acquire a monitoring image of the reactor
  • An image analysis module 502 is used to perform image analysis on the monitoring image to obtain analysis results
  • An alarm module 503 is used to output an alarm message when the analysis result meets the alarm condition
  • the alarm condition includes at least one of the following:
  • the number parameter of the black spot areas in the monitoring image is greater than or equal to a first threshold
  • the area parameter of the black spot area in the monitoring image is greater than or equal to a second threshold
  • the first similarity parameter corresponding to the monitoring image is less than or equal to a third threshold value, the first similarity parameter being used to characterize the image similarity between the monitoring image and a reference image, the reference image being an image captured in the reactor in a non-atomized state;
  • the monitoring image includes highlighted pixels located within a preset interval, and the preset interval is used to characterize a pixel interval corresponding to a first position in the monitoring image, wherein the first position is a position in the reactor for placing silicon rods, and the highlighted pixels are pixels in the monitoring image whose corresponding pixel values are greater than a pixel threshold.
  • the image analysis module 502 includes:
  • a grayscale processing submodule used for performing grayscale processing on the monitoring image to obtain a grayscale image
  • a connected domain analysis submodule configured to perform a connected domain analysis on the silicon rod image included in the grayscale image, and determine at least one black spot block included in the silicon rod image;
  • the black spot positioning submodule is used to determine the number parameter of the black spot area in the monitoring image and the area parameter of the black spot area in the monitoring image according to the at least one black spot image block.
  • the polysilicon monitoring device 500 further includes:
  • An optical signal detection module used to detect a reference optical signal emitted by a reference device in the reaction furnace at a current moment, and obtain detection signal data;
  • An optical attenuation determination module configured to determine an optical attenuation parameter according to a difference between the detection signal data and the reference signal data corresponding to the reference optical signal, wherein the optical attenuation parameter is used to characterize an optical energy attenuation amplitude of the reference optical signal;
  • the atomization warning module is used to generate atomization information when the light decay parameter is greater than or equal to the light decay threshold.
  • the polysilicon monitoring device 500 further includes:
  • a parameter adjustment determination module used to determine an adjustment parameter according to the optical attenuation parameter and a preset conversion coefficient
  • the reactor control module is used to adjust the input amount of the input raw material of the reactor according to the adjustment parameters.
  • the polysilicon monitoring device 500 further includes:
  • a first re-test module used to obtain a first optical attenuation value corresponding to a reference optical signal emitted by the reference device at a first detection time, and a second optical attenuation value corresponding to a reference optical signal emitted by the reference device at a second detection time;
  • a second rechecking module configured to determine a difference between the first optical attenuation value and the second optical attenuation value as an optical attenuation variation parameter
  • the recheck warning module is used to generate an atomization alarm when the light attenuation change parameter is less than or equal to a change threshold, and the atomization alarm is used to indicate the position of the reaction furnace.
  • the polysilicon monitoring device 500 further includes:
  • a reference detection module is used to detect the reference light signal emitted by the reference device at the initial time. Perform detection to obtain test signal data, wherein the initial time is the start-up time of the reactor;
  • a reference updating module is used to update the reference signal data according to the test signal data.
  • the polysilicon monitoring device 500 further includes:
  • the instruction transmission module is used to send a detection instruction to the reference device, so that the reference device sends the reference light signal based on the detection instruction.
  • the polysilicon monitoring device 500 further includes:
  • a fogging degree analysis module configured to obtain at least two second similarity parameters according to the monitoring image and at least two fogging image groups when the first similarity parameter is less than or equal to a third threshold value, wherein the at least two second similarity parameters correspond to the at least two fogging image groups one by one, wherein the at least two fogging image groups respectively indicate different fogging degrees of the reactor, and the second similarity parameters are used to characterize the image similarity between the corresponding fogging image group and the monitoring image;
  • a fogging degree determination module configured to determine a target fogging image group from among the at least two fogging image groups, wherein the second similarity parameter corresponding to the target fogging image group is a similarity parameter having the largest value among the at least two second similarity parameters;
  • the atomization processing module is used to output atomization warning information associated with the target atomized image group.
  • the polysilicon monitoring device 500 further includes:
  • a first atomization control module used to determine the control information of the reactor according to the atomization warning information
  • a second atomization control module used to control the reaction parameters of the reactor according to the control information
  • the reaction parameters include at least one of a current value, a ratio of a hydrogen input amount to a trichlorosilane input amount.
  • the polysilicon monitoring device 500 further includes:
  • An image binary processing module is used to perform binary processing on the monitoring image to obtain a binary image
  • a spacing determination module is used to determine a pixel spacing according to the binary image, wherein the pixel spacing is used to characterize the shortest distance between a pixel area corresponding to a first silicon rod and a pixel area corresponding to a second silicon rod, wherein the first silicon rod and the second silicon rod are two silicon rods adjacently arranged in the reaction furnace.
  • a spacing correction module configured to correct the pixel spacing according to a correction parameter to obtain a first spacing, wherein the correction parameter is used to characterize a physical length corresponding to a unit pixel block in the monitoring image, and the first spacing is used to characterize a spacing between the first silicon rod and the second silicon rod at a current moment;
  • a silicon rod parameter determination module is used to determine a target parameter according to the silicon rod spacing and an initial spacing, wherein the initial spacing is used to characterize the spacing between the first silicon rod and the second silicon rod at an initial moment, the initial moment is the start-up moment of the reactor, and the target parameter is used to characterize the diameter of the silicon rod in the reactor at the current moment.
  • the distance determination module includes:
  • a feature recognition submodule configured to perform feature recognition on the binary image to determine a first pixel point set and a second pixel point set, wherein the first pixel point set is a set of pixel points corresponding to an edge portion of the first silicon rod close to the second silicon rod, and the second pixel point set is a set of pixel points corresponding to an edge portion of the second silicon rod close to the first silicon rod;
  • the spacing determination submodule is used to determine the pixel spacing according to the first pixel point set and the second pixel point set.
  • the distance determination submodule includes:
  • a dot distance calculation unit used to calculate a dot distance between a first pixel point and a second pixel point located at the same pixel height, wherein the first pixel point is any pixel point in the first pixel point set, and the second pixel point is any pixel point in the second pixel point set;
  • the spacing determination unit is used to determine the average of the point spacings corresponding to a plurality of pixel heights as the pixel spacing.
  • the polysilicon monitoring device 500 further includes:
  • a correction image acquisition module used to acquire a correction image of the reactor in an initial period
  • a correction image analysis module configured to perform image analysis on the correction image to obtain a target silicon core spacing, wherein the target silicon core spacing is used to characterize a spacing between a pixel point set corresponding to the first silicon rod silicon core and a pixel point set corresponding to the second silicon rod silicon core;
  • a correction parameter determination module is used to determine the correction parameter according to a silicon core reference spacing and the silicon core target spacing, wherein the silicon core reference spacing is used to characterize the first silicon rod silicon core and the The physical distance between the silicon cores of the second silicon rods.
  • the initial spacing is determined based on a first reference image acquired at an initial moment, and a memory space occupied by the first reference image and a memory space occupied by the monitoring image are both greater than or equal to a memory threshold.
  • the polysilicon monitoring device 500 further includes:
  • An image comparison module used for comparing the silicon rod image included in the monitoring image with a second reference image for image similarity to obtain a comparison result, wherein the second reference image is used to characterize the bulk silicon rod;
  • a gap positioning module configured to, when the comparison result indicates that the reactor includes loose silicon rods, perform a connected domain analysis on the silicon rod image included in the monitoring image to determine N gap blocks, wherein the gap blocks are used to indicate gaps between multiple silicon particles on the surface of the silicon rod, where N is a positive integer;
  • the quality analysis module is used to determine the quality information of the silicon rods in the reactor according to the block areas of the M gap blocks, wherein the M gap blocks are the first M gap blocks arranged from large to small in block areas among the N gap blocks.
  • the image acquisition module is specifically used to:
  • a monitoring image of the reactor is acquired.
  • the polysilicon monitoring device 500 further includes:
  • a pixel comparison module configured to compare the pixel value of the highlighted pixel with the pixel values of a plurality of reference points for similarity when the pixel coordinates of the highlighted pixel are within a preset interval, so as to obtain a plurality of third similarity parameters, wherein the plurality of third similarity parameters correspond to the plurality of reference points one by one, and each of the plurality of reference points corresponds to a silicon rod bright spot in a temperature interval;
  • a target point determination module used to determine the reference point corresponding to the third similarity parameter with the largest value as the target point
  • the bright spot information determination module is used to determine the bright spot information of the highlight pixel point located in the preset interval according to the bright spot information of the target point, wherein the bright spot information includes the temperature information of the corresponding silicon rod bright spot and the processing flow information of the corresponding silicon rod bright spot.
  • the polysilicon monitoring device 500 further includes:
  • a data iteration module is used to store the highlighted pixel point as a new reference point when the plurality of third similarity parameters are all less than or equal to a similarity threshold.
  • the polysilicon monitoring device 500 further includes:
  • a gas concentration acquisition module used to acquire the hydrogen concentration and hydrogen chloride concentration in the reduction workshop where the reactor is located;
  • the gas leakage warning module is used to output gas leakage information when the hydrogen concentration is greater than a first preset concentration value and/or the hydrogen chloride concentration is greater than a second preset concentration value.
  • the polysilicon monitoring device 500 further includes:
  • a route information acquisition module used to acquire the initial inspection route of the inspection robot and the operating status information of the reactor, wherein the operating status information includes the start and stop data of the reactor;
  • a first inspection route determination module used to determine a first inspection route of the inspection robot according to the initial inspection route and the operation status information, wherein the first inspection route includes P inspection points arranged in order, where P is an integer greater than 1, and each of the inspection points corresponds to a reactor in a startup state;
  • the inspection control module is used to control the inspection robot to stop in sequence between the P inspection points based on the first inspection route.
  • the polysilicon monitoring device 500 further includes:
  • a second inspection route determination module configured to determine, when the first inspection route overlaps with an obstacle area, a second inspection route according to the obstacle area, the initial inspection route and the operation status information, wherein the second inspection route includes at least some of the P points to be inspected, and the second inspection route does not include the obstacle area, wherein the obstacle area is an area corresponding to an obstacle that hinders the movement of the inspection robot;
  • the inspection control module is also used to control the inspection robot to continue inspection according to the second inspection route.
  • the obstacle area is determined according to the moving speed of the obstacle and/or the volume of the obstacle.
  • the polysilicon monitoring device 500 provided in the embodiment of the present application can implement each process in the above method embodiment, and will not be described again here to avoid repetition.
  • Figure 6 is a structural diagram of an electronic device provided in an embodiment of the present application.
  • the electronic device includes: a processor 601, a memory 602, and a program 6021 stored in the memory 602 and executable on the processor 601.
  • program 6021 When program 6021 is executed by processor 601, it can implement any steps in the method embodiment corresponding to Figure 1 and achieve the same beneficial effects, which will not be repeated here.
  • An embodiment of the present application also provides a readable storage medium, on which a computer program is stored.
  • a computer program is stored.
  • any step in the method embodiment corresponding to FIG. 1 can be implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
  • the computer-readable storage medium of the embodiment of the present application may adopt any combination of one or more computer-readable media.
  • the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium.
  • the computer-readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above.
  • a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.
  • Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
  • the program code contained on the storage medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, radio frequency (RF), etc., or any suitable combination of the above.
  • any appropriate medium including but not limited to wireless, wire, optical cable, radio frequency (RF), etc., or any suitable combination of the above.
  • Computer program code for performing the operations of the present application may be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages.
  • the program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or terminal.
  • the remote computer may be connected to the user's computer via any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).
  • LAN Local Area Network
  • WAN Wide Area Network

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Emergency Management (AREA)
  • Signal Processing (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

