Laboratory alarm method, device and medium
By constructing a digital twin model of the laboratory and a data interface mapping relationship, the laboratory status can be monitored in real time, and sudden parameters can be identified and alarms triggered. This solves the problems of real-time monitoring and misjudgment in the laboratory, and improves the safety and operational efficiency of the laboratory.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-16
- Publication Date
- 2026-03-13
AI Technical Summary
Existing technologies are insufficient for real-time monitoring and automated alarms in laboratories, and traditional monitoring methods suffer from problems such as misjudgment and high costs.
By constructing a digital twin model of the laboratory, the data interface mapping relationship is determined, real-time parameters of the experimental scenario are obtained in real time, and compared with historical parameters. Combined with the correlation diagram, sudden parameter changes are identified and alarm prompts are triggered.
It enables precise monitoring and analysis of laboratory status, reduces false alarms, ensures the real-time nature and accuracy of alarms, and improves laboratory safety and operational efficiency.
Smart Images
Figure CN121661791A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a laboratory alarm method, device, and medium, belonging to the field of data processing technology. Background Technology
[0002] With the rapid development of science and technology, laboratories, as important venues for scientific research and technological innovation, provide a space for students, teachers, and other researchers to study and conduct research. However, due to the potential hazards posed by equipment and instruments within laboratories due to the research environment, traditional laboratory management often relies on manual inspections and periodic maintenance to ensure safety. This approach struggles to achieve real-time monitoring and effective early warning of the laboratory environment and equipment. Furthermore, this method is difficult to automate and is prohibitively costly. Therefore, automating laboratory monitoring and alarm systems is of paramount importance.
[0003] Current laboratory monitoring typically relies on sensors or cameras to monitor equipment in use and the laboratory environment, identifying anomalies in specific equipment or the overall environment. Alternatively, it may use operational data to monitor physical equipment for abnormalities. However, laboratories often use standardized circuits, water systems, or other resources, providing separate resources for different physical devices, and there may be overlap between different experimental procedures. If a piece of physical equipment malfunctions, it not only affects its own operation but also negatively impacts other equipment involved in the experimental process, ultimately affecting the entire laboratory's operation. Current monitoring methods based solely on sensors, cameras, or physical equipment operational status struggle to provide reliable alarms. Furthermore, independent monitoring methods, which rely on separate monitoring results from different physical devices to assess hazardous situations, are prone to misjudgments, resulting in low alarm reliability. Summary of the Invention
[0004] This invention addresses the shortcomings of existing technologies by providing a laboratory alarm method, device, and medium.
[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:
[0006] This specification provides one or more embodiments of a laboratory alarm method, the method comprising:
[0007] Acquire the collected data from the laboratory to be tested in order to construct a digital twin model of the laboratory to be tested;
[0008] Based on the correlation between the various experimental scenarios in the laboratory to be tested, the data interface mapping relationship of the digital twin model is determined;
[0009] Based on the digital twin model, the activated physical equipment of the laboratory to be tested is determined, and the current experimental scenario of the laboratory to be tested is determined based on the activated physical equipment.
[0010] Based on the data interface mapping relationship, the real-time parameters of the current experimental scenario are obtained in real time, and the real-time parameters and the historical parameters corresponding to the real-time parameters are compared to determine the mutation parameters of the current experimental scenario.
[0011] Based on the mutation parameters and the correlation diagram, the linked experimental scenario corresponding to the mutation parameters is determined, and the change trend information of the mutation parameters is obtained, so as to provide alarm prompts for the current experimental scenario and the linked experimental scenario according to the change trend information.
[0012] Optionally, in one or more embodiments of this specification, the step of acquiring the collected data of the laboratory to be tested to construct a digital twin model of the laboratory to be tested specifically includes:
[0013] Obtain the architectural design drawings corresponding to the laboratory to be tested, and establish a building information model of the laboratory to be tested based on the architectural design drawings;
[0014] The type and location of the physical equipment in the laboratory to be tested are obtained, and the three-dimensional model of each physical equipment is called according to the type of each physical equipment. Based on the location of the physical equipment, the three-dimensional model of the physical equipment is added to the building information model to obtain static scene environment data.
[0015] The monitoring images of the laboratory to be tested are acquired by the pre-set monitoring equipment, and the dynamic scene environment data of the laboratory to be tested is determined based on the difference image between consecutive monitoring images.
[0016] The static scene environment data, the dynamic scene environment data, and the operating data of each physical device are spatiotemporally aligned to construct a digital twin model of the laboratory to be tested.
[0017] Optionally, in one or more embodiments of this specification, the data interface mapping relationship of the digital twin model is determined based on the correlation between various experimental scenarios in the laboratory to be tested, specifically including:
[0018] Based on the historical operation records corresponding to each experimental scenario in the laboratory to be tested, the experimental constraint data corresponding to each experimental scenario is determined; wherein, the experimental constraint data includes: experimental procedures and experimental limitations.
[0019] Based on the experimental constraint data, the dependency and interaction relationships between each experimental scenario are determined, and based on the dependency and interaction relationships, an association graph between each experimental scenario is constructed.
[0020] Based on the relationship graph, the interaction parameters and function execution parameters of each experimental scenario are determined, and the data interface matching conditions corresponding to each experimental scenario are determined according to the interaction parameters and function execution parameters.
[0021] Based on the data interface matching conditions, query the information of each data interface to determine one or more specified data interfaces corresponding to each experimental scenario.
