A laboratory safety warning system

Abstract

The present application relates to the technical field of laboratory safety, and more particularly to a laboratory safety early warning system, comprising a registered module, a portrait tracking module, a digital twin module, a general monitoring module, a monitoring analysis module for determining the monitoring mode of the laboratory, and a monitoring control module, wherein the monitoring control module determines a safety warning library according to the experimental content of each experimenter to identify abnormal behavior of the experimenter, determines the running experimental equipment according to the experimental content and the experimental video to monitor and update the corresponding part of the digital twin model in real time, and controls the general monitoring module to perform general monitoring on the experimental area to determine the influence representation trend of the non-experimental area on the experimental area to determine whether to perform safety early warning. The present application solves the problems of incomplete monitoring, untimely early warning, difficult accident cause tracing, and unreasonable resource utilization in the laboratory safety management of the prior art.

Description

Technical Field

[0001] The present invention relates to the field of laboratory safety technology, and in particular to a laboratory safety early warning system. Background Art

[0002] In various laboratories, such as chemical laboratories, biological laboratories, and physical laboratories, safety risks always exist because the experimental process often involves hazardous chemicals, complex experimental equipment, and special experimental environment requirements. Traditional laboratory safety management relies mostly on manual inspections and simple sensor monitoring, which has many problems such as incomplete monitoring, untimely warnings, and difficulty in tracing the causes of accidents. For example, it is difficult to monitor the operating specifications of experimental personnel in real time, the overall and local changes in the laboratory environment cannot be accurately grasped, and it is difficult to quickly determine the starting link of the accident and the relevant responsible persons after the accident occurs. All of these urgently require a more intelligent, efficient, and comprehensive laboratory safety early warning system.

[0003] Chinese patent publication number CN116580350A discloses a laboratory safety monitoring and early warning method and system, which relates to the field of laboratory safety technology. First, the video to be detected in the laboratory is obtained, and several key frame images are collected for preprocessing to lay the foundation for contour extraction and equipment detection. Then, based on the preprocessed key frame images, the human body contour is extracted using the background difference method, and feature extraction and behavior classification are performed to identify abnormal behavior of personnel. At the same time, the key frame images are input into the preset equipment recognition model, and the abnormal status of the equipment is identified based on the established baseline information about the laboratory equipment. Finally, when abnormal behavior of personnel / abnormal status of equipment is detected, the corresponding alarm information is pushed to the management personnel. Thus, abnormal behavior of personnel and abnormal status of equipment that do not meet the safety requirements of scientific research laboratories are automatically monitored and timely reminded, which reduces the workload of laboratory safety management personnel and avoids the occurrence of major laboratory safety accidents. It can be seen that the invention has the following problems:

[0004] In terms of laboratory safety management, monitoring is incomplete, early warning is not timely, accident causes are difficult to trace, and resource utilization is unreasonable. Summary of the Invention

[0005] To this end, the present invention provides a laboratory safety early warning system to overcome the problems of incomplete monitoring, untimely early warning, difficulty in tracing the cause of accidents, and irrational resource utilization in the existing technology in laboratory safety management. Through the collaborative work of multiple modules, comprehensive monitoring and intelligent early warning of personnel operations, equipment status, and environmental conditions can be achieved, and complete data support can be provided for accident tracing, while optimizing resource utilization and improving management efficiency.

[0006] To achieve the above objectives, the present invention provides a laboratory safety early warning system, comprising:

[0007] The registration module is used to record the entry time, experimental content and departure time of the experimenter into the laboratory;

[0008] Portrait tracking module, used to identify, track and record the experimental videos of each experimenter in the laboratory;

[0009] Digital twin module, used to generate a digital twin model of the laboratory;

[0010] A general monitoring module is used to obtain real-time environmental monitoring data of the laboratory and the experimental area, including temperature data, humidity data, and smoke data;

[0011] A monitoring and analysis module, which is connected to the registration module and is used to determine the laboratory monitoring method based on whether there are experimenters in the laboratory;

[0012] The monitoring method includes using a portrait tracking module to identify and track the portraits of each experimenter in the laboratory and shoot experimental videos of each experimenter, and using a digital twin module to generate a digital twin model of the laboratory;

[0013] A monitoring and control module is respectively connected to the registration module, the portrait tracking module, the digital twin module, the general monitoring module and the monitoring and analysis module, and is used to determine the safety warning library according to the experimental content of each experimenter to identify the abnormal behavior of the experimenter, determine the running experimental equipment according to the experimental content and the experimental video to perform real-time monitoring and update of the partial digital twin model corresponding to the running experimental equipment, and control the general monitoring module to perform general monitoring of the experimental area to determine the impact characterization trend of the non-experimental area on the experimental area to determine whether to issue a safety warning.

[0014] Furthermore, the general monitoring module includes:

[0015] The overall monitoring unit is set at a fixed location in the laboratory to obtain the laboratory's environmental monitoring data in real time;

[0016] The mobile monitoring unit is configured to be mobile and is used to obtain environmental monitoring data of the experimental area in real time.

[0017] Furthermore, the monitoring and analysis module determines the monitoring mode of the laboratory according to whether there is an experimenter in the laboratory to control the portrait tracking module to start working according to the monitoring mode, including:

[0018] If there are experimenters in the laboratory, the monitoring method is to control the portrait tracking module to perform portrait recognition and tracking on each experimenter in the laboratory and to shoot an experiment video of each experimenter;

[0019] If there is no experimenter in the laboratory, it is determined that the monitoring method is to control the digital twin module to generate a digital twin model in the laboratory to monitor the laboratory, and the portrait tracking module is not controlled to start working.