本申请提供一种多晶硅监控方法、装置及相关设备，其中，所述方法包括：获取反应炉的监控图像；对所述监控图像进行图像分析，获得分析结果；在所述分析结果满足报警条件的情况下，输出报警信息；所述报警条件包括黑斑区域的数目参数大于或等于第一阈值；黑斑区域的面积参数大于或等于第二阈值；所述监控图像对应的第一相似参数小于或等于第三阈值；所述监控图像包括的高亮像素点位于预设区间内中的至少一项。

Description

一种多晶硅监控方法、装置及相关设备
相关申请的交叉引用
本申请主张在2022年11月14日在中国提交的中国专利申请No.202211425068.5的优先权，其全部内容通过引用包含于此。
技术领域
本申请涉及多晶硅生产技术领域，具体涉及一种多晶硅监控方法、装置及相关设备。
背景技术
多晶硅还原工序是生产过程中的关键环节，还原车间内包括多个反应炉，为了对多个反应炉中每一反应炉内反应状况进行及时监测，需要现场人员对反应炉进行定期巡检，以监控反应炉内是否存在雾化、黑斑、硅棒亮点等硅棒异常问题。
应用中发现，采用上述人工巡检的方式对反应炉内的硅棒反应状况进行监测，受人为因素的干扰较大，易出现误判或漏判的情况，也就是说，相关技术对硅棒异常状况的监控效果较差。
发明内容
本申请实施例的目的在于提供一种多晶硅监控方法、装置和相关设备，用于解决相关技术在监控硅棒异常状况过程中存在的监控效果差的问题。
第一方面，本申请实施例提供一种多晶硅监控方法，所述方法包括：
获取反应炉的监控图像；
对所述监控图像进行图像分析，获得分析结果；
在所述分析结果满足报警条件的情况下，输出报警信息；
其中，所述报警条件包括以下至少一项：
所述监控图像中黑斑区域的数目参数大于或等于第一阈值；
所述监控图像中黑斑区域的面积参数大于或等于第二阈值；
所述监控图像对应的第一相似参数小于或等于第三阈值，所述第一相似参数用于表征所述监控图像和基准图像之间的图像相似度，所述基准图像为所述反应炉内在无雾化的状态下采集的图像；
所述监控图像包括的高亮像素点位于预设区间内，所述预设区间用于表征第一位置在所述监控图像内对应的像素区间，所述第一位置为所述反应炉内用于放置硅棒的位置，所述高亮像素点为所述监控图像中对应像素值大于像素阈值的像素点。
可选的，所述对所述监控图像进行图像分析，获得分析结果，包括：
对所述监控图像进行灰度化处理，获得灰度图像；
对所述灰度图像包括的硅棒图像进行连通域分析，确定所述硅棒图像包括的至少一个黑斑图块；
根据所述至少一个黑斑图块，确定所述监控图像中黑斑区域的数目参数和所述监控图像中黑斑区域的面积参数。
可选的，所述方法还包括：
对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据；
根据所述检测信号数据与所述基准光信号对应的基准信号数据之间的差异，确定光衰参数，其中，所述光衰参数用于表征所述基准光信号的光能量衰减幅度；
在所述光衰参数大于或等于光衰阈值的情况下，生成雾化信息。
可选的，所述生成雾化信息之后，所述方法还包括：
根据所述光衰参数和预设置的转换系数，确定调整参数；
根据所述调整参数，对所述反应炉的输入原料的输入量进行调整。
可选的，所述根据所述调整参数，对所述反应炉的输入原料的输入量进行调整之后，所述方法还包括：
获取所述参照设备在第一检测时刻发出的基准光信号对应的第一光衰值、以及所述参照设备在第二检测时刻发出的基准光信号对应的第二光衰值；
将所述第一光衰值和所述第二光衰值之间的差值确定为光衰变化参数；
在所述光衰变化参数小于或等于变化阈值的情况下，生成雾化警报，所 述雾化警报用于指示所述反应炉的位置。
可选的，所述对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据之前，所述方法还包括：
对所述参照设备在初始时刻发出的基准光信号的进行检测，得到测试信号数据，所述初始时刻为所述反应炉的启动时刻；
根据所述测试信号数据对所述基准信号数据进行更新。
可选的，所述对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据之前，所述方法还包括：
向所述参照设备发送检测指令，以使所述参照设备基于所述检测指令发出所述基准光信号。
可选的，所述输出报警信息之后，所述方法还包括：
在所述第一相似参数小于或等于第三阈值的情况下，根据所述监控图像和至少两个雾化图像组，获得至少两个第二相似参数，所述至少两个第二相似参数与所述至少两个雾化图像组一一对应，其中，所述至少两个雾化图像组分别指示所述反应炉的不同雾化程度，所述第二相似参数用于表征对应的雾化图像组与所述监控图像之间的图像相似度；
在所述至少两个雾化图像组中确定目标雾化图像组，其中，所述目标雾化图像组对应的第二相似参数为所述至少两个第二相似参数中数值最大的相似参数；
输出所述目标雾化图像组关联的雾化警示信息。
可选的，所述输出所述目标雾化图像组关联的雾化警示信息之后，所述方法还包括：
根据所述雾化警示信息，确定所述反应炉的调控信息；
根据所述调控信息对所述反应炉的反应参数进行调控；
其中，所述反应参数包括电流值、氢气输入量和三氯氢硅输入量的比值中的至少一项。
可选的，所述获取反应炉的监控图像之后，所述方法还包括：
对所述监控图像进行二值化处理，获得二值图像；
根据所述二值图像，确定像素间距，其中，所述像素间距用于表征对应 第一硅棒的像素区域和对应第二硅棒的像素区域之间的最短距离，所述第一硅棒和所述第二硅棒为所述反应炉内相邻设置的两个硅棒；
根据修正参数对所述像素间距进行修正，获取第一间距，其中，所述修正参数用于表征所述监控图像中单位像素块对应的物理长度，所述第一间距用于表征当前时刻所述第一硅棒和所述第二硅棒之间的间距；
根据所述硅棒间距和初始间距，确定目标参数，其中，所述初始间距用于表征初始时刻所述第一硅棒和所述第二硅棒之间的间距，所述初始时刻为所述反应炉的启动时刻，所述目标参数用于表征当前时刻所述反应炉内的硅棒直径。
可选的，所述根据所述二值图像，确定像素间距，包括：
对所述二值图像进行特征识别，确定第一像素点集和第二像素点集，其中，所述第一像素点集为所述第一硅棒的靠近所述第二硅棒的边缘部分对应的像素点集合，所述第二像素点集为所述第二硅棒的靠近所述第一硅棒的边缘部分对应的像素点集合；
根据所述第一像素点集和所述第二像素点集，确定像素间距。
可选的，所述根据所述第一像素点集和所述第二像素点集，确定像素间距，包括：
计算位于同一像素高度的第一像素点和第二像素点之间的点间距，其中，所述第一像素点为所述第一像素点集中的任一像素点，所述第二像素点为所述第二像素点集中的任一像素点；
将多个像素高度分别对应的点间距的均值确定为所述像素间距。
可选的，所述根据修正参数对所述像素间距进行修正，获得第一间距之前，所述方法还包括：
获取所述反应炉在初始时段的校正图像；
对所述校正图像进行图像分析，获得硅芯目标间距，其中，所述硅芯目标间距用于表征对应所述第一硅棒硅芯的像素点集和对应所述第二硅棒硅芯的像素点集之间的间距；
根据硅芯基准间距和所述硅芯目标间距，确定所述修正参数，其中，所述硅芯基准间距用于表征所述第一硅棒硅芯和所述第二硅棒硅芯之间的物理 距离。
可选的，所述初始间距根据初始时刻获取的第一参考图像确定，所述第一参考图像占用的内存空间和所述监控图像占用的内存空间均大于或等于内存阈值。
可选的，所述获取反应炉的监控图像之后，所述方法还包括：
将所述监控图像包括的硅棒图像和第二参考图像进行图像相似度比较，获得比较结果，其中，所述第二参考图像用于表征疏松料硅棒；
在所述比较结果指示所述反应炉内包括疏松料硅棒的情况下，对所述监控图像包括的硅棒图像进行连通域分析，确定N个间隙图块，所述间隙图块用于指示硅棒表面的多个硅颗粒之间的间隙，N为正整数；
根据M个间隙图块的图块面积确定所述反应炉内硅棒的质量信息，其中，所述M个间隙图块为所述N个间隙图块中图块面积从大到小排列的前M个间隙图块。
可选的，所述获取反应炉的监控图像，包括：
在所述反应炉的红外图像指示所述第一位置的温度与所述反应炉内的硅棒平均温度之差大于或等于温差阈值的情况下，和/或，在所述反应炉的硅粉含量的增幅速率大于或等于增速阈值的情况下，获取反应炉的监控图像。
可选的，所述输出报警信息之后，所述方法还包括：
在所述高亮像素点的像素坐标位于预设区间内的情况下，将所述高亮像素点的像素值与多个参考点的像素值分别进行相似度比较，获得多个第三相似参数，所述多个第三相似参数和所述多个参考点一一对应，所述多个参考点中的每一参考点对应一个温度区间的硅棒亮点；
将数值最大的所述第三相似参数对应的参考点确定为目标点；
根据所述目标点的亮点信息，确定位于所述预设区间的高亮像素点的亮点信息，其中，所述亮点信息包括对应的硅棒亮点的温度信息、以及对应的硅棒亮点的处理流程信息。
可选的，根据所述目标点的亮点信息，确定位于所述预设区间的高亮像素点的亮点信息之后，所述方法还包括：
在所述多个第三相似参数均小于或等于相似阈值的情况下，将所述高亮 像素点作为新的参考点进行存储。
可选的，所述方法还包括：
获取所述反应炉所在还原车间内的氢气浓度和氯化氢浓度；
当所述氢气浓度大于第一预设浓度值和/或所述氯化氢浓度大于第二预设浓度值时，输出气体泄漏信息。
可选的，所述方法还包括：
获取巡检机器人的初始巡检路线和反应炉的运行状态信息，所述运行状态信息包括反应炉的启停数据；
根据所述初始巡检路线和所述运行状态信息，确定所述巡检机器人的第一巡检路线，所述第一巡检路线上包括按次序排列的P个待巡检点，P为大于1的整数，每一所述待巡检点对应一个处于启动状态的反应炉；
基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留。
可选的，所述基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留之后，所述方法还包括：
在所述第一巡检路线与障碍区域重叠的情况下，根据所述障碍区域、所述初始巡检路线和所述运行状态信息确定第二巡检路线，所述第二巡检路线上包括所述P个待巡检点中的至少部分待巡检点，且所述第二巡检路线不包括所述障碍区域，其中，所述障碍区域为阻碍所述巡检机器人移动的障碍物对应的区域；
按照所述第二巡检路线，控制所述巡检机器人继续巡检。
可选的，所述障碍区域根据所述障碍物的移动速度和/或所述障碍物的体积确定。
第二方面，本申请实施例还提供一种多晶硅监控装置，所述装置包括：
图像获取模块，用于获取反应炉的监控图像；
图像分析模块，用于对所述监控图像进行图像分析，获得分析结果；
报警模块，用于在所述分析结果满足报警条件的情况下，输出报警信息；
其中，所述报警条件包括以下至少一项：
所述监控图像中黑斑区域的数目参数大于或等于第一阈值；
所述监控图像中黑斑区域的面积参数大于或等于第二阈值；
所述监控图像对应的第一相似参数小于或等于第三阈值，所述第一相似参数用于表征所述监控图像和基准图像之间的图像相似度，所述基准图像为所述反应炉内在无雾化的状态下采集的图像；
所述监控图像包括的高亮像素点位于预设区间内，所述预设区间用于表征第一位置在所述监控图像内对应的像素区间，所述第一位置为所述反应炉内用于放置硅棒的位置，所述高亮像素点为所述监控图像中对应像素值大于像素阈值的像素点。
第三方面，本申请实施例还提供一种电子设备，包括处理器、存储器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述计算机程序被所述处理器执行时实现上述的多晶硅监控方法的步骤。
第四方面，本申请实施例还提供一种计算机可读存储介质，所述计算机可读存储介质上存储有计算机程序，所述计算机程序被处理器执行时实现上述的多晶硅监控方法的步骤。
在本申请实施例中，通过图像区域监测、图像相似度比较以及图像像素值监测等方式，对硅棒生产过程中导致不合格硅棒产生的硅棒黑斑、反应炉雾化以及硅棒亮点等问题进行量化监控，以在分析结果满足报警条件的情况下，及时输出报警信息，提醒巡检人员对反应炉的异常生产状况进行处置，降低不合格硅棒的产出率，提升硅棒所获得的监控效果。
附图说明
图1是本申请实施例提供的一种多晶硅监控方法的流程示意图；
图2是本申请实施例提供的一种反应炉内多个硅棒的示意图；
图3是本申请实施例提供的一种第一硅棒和第二硅棒的示意图；
图4是本申请实施例提供的经二值化处理后的监控图像的示意图；
图5是本申请实施例提供的一种多晶硅监控装置的结构示意图；
图6是本申请实施例提供的一种电子设备的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请实施例提供一种多晶硅监控方法，参见图1，图1是本申请实施例提供的一种多晶硅监控方法的流程示意图，如图1所示，包括以下步骤：
