An on-line monitoring method and system applied to semiconductor production

Abstract

The application provides an online monitoring method and system applied to semiconductor production, wherein the method comprises the following steps: acquiring acquisition data of acquisition nodes corresponding to various devices of semiconductor production through a LonWorks field bus; analyzing the acquisition data to determine whether an abnormality occurs in the various devices; when the abnormality occurs, sending abnormality reminding information to various monitoring nodes corresponding to the device where the abnormality occurs and acquiring video data through a video acquisition module corresponding to the various devices; and generating an emergency treatment record based on the abnormality reminding information and the video data. The online monitoring method applied to semiconductor production can timely and efficiently determine the running state of the various devices and effectively monitor the semiconductor production process.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of online monitoring technology, in particular to an online monitoring method and system applied to semiconductor production. BACKGROUND

[0002] Semiconductor manufacturing production refers to the process for manufacturing semiconductor devices; its process is relatively complex, including a multi-step sequence of photolithography and chemical processing steps (such as surface passivation, thermal oxidation, planar diffusion and junction isolation), in which electronic circuit semiconductor materials are gradually formed on a wafer made of pure silicon. Various mechanical equipment (such as single crystal furnace, vapor phase epitaxy furnace, photolithography machine, reactive ion etching system, wafer thinning machine, wafer dicing machine, wire bonding machine, etc.) is used in the production process, and how to ensure the safe and effective operation of each device in the semiconductor production process is the premise for ensuring the quality of semiconductor production, so how to timely and efficiently determine the running state of each device and effectively monitor the semiconductor production process is a technical problem to be solved. SUMMARY

[0003] One of the purposes of the present application is to provide an online monitoring method applied to semiconductor production, which timely and efficiently determines the running state of each device and effectively monitors the semiconductor production process.

[0004] The online monitoring method applied to semiconductor production provided by the embodiment of the present application comprises:

[0005] Obtaining the collection data of the collection nodes corresponding to each device of the semiconductor production through the LonWorks field bus;

[0006] Analyzing the collection data to determine whether an abnormality occurs in each device;

[0007] When an abnormality occurs, sending an abnormality reminding information to each monitoring node corresponding to the device where the abnormality occurs and obtaining the video data through the video collection module corresponding to each device;

[0008] Generating an emergency handling record based on the abnormality reminding information and the video data.

[0009] Preferably, the collection data of the collection nodes corresponding to each device of the semiconductor production is obtained through the LonWorks field bus, comprising:

[0010] Sending a data acquisition request to the collection node through the LonWorks field bus;

[0011] After receiving the response of the collection node to the data acquisition request, obtaining the data packet numbering rule of the collection node through the control node of the LonWorks field bus;

[0012] Based on the data packet numbering rule, the received first data packet corresponding to the collection data of the collection node is checked;

[0013] Based on the checking result, a secondary data acquisition request is constructed;

[0014] The second data packet sent by the collection node for the secondary data acquisition request is received;

[0015] The first data packet and the second data packet are parsed, and the collection data is obtained.

[0016] Preferably, the collection data is analyzed to determine whether each device has an abnormality, including:

[0017] Determine the device information corresponding to the device;

[0018] Based on the device information, the feature extraction rule corresponding to the device and the abnormality analysis sub-library are extracted from the preset analysis library;

[0019] Based on the feature extraction rule, the collection data is feature extracted to obtain a plurality of feature values;

[0020] Based on the plurality of feature values, an analysis feature set is constructed;

[0021] The analysis result corresponding to the analysis feature set is retrieved from the abnormality analysis sub-library;

[0022] The analysis result is parsed to determine whether the device has an abnormality and the information of the abnormality.

[0023] Preferably, when the abnormality occurs, the abnormality reminding information is sent to each monitoring node corresponding to the device having the abnormality, and the video data is obtained through the video acquisition module corresponding to each device, including:

[0024] The information of the abnormality is parsed to determine the component corresponding to the abnormality and the position of the component in the device;

[0025] Based on the component and the position of the component in the device, a preset video acquisition module retrieval table is queried to determine the corresponding video acquisition module and obtain the video data.

[0026] Preferably, based on the abnormality reminding information and the video data, an emergency handling record is generated, including:

[0027] The abnormality reminding information is parsed to determine each reminding data item and the abnormality code corresponding to the abnormality;

[0028] Based on the abnormality code, the corresponding emergency handling record generation template is retrieved;

[0029] Each reminding data item is filled into the first position corresponding to each reminding data in the emergency handling record generation template.

[0030] clipping the video data to obtain the on-site processing video of the processing personnel;

[0031] filling the start time of the on-site processing video into a preset second position in the emergency processing record generation template;

[0032] filling the on-site processing video into a preset third position in the emergency processing record generation template.

[0033] Preferably, the clipping the video data to obtain the on-site processing video of the processing personnel comprises:

[0034] Step S1: extracting frame images of the video data according to a preset interval number to obtain first to-be-analyzed images;

[0035] Step S2: sequentially performing personnel identification on the first to-be-analyzed images, and setting preset first statistical parameter and second statistical parameter values to initial values when a preset processing personnel appears;

[0036] Step S3: when the processing personnel is also identified in the next first to-be-analyzed image, the first statistical parameter is incremented by one;

[0037] Step S4: when the processing personnel is not identified in the next first to-be-analyzed image, the second statistical parameter is incremented by one;

[0038] Step S5: cyclically executing the step S3 and the step S4 until the first statistical parameter value reaches a first threshold value within a preset first time interval;

[0039] Step S6: setting a point at which the previous first statistical parameter is set to the initial value as a start point of the on-site processing video;

[0040] In the cyclic execution process of the step S5, when the second statistical parameter reaches a preset second threshold value within a preset second time interval, the first statistical parameter and the second statistical parameter are set to the initial values; and the first time interval is greater than the second time interval.