[0020] Furthermore, the monitoring control module determines whether to send a safety warning message based on the identity and number of the experimenters identified by the portrait tracking module and the experimenters recorded in the registration module, including:

[0021] If the comparison result shows that the identity and number of the identified experimenters are consistent with the recorded experimenters, it is determined that the safety warning information will not be sent;

[0022] If the comparison result shows that the identity and / or number of the identified experimenters do not match the recorded experimenters, it is determined that a safety warning message is sent to the experimenters recorded in the registration module.

[0023] Furthermore, the monitoring and control module determines the corresponding safety warning library based on the experimental content recorded by each experimenter in the registration module and trains a classification model based on the safety warning library to obtain a safety warning classification model. In addition, the monitoring and control module divides the experimental video into several frames and then determines several key frames, and extracts human images in each key frame to input each human image into the corresponding safety warning classification model to identify abnormal behaviors of the corresponding experimenter.

[0024] Furthermore, the monitoring control module determines the similarity of each frame image according to the machine learning model to determine a number of identical frame groups and selects a frame with the highest definition in each identical frame group as a key frame;

[0025] The similarity between the frames in the same frame group should be greater than or equal to a preset similarity, and the number of key frames is equal to the number of the same frame group.

[0026] Furthermore, the monitoring and control module determines the operating experimental equipment according to the experimental content and the experimental video and controls the digital twin module to monitor the operating experimental equipment in real time to update the corresponding partial digital twin model in real time, and generates a digital twin model of the laboratory according to the partial digital twin model to determine the safety characterization status of the laboratory, including:

[0027] If all devices in the real-time updated digital twin model are normal, the impact characterization trend of the non-experimental area on the experimental area is determined based on the environmental monitoring data of the experimental area to determine the safety characterization status of the laboratory;

[0028] If there is an abnormal device in the digital twin model updated in real time, the safety representation state is determined to be a dangerous state and an alarm signal is issued.

[0029] Furthermore, the digital twin model updated in real time by the monitoring and control module consists of a partial digital twin model and a fixed digital twin model;

[0030] Among them, the monitoring and control module monitors the running experimental equipment in real time to update the partial digital twin model in real time, and the monitoring and control module monitors and updates the fixed digital twin model of the non-running experimental equipment when the digital twin model is turned on.

[0031] Furthermore, the monitoring and control module determines the impact characterization trend of the non-experimental area on the experimental area based on the environmental monitoring data of the experimental area to determine the safety characterization status of the laboratory, including:

[0032] If the environmental monitoring data of the experimental area is not within the standard environmental data range of the experimental area, it is determined that the impact characterization trend of the non-experimental area on the experimental area is a negative impact trend and the safety characterization state is determined to be a dangerous state. The monitoring and control module controls the digital twin module to update the partial digital twin model and the fixed digital twin model and issue an alarm signal;

[0033] If the environmental monitoring data of the experimental area is within the standard environmental data range of the experimental area, it is determined that the impact characterization trend of the non-experimental area on the experimental area is a hidden impact trend and the safety characterization state is determined to be a safe state.

[0034] Furthermore, the monitoring and control module determines the storage location of the experimental preparation based on the experimental video and determines the storage location as the experimental area.

[0035] Compared with the prior art, the beneficial effects of the present invention are that the laboratory safety early warning system provided by the present invention accurately records the entry and exit time, experimental content and other information of the experimenter through the registration module, which is convenient for traceability management; the portrait tracking module effectively identifies and tracks the experimenter and records the experimental video, which is conducive to the subsequent review and analysis of the experimental process; the laboratory digital twin model generated by the digital twin module, combined with the monitoring method determined by the monitoring and analysis module, can intuitively present the operating status of the laboratory; the environmental monitoring data of the laboratory and the experimental area are obtained in real time through the general monitoring module, providing a basis for environmental safety assessment; the monitoring and control module determines the safety warning library according to the experimental content to accurately identify the abnormal behavior of the experimenter, and at the same time operates the experimental equipment based on the experimental content and experimental video and updates part of the digital twin model content, and also controls the general monitoring module to analyze the impact of the non-experimental area on the experimental area. Trends in the characterization of laboratory safety early warnings are achieved comprehensively and intelligently, effectively preventing accidents, ensuring the safety of experimental personnel, the smooth progress of experiments and the stability of the laboratory environment;

[0036] Furthermore, the overall monitoring unit of the ordinary monitoring module is fixed at a specific location in the laboratory and is always turned on. It can continuously and stably collect environmental monitoring data for the entire laboratory, providing a reliable basis for comprehensively understanding the overall environmental conditions of the laboratory, helping to promptly detect situations such as abnormal fluctuations in overall temperature, humidity imbalance or hidden smoke hazards, and ensure the basic safety of the laboratory environment. The mobile monitoring unit is flexible. When the monitoring and control module determines that there is an experimental area, it can accurately move to the location of the experimental area to obtain its environmental monitoring data. This can not only perform accurate environmental monitoring of the core area of ​​the experiment, such as detecting special temperature and humidity requirements or tiny smoke changes in the local area of ​​the experimental area, but also effectively make up for the monitoring blind spots of fixed monitoring points, and realize more detailed and targeted environmental monitoring of the experimental area. The two work together to greatly improve the comprehensiveness, accuracy and flexibility of laboratory environmental monitoring, and build a solid data monitoring defense line for the safe operation of the laboratory.