步骤101、获取反应炉的监控图像。
其中，所述监控图像包括第一位置的图像，所述第一位置为反应炉内用于放置硅棒的位置。
所述监控图像可以为图像采集设备经由硅棒反应炉的视镜拍摄得到的炉内图像(指彩色图像)，所述图像采集设备可以为正对硅棒反应炉的视镜设置的固定相机，也可以为巡检机器人上装配的相机(巡检机器人按预设程序周期性巡检反应车间内多个反应炉)。
示例性的，所述监控图像的采集频次可以为1秒1次、5秒1次或1分钟1次等，用户可根据实际需求适应性选择监控图像的采集频次，本申请实施例对此并不加以限定。
需要说明的是，反应炉内设置有多个用于插设硅棒硅芯的凹槽，第一位置即用于指示凹槽所在位置，也即硅棒硅芯插设于凹槽内之后，硅棒伸出凹槽部分的下端位置。
步骤102、对所述监控图像进行图像分析，获得分析结果。
其中，所述分析结果至少包括所述监控图像中黑斑区域的数目参数、所述监控图像中黑斑区域的面积参数，所述监控图像对应的第一相似参数和所述监控图像包括的高亮像素点等数据。
其中，所述第一相似参数用于表征所述监控图像和基准图像之间的图像相似度，所述基准图像为所述反应炉内在无雾化的状态下采集的图像；所述高亮像素点为所述监控图像中对应像素值大于像素阈值的像素点。
示例性的，可以通过对监控图像进行图像分析，以将监控图像中包括的背景图块、硅棒正常图块、硅棒黑斑图块进行区分，并基于区分结果确定硅棒黑斑图块对应的数目参数和面积参数，其中，硅棒黑斑图块对应的数目参 数(即所述监控图像中黑斑区域的数目参数)可理解为监控图像中包括的多个硅棒黑斑图块的总数目，硅棒黑斑图块对应的面积参数(即所述监控图像中黑斑区域的面积参数)可理解为监控图像中包括的多个硅棒黑斑图块中每一硅棒黑斑图块的图块面积。
举例来说，可基于连通域算法完成对监控图像的图像分析，上述连通域算法可以为两遍扫描法(Two-Pass)或种子填充法(Seed-Filling)。
示例性的，高亮像素点的获取过程可以为：通过对监控图像进行灰度化处理，获得灰度图像；再对灰度图像进行二值化处理，以获得所述高亮像素点。
在对监控图像进行灰度化处理的过程中，可以采用最大值法进行图像灰度处理，也可以采用平均值法进行图像灰度处理，还可以采用加权平均值法进行图像灰度处理，本申请实施例对图像灰度处理所采用的具体算法并不加以限定。
在对灰度图像进行二值化处理过程中，可以应用全局阈值的方式(即灰度图像中每一像素点的像素值均同前述像素阈值进行比较，此时，灰度图像中每一像素点所比较的像素阈值相同)，也可以应用局部自适应动态阈值算法实现(即动态计算每一像素点所要进行比较的像素阈值，此时，灰度图像中每一像素点所比较的像素阈值可以相同，也可以不同)，本申请实施例对此也不加以限定。
步骤103、在所述分析结果满足报警条件的情况下，输出报警信息。
其中，所述报警条件包括以下至少一项：
所述监控图像中黑斑区域的数目参数大于或等于第一阈值；
所述监控图像中黑斑区域的面积参数大于或等于第二阈值；
所述监控图像对应的第一相似参数小于或等于第三阈值；
所述监控图像包括的高亮像素点位于预设区间内，所述预设区间用于表征第一位置在所述监控图像内对应的像素区间，所述第一位置为所述反应炉内用于放置硅棒的位置。
如上，针对黑斑问题的监测，通过第一阈值和第二阈值的配合设置，可对硅棒内黑斑过度蔓延导致的硅棒异常生长情况进行量化，将黑斑数量过多 的硅棒识别为处于异常生长状态的硅棒和/或黑斑面积过大的硅棒识别为处于异常生长状态的硅棒，并相应生成报警信息，以提醒还原车间内的工作人员对处于异常生长状态的硅棒进行及时处置(如提高反应炉内的氢气量，或者，降低反应炉内的温度等处置措施)，避免反应炉内的黑斑问题出现进一步恶化。
需要指出的是，上述面积参数可以理解为所述监控图像包括的多个黑斑区域中区块面积最大的黑斑区域的区块面积，也可以理解为所述监控图像包括的多个黑斑区域的区块面积之和，用户可根据实际需求确定上述面积参数的指代含义，本申请实施例对此并不加以限定。
举例来说，若监控图像包括2个黑斑区域，其中一个黑斑区域的区块面积为25，另一个黑斑区域的区块面积为20，则前述面积参数可以为25或45。
针对雾化问题的监测，可通过比较监控图像和基准图像之间的图像相似度，也即确定第一相似参数，并对第一相似参数和第三阈值进行比较的方式，以确定反应炉内是否出现雾化状况，通过采用上述自动化监测的方式，对人工巡检的监控方式进行替换，以规避人为因素的干扰，提升反应炉在雾化监测方面所获得的监控效果。
针对硅棒亮点的监测，可将位于预设区间内的高亮像素点称为异常点，在输出指示异常点的报警信息后，通过在监控图像上对所述异常点进行标记的措施，便利用户(如：反应炉所在车间的巡检人员)对异常点的定位，加快异常点(也即硅棒亮点)的处置效率。
其中，硅棒亮点可理解为硅棒下端位置(也即硅棒在反应炉内的插设位置)因生产故障问题而产生的异常高温点，此异常高温点会在视界中以明亮光点的形式呈现，若不及时识别该异常高温点并加以处置，硅棒下端部会出现高温熔断情况，这将造成硅棒倾倒的事故，并带来较为严重的经济损失。
在一可选的实施方式中，在确定异常点的过程中，除了应用预设区间对多个高亮像素点进行筛分的方式外，还可以应用数量阈值的方式对多个高亮像素点进行筛分，即在多个高亮像素点中，将相邻的多个高亮像素点确定为一候选点组，并将候选点组内高亮像素点数量小于或等于数量阈值的候选点组滤除，以进一步提升所确定异常点的准确性。
在另一实施方式中，可以先基于数量阈值对多个高亮像素点进行初步筛分，再应用预设区间对初步筛分后的多个高亮像素点作进一步筛分，以最终确定异常点，这能显著提升所确定异常的准确性。
在一些实施方式中，若监控图像为巡检机器人上装配的相机采集的图像，则输出报警信息，且工作人员对处于异常生长状况的硅棒进行处置后，巡检机器人可以返回前述报警信息指示的反应炉视镜处，对所述反应炉内的硅棒进行复检，若复检通过，则恢复周期性巡检模式，若复检不通过，则再次生成报警信息。
需要说明的是，所输出的报警信息包括黑斑问题、雾化问题和硅棒亮点问题中的至少一项。
在本申请实施例中，通过图像区域监测、图像相似度比较以及图像像素值监测等方式，对硅棒生产过程中导致不合格硅棒产生的硅棒黑斑、反应炉雾化以及硅棒亮点等问题进行量化监控，以在分析结果满足报警条件的情况下，及时输出报警信息，提醒巡检人员对反应炉的异常生产状况进行处置，降低不合格硅棒的产出率，提升硅棒所获得的监控效果。
可选的，所述对所述监控图像进行图像分析，获得分析结果，包括：
对所述监控图像进行灰度化处理，获得灰度图像；
对所述灰度图像包括的硅棒图像进行连通域分析，确定所述硅棒图像包括的至少一个黑斑图块；
根据所述至少一个黑斑图块，确定所述监控图像中黑斑区域的数目参数和所述监控图像中黑斑区域的面积参数。
其中，对监控图像进行灰度化处理，一方面是为了便利后续的连通域分析过程，确保连通域分析的准确性；另一方面则是为了对监控图像中存在的背景噪声进行滤除。
示例性的，对所述灰度图像包括的硅棒图像进行连通域分析，确定所述硅棒图像包括的至少一个黑斑图块的过程可以为：
对所述灰度图像进行第一连通域分析，确定所述灰度图像包括的硅棒图像；
对所述灰度图像包括的硅棒图像进行反向二值化处理和第二连通域分析， 确定所述硅棒图像包括的至少一个黑斑图块。
在上述过程中，先确定所述灰度图像包括的硅棒图像是为了避免背景图块带来的干扰，提高黑斑图块的提取效率和提取准确度。
对所述灰度图像包括的硅棒图像进行反向二值化处理的过程可以为，将所述灰度图像包括的硅棒图像中每一像素块的像素值与像素阈值进行比较，并将大于像素阈值的像素块的像素值设为0(即黑色)，而将小于或等于像素阈值的像素块的像素值设为255(即白色)。
在第二连通域分析过程中，相邻的多个0值像素块将被确定为一个黑斑图块，此时可相应获得所述监控图像中黑斑区域的数目参数，而后通过统计每一黑斑图块包括的像素块数量，即可确定所述监控图像中黑斑区域的面积参数。
可选的，所述方法还包括：
对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据；
根据所述检测信号数据与所述基准光信号对应的基准信号数据之间的差异，确定光衰参数，其中，所述光衰参数用于表征所述基准光信号的光能量衰减幅度；
在所述光衰参数大于或等于光衰阈值的情况下，生成雾化信息。
需要指出的是，在还原车间中存在多个反应炉，也即存在多个硅棒反应炉，对于前述多个反应炉来说，每一反应炉内设置有至少一个参照设备。
示例性的，所述参照设备可以为设置于反应炉内的红外线发射器。
还原车间内设置有至少一个巡检机器人，所述巡检机器人根据预设置的巡检路线依次对还原车间内的多个反应炉进行巡视，在巡检机器人移动至目标巡检位置(即巡检机器人上所携带信号接收组件与反应炉的视镜相对的位置)时，巡检机器人将对反应炉内的参照设备在当前时刻所发出的基准光信号进行接收，并基于实际接收的光信号形成检测信号数据，所述检测信号数据至少包括所述巡检机器人所接收光信号的光能量。
需要指出的是，还原车间内除了设置至少一个巡检机器人外，还设置有中控服务器，中控服务器与巡检机器人通讯连接并构成巡检系统；巡检机器 人会将巡检过程中检测的多个检测信号数据上传至中控服务器进行存储和数据处理。
所述基准信号数据至少包括所述基准光信号发射时的光能量，在基准光信号的传输过程中，受反应炉内的硅粉干扰，基准光信号的光能量会逐渐衰减，且随着反应炉内的硅粉弥散程度的加剧，基准光信号的光能量衰减幅度也将增加，因此，利用基准光信号的光能量衰减幅度与反应炉内的硅粉弥散程度呈正相关关系的特性，可通过计算光衰参数的方式，间接确定反应炉的硅粉弥散程度。
上述光衰参数的确定操作，可以由巡检机器人执行，也可以由中控服务器执行，应用中，可以采用中控服务器计算得到所述光衰参数，以降低巡检机器人的能耗，延长巡检机器人的使用寿命。
在光衰参数大于或等于光衰阈值的情况下，即指示反应炉内的硅粉弥散程度超过临界标准，此时可判定反应炉内存在雾化状况，也即输出雾化信息，以指引巡检人员对存在雾化问题的反应炉进行及时处置。
需要强调的是，通过雾化信息和用于指示雾化问题的报警信息的双重监控，可降低漏判雾化问题的概率，提升对反应炉雾化监测的准确性。
进一步的，所述生成雾化信息之后，所述方法还包括：
根据所述光衰参数和预设置的转换系数，确定调整参数；
根据所述调整参数，对所述反应炉的输入原料的输入量进行调整。
在雾化问题存在时，可采用降低反应炉内的硅物质的输入量的方式，以对雾化状况进行处置，因此，所述转换系数应理解为，基准光信号的光能量衰减幅度与硅物质的输入量的降幅之间的关联系数，转换系数可通过拟合不同雾化状况下对应硅物质输入量的降幅与光能量衰减幅度之间的关联关系得到；前述硅物质可理解为三氯氢硅。
调整反应炉的输入原料的输入量的过程可以为：基于调整参数所指示的硅物质的输入量的降幅，对反应炉的硅物质的输入量进行降低，例如，将反应炉内的三氯氢硅的输入量下调200kg/h。
需要指出的是，上述调整操作可以由中控服务器下发第一雾化调控指令给巡检机器人，经由巡检机器人执行完成；也可以由中控服务器下发第二雾 化调控指令给反应炉的输入原料的控制组件，经由所述控制组件执行完成。
如上，通过在反应炉内设置参照设备，采集参照设备所发出基准光信号的检测信号数据，并比较检测信号数据和基准信号数据之间的差异，以确定当前时刻反应炉内的光衰参数，在光衰参数大于或等于光衰阈值的情况下，即可认定反应炉内存在雾化状况，通过应用转换系数对所获得光衰参数进行处理，以确定用于调整反应炉的输入原料的输入量的调整参数，抑制反应炉内的雾化情况；相较于人工巡检的方式来说，应用上述自动化检测雾化状况并处置的方式，可规避人为因素的干扰，提升反应炉在雾化监控方面所获得的监控效果。
可选的，所述根据所述调整参数，对所述反应炉的输入原料的输入量进行调整之后，所述方法还包括：
获取所述参照设备在第一检测时刻发出的基准光信号对应的第一光衰值、以及所述参照设备在第二检测时刻发出的基准光信号对应的第二光衰值；
将所述第一光衰值和所述第二光衰值之间的差值确定为光衰变化参数；
在所述光衰变化参数小于或等于变化阈值的情况下，生成雾化警报，所述雾化警报用于指示所述反应炉的位置。
如上，在确定反应炉内存在雾化状况，且基于预设逻辑对反应炉内的雾化装置进行处置后，也即根据所述调整参数对所述反应炉的输入原料的输入量进行调整后，为观测调控手段是否有效解决的反应炉内的雾化状况，巡检机器人会在反应炉的巡检位置停留，停留时间为预设时间(例如：1分钟、3分钟等)，在停留过程中，巡检机器人将获取参照设备在第一检测时刻发出的基准光信号并形成第一信号数据，以及参照设备在第二检测时刻发出的基准光信号并形成第二信号数据。