[0022] The specified data interfaces of each experimental scenario are mapped to the virtual data interfaces of the digital twin model to determine the data interface mapping relationship of the digital twin model.
[0023] Optionally, in one or more embodiments of this specification, determining the activated physical equipment of the laboratory to be tested based on the digital twin model, and determining the current experimental scenario of the laboratory to be tested based on the activated physical equipment, specifically includes:
[0024] Based on the current time and the experimental schedule of the laboratory to be tested, the planned experimental scenario of the laboratory to be tested is determined, and the planned physical equipment to be started based on the planned experimental scenario is compared with the physical equipment that has been started.
[0025] If they match, then the planned experimental scenario is determined to be the current experimental scenario;
[0026] If there is a discrepancy, the running status of the activated physical devices and each activated physical device is input into a preset identification model to determine the current experimental scenario of the laboratory to be tested based on the preset identification model.
[0027] Optionally, in one or more embodiments of this specification, before inputting the started physical devices and their operating states into a preset identification model, the method further includes:
[0028] Based on the recognition target, training data, and model performance of the preset recognition model, the model training task corresponding to the preset recognition model is determined.
[0029] Obtain the layer structure information corresponding to the model training task, and determine the hardware computing resource requirements corresponding to the layer structure information based on the matching relationship between the layer structure information and historical training data.
[0030] Based on the hardware resource computing power requirements corresponding to each layer structure information and the hardware resources of the heterogeneous device cluster corresponding to the model training task, the model cutting layers and horizontal cutting points corresponding to the model training task are determined; wherein, the horizontal cutting points are used to define the boundaries of each model cutting layer.
[0031] Based on the model segmentation layers and the horizontal segmentation points, multiple computation graphs corresponding to the model training task are determined. Based on each computation graph, the heterogeneous device cluster is statically partitioned to obtain multiple task execution units; wherein, the task execution units are composed of heterogeneous devices.
[0032] Obtain the real-time configuration information of the current hardware cluster, and dynamically adjust the heterogeneous device composition within each task execution unit based on the real-time configuration information.
[0033] Optionally, in one or more embodiments of this specification, real-time parameters of the current experimental scenario are obtained in real time based on the data interface mapping relationship, and the real-time parameters are compared with historical parameters corresponding to the real-time parameters to determine the mutation parameters of the current experimental scenario, specifically including:
[0034] Based on the data interface mapping relationship, the real-time parameters of the current experimental scenario are obtained in real time, and the historical parameters corresponding to the real-time parameters are obtained in chronological order.
[0035] Obtain the first difference between the real-time parameter and the adjacent historical parameter, and obtain multiple second differences between two adjacent historical parameters in sequence based on the time sequence;
[0036] By comparing the multiple second differences, the historical range and direction of change of each historical parameter can be determined;
[0037] The current range and direction of change of the real-time parameter are determined by comparing the first difference with the adjacent second difference.
[0038] Based on the comparison between the current range of change and the historical range of change, as well as the current direction of change and the historical direction of change, the mutation parameters of the current experimental scenario are determined.
[0039] Optionally, in one or more embodiments of this specification, based on the mutation parameters and the correlation diagram, a linked experimental scenario corresponding to the mutation parameters is determined, specifically including:
[0040] Based on the attribute information of the mutation parameter and the acquisition device corresponding to the mutation parameter, the direct influence range of the mutation parameter is determined, the activated physical devices located within the direct influence range are obtained, and the experimental process node corresponding to the activated physical devices in the current scene is determined.
[0041] Based on the experimental process nodes and the correlation diagram, the direct linkage experimental scenarios corresponding to the mutation parameters are determined.
[0042] Starting from the experimental process node, the connection path is traced along the relationship graph to determine the inverse connection path that has a reverse effect on the mutation parameter based on the relationship between each path in the connection path.
[0043] The experimental scenario corresponding to the influence connection path is taken as the indirect linkage experimental scenario, and the direct linkage experimental scenario and the indirect linkage experimental scenario are taken as the linkage experimental scenario corresponding to the mutation parameter.
[0044] Optionally, in one or more embodiments of this specification, obtaining the change trend information of the mutation parameter to provide alarm prompts for the current experimental scenario and the linked experimental scenario based on the change trend information, specifically including:
[0045] The time series data of the mutation parameters are fitted to obtain the fitting curve corresponding to the mutation parameters;
[0046] Based on the parameter and shape information of the fitted curve, the variation trend information of the mutation parameter is determined; wherein, the variation trend information includes: inflection point information, rate of change, and trend direction;
[0047] If, based on the change trend information, it is determined that the time it takes for the mutation parameter to reach the preset parameter threshold is less than a preset time threshold, then the corresponding alarm mechanism is triggered.
[0048] Based on the alarm mechanism, an audible and visual alarm is triggered for the current experimental scenario, and an alarm prompt is given for the linked experimental scenario.
[0049] This specification provides one or more embodiments of a laboratory alarm device, the device comprising:
[0050] At least one processor; and,
[0051] A memory communicatively connected to the at least one processor; wherein,
[0052] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods described above.
[0053] This specification provides one or more embodiments of a non-volatile computer storage medium storing computer-executable instructions, the computer-executable instructions being configured to execute any of the methods described above.