[0037] Furthermore, the monitoring and analysis module can intelligently determine the monitoring method based on whether there are experimenters in the laboratory. When there are experimenters, the portrait tracking module is quickly activated to shoot experimental videos with the help of portrait recognition and tracking technology, and the number of shots can be determined based on the number of people. This not only helps to monitor the operating behavior of experimenters in real time and promptly discover possible illegal operations or abnormal conditions, but also provides intuitive and detailed image data for subsequent experimental analysis and accident tracing. When there are no experimenters in the laboratory, it is reasonable to switch to the digital twin module to generate a laboratory model for monitoring, avoiding unnecessary waste of portrait tracking resources. The digital twin model effectively monitors equipment safety and ensures that the operating status of the laboratory is still under control when no one is on duty. This greatly improves the intelligence level, resource utilization efficiency, overall safety and reliability of laboratory monitoring, and realizes accurate, efficient and adaptive monitoring and management of the laboratory under different personnel status.

[0038] Furthermore, the monitoring and control module cleverly integrates the data information of the registration module and the portrait tracking module. By rigorously comparing the identities and number of identified experimenters with the recorded experimenter information, it accurately judges the personnel situation in the laboratory. When the two are consistent, the normal operation order of the laboratory is maintained, no warning information is sent, and unnecessary interference is avoided. Once a discrepancy occurs, a safety warning information is immediately sent to the experimenter recorded in the registration module. This mechanism effectively prevents illegal intrusion by non-experimental personnel, greatly improves the safety and confidentiality of the laboratory, and protects the experimental process from external interference and potential threats. At the same time, it also provides experimenters with a safe and focused experimental environment, making laboratory management more standardized, orderly, and intelligent, effectively reducing the risk of safety accidents caused by poor personnel management, and strongly promoting the efficient operation and improvement of the laboratory safety management system.

[0039] Furthermore, the monitoring and control module accurately determines the corresponding safety warning library based on the experimental content recorded by each experimenter in the registration module, and uses this to train the classification model to obtain a safety warning classification model. This measure enables the system to conduct in-depth cognition and learning of the potential risks of different experimental types, and build a highly adaptive safety assessment system; then, the experimental video is divided into several frames and the key frames are determined, and the human body images are extracted and input into the safety warning classification model, so as to achieve effective identification of abnormal behaviors of the corresponding experimenters; this intelligent and refined processing method can keenly capture operational behaviors that do not meet experimental specifications in complex experimental scenarios, such as violations of titration operation procedures, incorrect grinding techniques, etc., and issue alarms in time to effectively prevent safety accidents caused by improper human operations, greatly improving the standardization and safety of laboratory operations and the ability to control various experimental risks, and providing solid and powerful technical support and guarantee for the safe and stable operation of the laboratory;

[0040] Further, the monitoring control module determines the similarity of each frame of image by means of the machine learning model, accurately divides the same frame group, and skillfully selects the frame with the highest clarity in each group as the key frame. This operation not only effectively reduces data redundancy and improves data processing efficiency, but also provides high-quality and representative image samples for subsequent abnormal behavior recognition. In addition, by reasonably setting the preset similarity, the accuracy of the abnormal behavior recognition result is guaranteed, and the optimization of the allocation of computing resources is realized, avoiding excessive consumption of resources.

[0041] Further, the monitoring control module divides the digital twin model into a partial digital twin model and a fixed digital twin model, monitors and updates the partial digital twin model in real time for the running experimental equipment, so that the dynamic changes in the equipment operation can be immediately reflected in the model, providing a reliable basis for accurately grasping the real-time status of the equipment. The fixed digital twin model of the non-running experimental equipment is monitored and updated when the digital twin module is just started, effectively ensuring the accuracy and timeliness of the state information of the non-running equipment. Then, it is updated according to the preset update frequency, which balances the consumption of system resources while ensuring that the model data is not outdated. This comprehensive, intelligent and flexible digital twin model updating strategy greatly improves the integrity, timeliness and accuracy of laboratory equipment state monitoring, providing strong technical support and guarantee for laboratory safety management, equipment maintenance and overall operational efficiency improvement.

[0042] Further, the monitoring control module can determine the influence trend of the non-experimental area on the experimental area according to the environmental monitoring data of the experimental area, and then determine the safety representation state of the laboratory. When the environmental monitoring data of the experimental area exceeds the standard environmental data interval, it is determined that the influence trend is negative and the safety representation state is dangerous. The digital twin module is immediately controlled to update the partial digital twin model and the fixed digital twin model, and an alarm signal is sent. This series of measures can enable laboratory managers to know about abnormal situations in time and take prompt measures to effectively prevent the danger from further expanding and maximize the safety of the laboratory. When the environmental monitoring data is within the standard interval, it is determined that the influence trend is implicit and the safety state is achieved, realizing the meticulous control of the laboratory environment state. At the same time, the storage location of the experimental preparation is accurately determined through the experimental video and set as the experimental area. Based on this, the environmental data is analyzed and judged, so that the whole monitoring and judgment mechanism is more in line with the actual experimental situation, greatly improving the scientificity, accuracy and early warning ability of potential risks of laboratory environmental safety monitoring, and building a solid protection line for the safe and stable development of various experiments in the laboratory. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The connection diagram of the laboratory safety early warning system of the embodiment of the present application;

[0044] Figure 2 Schematic diagram of a common monitoring module according to an embodiment of the present invention;

[0045] Figure 3 This is a working step diagram of the laboratory safety early warning system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0047] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0048] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0049] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0050] See also Figure 1 As shown in FIG, it is a connection diagram of a laboratory safety early warning system according to an embodiment of the present invention. An embodiment of the present invention provides a laboratory safety early warning system, comprising:

[0051] The registration module is used to record the entry time, experimental content and departure time of the experimenter into the laboratory;

[0052] A portrait tracking module is used to identify, track, and record the experimental videos of each experimenter in the laboratory. In implementation, at least two cameras are set up, at least one of which can monitor the laboratory door to identify people entering and leaving.