通过比较第一信号数据和基准信号数据之间的差异，可确定前述第一光衰值，也即参照设备在第一检测时刻所发出基准光信号的光能量衰减幅度；同理，通过比较第二信号数据和基准信号数据之间的差异，可确定前述第二光衰值，也即参照设备在第二检测时刻所发出基准光信号的光能量衰减幅度。
需要说明的是，前述第一检测时刻和所述第二检测时刻为前述预设时间内的任意两个不同的时刻。
将第一光衰值和第二光衰值之间的差值确定为光衰变化参数，若光衰变化参数小于或等于变化阈值，则指示反应炉的雾化状况在调整后的改善效果不佳，因此需通过生成雾化警报的方式，指示巡检人员前往反应炉处进行人工处置。若光衰变化参数大于所述变化阈值，则指示反应炉的雾化状况在调整后的改善效果符合预期，因此，巡检机器人将离开反应炉的巡检位置，并基于预设的巡检程序对前述多个反应炉进行依次巡检。
需要说明的是，在应用中，雾化警报可以采用蜂鸣器蜂鸣、指示灯亮、指示灯闪烁、语音播报等措施中的至少一种实现，本申请实施例对此并不加以限定。
可选的，所述对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据之前，所述方法还包括：
对所述参照设备在初始时刻发出的基准光信号的进行检测，得到测试信号数据，所述初始时刻为所述反应炉的启动时刻；
根据所述测试信号数据对所述基准信号数据进行更新。
如上所述，在反应炉进行单次硅还原反应之前，可以采集测试信号数据，并利用测试信号数据对基准信号数据进行更新，以适配实际应用中，视镜模糊导致的光能量衰减或参照设备故障导致的光能量衰减，确保后续所获得的光衰参数能具备较高的准确性。
需要指出的是，前述初始时刻和当前时刻位于同一次硅还原反应的时间周期中。也就说是，可以在进行一次硅还原反应之前，即获取一次测试信号数据，并将所获取的最新的测试信号数据作为基准信号数据应用，以保障每次所计算的光衰参数具备较高的准确性。
可选的，所述对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据之前，所述方法还包括：
向所述参照设备发送检测指令，以使所述参照设备基于所述检测指令发出所述基准光信号。
如上所述，经由巡检机器人或中控服务器控制参照设备的基准光信号的发出动作，以适配实际应用中巡检机器人前往每个反应炉的巡检时间不固定的情况，即仅在巡检机器人位于目标巡检位置时，参照设备才发出所述基准 光信号，这不仅能规避巡检机器人的巡视时间与参照设备的基准光信号发出时间不匹配的问题，还能极大降低参照设备的基准光信号的发出次数，因此能降低参照设备的能耗，提升参照设备的使用寿命。
可选的，所述输出报警信息之后，所述方法还包括：
在所述第一相似参数小于或等于第三阈值的情况下，根据所述监控图像和至少两个雾化图像组，获得至少两个第二相似参数，所述至少两个第二相似参数与所述至少两个雾化图像组一一对应，其中，所述至少两个雾化图像组分别指示所述反应炉的不同雾化程度，所述第二相似参数用于表征对应的雾化图像组与所述监控图像之间的图像相似度；
在所述至少两个雾化图像组中确定目标雾化图像组，其中，所述目标雾化图像组对应的第二相似参数为所述至少两个第二相似参数中数值最大的相似参数；
输出所述目标雾化图像组关联的雾化警示信息。
反应炉的雾化程度存在差异，例如，存在轻度雾化、中度雾化、重度雾化三类，通过收集不同类别雾化程度所对应的雾化图像，即可形成前述至少两个雾化图像组，需要强调的是，每一雾化图像组所包括的雾化图像的数量可以相同，也可以不同；对于任一雾化图像组来说，其内所包括的雾化图像的数量大于或等于1。
雾化图像组对应的第二相似参数可理解为，雾化图像包括的多个雾化图像分别与监控图像进行图像相似度计算所分别获得多个相似值的均值。举例来说，若设定某一雾化图像组包括一号雾化图像、二号雾化图像和三号雾化图像，其中，一号雾化图像和监控图像之间的图像相似值为A1，二号雾化图像和监控图像之间的图像相似值为A2，三号雾化图像和监控图像之间的图像相似值为A3，则该雾化图像组对应的相似参数为(A1+A2+A3)÷3。
在一些实施方式中，在输出雾化警示信息后，可根据雾化警示信息中携带的雾化处理指令，触发预设置的处理流程对反应炉内的雾化状况进行自动化处置。
在一些实施方式中，在输出雾化警示信息后，也可以向用户反馈所述雾化警示信息，以使用户根据雾化警示信息确定所述反应炉的位置，并对所述 反应炉的雾化状况进行及时处置。
可选的，所述输出所述目标雾化图像组关联的雾化警示信息之后，所述方法还包括：
根据所述雾化警示信息，确定所述反应炉的调控信息；
根据所述调控信息对所述反应炉的反应参数进行调控；
其中，所述反应参数包括电流值、氢气输入量和三氯氢硅输入量的比值中的至少一项。
可以将雾化警示信息对应的历史调控信息确定为所述反应炉的调控信息，也可以参照雾化警示信息对应的历史调控信息确定所述反应炉的调控信息。
示例性的，在雾化警示信息指示反应炉当前雾化状况与轻度雾化状况较为相似的情况下，调控信息可以仅指示调整氢气输入量和三氯氢硅输入量的比值；而在雾化警示信息指示反应炉当前雾化状况与中重度雾化状况较为相似的情况下，调控信息不仅会指示调整氢气输入量和三氯氢硅输入量的比值，还会指示对反应炉的电流值进行降低(可通过停加电流的方式实现)。
可选的，所述获取反应炉的监控图像之后，所述方法还包括：
对所述监控图像进行二值化处理，获得二值图像；
根据所述二值图像，确定像素间距，其中，所述像素间距用于表征对应第一硅棒的像素区域和对应第二硅棒的像素区域之间的最短距离，所述第一硅棒和所述第二硅棒为所述反应炉内相邻设置的两个硅棒；
根据修正参数对所述像素间距进行修正，获取第一间距，其中，所述修正参数用于表征所述监控图像中单位像素块对应的物理长度，所述第一间距用于表征当前时刻所述第一硅棒和所述第二硅棒之间的间距；
根据所述硅棒间距和初始间距，确定目标参数，其中，所述初始间距用于表征初始时刻所述第一硅棒和所述第二硅棒之间的间距，所述初始时刻为所述反应炉的启动时刻，所述目标参数用于表征当前时刻所述反应炉内的硅棒直径。
其中，所述第一硅棒和第二硅棒可理解为反应炉包括的多个硅棒中距离反应炉视镜最近的两个硅棒。
示例性的，反应炉中的多个硅棒的布设情况可以如图2(图2中阴影圆 圈用于指示反应炉内的硅棒)所示，图2中虚线框所圈中的两个硅棒分别为第一硅棒和第二硅棒。
如上，通过对监控图像进行图像分析，获得当前时段第一硅棒和第二硅棒在监控图像内的第一间距，并比较初始间距和第一间距之间的差异，以确定目标参数的自动化监测方式，替代人工目视的监测方式，避免人为因素的干扰，使所监测得到的硅棒直径具备较高的准确性，提升对硅棒直径变化速率的监测效果。
需要说明的是，随着硅棒的生长(也即硅棒的直径增长)，第一硅棒和第二硅棒之间的间距会逐渐缩短，因此，第一间距小于或等于所述初始间距。
示例性的，可以通过初始间距减去第一间距，以获得间距差值，也即第一硅棒或第二硅棒的直径增长量，应用该直径增长量除以时间差值(即初始时段和当前时段之间的时间差)，即可获得第一硅棒或第二硅棒的直径增长速率，应用该直径增长量加上硅芯的直径，即可获得第一硅棒或第二硅棒在当前时段的直径。
示例性的，获得二值图像的过程可以为：
对所述监控图像进行灰度处理，获得灰度图像，将所述灰度图像包括的硅棒图像中每一像素块的像素值与像素阈值进行比较，并将大于像素阈值的像素块的像素值设为0(即黑色)，而将小于或等于像素阈值的像素块的像素值设为255(即白色)，以形成所述二值图像。
在获得前述二值图像后，基于特征识别算法(也可应用目标检测算法)，在前述二值图像中确定对应所述第一硅棒的像素区域和对应所述第二硅棒的像素区域，之后，根据所述第一硅棒的像素区域和对应所述第二硅棒的像素区域确定所述像素间距。
在一示例中，可以将多个点间距中最小的点间距确定为所述像素间距，其中，多个点间距中的每一点间距用于指代第一目标点和第二目标点之间的间距，所述第一目标点为对应所述第一硅棒的像素区域中的任一像素点，所述第二目标点为对应所述第二硅棒的像素区域中的任一像素点。
举例来说，若对应所述第一硅棒的像素区域包括N个像素点，对应所述第二硅棒的像素区域包括M个像素点，则可获得N×M个点间距。
在另一示例中，可以将平行的第一边缘线和第二边缘线之间的距离确定为所述像素间距，其中，第一边缘线为对第一像素区域靠近的第二像素区域的一侧边缘进行拟合获得的直线，第二边缘线为对第二像素区域靠近的第一像素区域的一侧边缘进行拟合获得的直线，第一像素区域为对应第一硅棒的像素区域，第二像素区域为对应第二硅棒的像素区域。
由于硅棒的设置位置偏差、图像的拍摄角度偏差、光线折射等因素的干扰，使得前述像素间距与当前时段中第一硅棒和第二硅棒的物理距离存在偏差，因此，在获得像素间距后，会基于修正参数对像素间距进行数据修正，以使所获得第一间距具备较优的数据准确度。
可选的，所述根据所述二值图像，确定像素间距，包括：
对所述二值图像进行特征识别，确定第一像素点集和第二像素点集，其中，所述第一像素点集为所述第一硅棒的靠近所述第二硅棒的边缘部分对应的像素点集合，所述第二像素点集为所述第二硅棒的靠近所述第一硅棒的边缘部分对应的像素点集合；
根据所述第一像素点集和所述第二像素点集，确定像素间距。
在一些实施方式中，获得第一像素点集和第二像素点集后，可以基于直线拟合的方式，根据第一像素点集和第二像素点集分别确定第一边缘线和第二边缘线，随后将第一边缘线和第二边缘线之间的距离确定为所述像素间距。
在另一些实施方式中，获得第一像素点集和第二像素点集后，也可以对像素点集中多个像素点进行均值计算的方式，获得第一中心点，同理基于均值计算的方式获得第二像素点集的第二中心点，随后将第一中心点和第二中心点之间的距离确定为所述像素间距。
可选的，所述根据所述第一像素点集和所述第二像素点集，确定像素间距，包括：
计算位于同一像素高度的第一像素点和第二像素点之间的点间距，其中，所述第一像素点为所述第一像素点集中的任一像素点，所述第二像素点为所述第二像素点集中的任一像素点；
将多个像素高度分别对应的点间距的均值确定为所述像素间距。
通过获取多个像素高度中每一像素高度对应的点间距，并计算多个点间 距之间的平均值，以提升所获得像素间距的准确度。
其中，多个像素高度所对应的像素区间为第一像素点集对应的像素区间和第二像素点集对应的像素区间的交集，例如，若第一像素点集对应的像素区间为[50，105]，第一像素点集对应的像素区间为[60，125]，则多个像素高度所对应的像素区间为[60，105]。
如上，先基于多个像素高度所对应的像素区间的设置，以过滤硅棒端部对应的多个像素点造成的数据干扰(硅棒端部的直径增长速率较硅棒整体的直径增长更快，且硅棒端部处的硅晶体长势不一，存在较为明显的凹凸区域)，再应用均值计算的方式，确保所获得像素间距可准确指示第一硅棒和第二硅棒当前时段内在监控图像中的间距。
可选的，所述根据修正参数对所述像素间距进行修正，获得第一间距之前，所述方法还包括：
获取所述反应炉在初始时段的校正图像；
对所述校正图像进行图像分析，获得硅芯目标间距，其中，所述硅芯目标间距用于表征对应所述第一硅棒硅芯的像素点集和对应所述第二硅棒硅芯的像素点集之间的间距；
根据硅芯基准间距和所述硅芯目标间距，确定所述修正参数，其中，所述硅芯基准间距用于表征所述第一硅棒硅芯和所述第二硅棒硅芯之间的物理距离。
由于第一硅棒的设置位置和第二硅棒的设置位置相对固定，因此，在获取监控图像之前，可测算得到所述硅芯基准间距；之后，通过对初始时段的校正图像进行图像分析，以获得硅芯目标间距，也即校正图像中第一硅棒硅芯和第二硅棒硅芯的像素距离，通过硅芯基准间距除以硅芯目标间距，即可获得前述修正参数。
应用中，可以设置每一次开炉生产时即对所述修正参数进行重新测算，以减轻硅芯放置误差、光线干扰等因素的不利影响，进一步提升后续所获得目标参数的准确度。
可选的，所述初始间距根据初始时刻获取的第一参考图像确定，所述第一参考图像占用的内存空间和所述监控图像占用的内存空间均大于或等于内 存阈值。
针对图像采集设备在初始时段的单次采集过程中会采集多张原始图像的情况，经由在所采集多张原始图像确定多个参考图像的方式，滤除多张原始图像中清晰度低的原始图像(也即滤除内存空间小于内存阈值的原始图像)，而后将多个参考图像中采集时间最早的参考图像确定为所述第一参考图像(也即校正图像)，以降低硅晶体在硅芯表面沉降带来的直径干扰，使所确定修正参数具备较高的准确度。
示例性的，上述内存阈值可以为50kb，实际应用中，也可以根据实际需求对所述内存阈值进行适应性调整，本申请实施例对内存阈值的具体取值并不加以限定。