[0054] The beneficial effects of this invention are: It utilizes a digital twin model to accurately simulate the laboratory, and combines this with a correlation diagram to determine data interface mapping relationships, making the monitoring and analysis of experimental scenarios more precise. Based on the data interface mapping relationship, real-time parameters of the current experimental scenario are acquired and compared with historical parameters, enabling timely detection and location of sudden parameter changes, ensuring real-time alarm response. Furthermore, by accurately identifying sudden parameters and the correlation diagram, the direct and indirect linkages of the sudden parameters to the experimental scenarios are determined, reducing misjudgments and omissions. By determining the changing trend of sudden parameters and identifying when they may exceed safe ranges or expected values, an alarm can be triggered immediately. This timely feedback mechanism helps laboratory personnel respond quickly, reducing losses and risks. Attached Figure Description
[0055] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:
[0056] Figure 1 This is a schematic diagram of a laboratory alarm method provided in the embodiments of this specification;
[0057] Figure 2 This specification provides a schematic diagram illustrating the construction logic of a digital twin model in an application scenario, as illustrated in an embodiment of the present invention.
[0058] Figure 3 This is a schematic diagram of the structure of a laboratory alarm device provided in the embodiments of this specification;
[0059] Figure 4 This is a schematic diagram of the structure of a non-volatile storage medium provided in the embodiments of this specification. Detailed Implementation
[0060] This specification provides a laboratory alarm method, device, and medium through its embodiments.
[0061] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.
[0062] like Figure 1 As shown in the diagram, this specification provides a schematic flowchart of a laboratory alarm method. Figure 1 As can be seen, in one or more embodiments of this specification, a laboratory alarm method includes:
[0063] S101: Acquire the collected data of the laboratory to be tested in order to construct a digital twin model of the laboratory to be tested.
[0064] Existing technologies often focus on monitoring and managing individual devices or single processes, lacking a comprehensive understanding and control of the entire laboratory system, and suffer from problems such as untimely and incomplete data acquisition. Therefore, to facilitate real-time reflection of the laboratory's environmental status and the operational status of physical equipment, thereby achieving comprehensive laboratory monitoring, this specification describes an embodiment that acquires scene environment data and physical equipment data of the laboratory under test using pre-set acquisition devices. This data is then fused to construct a digital twin model of the laboratory under test. The digital twin model can reflect the laboratory's environmental status and the operational status of physical equipment in real time, facilitating timely detection and response by management personnel, improving laboratory safety and operational efficiency. Furthermore, based on the digital twin model, overall monitoring of multi-source data for the entire laboratory is achieved, avoiding the difficulty in comprehensively understanding laboratory risks caused by focusing on individual devices or single processes, and helping to identify and resolve problems early.
[0065] Specifically, such as Figure 2 As shown in one or more embodiments of this specification, acquiring the collected data of the laboratory to be tested in order to construct a digital twin model of the laboratory to be tested specifically includes the following process:
[0066] First, the architectural design drawings corresponding to the laboratory to be tested are obtained, and a building information model (BIM) of the laboratory is established based on these drawings. The types and locations of the physical equipment within the laboratory are determined, and the 3D models of each physical device are retrieved according to its type. It should be noted that the BIM can be built using existing modeling tools based on Luban Civil Engineering Technology, a computer-aided design (CAD) platform-based technology for automatic quantity calculation and management. Using architectural design drawings and Luban Civil Engineering Technology to build the BIM ensures the accuracy of the laboratory's structural integrity. Then, the 3D models of the physical equipment are added to the BIM based on their locations to obtain static scene environment data. Simultaneously, the 3D models of the physical equipment are retrieved based on their type and location and integrated into the BIM, ensuring that the digital twin model highly replicates the actual laboratory conditions in physical space. Finally, monitoring images of the laboratory to be tested are acquired using pre-installed monitoring equipment. Dynamic scene environment data of the laboratory is determined based on the difference between consecutive frames of monitoring images. By acquiring monitoring images of the laboratory through pre-installed monitoring equipment and determining dynamic scene environment data based on the difference images of consecutive frames, the digital twin model can reflect the current state of the laboratory in real time. With the real-time acquisition and updating of physical equipment data, the digital twin model can dynamically reflect changes in the laboratory, including equipment operating status and environmental parameters. By combining static and dynamic scene environment data, the scene environment data of the laboratory to be tested is obtained. Then, based on the location and type of physical equipment, corresponding sensors are planned and deployed. Simultaneously, internal operating status data is acquired through the equipment's own communication interface. The external data from the sensors (data collected by external sensors) and the equipment's internal operating data (operating data provided by the equipment's internal interface) are spatiotemporally synchronized and semantically associated with the laboratory's 3D scene model to construct a mapping of the full state of the physical entity of the laboratory to be tested. That is, as shown... Figure 2 The required types and locations of pre-set acquisition devices are determined. External data collected by each type of pre-set acquisition device is used as the first physical device data. Then, based on the operational data of each physical device, internal operational status data is determined as the second physical device data. By fusing the location and scene environment data corresponding to the first and second physical device data, a digital twin model of the laboratory to be tested is constructed. In this process, by combining static scene environment data such as building structure and equipment layout with dynamic scene environment data such as real-time changes in monitoring images, and the operational data of the physical devices, a comprehensive digital twin model reflecting the laboratory's status is constructed. This allows the digital twin model to obtain comprehensive monitoring information and achieves information fusion, improving the reliability of laboratory testing.
[0067] S102: Determine the data interface mapping relationship of the digital twin model based on the correlation between the experimental scenarios in the laboratory to be tested.