[0053] A digital twin module is used to generate a digital twin model of the laboratory. In implementation, the digital twin module includes several binocular cameras to determine a complete digital twin model of the laboratory.

[0054] A general monitoring module is used to obtain real-time environmental monitoring data of the laboratory and the experimental area, including temperature data, humidity data, and smoke data;

[0055] A monitoring and analysis module, which is connected to the registration module and is used to determine the laboratory monitoring method based on whether there are experimenters in the laboratory;

[0056] The monitoring method includes using a portrait tracking module to identify and track each experimenter in the laboratory and capture each experimenter's experimental video, and using a digital twin module to generate a digital twin model of the laboratory. During implementation, it also includes obtaining environmental monitoring data of the experimental area and / or laboratory through a general monitoring module.

[0057] a monitoring and control module, which is respectively connected to the registration module, the portrait tracking module, the digital twin module, the general monitoring module, and the monitoring and analysis module, and is used to determine a safety warning library based on the experimental content of each experimenter to identify abnormal behavior of the experimenter, determine the operating experimental equipment based on the experimental content and the experimental video to perform real-time monitoring and update of the digital twin model corresponding to the operating experimental equipment, and control the general monitoring module to perform general monitoring of the experimental area to determine the impact of the non-experimental area on the experimental area to determine whether to issue a safety warning;

[0058] The safety warning library includes precautions and error demonstrations corresponding to the experimental content.

[0059] It can be understood that the laboratory safety early warning system provided by the present invention accurately records the entry and exit time, experimental content and other information of the experimenter through the registration module, which is convenient for traceability management; the portrait tracking module effectively identifies and tracks the experimenter and records the experimental video, which is conducive to the subsequent review and analysis of the experimental process; the laboratory digital twin model generated by the digital twin module, combined with the monitoring method determined by the monitoring and analysis module, can intuitively present the operating status of the laboratory; the environmental monitoring data of the laboratory and experimental area are obtained in real time through the general monitoring module to provide a basis for environmental safety assessment; the monitoring and control module determines the safety warning library according to the experimental content to accurately identify the abnormal behavior of the experimenter, and at the same time operates the experimental equipment based on the experimental content and experimental video and updates part of the digital twin model content, and also controls the general monitoring module to analyze the impact of the non-experimental area on the experimental area. Characterization trend, thereby comprehensively and intelligently realizing laboratory safety early warning, effectively preventing accidents, ensuring the safety of experimental personnel, the smooth progress of experiments and the stability of the laboratory environment.

[0060] See also Figure 2 As shown in FIG, it is a schematic diagram of a common monitoring module according to an embodiment of the present invention. Specifically, the common monitoring module includes:

[0061] The overall monitoring unit is set at a fixed position in the laboratory to obtain the laboratory's environmental monitoring data in real time; it is understandable that the sensors for obtaining the environmental monitoring data are any of the existing technologies;

[0062] The mobile monitoring unit is configured to be mobile and is used to obtain environmental monitoring data of the experimental area in real time. In implementation, it can be configured as a patrol vehicle equipped with sensors for obtaining environmental monitoring data. It can also be configured to be any movable shape, as long as it can be moved to the experimental area and obtain the environmental monitoring data of the experimental area.

[0063] It is understandable that the overall monitoring unit in the ordinary monitoring module remains in an on state to obtain real-time environmental monitoring data of the laboratory; when the mobile monitoring unit determines that there is an experimental area, the monitoring control module controls the mobile monitoring unit to move to the location of the experimental area.

[0064] It is understandable that the overall monitoring unit of the ordinary monitoring module is fixed at a specific location in the laboratory and is always turned on. It can continuously and stably collect the overall environmental monitoring data of the laboratory, providing a reliable basis for fully understanding the overall environmental conditions of the laboratory, and helping to promptly detect situations such as abnormal overall temperature fluctuations, humidity imbalance or smoke hazards, to ensure the basic environmental safety of the laboratory; while the mobile monitoring unit is flexible. When the monitoring and control module determines that there is an experimental area, it can be accurately moved to the location of the experimental area to obtain its environmental monitoring data. This can not only perform accurate environmental monitoring of the core area of ​​the experiment, such as detecting special temperature and humidity requirements or tiny smoke changes in the local experimental area, but also effectively make up for the monitoring blind spots of fixed monitoring points, and realize more detailed and targeted environmental monitoring of the experimental area. The two work together to greatly improve the comprehensiveness, accuracy and flexibility of laboratory environmental monitoring, and build a solid data monitoring line of defense for the safe operation of the laboratory.