为方便理解，示例说明如下：
在初始时段，设定测算得到的硅芯基准间距为L1，将采集时间最早且图像大小超过50Kb的原始图像确定为第一参考图像，对第一参考图像进行图像分析，获得校正图像中第一硅棒硅芯和第二硅棒硅芯之间的硅芯目标间距a，此时可计算得到修正参数rate＝L1÷a。
在当前时段，对图像大小超过50Kb的监控图像进行图像分析，获得监控图像中第一硅棒和第二硅棒之间的第一间距b，此时可计算得到当前时段第一硅棒和第二硅棒之间的物理间距L2＝b×rate。
如图3所示，在假定反应炉内每一硅棒的生长状况均相同的情况下，可计算得到目标参数R＝(L1-L2)+r，其中，r用于指示硅棒的硅芯直径。
可选的，所述获取反应炉的监控图像之后，所述方法还包括：
将所述监控图像包括的硅棒图像和第二参考图像进行图像相似度比较，获得比较结果，其中，所述第二参考图像用于表征疏松料硅棒；
在所述比较结果指示所述反应炉内包括疏松料硅棒的情况下，对所述监控图像包括的硅棒图像进行连通域分析，确定N个间隙图块，所述间隙图块用于指示硅棒表面的多个硅颗粒之间的间隙，N为正整数；
根据M个间隙图块的图块面积确定所述反应炉内硅棒的质量信息，其中，所述M个间隙图块为所述N个间隙图块中图块面积从大到小排列的前M个间隙图块。
如上，在比较结果指示待检测硅棒为疏松料硅棒的情况下，可以通过对监控图像作进一步分析，以判别待检测硅棒的疏松状况，并相应生成警示信息，以指导工作人员对待检测硅棒进行匹配待检测硅棒疏松状况的处置措施。
在对所述监控图像包括的硅棒图像进行连通域分析之前，可以通过对监控图像进行动态阈值二值化处理，以过滤监控图像中包括的部分图像噪声，提升之后所进行的连通域分析的分析准确度。
如图4所示，经过动态阈值二值化处理后，可对监控图像包括的硅棒图像内的间隙图块进行清楚显示，图4中虚线框所示出图块即为间隙图块，所示间隙图块用于指示疏松料硅棒表面各颗粒之间的颗粒间隙。
需要指出的是，实际应用中，经过动态阈值二值化处理后，间隙图块应为白色图块，非间隙图块的区域部分则为黑色图块。
示例性的，在N大于或等于5的情况下，M的具体数值可以为5。
根据M个间隙图块的图块面积确定所述待检测硅棒的质量信息的过程可以为：
计算M个间隙图块的图块面积的平均值，以获得均值参数；
将均值参数与基准参数之比确定为质量参数；
根据质量参数生成所述质量信息。
其中，基准参数可理解为疏松料硅棒表面各颗粒之间的颗粒间隙所能达到的最大间隙面积，在所述质量参数小于或等于35％的情况下，所述质量信息用于指示所述待检测硅棒的菜花状况处于前期阶段；在所述质量参数为35％～65％的情况下，所述质量信息用于指示所述待检测硅棒的菜花状况处于中期阶段；在所述质量参数为65％～85％的情况下，所述质量信息用于指示所述待检测硅棒的菜花状况处于后期阶段；在所述质量参数大于85％的情况下，所述质量信息用于指示所述待检测硅棒的菜花状况处于末期阶段。
可选的，所述获取反应炉的监控图像，包括：
在所述反应炉的红外图像指示所述第一位置的温度与所述反应炉内的硅棒平均温度之差大于或等于温差阈值的情况下，和/或，在所述反应炉的硅粉含量的增幅速率大于或等于增速阈值的情况下，获取反应炉的监控图像。
如上，由于红外图像的图像分析难度较监控图像的图像分析难度更低， 所耗费的计算资源更少，因此，通过红外相机周期性获取红外图像，并对红外图像进行图像分析，仅在红外图像分析结果指示所述第一位置的温度与所述硅棒的平均温度之差大于或等于温度差值的情况下，才继续采集监控图像的方式，可降低监控图像的采集频次，进而降低硅棒亮点监测过程中的能耗，使硅棒亮点监测的成本降低。
需要说明的是，所述第一位置的温度可通过统计前述预设区间在红外图像中的光谱能量的均值获得，所述硅棒的平均温度可通过统计硅棒对应的目标区间在红外图像中的光谱能量的均值获得。
至于反应炉的硅粉含量，则可通过反应炉内预设置的硅粉含量传感器实时或周期性采集得到，例如，可以将硅粉含量传感器设置于反应炉的尾气管中，硅粉含量传感器可将反应炉的硅粉含量转换为电子信号(如电压信号/电流信号等)，随着反应炉内硅粉含量的升高，硅粉含量传感器对应输出的电子信号也会相应增高。
需要指出的是，在反应炉单次应用过程中，无雾化状态时，反应炉内的硅粉含量会有序升高，而随着雾化状况的出现，反应炉内的硅粉含量会出现激增的情况，基于上述雾化状况与反应炉内的硅粉含量的关联特性，可采用了监测反应炉的硅粉含量的增幅速率的方式，来对初步判定反应炉内是否存在雾化状况。
具体的，对反应炉的单次反应时间(一般为100小时，实际中也可根据实际需求进行适应性调整)进行均分处理，获得N个监测周期，在某一监测周期结束后，即相应计算该监测周期内反应炉的硅粉含量的增幅速率，并将所计算的增幅速率与增速阈值进行比较，在所计算的增幅速率大于或等于增速阈值时，则生成雾化预警信息。示例性的，若设定反应炉的单次反应时间为100小时，N的取值为3600，则每一监测周期的时间长度为1分钟；应用中，随着N的数值增加，所生成的雾化预警信息的时效性也会相应增加。
上述增速阈值可由用户根据经验确定，也可以通过统计反应炉处于雾化状态下的硅粉含量的第一增速和反应炉处于非雾化状态下的硅粉含量的第二增速确定，此时所确定的增速阈值大于所述第二增速，且小于所述第一增速。
可选的，所述输出报警信息之后，所述方法还包括：
在所述高亮像素点的像素坐标位于预设区间内的情况下，将所述高亮像素点的像素值与多个参考点的像素值分别进行相似度比较，获得多个第三相似参数，所述多个第三相似参数和所述多个参考点一一对应，所述多个参考点中的每一参考点对应一个温度区间的硅棒亮点；
将数值最大的所述第三相似参数对应的参考点确定为目标点；
根据所述目标点的亮点信息，确定位于所述预设区间的高亮像素点的亮点信息，其中，所述亮点信息包括对应的硅棒亮点的温度信息、以及对应的硅棒亮点的处理流程信息。
应用中，不同温度区间的硅棒亮点在图像所对应的像素值不同，且处理流程也有所差异，因此，通过在历史生产时期内收集不同温度对应的参考点信息以及对应的处理流程信息，并将当前生产时期内所确定异常点与参考点进行相似度比较，以确定与当前生产时期内所确定异常点最为接近的参考点，也即确定目标点，随后参照目标点的亮点信息，相应确定异常点的亮点信息。
基于上述方式，可以有效简化当前生产时期内所确定异常点的亮点信息确定流程，提升异常点的处理效率，使异常点所对应硅棒亮点带来的不利影响进一步降低。
需要说明的是，在一示例中，可以将目标点的亮点信息直接作为异常点的亮点信息使用；而在另一示例中，可以将目标点的亮点信息作为异常点的亮点信息的初始模板使用，令用户可根据实际情况对该初始模板进行修订或增删，以使所确定异常点的亮点信息与实际的硅棒亮点的温度以及处理流程相适配。
上述处理流程包括但不限于：改变反应炉内的温度值、改变反应炉内氢气和三氯氢硅的比值、改变反应炉内氢气输入量、改变反应炉内的电流值、改变反应炉内的三氯氢硅输入量以及停炉处理。
在一些可选的实施方式中，也可以将每一参考点对应的灰度图像与第一参考图像对应的灰度图像进行图像相似度比较，并获得每一参考点对应的灰度图像的图像相似参数，而后在多个图像相似参数将数值最大的图像相似参数对应的参考点确定为目标点。
可选的，根据所述目标点的亮点信息，确定位于所述预设区间的高亮像 素点的亮点信息之后，所述方法还包括：
在所述多个第三相似参数均小于或等于相似阈值的情况下，将所述高亮像素点作为新的参考点进行存储。
如上所述，在将异常点和多个参考点进行比较，并获得多个参考点中每一参考点对应的相似参数后，若所获得多个相似参数均小于或等于相似阈值，即当前生产时期所采集的异常点及监控图像与数据库存储的多个参考点的数据存在较大差异，此时可通过将所述异常点作为新的参考点存储于数据库内，以迭代式丰富数据库内存储的参考点数据，使后续所确定的异常点的亮点信息的数据准确性得到进一步提升。
可选的，所述方法还包括：
获取所述反应炉所在还原车间内的氢气浓度和氯化氢浓度；
当所述氢气浓度大于第一预设浓度值和/或所述氯化氢浓度大于第二预设浓度值时，输出气体泄漏信息。
示例性的，巡检机器人包括行走机构和机械臂，监测氢气和氯化氢浓度的气体传感器可以设置在行走机构上，也可以设置在机械臂上。
进一步地，设置在机械臂上可以更好的检测反应炉的泄露情况，因为机械臂需要伸向反应炉上的视镜处，气体传感器设置在机械臂上的同时，可以是气体传感器靠近视镜，而视镜本身就是检测反应炉是否发生气体泄露的关键部位。
在本申请实施例中，通过监测还原车间内的氢气浓度和氯化氢浓度；当氢气浓度大于第一预设浓度值和/或氯化氢浓度大于第二预设浓度值时，输出气体泄漏信息，以指示出现泄漏情况的反应炉位置和泄露状况，确保还原车间的安全作业。
可选的，所述方法还包括：
获取巡检机器人的初始巡检路线和反应炉的运行状态信息，所述运行状态信息包括反应炉的启停数据；
根据所述初始巡检路线和所述运行状态信息，确定所述巡检机器人的第一巡检路线，所述第一巡检路线上包括按次序排列的P个待巡检点，P为大于1的整数，每一所述待巡检点对应一个处于启动状态的反应炉；
基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留。
需要进行说明的是，初始巡检路线上的待巡检点的数量可以与还原车间在反应炉的数量相一致，且待巡检点与反应炉一一对应设置。
需要进行说明的是，根据初始巡检路线和运行状态信息确定第一巡检路线，是通过获知初始巡检路线上每一待巡检点对应的反应炉的运行状态信息，例如每一待巡检点对应的反应炉处于启动状态还是处于停止状态，保留对应的反应炉处于启动状态下的待巡检点，再将这些对应的反应炉处于启动状态下的待巡检点规划起来得到第一巡检路线。
进一步地，第一巡检路线对应的第一停留监测顺序，第一停留监测顺序表示第一巡检路线上的待巡检点之间具有的停留检测顺序。
应理解，巡检机器人调度平台获取到的初始巡检路线和还原车间内的反应炉的运行状态信息确定出第一巡检路线，机器调度平台按照第一巡检路线控制巡检机器人在还原车间内进行巡检；这样利用规划的具有先后监测顺序的路线进行监测，对需要监测的反应炉进行监测，对需要优先监测的反应炉优先监测，从而实现及时监测还原车间内多个反应炉内产品的运行状态。
可选的，所述基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留之后，所述方法还包括：
在所述第一巡检路线与障碍区域重叠的情况下，根据所述障碍区域、所述初始巡检路线和所述运行状态信息确定第二巡检路线，所述第二巡检路线上包括所述P个待巡检点中的至少部分待巡检点，且所述第二巡检路线不包括所述障碍区域，其中，所述障碍区域为阻碍所述巡检机器人移动的障碍物对应的区域；
按照所述第二巡检路线，控制所述巡检机器人继续巡检。
如上，通过检测第一巡检路线是否存在阻碍巡检机器人移动的障碍区域；并在第一巡检路线与障碍区域至少部分重叠的情况下，在排除阻碍区域的前提下，根据初始巡检路线和运行状态信息确定第二巡检路线，以使巡检机器人在巡检过程中避开障碍物，提高巡检时的工作效率和安全性。
进一步的，所述障碍区域根据所述障碍物的移动速度和/或所述障碍物的 体积确定。
如上，在障碍物为静止的物体时，基于障碍物的体积即可确定障碍区域的大小；而在障碍物为运动中的物体时，则需根据障碍物的移动速度以及障碍物的体积，来共同确定障碍物对应的障碍区域，以降低巡检机器人与障碍物发生碰撞的概率，保障巡检机器人的使用安全。
在一些示例中，获取障碍物的移动速度可以根据巡检机器人本体上安装红外传感器获得，也可以根据巡检机器人本体上安装雷达传感器获取，还可以根据巡检机器人本体上安装的摄像头获得。
当在巡检机器人本体上安装红外传感器时，红外传感器可以是两个，也可以是多个。当红外传感器的数量为两个时，两个红外传感器间隔一定距离设置。
当在巡检机器人本体上安装摄像头时，巡检机器人获取其行进路线上的图像信息，再利用帧间差分法，去计算其行进路线上障碍物的移动速度。
参见图5，图5是本申请实施例提供的多晶硅监控装置500的结构示意图，如图5所示，多晶硅监控装置500包括：
图像获取模块501，用于获取反应炉的监控图像；
图像分析模块502，用于对所述监控图像进行图像分析，获得分析结果；
报警模块503，用于在所述分析结果满足报警条件的情况下，输出报警信息；
其中，所述报警条件包括以下至少一项：
所述监控图像中黑斑区域的数目参数大于或等于第一阈值；
所述监控图像中黑斑区域的面积参数大于或等于第二阈值；
所述监控图像对应的第一相似参数小于或等于第三阈值，所述第一相似参数用于表征所述监控图像和基准图像之间的图像相似度，所述基准图像为所述反应炉内在无雾化的状态下采集的图像；