[0068] After constructing the digital twin model of the laboratory to be tested based on the above step S101, in order to obtain data from the correct data source to drive and update the digital twin model, and to solve the data silo problem that may be caused by equipment resource sharing and process overlap when constructing the digital twin model of the laboratory to be tested, this embodiment of the specification determines the correlation between various experimental scenarios in the laboratory to be tested, and determines the correlation diagram of each experimental scenario. Based on the correlation diagram, the data interface mapping relationship of the digital twin model is determined, thereby realizing the acquisition of data from multiple experimental scenarios. By determining the correlation between various experimental scenarios, it can be ensured that the digital twin model, when reflecting the actual situation of the laboratory, not only focuses on individual scenarios, but also on the interaction and dependency between scenarios, thereby improving the integrity and accuracy of the model. Furthermore, the correlation diagram helps to clarify the data flow and interaction methods between different experimental scenarios, providing clear guidance for the data interface mapping relationship of the digital twin model. In the laboratory, there may be complex interrelationships and influences between different experimental scenarios. The correlation diagram can clearly show these complex relationships and support in-depth analysis of complex scenarios. The digital twin model constructed based on the correlation diagram can provide more comprehensive and accurate laboratory status information. It helps to clarify the dependencies and interaction methods between the various components in the digital twin model, thereby improving the maintainability of the model.
[0069] Specifically, in one or more embodiments of this specification, the data interface mapping relationship of the digital twin model is determined based on the correlation between various experimental scenarios in the laboratory to be tested, which specifically includes the following process:
[0070] First, based on the historical operation records corresponding to each experimental scenario in the laboratory under test, the experimental constraint data corresponding to each experimental scenario is determined. This constraint data includes the experimental procedure and experimental limitations. Then, based on the experimental procedure and limitations, the dependency and interaction relationships between each experimental scenario are determined, and a correlation graph is established based on these relationships. The interaction parameters and functional execution parameters of each experimental scenario are determined based on the correlation graph, and the data interface matching conditions corresponding to each experimental scenario are determined based on these parameters. Interaction parameters refer to data items or state quantities that are transmitted, exchanged, or have an impact between different experimental scenarios. These parameters can be obtained based on dependency and interaction relationship analysis. For example, when the output product of experimental scenario A is the input raw material of experimental scenario B, the material composition concentration and temperature are interaction parameters. Functional execution parameters refer to data items or state quantities that need to be monitored and controlled to ensure the normal and compliant operation of a single experimental scenario. For example, if a safety condition requires the indoor oxygen concentration to be higher than 19.5%, then the ambient oxygen content is a functional execution parameter. Data interface matching conditions are technical requirements specifications for physical data interfaces in order to accurately obtain or control the aforementioned interaction parameters and function execution parameters. They include constraints such as data types, communication protocols, and physical interface forms.
[0071] The data interface information is queried according to the data interface matching conditions to determine one or more specified data interfaces corresponding to each experimental scenario. Then, the specified data interfaces of each experimental scenario are mapped to the virtual data interfaces of the digital twin model to determine the data interface mapping relationship of the digital twin model.
[0072] In the above process, the experimental procedures and constraints are determined based on the historical operation records of each experimental scenario in the laboratory under test. This method ensures that the relationships and mappings are determined based on actual operational data, thereby improving the accuracy and reliability of subsequent analysis. By analyzing the experimental procedures and constraints, a comprehensive understanding of the dependencies and interactions between various experimental scenarios can be achieved, including direct and indirect dependencies, constraints, and influences, which helps to consider all possible scenarios and situations when constructing the digital twin model. Determining the specified data interface by querying and matching data interface information allows the system to flexibly adapt to different experimental scenarios and data requirements. Simultaneously, when the laboratory or experimental scenario changes, the relationship graph and mappings can be easily updated, achieving system scalability. Furthermore, determining the relationships and mappings through a systematic approach reduces the trial-and-error and debugging costs during the digital twin model development process. At the same time, a clear relationship graph and mappings also reduce the maintenance costs of the model in its later stages.
[0073] S103: Based on the digital twin model, determine the activated physical equipment of the laboratory to be tested, and determine the current experimental scenario of the laboratory to be tested based on the activated physical equipment.
[0074] To determine the current experimental scenario and perform precise analysis based on data acquired from the digital twin model, this embodiment of the specification identifies the activated physical devices in the laboratory to be tested using the digital twin model. By receiving real-time sensor data from the physical devices, the digital twin model can instantly reflect the operational status of the devices. Therefore, when the model detects that certain physical devices have been activated, it can quickly and accurately determine the current experimental scenario of the laboratory. This real-time performance and accuracy are particularly important for experimental environments requiring rapid response and precise control. Based on the above step S102, it is clear that the digital twin model not only focuses on the operational status of individual devices but also comprehensively considers the interactions and dependencies between multiple devices through relationship diagrams and data interface mappings. Therefore, when determining the current experimental scenario, it can comprehensively and holistically consider various factors, improving the accuracy and reliability of the judgment.
[0075] Specifically, in one or more embodiments of this specification, the activated physical equipment of the laboratory to be tested is determined based on a digital twin model, and the current experimental scenario of the laboratory to be tested is determined based on the activated physical equipment, specifically including:
[0076] First, based on the current time and the laboratory's experimental schedule, the planned experimental scenario for the laboratory to be tested is determined. This is then compared with the planned physical equipment to be started and the already started physical equipment corresponding to the planned experimental scenario. If they match, the planned experimental scenario is determined as the current experimental scenario. If they do not match, the already started physical equipment and their operational status are input into a preset recognition model, which then determines the current experimental scenario for the laboratory to be tested.