[0041] Preferably, when there are multiple video acquisition modules for shooting, the clipping the video data to obtain the on-site processing video of the processing personnel comprises:

[0042] determining whether there are multiple frame images containing the maintenance personnel at the same time;

[0043] when there are, determining an included angle between a facing vector representing a facing direction of the maintenance personnel in each frame image and a center vector of a video acquisition module corresponding to the frame image;

[0044] determining a first score value based on the included angle;

[0045] Determine the proportion of the repair personnel being blocked in each frame image;

[0046] Determine a second score value based on the proportion;

[0047] Determine a total score value based on the first score value and the second score value;

[0048] Based on the total score value, screen multiple frame images within the same time to determine images for constructing the on-site processing video.

[0049] Preferably, based on the abnormal prompt information and the video data, the emergency treatment record is generated, and further comprises:

[0050] Analyze the on-site processing video to determine the replacement part information and the use order of the parts;

[0051] Generate a first mark based on the part information and the use order of the parts;

[0052] Based on the first mark, mark the completed emergency treatment record generation template;

[0053] Among them, analyzing the on-site processing video to determine the replacement part information and the use order of the parts comprises:

[0054] Identify each frame image of the on-site processing video, and determine the frame image with the first mark area as a second analysis image;

[0055] Extract the first mark area in each second analysis image to obtain each third analysis image;

[0056] Based on the preset part recognition model, identify the third analysis image to obtain the part code set corresponding to each third analysis image;

[0057] Based on the part code set, screen each third analysis image to obtain multiple fourth analysis images;

[0058] Based on the part code set corresponding to each fourth analysis image, construct a total part code set;

[0059] Compare the part code sets corresponding to two adjacent fourth analysis images in sequence to determine the difference between the two part code sets;

[0060] Based on the difference between the part code sets corresponding to adjacent fourth analysis images, determine the use order of the parts.

[0061] Preferably, based on the abnormal prompt information and the video data, the emergency treatment record is generated, and further comprises:

[0062] Analyze the on-site processing video to determine the replacement part information and the use order of the parts;

[0063] generate the second identification based on the accessory information of replacement use and the accessory information replaced from the equipment;

[0064] based on the second identification, mark the emergency treatment record generation template filled in;

[0065] Wherein, the accessory information replaced from the equipment is determined by analyzing the on-site treatment video, comprising:

[0066] The frame images of the on-site treatment video are identified to determine the frame images with the second mark area as the fifth analysis image;

[0067] The second mark area in each fifth analysis image is extracted to obtain each sixth analysis image;

[0068] Based on the preset accessory identification model, the sixth analysis image is identified to obtain the accessory code set corresponding to each sixth analysis image;

[0069] Based on the accessory code set of the last sixth analysis image of the on-site treatment video, the accessory information replaced from the equipment is determined.

[0070] The application also provides an online monitoring system applied to semiconductor production, comprising:

[0071] The acquisition module is used for acquiring the acquisition data of the acquisition node corresponding to each device of semiconductor production through the LonWorks field bus;

[0072] The analysis module is used for analyzing the acquisition data to determine whether each device has an abnormality;

[0073] The processing module is used for sending abnormality reminding information to each monitoring node corresponding to the device with an abnormality and acquiring video data through the video acquisition module corresponding to each device when the abnormality occurs;

[0074] The generation module is used for generating an emergency treatment record based on the abnormality reminding information and the video data.

[0075] Other features and advantages of the present application will be further described in the following description, and some will become apparent from the description, or will be learned by practice of the present application. The purposes and other advantages of the present application can be realized and obtained by the structure particularly pointed out in the written description, claims, and drawings.

[0076] The technical solutions of the present application will be further described in detail below with the help of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0077] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0078] Figure 1 a schematic diagram of an online monitoring method applied to semiconductor production in an embodiment of the application;

[0079] Figure 2 a schematic diagram of a generation flow of on-site processing video in an embodiment of the application;

[0080] Figure 3 a schematic diagram of an online monitoring system applied to semiconductor production in an embodiment of the application. DETAILED DESCRIPTION

[0081] The preferred embodiments of the application will be described below in conjunction with the accompanying drawings, which should be understood as merely exemplary and explanatory, and are not intended to limit the application.

[0082] An online monitoring method applied to semiconductor production is provided in the embodiments of the application, as shown in the accompanying drawings, comprising: Figure 1

[0083] Step 1: Obtain the collection data of the collection nodes corresponding to each device of semiconductor production through LonWorks field bus;

[0084] Step 2: Analyze the collection data to determine whether each device has an abnormality;

[0085] Step 3: When an abnormality occurs, send abnormality reminding information to each monitoring node corresponding to the device having the abnormality and obtain video data through the video collection module corresponding to each device;

[0086] Step 4: Generate emergency processing records based on the abnormality reminding information and the video data.