所述监控图像包括的高亮像素点位于预设区间内，所述预设区间用于表征第一位置在所述监控图像内对应的像素区间，所述第一位置为所述反应炉内用于放置硅棒的位置，所述高亮像素点为所述监控图像中对应像素值大于像素阈值的像素点。
可选的，所述图像分析模块502，包括：
灰度处理子模块，用于对所述监控图像进行灰度化处理，获得灰度图像；
连通域分析子模块，用于对所述灰度图像包括的硅棒图像进行连通域分析，确定所述硅棒图像包括的至少一个黑斑图块；
黑斑定位子模块，用于根据所述至少一个黑斑图块，确定所述监控图像中黑斑区域的数目参数和所述监控图像中黑斑区域的面积参数。
可选的，所述多晶硅监控装置500还包括：
光信号检测模块，用于对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据；
光衰确定模块，用于根据所述检测信号数据与所述基准光信号对应的基准信号数据之间的差异，确定光衰参数，其中，所述光衰参数用于表征所述基准光信号的光能量衰减幅度；
雾化警示模块，用于在所述光衰参数大于或等于光衰阈值的情况下，生成雾化信息。
可选的，所述多晶硅监控装置500还包括：
调参确定模块，用于根据所述光衰参数和预设置的转换系数，确定调整参数；
反应炉调控模块，用于根据所述调整参数，对所述反应炉的输入原料的输入量进行调整。
可选的，所述多晶硅监控装置500还包括：
第一复检模块，用于获取所述参照设备在第一检测时刻发出的基准光信号对应的第一光衰值、以及所述参照设备在第二检测时刻发出的基准光信号对应的第二光衰值；
第二复检模块，用于将所述第一光衰值和所述第二光衰值之间的差值确定为光衰变化参数；
复检警示模块，用于在所述光衰变化参数小于或等于变化阈值的情况下，生成雾化警报，所述雾化警报用于指示所述反应炉的位置。
可选的，所述多晶硅监控装置500还包括：
基准检测模块，用于对所述参照设备在初始时刻发出的基准光信号的进 行检测，得到测试信号数据，所述初始时刻为所述反应炉的启动时刻；
基准更新模块，用于根据所述测试信号数据对所述基准信号数据进行更新。
可选的，所述多晶硅监控装置500还包括：
指令传输模块，用于向所述参照设备发送检测指令，以使所述参照设备基于所述检测指令发出所述基准光信号。
可选的，所述多晶硅监控装置500还包括：
雾化程度分析模块，用于在所述第一相似参数小于或等于第三阈值的情况下，根据所述监控图像和至少两个雾化图像组，获得至少两个第二相似参数，所述至少两个第二相似参数与所述至少两个雾化图像组一一对应，其中，所述至少两个雾化图像组分别指示所述反应炉的不同雾化程度，所述第二相似参数用于表征对应的雾化图像组与所述监控图像之间的图像相似度；
雾化程度判定模块，用于在所述至少两个雾化图像组中确定目标雾化图像组，其中，所述目标雾化图像组对应的第二相似参数为所述至少两个第二相似参数中数值最大的相似参数；
雾化处理模块，用于输出所述目标雾化图像组关联的雾化警示信息。
可选的，所述多晶硅监控装置500还包括：
第一雾化调控模块，用于根据所述雾化警示信息，确定所述反应炉的调控信息；
第二雾化调控模块，用于根据所述调控信息对所述反应炉的反应参数进行调控；
其中，所述反应参数包括电流值、氢气输入量和三氯氢硅输入量的比值中的至少一项。
可选的，所述多晶硅监控装置500还包括：
图像二值处理模块，用于对所述监控图像进行二值化处理，获得二值图像；
间距确定模块，用于根据所述二值图像，确定像素间距，其中，所述像素间距用于表征对应第一硅棒的像素区域和对应第二硅棒的像素区域之间的最短距离，所述第一硅棒和所述第二硅棒为所述反应炉内相邻设置的两个硅 棒；
间距修正模块，用于根据修正参数对所述像素间距进行修正，获取第一间距，其中，所述修正参数用于表征所述监控图像中单位像素块对应的物理长度，所述第一间距用于表征当前时刻所述第一硅棒和所述第二硅棒之间的间距；
硅棒参数确定模块，用于根据所述硅棒间距和初始间距，确定目标参数，其中，所述初始间距用于表征初始时刻所述第一硅棒和所述第二硅棒之间的间距，所述初始时刻为所述反应炉的启动时刻，所述目标参数用于表征当前时刻所述反应炉内的硅棒直径。
可选的，所述间距确定模块，包括：
特征识别子模块，用于对所述二值图像进行特征识别，确定第一像素点集和第二像素点集，其中，所述第一像素点集为所述第一硅棒的靠近所述第二硅棒的边缘部分对应的像素点集合，所述第二像素点集为所述第二硅棒的靠近所述第一硅棒的边缘部分对应的像素点集合；
间距确定子模块，用于根据所述第一像素点集和所述第二像素点集，确定像素间距。
可选的，所述间距确定子模块，包括：
点距计算单元，用于计算位于同一像素高度的第一像素点和第二像素点之间的点间距，其中，所述第一像素点为所述第一像素点集中的任一像素点，所述第二像素点为所述第二像素点集中的任一像素点；
间距确定单元，用于将多个像素高度分别对应的点间距的均值确定为所述像素间距。
可选的，所述多晶硅监控装置500还包括：
校正图像获取模块，用于获取所述反应炉在初始时段的校正图像；
校正图像分析模块，用于对所述校正图像进行图像分析，获得硅芯目标间距，其中，所述硅芯目标间距用于表征对应所述第一硅棒硅芯的像素点集和对应所述第二硅棒硅芯的像素点集之间的间距；
修正参数确定模块，用于根据硅芯基准间距和所述硅芯目标间距，确定所述修正参数，其中，所述硅芯基准间距用于表征所述第一硅棒硅芯和所述 第二硅棒硅芯之间的物理距离。
可选的，所述初始间距根据初始时刻获取的第一参考图像确定，所述第一参考图像占用的内存空间和所述监控图像占用的内存空间均大于或等于内存阈值。
可选的，所述多晶硅监控装置500还包括：
图像比较模块，用于将所述监控图像包括的硅棒图像和第二参考图像进行图像相似度比较，获得比较结果，其中，所述第二参考图像用于表征疏松料硅棒；
间隙定位模块，用于在所述比较结果指示所述反应炉内包括疏松料硅棒的情况下，对所述监控图像包括的硅棒图像进行连通域分析，确定N个间隙图块，所述间隙图块用于指示硅棒表面的多个硅颗粒之间的间隙，N为正整数；
质量分析模块，用于根据M个间隙图块的图块面积确定所述反应炉内硅棒的质量信息，其中，所述M个间隙图块为所述N个间隙图块中图块面积从大到小排列的前M个间隙图块。
可选的，所述图像获取模块，具体用于：
在所述反应炉的红外图像指示所述第一位置的温度与所述反应炉内的硅棒平均温度之差大于或等于温差阈值的情况下，和/或，在所述反应炉的硅粉含量的增幅速率大于或等于增速阈值的情况下，获取反应炉的监控图像。
可选的，所述多晶硅监控装置500还包括：
像素点比较模块，用于在所述高亮像素点的像素坐标位于预设区间内的情况下，将所述高亮像素点的像素值与多个参考点的像素值分别进行相似度比较，获得多个第三相似参数，所述多个第三相似参数和所述多个参考点一一对应，所述多个参考点中的每一参考点对应一个温度区间的硅棒亮点；
目标点确定模块，用于将数值最大的所述第三相似参数对应的参考点确定为目标点；
亮点信息确定模块，用于根据所述目标点的亮点信息，确定位于所述预设区间的高亮像素点的亮点信息，其中，所述亮点信息包括对应的硅棒亮点的温度信息、以及对应的硅棒亮点的处理流程信息。
可选的，所述多晶硅监控装置500还包括：
数据迭代模块，用于在所述多个第三相似参数均小于或等于相似阈值的情况下，将所述高亮像素点作为新的参考点进行存储。
可选的，所述多晶硅监控装置500还包括：
气体浓度获取模块，用于获取所述反应炉所在还原车间内的氢气浓度和氯化氢浓度；
漏气警示模块，用于当所述氢气浓度大于第一预设浓度值和/或所述氯化氢浓度大于第二预设浓度值时，输出气体泄漏信息。
可选的，所述多晶硅监控装置500还包括：
路线信息获取模块，用于获取巡检机器人的初始巡检路线和反应炉的运行状态信息，所述运行状态信息包括反应炉的启停数据；
第一巡检路线确定模块，用于根据所述初始巡检路线和所述运行状态信息，确定所述巡检机器人的第一巡检路线，所述第一巡检路线上包括按次序排列的P个待巡检点，P为大于1的整数，每一所述待巡检点对应一个处于启动状态的反应炉；
巡检控制模块，用于基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留。
可选的，所述多晶硅监控装置500还包括：
第二巡检路线确定模块，用于在所述第一巡检路线与障碍区域重叠的情况下，根据所述障碍区域、所述初始巡检路线和所述运行状态信息确定第二巡检路线，所述第二巡检路线上包括所述P个待巡检点中的至少部分待巡检点，且所述第二巡检路线不包括所述障碍区域，其中，所述障碍区域为阻碍所述巡检机器人移动的障碍物对应的区域；
所述巡检控制模块还用于：按照所述第二巡检路线，控制所述巡检机器人继续巡检。
可选的，所述障碍区域根据所述障碍物的移动速度和/或所述障碍物的体积确定。
本申请实施例提供的多晶硅监控装置500能够实现上述方法实施例中的各个过程，为避免重复，这里不再赘述。
请参见图6，图6是本申请实施例提供的一种电子设备的结构示意图，如图6所示，电子设备包括：处理器601、存储器602及存储在存储器602上并可在处理器601上运行的程序6021。
程序6021被处理器601执行时可实现图1对应的方法实施例中的任意步骤及达到相同的有益效果，此处不再赘述。
本领域普通技术人员可以理解实现上述实施例方法的全部或者部分步骤是可以通过程序指令相关的硬件来完成，所述的程序可以存储于一可读取介质中。
本申请实施例还提供一种可读存储介质，所述可读存储介质上存储有计算机程序，所述计算机程序被处理器执行时可实现上述图1对应的方法实施例中的任意步骤，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例的计算机可读存储介质，可以采用一个或多个计算机可读的介质的任意组合。计算机可读介质可以是计算机可读信号介质或者计算机可读存储介质。计算机可读存储介质例如可以是电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。计算机可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式计算机磁盘、硬盘、随机存取存储器(Random Access Memory，RAM)、只读存储器(Read-Only Memory，ROM)、可擦式可编程只读存储器(Erasable Programmable Read-Only Memory，EPROM，也称为闪存)、光纤、便携式紧凑磁盘只读存储器(Compact Disc Read-Only Memory，CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。在本文件中，计算机可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。
计算机可读的信号介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了计算机可读的程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。计算机可读的信号介质还可以是计算机可读存储介质以外的任何计算机可读介质，该计算机可读介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。
存储介质上包含的程序代码可以用任何适当的介质传输，包括但不限于无线、电线、光缆、射频(Radio Frequency，RF)等等，或者上述的任意合适的组合。
可以以一种或多种程序设计语言或其组合来编写用于执行本申请操作的计算机程序代码，所述程序设计语言包括面向对象的程序设计语言—诸如Java、Smalltalk、C++，还包括常规的过程式程序设计语言—诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或终端上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络——包括局域网(Local Area Network，LAN)或广域网(Wide Area Network，WAN)—连接到用户计算机，或者，可以连接到外部计算机(例如利用因特网服务提供商来通过因特网连接)。
以上所述是本申请实施例的可选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本申请所述原理的前提下，还可以作出若干改进和润饰，这些改进和润饰也应视为本申请的保护范围。