[0077] First, the planned experimental scenario is quickly determined by comparing the current time with the experimental schedule. This is an efficient method that rapidly narrows down the search scope. If the plan matches the actual situation, the current experimental scenario is directly determined, avoiding unnecessary computation and resource consumption. When the plan and the actual situation do not match, a pre-built identification model is introduced, using the operating status of already started physical equipment as input for accurate identification. This method fully considers the actual operating conditions of the equipment, improving the accuracy of determining the current experimental scenario. Compared to traditional manual judgment or simple automated judgment, this method reduces errors caused by human factors. Through automated data comparison and model recognition, the accuracy and reliability of the judgment are improved.
[0078] Furthermore, in one or more embodiments of this specification, before inputting the started physical devices and their operating states into the preset identification model, the method further includes:
[0079] First, based on the recognition target, training data, and model performance of the pre-defined recognition model, the model training task corresponding to the pre-defined recognition model is determined. Then, the layer structure information corresponding to this model training task is obtained, including the model's layer structure composition and layer parameter configuration. Based on the matching relationship between the layer structure information and historical training data, the historical training data closest to the current inference performance requirements is determined, and the reference item with the closest resource consumption characteristics is obtained, thereby estimating the hardware computing power resource requirements for a single inference operation of the layer structure information. The hardware computing power resource requirements include the number of floating-point operations and memory requirements. Then, based on the hardware computing power requirements corresponding to each layer information and the hardware resources of the heterogeneous device cluster corresponding to the model training task, the model segmentation layers and horizontal segmentation points corresponding to the model training task are determined; it should be noted that the horizontal segmentation points are used to define the boundaries of each model segmentation layer. Understandably, after obtaining the hardware computing resource requirements needed for the layer structure information, this is matched and analyzed against the available resources of each device in the target heterogeneous hardware cluster, such as a cluster containing CPUs, GPUs, NPUs, etc., to obtain the segmentation layers and the partitioning points between devices for each layer. Then, based on the model segmentation layers and lateral partitioning points, multiple computation graphs corresponding to the model training tasks are determined. Based on each computation graph, the heterogeneous device cluster is statically partitioned to obtain multiple task execution units. It should be noted that the task execution units are based on heterogeneous device composition. By obtaining the real-time configuration information of the current hardware cluster, the heterogeneous device composition within each task execution unit is dynamically adjusted based on this real-time configuration information. It should be noted that the real-time configuration information includes hardware characteristics and load characteristics. Hardware characteristics include the hardware type, quantity, operating speed, and memory size of devices such as CPUs, GPUs, and TPUs. Load characteristics include CPU utilization, memory usage, and network bandwidth.
[0080] During this process, by analyzing the model's layer structure information and historical training data, the hardware computing resources required for model training can be accurately estimated, including the number of floating-point operations and memory requirements. This helps to rationally allocate and utilize hardware resources in heterogeneous device clusters, avoiding resource waste and improving resource utilization. By obtaining the real-time configuration information of the current hardware cluster, the composition of heterogeneous devices within each task execution unit can be dynamically adjusted, enabling the system to make corresponding adjustments based on changes in the actual hardware environment, ensuring the smooth execution of model training tasks. Simultaneously, it also improves the system's adaptability to different hardware environments. This flexibility allows the system to make corresponding adjustments based on changes in the actual hardware environment, ensuring the smooth execution of model training tasks. Simultaneously, it also improves the system's adaptability to different hardware environments. This process of dynamically adjusting the initial logical computing units to architectural computing units, and further subdividing them into node computing units, allows for flexible adjustment of computing resource allocation based on real-time conditions. This not only supports the training needs of models of different scales but also enables rapid adaptation to increases or decreases in hardware resources. By combining static partitioning with dynamic adjustment, after static partitioning ensures that critical resource requirements are met, dynamic partitioning can flexibly respond to sudden demands or resource fluctuations, enabling the system to simultaneously meet the requirements of stability and flexibility. This approach reduces overall costs and improves cost-effectiveness while maintaining system performance.
[0081] S104: Based on the data interface mapping relationship, obtain the real-time parameters of the current experimental scenario in real time, and compare the real-time parameters with the historical parameters corresponding to the real-time parameters to determine the mutation parameters of the current experimental scenario.
[0082] In order to promptly identify abnormal data in the current experimental scenario and issue timely alarms to improve the safety of various experimental scenarios in the laboratory, this embodiment of the specification will obtain real-time parameters of the current experimental scenario in real time according to the data interface mapping relationship. Then, it will obtain historical data corresponding to the real-time parameters according to the time sequence, compare the current historical parameters with the real-time parameters, determine the sudden change parameters of the current experimental scenario, and use the sudden change parameters to realize the danger alarm of the experimental scenario.