[0087] The working principle and beneficial effects of the above technical solution are as follows:

[0088] ​The present application is to ensure timely and efficient monitoring of semiconductor production, and its monitoring means includes two aspects; the first aspect is to monitor the running state of the equipment, mainly: each device of semiconductor production is connected to the LonWorks field bus through the corresponding acquisition node, realizing the sending of each running state data of the device to the LonWorks field bus through the acquisition node, analyzing the collected data collected by the acquisition node to determine whether each device has an abnormality, and then realizing online monitoring of the running state of the equipment, the second aspect is to monitor the maintenance event of the equipment, when the equipment has an abnormality, an abnormality reminder information is sent to each monitoring node, and the monitoring node can be a fixed monitoring terminal or a mobile terminal held by a maintenance personnel; when the maintenance personnel determines that the equipment has an abnormality from the monitoring node, the on-site maintenance is carried out, the video data is obtained through the video acquisition module corresponding to each device, then the emergency treatment record is generated according to the abnormality reminder information and the video data, the monitoring of the maintenance process is realized, and the maintenance record is enriched, which is convenient for subsequent maintenance traceability; the comprehensive monitoring of the two aspects ensures the effective production and maintenance of the equipment. The collected data includes: the current, voltage of the equipment, the detection data of each sensor in the equipment, etc.

[0089] In one embodiment, the acquisition data of the acquisition node corresponding to each device of semiconductor production is obtained through the LonWorks field bus, including:

[0090] Sending a data acquisition request to the acquisition node through the LonWorks field bus;

[0091] After receiving the response of the acquisition node to the data acquisition request, the data packet numbering rule of the acquisition node is obtained through the control node of the LonWorks field bus; the data packet numbering rule of each acquisition node is different, so the data packet numbering rule needs to be obtained through the communication of the control node; the packet numbering includes two parts, the first part is the unique identification code of the acquisition node in the bus, and the second part is the data code of the current data packet, which is coded according to the packet time in order; therefore, the obtained data packet numbering rule includes the data code of the current time and the unique identification code;

[0092] Based on the data packet numbering rule, the first data packet corresponding to the received acquisition data of the acquisition node is verified; the data packet numbering rule is the numbering rule of the packet numbering of the data header; through the verification of the packet numbering, it can be determined whether the packet is lost; that is, whether the data packet is missing;

[0093] Based on the verification result, a secondary data acquisition request is constructed; when the verification result is that the data packet is missing, the packet numbering of the missing data packet is written into the secondary data acquisition request template to form the secondary data acquisition request;

[0094] receiving a second data packet sent by the collection node for the secondary data acquisition request;

[0095] parsing the first data packet and the second data packet to acquire the collection data.

[0096] The working principle and beneficial effects of the above technical solution are:

[0097] The data packet received by the primary data request is checked to determine whether to perform secondary data acquisition. The secondary data acquisition is mainly to supplement the primary data acquisition to avoid data packet loss. The primary data acquisition and the secondary data acquisition are combined to realize complete acquisition of the collection data, thereby ensuring effective implementation of the operation state monitoring of the equipment.

[0098] In order to analyze the collection data, in an embodiment, the collection data is analyzed to determine whether each device is abnormal, including:

[0099] determining device information corresponding to the device; the device information includes: device type, device model, device number, etc.;

[0100] Based on the device information, a feature extraction rule corresponding to the device and an abnormality analysis sub-library are extracted from a preset analysis library;

[0101] Based on the feature extraction rule, the collection data is subjected to feature extraction to obtain a plurality of feature values; the feature values include: voltage value, maximum voltage fluctuation value, current value, maximum current fluctuation value, average value and instantaneous maximum value of detection data of each sensor, etc.; the type and number of feature values to be extracted for each device are different, and the analysis library constructed in advance is used to determine the feature values to be extracted for each device, and then the collection data is extracted;

[0102] Based on the plurality of feature values, an analysis feature set is constructed; the feature values are arranged in order to form the analysis feature set;

[0103] The analysis result corresponding to the analysis feature set is called from the abnormality analysis sub-library; the abnormality analysis sub-library is constructed by professional personnel according to a large amount of data in advance;

[0104] The analysis result is parsed to determine whether the device is abnormal and the information of the abnormality; the analysis result shows whether the device is abnormal and the information of the abnormality, wherein the information of the abnormality includes: the position of the component that is abnormal and the component that is abnormal, etc.

[0105] In an embodiment, when the abnormality occurs, abnormality reminding information is sent to each monitoring node corresponding to the device that is abnormal, and video data is acquired through a video acquisition module corresponding to each device, including:

[0106] Analyzing the information of the exception, determining the component corresponding to the exception and the position of the component in the equipment;

[0107] Based on the component and the position of the component in the equipment, querying the preset video acquisition module call table to determine the corresponding video acquisition module and obtain the video data.

[0108] When monitoring the equipment through the video acquisition module, multiple video acquisition modules are usually installed to monitor the equipment, and the range or direction of the equipment that can be monitored by each video acquisition module is different. Through the analysis of the information of the exception, accurate video data is obtained, avoiding the cumbersome operation of simultaneously obtaining and analyzing the data of multiple video acquisition modules, and improving the analysis efficiency. The video acquisition module call table is constructed by professional personnel in advance according to the installation environment of the equipment, the positional relationship between the equipment and the video acquisition module and other factors.