[0065] See also Figure 3 As shown, it is a working step diagram of the laboratory safety early warning system according to an embodiment of the present invention. Specifically, the monitoring and analysis module determines the laboratory monitoring method according to whether there is an experimenter in the laboratory to control the portrait tracking module to start working according to the monitoring method, including:

[0066] If there are experimenters in the laboratory, the monitoring method is determined to be to control the portrait tracking module to start working to identify and track the portraits of each experimenter in the laboratory and to capture the experiment video of each experimenter. It is understandable that the portrait recognition and tracking technology is existing technology and will not be described in detail. In implementation, the portrait tracking module determines the number of experiment videos to be captured based on the number of experimenters in the laboratory.

[0067] If there is no experimenter in the laboratory, it is determined that the monitoring method is to control the digital twin module to generate a digital twin model in the laboratory to monitor the laboratory, and the portrait tracking module is not controlled to start working.

[0068] It is understandable that when there are people in the laboratory, the experimenter in the experiment can check whether the running experimental equipment is normal, so there is no need to turn on the digital twin module and determine whether the running experimental equipment is operating normally through the generated digital twin model; when there are no experimenters in the laboratory, the safety of the equipment in the laboratory can only be determined by turning on the digital twin module.

[0069] It is understandable that the monitoring and analysis module can intelligently determine the monitoring method based on whether there are experimenters in the laboratory. When there are experimenters, the portrait tracking module is quickly started, and the experimental video is shot with the help of portrait recognition and tracking technology. The number of shots can be determined based on the number of people. This not only helps to monitor the experimenter's operating behavior in real time and promptly discover possible illegal operations or abnormal conditions, but also provides intuitive and detailed image data for subsequent experimental analysis and accident tracing; when there are no experimenters in the laboratory, it is reasonable to switch to the digital twin module to generate a laboratory model for monitoring, avoiding unnecessary waste of portrait tracking resources, and effectively monitoring equipment safety through the digital twin model to ensure that the laboratory's operating status is still under control when unattended, greatly improving the intelligence level of laboratory monitoring, resource utilization efficiency, and overall safety and reliability, and realizing accurate, efficient, and adaptive monitoring and management of the laboratory under different personnel status.

[0070] Specifically, the monitoring control module determines whether to send a safety warning message based on the identity and number of the experimenters identified by the portrait tracking module and the experimenters recorded in the registration module, including:

[0071] If the comparison result shows that the identity and number of the identified experimenters are consistent with the recorded experimenters, it is determined that no safety warning information will be sent; it is understandable that the number of people in the laboratory and their respective identities at any time can be determined through the name registration module (fingerprint registration, face recognition can be used to enter the door, etc., and fingerprint and face recognition are required again when leaving), and the number of people in the laboratory and their respective identities can also be identified in real time through the portrait tracking module (face recognition can be used). Therefore, by comparing the number of people and identities, it can be determined whether there are non-experimental personnel entering, which can improve the safety of the experiment; in addition, when the name registration module or the portrait tracking module finds that the number of people has changed, it will trigger the monitoring and control module to compare the number and identities of the identified and recorded experimenters;

[0072] If the comparison result is that the identity and / or number of the identified experimenters do not match the recorded experimenters, it is determined that a safety warning message is sent to the experimenters recorded in the name registration module; in implementation, if there is a mismatch, a safety warning message is sent based on the currently registered experimenters entering the name registration module to remind the correct experimenter to confirm the identity of the people in the laboratory. In one implementation, the name registration module records that there should be 3 experimenters at the current moment, and the portrait tracking module finds that there are not equal to 3 experimenters, then it is necessary to send information to the 3 experimenters recorded in the name registration module; in one implementation, the name registration module records that there should be a and b in the laboratory at the current moment, but the portrait tracking module identifies a and c, then it is necessary to send information to a and b.

[0073] It is understandable that the monitoring and control module cleverly integrates the data information of the name registration module and the portrait tracking module, and accurately judges the laboratory personnel situation by rigorously comparing the identity and number of identified experimenters with the recorded experimenter information; when the two are consistent, the normal operation order of the laboratory is maintained, no warning information is sent, and unnecessary interference is avoided; and once a discrepancy occurs, a safety warning information is immediately sent to the experimenter recorded in the name registration module. This mechanism effectively prevents illegal intrusion by non-experimental personnel, greatly improves the safety and confidentiality of the laboratory, and protects the experimental process from external interference and potential threats. At the same time, it also provides experimenters with a safe and focused experimental environment, making laboratory management more standardized, orderly and intelligent, effectively reducing the risk of safety accidents caused by poor personnel management, and strongly promoting the efficient operation and improvement of the laboratory safety management system.

[0074] Specifically, the monitoring and control module determines the corresponding safety warning library based on the experimental content recorded by each experimenter in the registration module and trains the classification model based on the safety warning library to obtain a safety warning classification model. In addition, the monitoring and control module divides the experimental video into several frames and then determines several key frames, and extracts human images in each key frame to input each human image into the corresponding safety warning classification model to identify abnormal behavior of the corresponding experimenter.

[0075] It is understandable that each experiment has its corresponding precautions and error demonstrations, and the precautions and error demonstrations corresponding to each experiment are set as a safety warning library for the experimental content. For example, the experimental contents such as titration, grinding, coating, etc. are only different in specific raw materials, while the overall process, steps, precautions and error demonstrations are the same.