[0083] Specifically, in one or more embodiments of this specification, real-time parameters of the current experimental scenario are obtained in real time based on the data interface mapping relationship, and the real-time parameters are compared with historical parameters corresponding to the real-time parameters to determine the mutation parameters of the current experimental scenario, specifically including:
[0084] Based on the established data interface mapping relationship, real-time parameters of the current experimental scenario are obtained in real time, and historical parameters corresponding to the real-time parameters are obtained in chronological order. Then, the first difference between the real-time parameter and adjacent historical parameters is obtained, and multiple second differences between adjacent historical parameters are obtained sequentially in chronological order. By comparing multiple second differences, the historical range and direction of change of each historical parameter can be determined. Then, the current range and direction of change of the real-time parameter are determined by comparing the first difference with adjacent second differences. Based on the comparison results of the current range and historical range, and the current direction of change with historical direction of change, the parameter whose range or direction of change abruptly changes can be identified, which is the abrupt parameter of the current experimental scenario. Assume that a parameter is subjected to: historical parameter 3, historical parameter 2, historical parameter 1, and real-time parameter. This quantifies the change in the parameter, and the first difference is the difference between the real-time parameter and historical parameter 1, representing the change from the most recent historical moment to the current moment. Multiple second differences are calculated sequentially between adjacent historical parameters, such as the difference between historical parameter 1 and historical parameter 2, the difference between historical parameter 2 and historical parameter 3, and so on. These difference sequences represent the continuous trajectory of the parameter's change over a historical period. By analyzing multiple second differences, historical change data such as the historical range and direction of change can be determined. By comparing the first difference with the nearest second difference in historical change data, it is possible to determine whether there are abrupt changes in amplitude or direction. If a sudden change in the current range or direction of change of the parameter is observed, the parameter is identified as a sudden change parameter.
[0085] This process involves acquiring real-time parameters of the current experimental scenario and comparing them with historical parameters to promptly detect anomalies during the experiment. For example, in a chemical experiment, if the concentration of a reactant suddenly increases or decreases, this abrupt change may indicate an unexpected alteration in reaction conditions or an abnormality in the reaction process. This process can quickly capture such changes, providing timely alerts to experimenters.
[0086] S105: Based on the mutation parameter and the correlation diagram, determine the linkage experimental scenario corresponding to the mutation parameter, and obtain the change trend information of the mutation parameter, so as to provide alarm prompts for the current experimental scenario and the linkage experimental scenario according to the change trend information.
[0087] Since laboratories typically use unified circuits, water systems, or other resources, providing resources separately for different physical devices, and there may be equipment overlap between different experimental procedures, timely early warning responses are necessary to promptly identify adverse effects from the physical devices or experimental procedures involved, and to avoid affecting related experimental scenarios. In the embodiments of this specification, the linked experimental scenarios corresponding to the mutation parameters will be determined based on the mutation parameters and the correlation diagram determined in the above process.
[0088] In complex experimental systems, a single mutation parameter may simultaneously affect multiple experimental scenarios. Therefore, based on this process, when a parameter in an experimental scenario undergoes a mutation, a correlation diagram can be used to systematically analyze the interactions and synergies between these scenarios, quickly locating other experimental parameters or scenarios associated with the mutated parameter. This rapid linkage identification capability allows researchers to immediately assess the potential impact of the mutation, enabling them to make swift decisions, adjust experimental conditions, or take other countermeasures. By quickly identifying linked experimental scenarios and taking corresponding countermeasures, the number of experimental interruptions and repetitions can be reduced, thereby improving experimental efficiency. Furthermore, based on the analysis results of the correlation diagram, researchers can design subsequent experiments more effectively, avoiding blind experimentation and wasted resources.
[0089] Specifically, in one or more embodiments of this specification, the linkage experimental scenario corresponding to the mutation parameter is determined based on the mutation parameter and the correlation diagram, which specifically includes the following process:
[0090] First, based on the attribute information of the mutation parameter and the corresponding acquisition device, the direct impact range of the mutation parameter is determined. Then, the activated physical devices within the direct impact range are identified to determine the corresponding experimental process nodes within the current scenario. Next, based on the experimental process nodes and the correlation graph, the directly linked experimental scenarios corresponding to the mutation parameter are determined. Then, starting from the experimental process nodes, the connection paths in the correlation graph are traced. Based on the correlation relationships of each path, the inverse connection paths affecting the mutation parameter are determined, and the experimental scenarios corresponding to these influencing connection paths are used as indirect linked experimental scenarios. The directly linked experimental scenarios and the indirect linked experimental scenarios are then used as the linked experimental scenarios corresponding to the mutation parameter. For example: Suppose a new drug's stability test is being conducted in a pharmaceutical factory laboratory. During the experiment, the temperature control equipment suddenly malfunctions, causing a sharp rise in the experimental temperature, becoming a mutation parameter. First, based on the attribute information of the temperature control equipment and the collected temperature data, it is determined that the temperature mutation directly affects the ongoing stability test experiment. Then, the temperature control equipment is identified as a key node in the current experimental process, directly related to the new drug's stability test process. Based on the experimental process nodes and the correlation graph, the stability testing experiments of the current new drug can be identified as the directly related experimental scenario to temperature mutations. Then, starting from the experimental process nodes of the temperature control equipment, the connection paths in the correlation graph are traced. It is found that temperature mutations may also affect subsequent drug purification and packaging experiments, as these require specific temperature conditions. Based on the correlations in the connection paths, these experimental scenarios indirectly affected by temperature mutations are identified as indirectly linked experimental scenarios. Finally, the directly and indirectly linked experimental scenarios are combined to form the linked experimental scenarios corresponding to the temperature mutation parameters.
[0091] This process accurately identifies the attributes and acquisition devices of mutation parameters, as well as their specific locations within the experimental procedure, enabling precise determination of direct and indirect interconnected experimental scenarios and reducing misjudgments and omissions. Furthermore, utilizing a correlation graph for path tracing and correlation analysis allows for the rapid identification of all interconnected experimental scenarios related to the mutation parameters, improving the efficiency of problem diagnosis and response. It not only considers the direct impact of the mutation parameters but also determines indirect impacts by tracing the connection paths in the correlation graph, thus ensuring the comprehensiveness of the analysis.