[0109] In one embodiment, based on the exception prompt information and the video data, an emergency treatment record is generated, including:

[0110] Analyzing the exception prompt information, determining each prompt data item and the exception code corresponding to the exception; the prompt data item includes: the position of the equipment, the type of the equipment, the description information of the exception, etc.; the description information includes: damage of the electromagnetic valve, pressure anomaly, etc.;

[0111] Based on the exception code, the corresponding emergency treatment record generation template is called;

[0112] Fill each prompt data item into the first position corresponding to each prompt data in the emergency treatment record generation template;

[0113] Editing the video data to obtain the on-site processing video of the processing personnel;

[0114] Fill the start time of the on-site processing video into the second position in the emergency treatment record generation template;

[0115] Fill the on-site processing video into the third position in the emergency treatment record generation template.

[0116] The working principle and beneficial effects of the above technical solutions are:

[0117] Adding the video data into the emergency treatment record, so that the specific processing situation can be seen from the emergency treatment record during the later tracing, and the emergency treatment is more intuitive and accurate. In addition, the end time of the on-site processing video is determined by pushing back the time when the equipment is confirmed to be in normal production through the bus, and the time point when the last maintenance personnel appears in the video data is taken as the end time. The video data between the start time and the end time is taken as the on-site processing video.

[0118] In one embodiment, as shown in Figure 2 The video data is clipped to obtain the live processing video of the processing personnel, including:

[0119] Step S1: frames of the video data are extracted according to a preset interval number to obtain a first to-be-analyzed image; for example, the interval number is in a value range of [2, 30); the first to-be-analyzed image extracted from the video data by interval extraction further reduces the amount of data analysis and improves the analysis efficiency;

[0120] Step S2: personnel recognition is performed on the first to-be-analyzed image in sequence, and when a preset processing personnel appears, the preset first statistical parameter and the second statistical parameter value are both set to initial values; because the environment required for semiconductor production is relatively high and the dressing requirements of the factory are relatively standard, different colors or styles of dust-free clothes are usually used to distinguish the types of work in a dust-free workshop, for example, the dust-free clothes of production personnel are white and the dust-free clothes of maintenance personnel are brown, at this time, the maintenance personnel (i.e., the processing personnel) are distinguished by recognizing the color of the clothes of the personnel in the first analysis image;

[0121] Step S3: when the processing personnel are also recognized in the next first to-be-analyzed image, the first statistical parameter is incremented by one; when the processing personnel are not recognized, the second statistical parameter is incremented by one;

[0122] Step S4: step S3 is executed in a loop until the first statistical parameter value reaches a first threshold value within a preset first time interval;

[0123] Step S5: a point at which the previous first statistical parameter is set to the initial value is taken as a starting point of the live processing video;

[0124] In the loop execution process of step S4, when the second statistical parameter reaches a preset second threshold value within a preset second time interval, the first statistical parameter and the second statistical parameter are both set to the initial values; the first time interval is greater than the second time interval.

[0125] The working principle and beneficial effects of the above technical solution are as follows:

[0126] When the maintenance personnel are recognized through image recognition, the behavior of the personnel needs to be analyzed, the time when the personnel appear in the video is analyzed through the first statistical parameter and the second statistical parameter to determine whether the personnel are the processing personnel of the equipment; the second statistical parameter is added to the judgment process to reset the first statistical parameter and the second statistical parameter, thereby avoiding the interference of the maintenance personnel passing by the equipment multiple times on the behavior analysis and judgment, and improving the accuracy of the behavior analysis.

[0127] In one embodiment, when there are multiple video acquisition modules for shooting, the video data is clipped to obtain the live processing video of the processing personnel, including:

[0128] determining whether there are multiple frame images containing the maintenance personnel at the same time;

[0129] when there are, determining an included angle between a facing vector representing a facing direction of the maintenance personnel in each frame image and a center vector of a video capture module corresponding to the frame image; wherein the center vector of the video capture module is a center position of a capture lens of the video capture module as a starting point and is perpendicular to the lens surface and outward; and the facing vector is a center position of a torso of the maintenance personnel perpendicular to the human body torso and forward;

[0130] based on the included angle, determining a first score value; the smaller the included angle, the lower the first score value; the larger the included angle, the higher the score;

[0131] determining a proportion of the maintenance personnel being blocked in each frame image; the proportion of the maintenance personnel being blocked is the area of the maintenance personnel in the frame image and the area of the maintenance personnel in the actual posture

[0132] based on the proportion, determining a second score value; the larger the proportion, the smaller the second score value;

[0133] based on the first score value and the second score value, determining a total score value; the total score value is a weighted sum of the first score value and the second score value;

[0134] based on the total score value, screening multiple frame images within the same time to determine the image in the on-site processing video. Mainly, the frame image with the highest total score value is extracted as the on-site processing video;

[0135] The working principle and beneficial effects of the above technical solutions are:

[0136] When making the on-site processing video, only one video capture module is needed to capture video data within the same time, and the frame image is screened through the scoring method of multiple frame images; the clearer and more convenient image for tracing is selected through screening, and the screening mainly triggers from the angle of the facing direction of the maintenance personnel and the center vector of the video capture module, and whether the maintenance personnel is blocked and the blocking condition, from the angle to ensure that the maintenance personnel is as much as possible shot from the front and side, and from the blocking condition to ensure that the maintenance personnel can be seen completely. In addition, when making the on-site processing image, if the adjacent frame images are images captured by different video capture modules, a preset transition animation needs to be inserted between the two to indicate the switching of the video capture module.