[0076] In implementation, the classification models that can be used include: (1) Computer vision-based models: ① YOLO series models, such as YOLOv5. YOLO series models are widely used for their fast and accurate target detection capabilities. They can identify whether the experimenter performs tasks according to the established operating procedures, whether the correct tools are used, and whether safety regulations are followed. Through training with the safety warning library, the model can learn to identify correct and incorrect action patterns in various experimental operations, thereby detecting and warning incorrect actions in real time; for example, in chemical experiments, it can detect whether the dropper is used correctly, whether the reagent is poured in the specified area, etc.; the image and video data in the safety warning library can be used as additional training Data can enrich the model's understanding of incorrect actions; for example, after annotating images or video clips of different types of incorrect actions in the safety warning library and adding them to the training set, the YOLO model can learn these specific error patterns, thereby improving the ability to detect similar incorrect actions; ② CNN can automatically extract features from images. By learning and training a large amount of annotated correct and incorrect experimental action image data, CNN can learn the feature representations of different actions, and then classify new experimental action images to determine whether they are incorrect actions; for example, whether the operating gestures and positions of instruments in physical experiments are correct can be identified; CNN can be pre-trained or fine-tuned using image data in the safety warning library. By allowing CNN to learn the features of the incorrect action images in the safety warning library during the pre-training stage, the model can better capture the abnormal features in the experimental actions, and then more accurately identify incorrect actions in practical applications; (2) Models based on video analysis: ① Temporal Action Localization ion (temporal action localization) model, which focuses on locating and classifying actions in videos, and can understand the sequence and time span of actions in video content; through training, it can identify whether the actions at different stages of the experiment meet the specifications and locate the time segments where incorrect actions occur; for example, in biological experiments, it can identify whether the operations of different steps in the cell culture process are correct and the time when the incorrect actions occur; ②C3D is a video classification model based on 3D convolutional neural networks, which can simultaneously capture the spatiotemporal information of the video and model and classify the action sequences in the experimental process; by inputting continuous experimental operation video frames, the C3D model can learn the patterns of different actions in time and space, so as to judge whether the actions are correct; the video data in the safety warning library can provide more training samples for the C3D model, so that it can learn richer spatiotemporal features of incorrect actions. By learning and analyzing these data, the C3D model can establish a more accurate classification model and improve the recognition accuracy and recall rate of incorrect actions in the experimental process.

[0077] It is understandable that the monitoring and control module accurately determines the corresponding safety warning library based on the experimental content recorded by each experimenter in the registration module and uses it to train the classification model to obtain a safety warning classification model. This measure enables the system to conduct in-depth cognition and learning of the potential risks of different experimental types, and build a highly adaptive safety assessment system; then, the experimental video is divided into several frames and the key frames are determined, and the human body images are extracted and input into the safety warning classification model, so as to achieve effective identification of abnormal behaviors of the corresponding experimenters; this intelligent and refined processing method can keenly capture operational behaviors that do not comply with experimental specifications in complex experimental scenarios, such as violations of titration operation procedures, incorrect grinding techniques, etc., and issue alarms in time to effectively prevent safety accidents caused by improper human operations, greatly improving the standardization and safety of laboratory operations and the ability to control various experimental risks, and providing solid and powerful technical support and guarantee for the safe and stable operation of the laboratory.

[0078] Specifically, the monitoring control module determines the similarity of each frame image according to the machine learning model to determine a number of identical frame groups and selects a frame with the highest definition in each identical frame group as a key frame;

[0079] The similarity between the frames in the same frame group should be greater than or equal to a preset similarity, and the number of key frames is equal to the number of the same frame group.

[0080] It is understandable that the method of determining the similarity of images through a machine learning model is an existing technology and will not be elaborated on here. In implementation, the preset similarity affects the number of key frames. The greater the preset similarity, the more key frames there are and the more accurate the abnormal behavior recognition results are. In order to balance computing resources and the results of abnormal behavior recognition, the preset similarity ∈ [90%, 100%) is preferably set to 95%.

[0081] In implementation, (1) if the abnormal behavior of the experimenter is identified through the YOLO series model or convolutional neural network, this step is performed to extract several key frames in each experimental video; this method can reduce the waste of computer resources; (2) if the abnormal behavior of the experimenter is identified through the Temporal Act Localization model or the C3D model, this step is not performed to directly determine the abnormal behavior identification through the experimental video.

[0082] It is understandable that the monitoring and control module uses the machine learning model to determine the similarity of each frame image to accurately divide the same frame group, and cleverly selects the frame with the highest clarity in each group as the key frame. This operation not only effectively reduces data redundancy and improves data processing efficiency, but also provides high-quality and representative image samples for subsequent abnormal behavior identification; in addition, by reasonably setting the preset similarity, it not only ensures the accuracy of the abnormal behavior identification results but also achieves the optimal allocation of computing resources and avoids excessive resource consumption.

[0083] When using the YOLO series models or convolutional neural networks for abnormal behavior identification, the step of extracting key frames can significantly reduce the waste of computer resources and make the system run more efficiently and smoothly; for the Temporal Act Localization model or the C3D model, it can flexibly choose not to perform this step and directly determine the abnormal behavior identification through the experimental video, which reflects the adaptability and intelligence of the monitoring and control module in different model application scenarios, comprehensively enhances the accuracy and efficiency of the laboratory safety early warning system in identifying abnormal behaviors of experimenters, as well as the stability and reliability of the overall operation, and builds a solid technical defense line for laboratory safety assurance.

