Intelligent measurement laboratory environment monitoring and evaluation method and system

Abstract

The invention relates to an intelligent measurement laboratory environment monitoring and evaluation method and system, and belongs to the technical field of intelligent laboratory monitoring, and the system comprises a physical entity layer which collects physical perception data streams in real time through an Internet of Things sensor array; and the virtual platform layer is used for constructing a five-dimensional virtual mapping model based on a digital twinning technology and comprises the steps that the structural layer performs data standardization preprocessing and multi-source information feature extraction, and the intelligent layer uses an improved SwinTransform model to generate a personnel behavior classification probability matrix and a device state recognition probability matrix. The improved model optimizes feature extraction through a deep convolution kernel, a GeLU activation function and a simplified regularization structure; and the decision-making layer is used for calculating a laboratory operation risk index based on the probability matrix by adopting a decision-making level fusion algorithm, and triggering graded early warning when the index exceeds a preset threshold value. According to the method, the problems of metering laboratory data island, environmental response lag and weak risk early warning are solved, the personnel behavior recognition accuracy is improved, and the model convergence speed is increased.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart laboratory monitoring, more particularly, to a smart metrology laboratory environment monitoring and evaluation method and system. BACKGROUND

[0002] The current metrology laboratory generally has systematic defects such as fragmented equipment data, backward environmental monitoring means, lack of personnel behavior supervision, and weak risk early warning mechanism: traditional detection instruments are mostly in an offline independent working state, resulting in scattered storage of key signals such as voltage and current, forming a data island; environmental parameters rely on manual recording, and the response to sudden abnormalities such as temperature and humidity mutations and gas leaks is lagging, and the sensor coverage is insufficient; the operation standardization of experimental personnel lacks real-time supervision, and it is difficult to distinguish individual behavior in a multi-person collaborative scene, and equipment shielding leads to image recognition failure; the existing early warning mechanism relies on static thresholds and cannot adapt to the dynamic needs of different experimental scenes, ignores the coupling risks of personnel behavior, equipment status and environmental parameters, and causes substantial damage when the alarm is triggered. Although deep learning technology has made breakthroughs in image recognition, its application in metrology laboratory scenarios still faces two major bottlenecks: multi-source heterogeneous data processing obstacles and real-time and precision contradictions, and existing digital twin technology also cannot meet the dual needs of depth of virtual-real interaction and risk decision accuracy.

[0003] The prior art such as Chinese patent No. CN119046647A discloses an intelligent risk control method for multi-source data fusion, which, although optimizes risk decision through high-dimensional tensor conversion and cost-sensitive model, still has problems such as heterogeneous data processing being difficult to adapt to the strong space-time coupling characteristics of metrology laboratories, and real-time and precision contradictions leading to the inability to meet the millisecond-level risk response needs. SUMMARY

[0004] To solve the above technical problems, the present application proposes a smart metrology laboratory environment monitoring and evaluation method and system.

[0005] The technical solution of the present application is as follows: The present application proposes a smart metrology laboratory environment monitoring and evaluation system, which comprises: A physical entity layer, which collects physical perception data streams in real time through an Internet of Things sensor array deployed in the metrology laboratory; A virtual platform layer, which constructs a metrology laboratory virtual mapping model based on digital twin technology, comprising: A structure layer, which performs data standardization preprocessing and multi-source information feature extraction on the collected physical perception data streams; An intelligent layer, which inputs the multi-source information features into an improved SwinTransformer model, and outputs a metrology laboratory personnel behavior classification probability matrix and a device state recognition probability matrix; The decision layer adopts a decision-level fusion algorithm to calculate a laboratory operation risk index based on the personnel behavior classification probability matrix and the device state recognition probability matrix, and triggers a warning when the laboratory operation risk index is greater than a preset threshold.

[0006] Preferably, the Internet of Things sensor array comprises: The metering device monitoring unit: real-time acquisition of voltage, current, power factor electrical parameter signals; The personnel behavior capture unit: RGB-D camera deployed in key areas of the laboratory; The environmental perception unit: temperature and humidity sensor, light sensor, and gas concentration detector.

[0007] Preferably, the personnel behavior capture unit eliminates occlusion interference through multiple frames of images.

[0008] Preferably, the virtual mapping model of the metrology laboratory is constructed based on digital twinning technology, and the virtual mapping model is a five-dimensional model: Physical object: physical entity data obtained through the Internet of Things sensor array; Twin data: iterative optimization data of physical objects, virtual models, and service systems; Virtual model: four-dimensional digital mapping model integrating geometry, physics, behavior, and rules; Service system: monitoring services provided based on physical objects and virtual models; Data connection: communication link for real-time data transmission between the above-mentioned dimensions.

[0009] Preferably, the improved SwinTransformer model adds a Large Kernel Block module before the window multi-head self-attention W-MSA module, the cross-window attention SW-MSA module, and the patch merging Patch Merging of the original SwinTransformer.

[0010] Preferably, the Large Kernel Block module optimizes feature extraction through a triple mechanism, including: Using a deep convolution kernel to replace traditional small-size convolution kernel stacking to expand the receptive field and achieve multi-level feature fusion through single-layer convolution; Using GeLU activation function instead of ReLU activation function to preserve negative value feature information; Only preserving the Layer Normalization layer after deep convolution, and simplifying the regularization structure.

[0011] Preferably, the decision-level fusion algorithm is based on the metrology laboratory personnel behavior classification probability matrix and the device state identification probability matrix, comprehensively calculates the risk values of all personnel behaviors and device states in the laboratory, and calculates the environmental risk value collected by the environment perception unit, dynamically weights and fuses the personnel behavior risk value, the device state risk value and the environmental risk value, and obtains the laboratory operation risk index.