[0137] In one embodiment, based on the abnormal prompt information and the video data, an emergency handling record is generated, further comprising:

[0138] analyzing the on-site processing video to determine the replacement part information and the use order of the parts;

[0139] generate a first mark based on the accessory information and the use sequence of the accessory;

[0140] based on the first mark, mark the template of the completed emergency treatment record;

[0141] The method comprises the following steps:

[0142] The first mark area is identified by the maintenance personnel during on-site maintenance. For example, yellow cloth can be laid out, and the accessories to be replaced to the equipment can be placed on the yellow cloth one by one. The video acquisition module can capture the accessories on the yellow cloth. The area of the yellow cloth is the first mark area.

[0143] extract the first mark area in each second analysis image to obtain each third analysis image;

[0144] based on a preset accessory recognition model, recognize the third analysis image to obtain an accessory code set corresponding to each third analysis image. The accessory code set includes the codes of each accessory. If there are several accessories in the accessory code set, there are several accessory codes in the accessory code set.

[0145] based on the accessory code set, filter each third analysis image to obtain a plurality of fourth analysis images; if the accessory code sets corresponding to adjacent third analysis images are consistent, only one third analysis image is retained; that is, the accessory code sets corresponding to adjacent fourth analysis images are different.

[0146] based on the accessory code set corresponding to each fourth analysis image, construct an accessory code total set; the accessory code total set is the union set of each accessory code set;

[0147] compare the accessory code sets corresponding to adjacent fourth analysis images in sequence to determine the difference between the two accessory code sets;

[0148] based on the difference between the accessory code sets corresponding to adjacent fourth analysis images, determine the use sequence of the accessories. Generally, the accessory codes in the accessory code set decrease in sequence, so the use sequence of the accessories can be obtained according to the sequence of the decreasing accessory codes; when the difference is an increase, the increased accessory code is determined, and the corresponding accessory code is deleted from the arranged accessory use sequence;

[0149] The working principle and beneficial effects of the above technical solution are as follows:

[0150] The on-site processing video is analyzed, replacement of the used accessory information and the use order of the accessory are determined, the first mark is generated through the accessory information and the use order of the accessory, and the use of the accessory and the order are marked, so that the information recognition and marking of the use of the accessory are realized.

[0151] In addition, the first mark can further include a mark indicating whether the use order of the accessory is incorrect or non-standard, for example, the background of the information marking box of the use of the accessory is configured as green to indicate no error or standard, and configured as yellow to indicate error or non-standard. Specifically, the use order of the accessory is analyzed to determine whether the use order of the accessory conflicts, an analysis data set is constructed based on the use order of the accessory and the accessory code corresponding to the accessory, then the analysis data set is matched with a standard data set corresponding to each analysis result in an analysis library, an analysis result corresponding to the matched standard data set is extracted, and when the analysis result is standard, the marking box is filled with a first marking color; and when the analysis result is non-standard, the marking box is filled with a second marking color.

[0152] In one embodiment, based on the abnormal reminding information and the video data, an emergency treatment record is generated, and further includes:

[0153] The on-site processing video is analyzed, replacement of the used accessory information and the use order of the accessory are determined, the first mark is generated through the accessory information and the use order of the accessory, and the use of the accessory and the order are marked, so that the information recognition and marking of the use of the accessory are realized.

[0154] Based on the replacement of the used accessory information and the replacement of the accessory information from the equipment, a second mark is generated.

[0155] Based on the second mark, a completed emergency treatment record generation template is marked.

[0156] The on-site processing video is analyzed, replacement of the used accessory information and the use order of the accessory are determined, the first mark is generated through the accessory information and the use order of the accessory, and the use of the accessory and the order are marked, so that the information recognition and marking of the use of the accessory are realized.

[0157] Each frame image of the on-site processing video is identified, and a frame image with a second mark area is determined as a fifth to-be-analyzed image; the second mark area is an area marked by a brown or other color cloth laid by the maintenance personnel during maintenance, and the maintenance personnel places the accessories removed from the equipment on the area in sequence;

[0158] Each second mark area in each fifth to-be-analyzed image is extracted to obtain each sixth to-be-analyzed image.

[0159] Based on a preset accessory recognition model, each sixth to-be-analyzed image is identified to obtain an accessory code set corresponding to each sixth to-be-analyzed image.

[0160] Based on the accessory code set of the last sixth to-be-analyzed image of the on-site processing video, the replacement of the accessory information from the equipment is determined.

[0161] The working principle and beneficial effects of the above technical solutions are:

[0162] Through the identification of the accessories disassembled from the maintenance, the record of the accessories disassembled from the maintenance is realized, the record can be compared with the accessories replaced, whether the difference is determined, and when the difference exists, the alarm information is sent to the monitoring node.