[0092] After determining the mutation parameters of the current experimental scenario in the laboratory to be tested according to the above steps, in order to provide alarm prompts for the laboratory, this embodiment of the specification will fit each mutation parameter to determine the change trend information of each mutation parameter, and then provide alarm prompts for the current experimental scenario and the linked experimental scenario based on the change trend information. The change trend information obtained by fitting can predict the future change trend of the mutation parameter. Once the change trend of the mutation parameter is determined by fitting and it is found that it may exceed the safe range or expected value, an alarm prompt can be triggered immediately. This timely feedback mechanism helps experimental personnel to respond quickly and reduce losses or risks. In addition, this process not only focuses on the change of a single mutation parameter, but also considers multiple parameters in the linked experimental scenario through a correlation diagram. Therefore, the alarm prompts can cover a wider range of experimental scenarios, ensuring the stability and safety of the entire experimental system.
[0093] Specifically, in one or more embodiments of this specification, obtaining the change trend information of the mutation parameter, and providing alarm prompts for the current experimental scenario and the linked experimental scenario based on the change trend information, specifically includes the following process:
[0094] First, the time series data of the mutation parameter is fitted, for example, using the least squares method, to obtain a fitted curve corresponding to the mutation parameter. Then, based on the parameter and shape information of the fitted curve, the trend information of the mutation parameter is determined. This trend information includes inflection point information, rate of change, and trend direction. Fitting the time series data of the mutation parameter using the least squares method yields a fitted curve that accurately reflects the parameter's change pattern. This allows for precise determination of the mutation parameter's trend, including inflection points, rate of change, and trend direction, based on the parameter and shape information of the fitted curve. If the time it takes for the mutation parameter to reach a preset parameter threshold is less than the preset threshold, a corresponding alarm mechanism is triggered. The alarm mechanism then triggers an audible and visual alarm for the current experimental scenario and provides alarm prompts for linked experimental scenarios. By analyzing the trend information, the time for the mutation parameter to reach the preset parameter threshold is predicted. It can be understood that the trend information includes, for example, instantaneous rate of change, average rate of change, acceleration, and indicators of linear growth, exponential growth, or growth following a certain function. Therefore, based on the observed change pattern, a matching mathematical model can be used to represent the trend. By substituting the current parameter value and the preset threshold into this mathematical model, the time it takes for the mutation parameter to reach the preset threshold can be predicted. If this time is less than the preset threshold, it indicates that the parameter is changing rapidly and may exceed the safe range or expected value. Triggering an alarm mechanism at this point provides early warning to the experimenters, giving them sufficient time to address potential problems or risks. This method not only focuses on the mutation parameter in the current experimental scenario but also considers linked experimental scenarios through a correlation diagram. Therefore, when the alarm mechanism is triggered, it will not only provide audible and visual alarms for the current experimental scenario but also alert related linked experimental scenarios. This helps experimenters fully understand the scope of the problem's impact and take appropriate countermeasures.
[0095] like Figure 3 As shown in the diagram, this specification provides a structural schematic of a laboratory alarm device. Figure 3 As can be seen, in one or more embodiments of this specification, a laboratory alarm device includes:
[0096] At least one processor; and,
[0097] A memory communicatively connected to the at least one processor; wherein,
[0098] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods described above.
[0099] like Figure 4 As shown in the diagram, this specification provides a schematic diagram of a non-volatile storage medium structure. Figure 4 As can be seen, in one or more embodiments of this specification, a non-volatile storage medium stores computer-executable instructions, which are capable of executing any of the methods described above.
[0100] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.
[0101] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.
[0102] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.
Claims
1. A laboratory alarm method, characterized in that, The method includes: Acquire the collected data from the laboratory to be tested in order to construct a digital twin model of the laboratory to be tested; Based on the correlation between the various experimental scenarios in the laboratory to be tested, the data interface mapping relationship of the digital twin model is determined; Based on the digital twin model, the activated physical equipment of the laboratory to be tested is determined, and the current experimental scenario of the laboratory to be tested is determined based on the activated physical equipment. Based on the data interface mapping relationship, the real-time parameters of the current experimental scenario are obtained in real time, and the real-time parameters and the historical parameters corresponding to the real-time parameters are compared to determine the mutation parameters of the current experimental scenario. Based on the mutation parameters and the correlation diagram, the linked experimental scenario corresponding to the mutation parameters is determined, and the change trend information of the mutation parameters is obtained, so as to provide alarm prompts for the current experimental scenario and the linked experimental scenario according to the change trend information.
2. The laboratory alarm method according to claim 1, characterized in that, The acquisition of data collected by the laboratory to be tested, in order to construct a digital twin model of the laboratory to be tested, specifically includes: Obtain the architectural design drawings corresponding to the laboratory to be tested, and establish a building information model of the laboratory to be tested based on the architectural design drawings; The type and location of the physical equipment in the laboratory to be tested are obtained, and the three-dimensional model of each physical equipment is called according to the type of each physical equipment. Based on the location of the physical equipment, the three-dimensional model of the physical equipment is added to the building information model to obtain static scene environment data. The monitoring images of the laboratory to be tested are acquired by the pre-set monitoring equipment, and the dynamic scene environment data of the laboratory to be tested is determined based on the difference image between consecutive monitoring images. The static scene environment data, the dynamic scene environment data, and the operating data of each physical device are spatiotemporally aligned to construct a digital twin model of the laboratory to be tested.