[0084] Specifically, the monitoring and control module determines the running experimental equipment according to the experimental content and the experimental video and controls the digital twin module to monitor the running experimental equipment in real time to update the corresponding partial digital twin model in real time, and generates a digital twin model of the laboratory according to the partial digital twin model to determine the safety characterization status of the laboratory, including:

[0085] If all devices in the real-time updated digital twin model are normal, the impact characterization trend of the non-experimental area on the experimental area is determined based on the environmental monitoring data of the experimental area to determine the safety characterization status of the laboratory;

[0086] If there is an abnormal device in the digital twin model updated in real time, the safety representation state is determined to be a dangerous state and an alarm signal is issued.

[0087] In practice, the alarm signals include sounding an alarm bell, lighting an alarm light, and sending an alarm message to the laboratory administrator and laboratory building security personnel.

[0088] It can be understood that the monitoring control module determines the running experimental equipment according to the experimental content and the experimental video, and drives the digital twin module to monitor the running experimental equipment in real time and update the corresponding part of the digital twin model, which not only makes the running state of the experimental equipment intuitively presented, but also provides the possibility for timely discovery of potential problems of the equipment; through deep analysis of the digital twin model, once an abnormal equipment is found, the safety representation state can be quickly determined as a dangerous state and a multi-element alarm signal including an alarm bell, a constant alarm lamp and information sent to the laboratory administrator and security personnel can be immediately sent to ensure that the dangerous situation can be known and handled in time; when the equipment is normal, the influence representation trend of the non-experimental area on the experimental area can be further analyzed in combination with the experimental area environment monitoring data, so that the safety representation state of the laboratory as a whole can be comprehensively and accurately determined, realizing the all-round and intelligent safety monitoring of the laboratory from equipment operation to environmental influence, greatly improving the timeliness, accuracy and efficiency of laboratory safety management, effectively reducing the risk of safety accidents, and providing a solid and reliable guarantee for the stable operation and personnel safety of the laboratory.

[0089] Specifically, the digital twin model updated by the monitoring control module in real time is composed of part of the digital twin model and the fixed digital twin model.

[0090] The monitoring control module monitors the running experimental equipment in real time to update the part of the digital twin model of the running experimental equipment, and monitors and updates the fixed digital twin model of the non-running experimental equipment each time the digital twin model is started (i.e. when the experimental staff leave the laboratory).

[0091] In implementation, if there is no running experimental equipment, the digital twin model is updated according to the preset update frequency, which is updated once every hour to once every two hours.

[0092] It can be understood that the monitoring control module divides the digital twin model into part of the digital twin model and the fixed digital twin model, monitors the running experimental equipment in real time and updates part of the digital twin model, so that the dynamic changes in the equipment operation can be reflected in the model in time, providing a reliable basis for accurately grasping the real-time status of the equipment; the fixed digital twin model of the non-running experimental equipment is monitored and updated when the digital twin module is just started, effectively ensuring the accuracy and timeliness of the state information of the non-running equipment, and then updated according to the preset update frequency, which not only ensures that the model data is not outdated, but also reasonably balances the consumption of system resources; such comprehensive, intelligent and flexible digital twin model updating strategy greatly improves the completeness, timeliness and accuracy of the laboratory equipment state monitoring, and provides strong technical support and guarantee for the improvement of laboratory safety management, equipment maintenance and overall operation efficiency.

[0093] Specifically, the monitoring control module determines the influence characterization trend of the non-experimental area on the experimental area according to the environmental monitoring data of the experimental area to determine the safety characterization state of the laboratory, including,

[0094] If the environmental monitoring data of the experimental area is not in the standard environmental data interval of the experimental area, it is determined that the influence characterization trend of the non-experimental area on the experimental area is a negative influence trend, and the safety characterization state is determined as a dangerous state, the monitoring control module controls the digital twin module to update part of the digital twin model and the fixed digital twin model and sends an alarm signal;

[0095] If the environmental monitoring data of the experimental area is in the standard environmental data interval of the experimental area, it is determined that the influence characterization trend of the non-experimental area on the experimental area is an implicit influence trend, and the safety characterization state is determined as a safe state.

[0096] It can be understood that the standard environmental data interval of the experimental area is input by the experimenter after preparing the experimental preparation.

[0097] Specifically, the monitoring control module determines the storage location of the experimental preparation (i.e. the experimental product prepared after each experiment of the experimenter) according to the experimental video and determines the storage location as the experimental area.

[0098] It can be understood that the monitoring control module can determine the influence characterization trend of the non-experimental area on the experimental area according to the environmental monitoring data of the experimental area, and further determine the safety characterization state of the laboratory; when the environmental monitoring data of the experimental area exceeds the standard environmental data interval, it is determined as a negative influence trend and the safety characterization state is determined as a dangerous state, and then the digital twin module is controlled to update part of the digital twin model and the fixed digital twin model, and an alarm signal is sent. This series of measures can enable laboratory managers to know about abnormal situations in time, take corresponding measures quickly, effectively avoid the further expansion of danger, and maximize the safety of the laboratory; when the environmental monitoring data is in the standard interval, it is determined as an implicit influence trend and a safe state, realizing the meticulous control of the laboratory environment state; at the same time, the storage location of the experimental preparation is accurately determined through the experimental video and set as the experimental area, and the analysis and judgment of the environmental data are based on this, so that the whole monitoring and judgment mechanism is more in line with the actual experimental situation, greatly improving the scientificity, accuracy and early warning ability of potential risks of the laboratory environment safety monitoring, and building a solid protection line for the safety and stability of the laboratory to carry out various experiments.