[0163] The application also provides an online monitoring system applied to semiconductor production, as shown in the figure, comprising: Figure 3

[0164] The acquisition module 1 is used for acquiring the acquisition data of the acquisition nodes corresponding to each device of the semiconductor production through the LonWorks field bus;

[0165] The analysis module 2 is used for analyzing the acquisition data and determining whether each device is abnormal;

[0166] The processing module 3 is used for sending the abnormal reminding information to each monitoring node corresponding to the device when the abnormality occurs and acquiring the video data through the video acquisition module corresponding to each device;

[0167] The generation module 4 is used for generating the emergency treatment record based on the abnormal reminding information and the video data.

[0168] In one embodiment, the acquisition module 1 acquires the acquisition data of the acquisition nodes corresponding to each device of the semiconductor production through the LonWorks field bus, and performs the following operations:

[0169] A data acquisition request is sent to the acquisition node through the LonWorks field bus;

[0170] After receiving the response of the acquisition node to the data acquisition request, the data packet number rule of the acquisition node is acquired through the control node of the LonWorks field bus;

[0171] Based on the data packet number rule, the first data packet corresponding to the acquisition data of the acquisition node is verified;

[0172] Based on the verification result, a secondary data acquisition request is constructed;

[0173] The second data packet sent by the acquisition node to the secondary data acquisition request is received;

[0174] The first data packet and the second data packet are parsed, and the acquisition data is acquired.

[0175] In one embodiment, the analysis module 2 analyzes the acquisition data and determines whether each device is abnormal, and performs the following operations:

[0176] ​determining device information corresponding to the device;

[0177] extracting feature extraction rules corresponding to the device and an abnormality analysis sub-library from a preset analysis library based on the device information;

[0178] performing feature extraction on the collected data based on the feature extraction rules to obtain a plurality of feature values;

[0179] constructing an analysis feature set based on the plurality of feature values;

[0180] calling an analysis result corresponding to the analysis feature set from the abnormality analysis sub-library;

[0181] analyzing the analysis result to determine whether the device has an abnormality and information of the abnormality.

[0182] In one embodiment, when an abnormality occurs, the processing module 3 sends abnormality reminding information to each monitoring node corresponding to the device where the abnormality occurs and acquires video data through a video acquisition module corresponding to each device, and performs the following operations:

[0183] analyzing the information of the abnormality to determine a component corresponding to the abnormality and a position of the component in the device;

[0184] querying a preset video acquisition module calling table based on the component and the position of the component in the device to determine a corresponding video acquisition module and acquire video data.

[0185] In one embodiment, the generating module 4 generates an emergency treatment record based on the abnormality reminding information and the video data, and performs the following operations:

[0186] analyzing the abnormality reminding information to determine each reminding data item and an abnormality code corresponding to the abnormality;

[0187] calling a corresponding emergency treatment record generation template based on the abnormality code;

[0188] filling each reminding data item into a first position corresponding to each reminding data in the emergency treatment record generation template;

[0189] editing the video data to obtain a field treatment video of a treatment personnel;

[0190] filling a start time of the field treatment video into a second position in the emergency treatment record generation template;

[0191] filling the field treatment video into a third position in the emergency treatment record generation template.

[0192] In one embodiment, the generating module 4 edits the video data to obtain a field treatment video of a treatment personnel, including:

[0193] Step S1: Extract frame images of the video data according to a preset interval, to obtain a first image to be analyzed;

[0194] Step S2: Perform personnel recognition on the first image to be analyzed in sequence, and when a preset processing personnel appears, set the preset first statistical parameter and the second statistical parameter value to initial values;

[0195] Step S3: When the processing personnel is also recognized in the next first image to be analyzed, the first statistical parameter is incremented by one;

[0196] Step S4: When the processing personnel is not recognized in the next first image to be analyzed, the second statistical parameter is incremented by one;

[0197] Step S5: Recursively execute step S3 and step S4 until the first statistical parameter value reaches a first threshold value within a preset first time interval;

[0198] Step S6: Set a point at which the previous first statistical parameter is set to the initial value as a starting point of the live processing video;

[0199] In the recursive execution process of step S5, when the second statistical parameter reaches a preset second threshold value within a preset second time interval, the first statistical parameter and the second statistical parameter are set to the initial values; the first time interval is greater than the second time interval.

[0200] In an embodiment, when there are multiple video acquisition modules for shooting, the generating module 4 performs editing on the video data to obtain the live processing video of the processing personnel, and performs the following operations:

[0201] Determine whether there are multiple frame images containing the maintenance personnel at the same time;

[0202] When there are, determine the included angle between the facing vector representing the facing of the maintenance personnel in each frame image and the center vector of the video acquisition module corresponding to the frame image;

[0203] Based on the included angle, determine a first score value;

[0204] Determine the proportion of the maintenance personnel being blocked in each frame image;

[0205] Based on the proportion, determine a second score value;

[0206] Based on the first score value and the second score value, determine a total score value;

[0207] Based on the total score value, screen the multiple frame images within the same time to determine the images to be used to construct the live processing video.