Claims (25)

  1. 一种多晶硅监控方法，所述方法包括：
    获取反应炉的监控图像；
    对所述监控图像进行图像分析，获得分析结果；
    在所述分析结果满足报警条件的情况下，输出报警信息；
    其中，所述报警条件包括以下至少一项：
    所述监控图像中黑斑区域的数目参数大于或等于第一阈值；
    所述监控图像中黑斑区域的面积参数大于或等于第二阈值；
    所述监控图像对应的第一相似参数小于或等于第三阈值，所述第一相似参数用于表征所述监控图像和基准图像之间的图像相似度，所述基准图像为所述反应炉内在无雾化的状态下采集的图像；
    所述监控图像包括的高亮像素点位于预设区间内，所述预设区间用于表征第一位置在所述监控图像内对应的像素区间，所述第一位置为所述反应炉内用于放置硅棒的位置，所述高亮像素点为所述监控图像中对应像素值大于像素阈值的像素点。
  2. 根据权利要求1所述的方法，其中，所述对所述监控图像进行图像分析，获得分析结果，包括：
    对所述监控图像进行灰度化处理，获得灰度图像；
    对所述灰度图像包括的硅棒图像进行连通域分析，确定所述硅棒图像包括的至少一个黑斑图块；
    根据所述至少一个黑斑图块，确定所述监控图像中黑斑区域的数目参数和所述监控图像中黑斑区域的面积参数。
  3. 根据权利要求1所述的方法，其中，所述方法还包括：
    对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据；
    根据所述检测信号数据与所述基准光信号对应的基准信号数据之间的差异，确定光衰参数，其中，所述光衰参数用于表征所述基准光信号的光能量衰减幅度；
    在所述光衰参数大于或等于光衰阈值的情况下，生成雾化信息。
  4. 根据权利要求3所述的方法，其中，所述生成雾化信息之后，所述方法还包括：
    根据所述光衰参数和预设置的转换系数，确定调整参数；
    根据所述调整参数，对所述反应炉的输入原料的输入量进行调整。
  5. 根据权利要求4所述的方法，其中，所述根据所述调整参数，对所述反应炉的输入原料的输入量进行调整之后，所述方法还包括：
    获取所述参照设备在第一检测时刻发出的基准光信号对应的第一光衰值、以及所述参照设备在第二检测时刻发出的基准光信号对应的第二光衰值；
    将所述第一光衰值和所述第二光衰值之间的差值确定为光衰变化参数；
    在所述光衰变化参数小于或等于变化阈值的情况下，生成雾化警报，所述雾化警报用于指示所述反应炉的位置。
  6. 根据权利要求3所述的方法，其中，所述对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据之前，所述方法还包括：
    对所述参照设备在初始时刻发出的基准光信号的进行检测，得到测试信号数据，所述初始时刻为所述反应炉的启动时刻；
    根据所述测试信号数据对所述基准信号数据进行更新。
  7. 根据权利要求3所述的方法，其中，所述对所述反应炉内的参照设备在当前时刻发出的基准光信号进行检测，获得检测信号数据之前，所述方法还包括：
    向所述参照设备发送检测指令，以使所述参照设备基于所述检测指令发出所述基准光信号。
  8. 根据权利要求1所述的方法，其中，所述输出报警信息之后，所述方法还包括：
    在所述第一相似参数小于或等于第三阈值的情况下，根据所述监控图像和至少两个雾化图像组，获得至少两个第二相似参数，所述至少两个第二相似参数与所述至少两个雾化图像组一一对应，其中，所述至少两个雾化图像组分别指示所述反应炉的不同雾化程度，所述第二相似参数用于表征对应的 雾化图像组与所述监控图像之间的图像相似度；
    在所述至少两个雾化图像组中确定目标雾化图像组，其中，所述目标雾化图像组对应的第二相似参数为所述至少两个第二相似参数中数值最大的相似参数；
    输出所述目标雾化图像组关联的雾化警示信息。
  9. 根据权利要求8所述的方法，其中，所述输出所述目标雾化图像组关联的雾化警示信息之后，所述方法还包括：
    根据所述雾化警示信息，确定所述反应炉的调控信息；
    根据所述调控信息对所述反应炉的反应参数进行调控；
    其中，所述反应参数包括电流值、氢气输入量和三氯氢硅输入量的比值中的至少一项。
  10. 根据权利要求1所述的方法，其中，所述获取反应炉的监控图像之后，所述方法还包括：
    对所述监控图像进行二值化处理，获得二值图像；
    根据所述二值图像，确定像素间距，其中，所述像素间距用于表征对应第一硅棒的像素区域和对应第二硅棒的像素区域之间的最短距离，所述第一硅棒和所述第二硅棒为所述反应炉内相邻设置的两个硅棒；
    根据修正参数对所述像素间距进行修正，获取第一间距，其中，所述修正参数用于表征所述监控图像中单位像素块对应的物理长度，所述第一间距用于表征当前时刻所述第一硅棒和所述第二硅棒之间的间距；
    根据所述硅棒间距和初始间距，确定目标参数，其中，所述初始间距用于表征初始时刻所述第一硅棒和所述第二硅棒之间的间距，所述初始时刻为所述反应炉的启动时刻，所述目标参数用于表征当前时刻所述反应炉内的硅棒直径。
  11. 根据权利要求10所述的方法，其中，所述根据所述二值图像，确定像素间距，包括：
    对所述二值图像进行特征识别，确定第一像素点集和第二像素点集，其中，所述第一像素点集为所述第一硅棒的靠近所述第二硅棒的边缘部分对应的像素点集合，所述第二像素点集为所述第二硅棒的靠近所述第一硅棒的边 缘部分对应的像素点集合；
    根据所述第一像素点集和所述第二像素点集，确定像素间距。
  12. 根据权利要求11所述的方法，其中，所述根据所述第一像素点集和所述第二像素点集，确定像素间距，包括：
    计算位于同一像素高度的第一像素点和第二像素点之间的点间距，其中，所述第一像素点为所述第一像素点集中的任一像素点，所述第二像素点为所述第二像素点集中的任一像素点；
    将多个像素高度分别对应的点间距的均值确定为所述像素间距。
  13. 根据权利要求10所述的方法，其中，所述根据修正参数对所述像素间距进行修正，获得第一间距之前，所述方法还包括：
    获取所述反应炉在初始时段的校正图像；
    对所述校正图像进行图像分析，获得硅芯目标间距，其中，所述硅芯目标间距用于表征对应所述第一硅棒硅芯的像素点集和对应所述第二硅棒硅芯的像素点集之间的间距；
    根据硅芯基准间距和所述硅芯目标间距，确定所述修正参数，其中，所述硅芯基准间距用于表征所述第一硅棒硅芯和所述第二硅棒硅芯之间的物理距离。
  14. 根据权利要求10所述的方法，其中，所述初始间距根据初始时刻获取的第一参考图像确定，所述第一参考图像占用的内存空间和所述监控图像占用的内存空间均大于或等于内存阈值。
  15. 根据权利要求1所述的方法，其中，所述获取反应炉的监控图像之后，所述方法还包括：
    将所述监控图像包括的硅棒图像和第二参考图像进行图像相似度比较，获得比较结果，其中，所述第二参考图像用于表征疏松料硅棒；
    在所述比较结果指示所述反应炉内包括疏松料硅棒的情况下，对所述监控图像包括的硅棒图像进行连通域分析，确定N个间隙图块，所述间隙图块用于指示硅棒表面的多个硅颗粒之间的间隙，N为正整数；
    根据M个间隙图块的图块面积确定所述反应炉内硅棒的质量信息，其中，所述M个间隙图块为所述N个间隙图块中图块面积从大到小排列的前M个 间隙图块。
  16. 根据权利要求1所述的方法，其中，所述获取反应炉的监控图像，包括：
    在所述反应炉的红外图像指示所述第一位置的温度与所述反应炉内的硅棒平均温度之差大于或等于温差阈值的情况下，和/或，在所述反应炉的硅粉含量的增幅速率大于或等于增速阈值的情况下，获取反应炉的监控图像。
  17. 根据权利要求1所述的方法，其中，所述输出报警信息之后，所述方法还包括：
    在所述高亮像素点的像素坐标位于预设区间内的情况下，将所述高亮像素点的像素值与多个参考点的像素值分别进行相似度比较，获得多个第三相似参数，所述多个第三相似参数和所述多个参考点一一对应，所述多个参考点中的每一参考点对应一个温度区间的硅棒亮点；
    将数值最大的所述第三相似参数对应的参考点确定为目标点；
    根据所述目标点的亮点信息，确定位于所述预设区间的高亮像素点的亮点信息，其中，所述亮点信息包括对应的硅棒亮点的温度信息、以及对应的硅棒亮点的处理流程信息。
  18. 根据权利要求17所述的方法，其中，根据所述目标点的亮点信息，确定位于所述预设区间的高亮像素点的亮点信息之后，所述方法还包括：
    在所述多个第三相似参数均小于或等于相似阈值的情况下，将所述高亮像素点作为新的参考点进行存储。
  19. 根据权利要求1所述的方法，其中，所述方法还包括：
    获取所述反应炉所在还原车间内的氢气浓度和氯化氢浓度；
    当所述氢气浓度大于第一预设浓度值和/或所述氯化氢浓度大于第二预设浓度值时，输出气体泄漏信息。
  20. 根据权利要求1所述的方法，其中，所述方法还包括：
    获取巡检机器人的初始巡检路线和反应炉的运行状态信息，所述运行状态信息包括反应炉的启停数据；
    根据所述初始巡检路线和所述运行状态信息，确定所述巡检机器人的第一巡检路线，所述第一巡检路线上包括按次序排列的P个待巡检点，P为大 于1的整数，每一所述待巡检点对应一个处于启动状态的反应炉；
    基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留。
  21. 根据权利要求20所述的方法，其中，所述基于所述第一巡检路线，控制所述巡检机器人在所述P个待巡检点之间依次停留之后，所述方法还包括：
    在所述第一巡检路线与障碍区域重叠的情况下，根据所述障碍区域、所述初始巡检路线和所述运行状态信息确定第二巡检路线，所述第二巡检路线上包括所述P个待巡检点中的至少部分待巡检点，且所述第二巡检路线不包括所述障碍区域，其中，所述障碍区域为阻碍所述巡检机器人移动的障碍物对应的区域；
    按照所述第二巡检路线，控制所述巡检机器人继续巡检。
  22. 根据权利要求21所述的方法，其中，所述障碍区域根据所述障碍物的移动速度和/或所述障碍物的体积确定。
  23. 一种多晶硅监控装置，所述装置包括：
    图像获取模块，用于获取反应炉的监控图像；
    图像分析模块，用于对所述监控图像进行图像分析，获得分析结果；
    报警模块，用于在所述分析结果满足报警条件的情况下，输出报警信息；
    其中，所述报警条件包括以下至少一项：
    所述监控图像中黑斑区域的数目参数大于或等于第一阈值；
    所述监控图像中黑斑区域的面积参数大于或等于第二阈值；
    所述监控图像对应的第一相似参数小于或等于第三阈值，所述第一相似参数用于表征所述监控图像和基准图像之间的图像相似度，所述基准图像为所述反应炉内在无雾化的状态下采集的图像；
    所述监控图像包括的高亮像素点位于预设区间内，所述预设区间用于表征第一位置在所述监控图像内对应的像素区间，所述第一位置为所述反应炉内用于放置硅棒的位置，所述高亮像素点为所述监控图像中对应像素值大于像素阈值的像素点。
  24. 一种电子设备，包括处理器、存储器及存储在所述存储器上并可在 所述处理器上运行的计算机程序，所述计算机程序被所述处理器执行时实现如权利要求1至22中任一项所述的多晶硅监控方法的步骤。
  25. 一种可读存储介质，所述可读存储介质上存储有计算机程序，所述计算机程序被处理器执行时实现如权利要求1至22中任一项所述的多晶硅监控方法的步骤。
PCT/CN2023/113706 2022-11-14 2023-08-18 一种多晶硅监控方法、装置及相关设备 WO2024103871A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211425068.5A CN115775229A (zh) 2022-11-14 2022-11-14 一种多晶硅监控方法、装置及相关设备
CN202211425068.5 2022-11-14