3. The laboratory alarm method according to claim 1, characterized in that, The step of determining the data interface mapping relationship of the digital twin model based on the correlation between various experimental scenarios in the laboratory to be tested specifically includes: Based on the historical operation records corresponding to each experimental scenario in the laboratory to be tested, the experimental constraint data corresponding to each experimental scenario is determined; wherein, the experimental constraint data includes: experimental procedures and experimental limitations. Based on the experimental constraint data, the dependency and interaction relationships between each experimental scenario are determined, and based on the dependency and interaction relationships, an association graph between each experimental scenario is constructed. Based on the relationship graph, the interaction parameters and function execution parameters of each experimental scenario are determined, and the data interface matching conditions corresponding to each experimental scenario are determined according to the interaction parameters and function execution parameters. Based on the data interface matching conditions, query the information of each data interface to determine one or more specified data interfaces corresponding to each experimental scenario. The specified data interfaces of each experimental scenario are mapped to the virtual data interfaces of the digital twin model to determine the data interface mapping relationship of the digital twin model.
4. A laboratory alarm method according to claim 1, characterized in that, Based on the digital twin model, the activated physical equipment of the laboratory to be tested is determined, and the current experimental scenario of the laboratory to be tested is determined based on the activated physical equipment, specifically including: Based on the current time and the experimental schedule of the laboratory to be tested, the planned experimental scenario of the laboratory to be tested is determined, and the planned physical equipment to be started based on the planned experimental scenario is compared with the physical equipment that has been started. If they match, then the planned experimental scenario is determined to be the current experimental scenario; If there is a discrepancy, the running status of the activated physical devices and each activated physical device is input into a preset identification model to determine the current experimental scenario of the laboratory to be tested based on the preset identification model.
5. A laboratory alarm method according to claim 4, characterized in that, Before inputting the started physical devices and their operating status into the preset identification model, the method further includes: Based on the recognition target, training data, and model performance of the preset recognition model, the model training task corresponding to the preset recognition model is determined. Obtain the layer structure information corresponding to the model training task, and determine the hardware computing resource requirements corresponding to the layer structure information based on the matching relationship between the layer structure information and historical training data. Based on the hardware resource computing power requirements corresponding to each layer structure information and the hardware resources of the heterogeneous device cluster corresponding to the model training task, the model segmentation layer and horizontal segmentation point corresponding to the model training task are determined; wherein, the horizontal segmentation point is used to define the boundary of each model segmentation layer. Based on the model segmentation layers and the horizontal segmentation points, multiple computation graphs corresponding to the model training task are determined. Based on each computation graph, the heterogeneous device cluster is statically partitioned to obtain multiple task execution units; wherein, the task execution units are composed of heterogeneous devices. Obtain the real-time configuration information of the current hardware cluster, and dynamically adjust the heterogeneous device composition within each task execution unit based on the real-time configuration information.
6. A laboratory alarm method according to claim 1, characterized in that, Based on the data interface mapping relationship, real-time parameters of the current experimental scenario are obtained in real time, and the real-time parameters are compared with the historical parameters corresponding to the real-time parameters to determine the mutation parameters of the current experimental scenario, specifically including: Based on the data interface mapping relationship, the real-time parameters of the current experimental scenario are obtained in real time, and the historical parameters corresponding to the real-time parameters are obtained in chronological order. Obtain the first difference between the real-time parameter and the adjacent historical parameter, and obtain multiple second differences between two adjacent historical parameters in sequence based on the time sequence; By comparing the multiple second differences, the historical range and direction of change of each historical parameter can be determined; The current range and direction of change of the real-time parameter are determined by comparing the first difference with the adjacent second difference. Based on the comparison between the current range of change and the historical range of change, as well as the current direction of change and the historical direction of change, the mutation parameters of the current experimental scenario are determined.
7. A laboratory alarm method according to claim 1, characterized in that, Based on the mutation parameters and the correlation diagram, the linked experimental scenarios corresponding to the mutation parameters are determined, specifically including: Based on the attribute information of the mutation parameter and the acquisition device corresponding to the mutation parameter, the direct influence range of the mutation parameter is determined, the activated physical devices located within the direct influence range are obtained, and the experimental process node corresponding to the activated physical devices in the current scene is determined. Based on the experimental process nodes and the correlation diagram, the direct linkage experimental scenarios corresponding to the mutation parameters are determined. Starting from the experimental process node, the connection path is traced along the relationship graph to determine the inverse connection path that has a reverse effect on the mutation parameter based on the relationship between each path in the connection path. The experimental scenario corresponding to the influence connection path is taken as the indirect linkage experimental scenario, and the direct linkage experimental scenario and the indirect linkage experimental scenario are taken as the linkage experimental scenario corresponding to the mutation parameter.
8. A laboratory alarm method according to claim 1, characterized in that, Obtain the trend information of the mutation parameters, and provide alarm prompts for the current experimental scenario and the linked experimental scenario based on the trend information, specifically including: The time series data of the mutation parameters are fitted to obtain the fitting curve corresponding to the mutation parameters; Based on the parameter and shape information of the fitted curve, the variation trend information of the mutation parameter is determined; wherein, the variation trend information includes: inflection point information, rate of change, and trend direction; If, based on the change trend information, it is determined that the time it takes for the mutation parameter to reach the preset parameter threshold is less than a preset time threshold, then the corresponding alarm mechanism is triggered. Based on the alarm mechanism, an audible and visual alarm is triggered for the current experimental scenario, and an alarm prompt is given for the linked experimental scenario.
9. A laboratory alarm device, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1-8.
10. A non-volatile storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are capable of performing the method described in any one of claims 1-8.