[0208] In one embodiment, the generating module 4 generates the emergency treatment record based on the abnormal prompt information and the video data, and further performs the following operations:

[0209] analyzing the on-site treatment video to determine the information of the replaced spare parts and the use sequence of the spare parts;

[0210] generating a first mark based on the information of the replaced spare parts and the use sequence of the spare parts;

[0211] annotating the template of the filled emergency treatment record based on the first mark;

[0212] The analyzing the on-site treatment video to determine the information of the replaced spare parts and the use sequence of the spare parts comprises:

[0213] identifying each frame image of the on-site treatment video, and determining a frame image with a first mark region as a second to-be-analyzed image;

[0214] extracting the first mark region in each second to-be-analyzed image to obtain a third to-be-analyzed image;

[0215] identifying the third to-be-analyzed image based on a preset spare part recognition model to obtain a spare part code set corresponding to each third to-be-analyzed image;

[0216] filtering each third to-be-analyzed image based on the spare part code set to obtain a fourth to-be-analyzed image;

[0217] constructing a total spare part code set based on the spare part code set corresponding to each fourth to-be-analyzed image;

[0218] sequentially comparing the spare part code sets corresponding to two adjacent fourth to-be-analyzed images to determine the difference between the two spare part code sets;

[0219] determining the use sequence of the spare parts based on the difference between the spare part code sets corresponding to the adjacent fourth to-be-analyzed images.

[0220] In one embodiment, the generating module 4 generates the emergency treatment record based on the abnormal prompt information and the video data, and further performs the following operations:

[0221] analyzing the on-site treatment video to determine the information of the replaced spare parts and the use sequence of the spare parts;

[0222] generating a second mark based on the information of the replaced spare parts and the information of the replaced spare parts from the equipment;

[0223] annotating the template of the filled emergency treatment record based on the second mark;

[0224] The analyzing the on-site treatment video to determine the information of the replaced spare parts and the use sequence of the spare parts comprises:

[0225] Identify each frame image of the on-site processing video, and determine the frame image with the second marking area as a fifth to-be-analyzed image;

[0226] Extract the second marking area in each fifth to-be-analyzed image to obtain each sixth to-be-analyzed image;

[0227] Based on the preset accessory recognition model, the sixth to-be-analyzed image is identified to obtain the accessory code set corresponding to each sixth to-be-analyzed image;

[0228] Based on the accessory code set of the last sixth to-be-analyzed image of the on-site processing video, the accessory information replaced from the equipment is determined.

[0229] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.

Claims

1. An on-line monitoring method applied to semiconductor production, characterized by, The application relates to a semiconductor production device abnormality processing method and device. Obtaining collection data of collection nodes corresponding to each device in semiconductor production through a LonWorks field bus; Analyzing the collection data to determine whether an abnormality occurs in each device; When an abnormality occurs, sending abnormality reminding information to each monitoring node corresponding to the device where the abnormality occurs and obtaining video data through a video collection module corresponding to each device; Generating an emergency processing record based on the abnormality reminding information and the video data; The method comprises the following steps: Analyzing the abnormality reminding information to determine each reminding data item and an abnormality code corresponding to the abnormality; Based on the abnormality code, a corresponding emergency processing record generation template is called; Each reminding data item is filled into a first position corresponding to each reminding data in the emergency processing record generation template; The video data is edited to obtain on-site processing video of a processing personnel; A start time of the on-site processing video is filled into a second position in the emergency processing record generation template; The on-site processing video is filled into a third position in the emergency processing record generation template; The method comprises the following steps: Step S1: according to a preset interval, frame images of the video data are extracted to obtain first to-be-analyzed images; Step S2: personnel identification is performed on the first to-be-analyzed images one by one, and when a preset processing personnel appears, preset first and second statistical parameter values are both set as initial values; Step S3: when the processing personnel is also identified in the next first to-be-analyzed image, the first statistical parameter is increased by one; Step S4: when the processing personnel is not identified in the next first to-be-analyzed image, the second statistical parameter is increased by one; Step S5: the steps S3 and S4 are cyclically executed until the first statistical parameter value reaches a first threshold value within a preset first time interval; Step S6: a point where the previous first statistical parameter is set as the initial value is taken as a start point of the on-site processing video; During the cyclic execution of the step S5, when the second statistical parameter reaches a preset second threshold value within a preset second time interval, the first statistical parameter and the second statistical parameter are both set as the initial values; and the first time interval is greater than the second time interval.

2. The on-line monitoring method for semiconductor production according to claim 1, wherein The method comprises the following steps: Sending a data acquisition request to the collection node through the LonWorks field bus; After receiving a response of the collection node to the data acquisition request, obtaining a data packet numbering rule of the collection node through a control node of the LonWorks field bus; Based on the data packet numbering rule, a first data packet corresponding to the collection data of the collection node is checked; Based on the checking result, a secondary data acquisition request is constructed; Receiving a second data packet sent by the collection node for the secondary data acquisition request; Parsing the first data packet and the second data packet to obtain the collection data.

3. The on-line monitoring method for semiconductor production according to claim 1, wherein The analysis of the collection data to determine whether each device is abnormal, comprising: Determine the device information corresponding to the device; Based on the device information, extract the feature extraction rule corresponding to the device and the abnormal analysis sub-library from the preset analysis library; Based on the feature extraction rule, the feature extraction is performed on the collection data to obtain a plurality of characteristic values; Based on a plurality of characteristic values, an analysis feature set is constructed; Retrieve the analysis result corresponding to the analysis feature set from the abnormal analysis sub-library; Parse the analysis result to determine whether the device is abnormal and the information of the abnormality.