Publications (1)

Publication Number Publication Date
WO2024103871A1 true WO2024103871A1 (zh) 2024-05-23

Family

ID=85389071

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/113706 WO2024103871A1 (zh) 2022-11-14 2023-08-18 一种多晶硅监控方法、装置及相关设备

Country Status (2)

Country Link
CN (1) CN115775229A (zh)
WO (1) WO2024103871A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115775229A (zh) * 2022-11-14 2023-03-10 新特能源股份有限公司 一种多晶硅监控方法、装置及相关设备
CN116740651B (zh) * 2023-08-11 2023-10-17 南京吾悦农业科技有限公司 基于智能决策的食用菌栽培监控方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105321289A (zh) * 2015-12-04 2016-02-10 中国建筑设计院有限公司 一种全天候监控图像智能分析与报警系统与方法
CN108762187A (zh) * 2018-05-29 2018-11-06 南通理工学院 一种基于mcgs的加热反应炉监控系统
CN111210462A (zh) * 2019-12-30 2020-05-29 视联动力信息技术股份有限公司 一种报警方法及装置
CN115775229A (zh) * 2022-11-14 2023-03-10 新特能源股份有限公司 一种多晶硅监控方法、装置及相关设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105321289A (zh) * 2015-12-04 2016-02-10 中国建筑设计院有限公司 一种全天候监控图像智能分析与报警系统与方法
CN108762187A (zh) * 2018-05-29 2018-11-06 南通理工学院 一种基于mcgs的加热反应炉监控系统
CN111210462A (zh) * 2019-12-30 2020-05-29 视联动力信息技术股份有限公司 一种报警方法及装置
CN115775229A (zh) * 2022-11-14 2023-03-10 新特能源股份有限公司 一种多晶硅监控方法、装置及相关设备

Also Published As

Publication number Publication date
CN115775229A (zh) 2023-03-10

Similar Documents

Publication Publication Date Title
WO2024103871A1 (zh) 一种多晶硅监控方法、装置及相关设备
US11736062B2 (en) Method and apparatus for determining operating state of photovoltaic array, device and storage medium
CN110031477A (zh) 基于图像监控数据的桥梁关键构件病害预警系统及方法
CN107063946B (zh) 视频生球粒径仪及其生球粒径状态检测方法
JP2006275743A (ja) 欠陥検査方法
CN110196252B (zh) 坡口缺陷检测算法及装置
CN115937097A (zh) 基于人工智能的工业机器人视觉图像识别系统
CN116388402B (zh) 一种应用于变电设备的异常报警分析方法
CN112183369A (zh) 一种用于变电站无人巡检的指针仪表读数识别方法
CN114187511A (zh) 一种输电全景大数据智能监控方法
CN205408031U (zh) 一种输变配电设备的智能监控系统
CN116229854A (zh) 一种液晶显示屏玻璃显示内容异常的监测装置及方法
CN111062920B (zh) 用于生成半导体检测报告的方法及装置
CN110633657B (zh) 一种高炉崩料预测方法
CN110414606B (zh) 输电线路通道可视化告警区域划分数量的确定方法
CN115830345A (zh) 一种雾化监测方法、装置及相关设备
CN116109554A (zh) 一种硅棒监测方法、装置及相关设备
CN115761628A (zh) 一种硅棒监控方法、装置及相关设备
CN113992151A (zh) 光伏阵列工作状态的确定方法、装置、设备及存储介质
CN113807209A (zh) 车位检测方法、装置、电子设备及存储介质
CN112305987A (zh) 一种基于物联网的水利自动化管理系统
CN115718003A (zh) 一种雾化检测方法、装置及相关设备
CN115690076A (zh) 硅棒的外观检测方法及相关设备
CN114459606B (zh) 一种热像仪对日保护方法、系统、智能终端以及存储介质
CN114509020A (zh) 炼钢转炉副枪弯曲程度的检测系统及方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23890310

Country of ref document: EP

Kind code of ref document: A1