[0099] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0100] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A laboratory safety early warning system, characterized in that: include: The registration module is used to record the entry time, experimental content and departure time of the experimenter into the laboratory; Portrait tracking module, used to identify, track and record the experimental videos of each experimenter in the laboratory; Digital twin module, used to generate a digital twin model of the laboratory; A general monitoring module is used to obtain real-time environmental monitoring data of the laboratory and the experimental area, including temperature data, humidity data, and smoke data; A monitoring and analysis module, which is connected to the registration module and is used to determine the laboratory monitoring method based on whether there are experimenters in the laboratory; The monitoring method includes using a portrait tracking module to identify and track the portraits of each experimenter in the laboratory and shoot experimental videos of each experimenter, and using a digital twin module to generate a digital twin model of the laboratory; a monitoring and control module, which is respectively connected to the registration module, the portrait tracking module, the digital twin module, the general monitoring module, and the monitoring and analysis module, and is used to determine a safety warning library based on the experimental content of each experimenter to identify abnormal behavior of the experimenter, determine the operating experimental equipment based on the experimental content and the experimental video to perform real-time monitoring and update of the digital twin model corresponding to the operating experimental equipment, and control the general monitoring module to perform general monitoring of the experimental area to determine the impact of the non-experimental area on the experimental area to determine whether to issue a safety warning; The monitoring and analysis module determines the monitoring mode of the laboratory according to whether there is an experimenter in the laboratory and controls the portrait tracking module to start working according to the monitoring mode, including: If there are experimenters in the laboratory, the monitoring method is to control the portrait tracking module to perform portrait recognition and tracking on each experimenter in the laboratory and to shoot an experiment video of each experimenter; If there is no experimenter in the laboratory, it is determined that the monitoring method is to control the digital twin module to generate a digital twin model in the laboratory to monitor the laboratory, and the portrait tracking module is not controlled to start working; The monitoring and control module determines the impact characterization trend of the non-experimental area on the experimental area based on the environmental monitoring data of the experimental area to determine the safety characterization status of the laboratory, including: If the environmental monitoring data of the experimental area is not within the standard environmental data range of the experimental area, it is determined that the impact characterization trend of the non-experimental area on the experimental area is a negative impact trend and the safety characterization state is determined to be a dangerous state. The monitoring and control module controls the digital twin module to update the partial digital twin model and the fixed digital twin model and issue an alarm signal; If the environmental monitoring data of the experimental area is within the standard environmental data range of the experimental area, it is determined that the impact characterization trend of the non-experimental area on the experimental area is a hidden impact trend and the safety characterization state is determined to be a safe state.

2. The laboratory safety early warning system according to claim 1, characterized in that: The common monitoring module includes: The overall monitoring unit is set at a fixed location in the laboratory to obtain the laboratory's environmental monitoring data in real time; The mobile monitoring unit is configured to be mobile and is used to obtain environmental monitoring data of the experimental area in real time.

3. The laboratory safety early warning system according to claim 1, characterized in that: The monitoring control module determines whether to send a safety warning message based on the identity and number of the experimenters identified by the portrait tracking module and the experimenters recorded in the registration module, including: If the comparison result shows that the identity and number of the identified experimenters are consistent with the recorded experimenters, it is determined that no safety warning information will be sent; If the comparison result shows that the identity and / or number of the identified experimenters do not match the recorded experimenters, it is determined that a safety warning message is sent to the experimenters recorded in the registration module.

4. The laboratory safety early warning system according to claim 1, characterized in that: The monitoring and control module determines the corresponding safety warning library based on the experimental content recorded by each experimenter in the registration module and trains a classification model based on the safety warning library to obtain a safety warning classification model. In addition, the monitoring and control module divides the experimental video into several frames and then determines several key frames, and extracts human images in each key frame to input each human image into the corresponding safety warning classification model to identify abnormal behavior of the corresponding experimenter.

5. The laboratory safety early warning system according to claim 4, characterized in that: The monitoring control module determines the similarity of each frame image according to the machine learning model to determine a number of identical frame groups and selects a frame with the highest definition in each identical frame group as a key frame; The similarity between the frames in the same frame group should be greater than or equal to a preset similarity, and the number of key frames is equal to the number of the same frame group.

6. The laboratory safety early warning system according to claim 1, characterized in that: The monitoring and control module determines the operating experimental equipment according to the experimental content and the experimental video and controls the digital twin module to monitor the operating experimental equipment in real time to update the corresponding partial digital twin model in real time, and generates a digital twin model of the laboratory according to the partial digital twin model to determine the safety characterization status of the laboratory, including: If all devices in the real-time updated digital twin model are normal, the impact characterization trend of the non-experimental area on the experimental area is determined based on the environmental monitoring data of the experimental area to determine the safety characterization status of the laboratory; If there is an abnormal device in the digital twin model updated in real time, the safety representation state is determined to be a dangerous state and an alarm signal is issued.

7. The laboratory safety early warning system according to claim 6, characterized in that: The digital twin model updated in real time by the monitoring and control module consists of a partial digital twin model and a fixed digital twin model; Among them, the monitoring and control module monitors the running experimental equipment in real time to update the partial digital twin model in real time, and the monitoring and control module monitors and updates the fixed digital twin model of the non-running experimental equipment when the digital twin model is turned on.

8. The laboratory safety early warning system according to claim 1, characterized in that: The monitoring and control module determines the storage location of the experimental preparation according to the experimental video and determines the storage location as the experimental area.

Images