4. The on-line monitoring method for semiconductor production according to claim 3, wherein When an abnormality occurs, send an abnormality reminder information to each monitoring node corresponding to the device that occurs an abnormality and obtain video data through the video collection module corresponding to each device, comprising: Parse the information of the abnormality to determine the component corresponding to the abnormality and the position of the component in the device; Based on the component and the position of the component in the device, query the preset video collection module retrieval table to determine the corresponding video collection module and obtain the video data.

5. The on-line monitoring method for semiconductor production according to claim 1, wherein When there are multiple video collection modules for shooting, the video data is edited to obtain the on-site processing video of the processing personnel, comprising: Determine whether there are multiple frame images containing maintenance personnel at the same time; When there are, determine the included angle between the facing vector representing the maintenance personnel in each frame image and the center vector of the video collection module corresponding to the frame image; Based on the included angle, determine a first score value; Determine the proportion of the maintenance personnel being blocked in each frame image; Based on the proportion, determine a second score value; Based on the first score value and the second score value, determine a total score value; Based on the total score value, filter multiple frame images at the same time to determine the images used to construct the on-site processing video.

6. The on-line monitoring method for semiconductor production according to claim 1, wherein The emergency processing record is generated based on the abnormality reminder information and the video data, further comprising: Parse the on-site processing video to determine the accessory information used for replacement and the use order of the accessories; Based on the accessory information and the use order of the accessories, generate a first identifier; Based on the first identifier, label the emergency processing record template filled out; Wherein, the parsing of the on-site processing video to determine the accessory information used for replacement and the use order of the accessories, comprising: Identify each frame image of the on-site processing video, and determine the frame image with the first label area as a second to-be-analyzed image; Extract the first label area in each second to-be-analyzed image to obtain each third to-be-analyzed image; Based on a preset accessory identification model, identify the third to-be-analyzed image to obtain an accessory code set corresponding to each third to-be-analyzed image; Based on the accessory code set, filter each third to-be-analyzed image to obtain a plurality of fourth to-be-analyzed images; Based on the accessory code set corresponding to each fourth to-be-analyzed image, construct an accessory code total set; Comparing two adjacent fourth to-be-analyzed image corresponding accessory code set in sequence, determine the difference between the two accessory code set; Based on the difference between the accessory code set corresponding to the adjacent fourth to-be-analyzed image, determine the use order of the accessory.

7. The on-line monitoring method for semiconductor production according to claim 1, wherein The generating emergency processing record based on the abnormal prompt information and the video data further includes: Parsing the on-site processing video to determine the accessory information replaced from the equipment; Generating a second identifier based on the accessory information replaced from the equipment; Based on the second identifier, labeling the emergency processing record generation template filled in; Wherein, the parsing the on-site processing video to determine the accessory information replaced from the equipment includes: Identifying each frame image of the on-site processing video, and determining the frame image with the second mark area as the fifth to-be-analyzed image; Extracting the second mark area in each fifth to-be-analyzed image to obtain each sixth to-be-analyzed image; Based on a preset accessory identification model, identifying the sixth to-be-analyzed image to obtain the accessory code set corresponding to each sixth to-be-analyzed image; Based on the accessory code set of the last sixth to-be-analyzed image appearing in the on-site processing video, determining the accessory information replaced from the equipment.

8. An on-line monitoring system applied to semiconductor production, characterized by, It includes: The acquisition module is used for acquiring the acquisition data of the acquisition node corresponding to each device of semiconductor production through the LonWorks field bus; The analysis module is used for analyzing the acquisition data to determine whether each device has an abnormality; The processing module is used for sending abnormal prompt information to each monitoring node corresponding to the device when an abnormality occurs and acquiring video data through the video acquisition module corresponding to each device; The generating module is used for generating an emergency processing record based on the abnormal prompt information and the video data; Wherein, the generating module generates an emergency processing record based on the abnormal prompt information and the video data, and performs the following operations: Parsing the abnormal prompt information to determine each prompt data item and the abnormality corresponding to the abnormality code; Based on the abnormal code, calling the corresponding emergency processing record generation template; Filling each prompt data item into the first position corresponding to each prompt data in the emergency processing record generation template; Editing the video data to obtain the on-site processing video of the processing personnel; Filling the start time of the on-site processing video into the second position in the emergency processing record generation template; Filling the on-site processing video into the third position in the emergency processing record generation template; Wherein, the generating module edits the video data to obtain the on-site processing video of the processing personnel, including: Step S1: extracting the frame image of the video data according to a preset interval number to obtain a first to-be-analyzed image; Step S2: sequentially performing personnel identification on the first to-be-analyzed image, and setting the preset first statistical parameter and second statistical parameter value to initial values when a preset processing personnel appears; Step S3: when the processing personnel is also identified in the next first to-be-analyzed image, the first statistical parameter is incremented by one; Step S4: when a processing person is not recognized in the next first image to be analyzed, the second statistical parameter is incremented by one; Step S5: steps S3 and S4 are executed in a loop until the first statistical parameter value reaches a first threshold value within a preset first time interval; Step S6: a point at which the previous first statistical parameter is set to an initial value is taken as a starting point of the live processing video; In the loop execution of step S5, when the second statistical parameter reaches a preset second threshold value within a preset second time interval, both the first statistical parameter and the second statistical parameter are set to the initial value; the first time interval is greater than the second time interval.

Images