[0012] In another aspect, the present application also provides a smart metrology laboratory environment monitoring and evaluation method, comprising the following steps: Collecting physical perception data streams through an Internet of Things sensor array; In the virtual platform layer, the structure layer is standardized for pretreatment and feature extraction; in the intelligent layer, the improved SwinTransformer model is used to generate personnel behavior and device state probability matrices; The decision layer calculates the laboratory operation risk index based on the probability matrix; when the risk index exceeds the threshold, a hierarchical early warning is triggered.

[0013] In another aspect, the present application also provides an electronic device having a computer program stored thereon, wherein the computer program is executed by a processor to implement a smart metrology laboratory environment monitoring and evaluation method according to any one of the embodiments of the present application.

[0014] In another aspect, the present application also provides a computer readable medium for storing one or more programs, wherein the one or more programs, when executed by one or more processors, cause the one or more processors to implement a smart metrology laboratory environment monitoring and evaluation method according to any one of the embodiments of the present application.

[0015] The present application has the following beneficial effects: Through the Internet of Things sensor array, equipment-environment-personnel full-element real-time monitoring is realized, the five-dimensional digital twin model breaks through the data island limit, and a laboratory operation situation panoramic map is formed, wherein the improved SwinTransformer model structure optimizes feature extraction by introducing a Large Kernel Block module, and significantly improves the recognition efficiency; The improved SwinTransformer model adopts a deep convolution kernel single layer to realize multi-level feature fusion, the GeLU activation function retains negative value features to enhance robustness, and the simplified Layer Normalization structure reduces the calculation load by 30%, so that the personnel behavior recognition accuracy is improved to 98.7%, the model convergence speed is accelerated by 40%, and the millisecond-level real-time response demand is met; A three-level dynamic early warning mechanism is constructed based on a multi-factor fusion algorithm: the weight is automatically increased when the environmental parameters mutate, the behavior weight is adaptively increased when the behavior is abnormal, the weight is decayed when the device is continuously abnormal, and the risk is quantified through the laboratory operation risk index, so that the early warning response time is shortened to 800 ms and the false alarm rate is reduced to below 5%. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 To improve the SwinTransformerBlock network structure diagram. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0018] It should be understood that the step numbers used herein are only for the convenience of description, and are not limited to the execution sequence of the steps.

[0019] It should be understood that the terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, unless otherwise clear from the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0020] The terms "include" and "contain" indicate the presence of the described features, whole, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, whole, steps, operations, elements, components and / or sets thereof.

[0021] The term "and / or" means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0022] Embodiment one: To make the purpose, technical solutions and advantages of the present application clearer, the following will combine specific embodiments of the present application, and refer to the accompanying Figure 1 The technical solutions of the present application are clearly and completely described.

[0023] To solve the problems in the prior art, the present embodiment provides a smart measurement laboratory environment monitoring and evaluation system, which comprises: A physical entity layer acquires physical perception data stream in real time through an Internet of Things sensor array deployed in a measurement laboratory; the Internet of Things sensor array comprises: Metering device monitoring unit: real-time acquisition of voltage, current, power factor electrical parameter signals; Personnel behavior capture unit: RGB-D camera deployed in key areas of the laboratory; the personnel behavior capture unit eliminates occlusion interference through multiple frames of images.

[0024] Environmental perception unit: temperature and humidity sensor, light sensor, and gas concentration detector.

[0025] Virtual platform layer: a virtual mapping model of the metrology laboratory is constructed based on digital twinning technology; the virtual mapping model is a five-dimensional model: Physical object: physical entity data obtained through an Internet of Things sensor array; Twin data: iterative optimization data of physical objects, virtual models, and service systems; Virtual model: a four-dimensional digital mapping model integrating geometry, physics, behavior, and rules; Service system: monitoring services provided based on physical objects and virtual models; Data connection: communication link for real-time data transmission between the above-mentioned dimensions.

[0026] In the structure layer of the virtual platform layer, the collected physical perception data stream is subjected to data standardization preprocessing and multi-source information feature extraction; the preprocessing includes data cleaning, normalization, and feature dimension reduction.

[0027] In the intelligent layer of the virtual platform layer, the multi-source information features are input into an improved SwinTransformer model, and a metrology laboratory personnel behavior classification probability matrix and a device state recognition probability matrix are output; the improved SwinTransformer model adds a Large Kernel Block module before the window multi-head self-attention W-MSA module, the cross-window attention SW-MSA module, and the patch merging Patch Merging of the original SwinTransformer. The specific optimization mechanisms include: A deep convolution kernel is used to replace the traditional small-size convolution kernel to stack and expand the receptive field; in this embodiment, the deep convolution kernel is a 13x13 large kernel, and multi-level feature fusion is achieved through single-layer convolution; A GeLU activation function is used to replace the ReLU activation function to retain negative value feature information; Only the Layer Normalization layer after the deep convolution is retained, and the regularization structure is simplified.

[0028] As Figure 1As shown, it is an improved SwinTransformerBlock network structure diagram. In the improved SwinTransformer model, after image processing, global information exchange is performed through the Large Kernel Block, and then information exchange between patches is performed in the Swin Transformer Block. The Large Kernel Block is added before the Swin Transformer Block of stage1, stage2 and stage3, which can enhance the representation ability of the feature map in the model before the network performs patching self-attention on the picture features. The activation function in the improved model does not use the most commonly used ReLU in CNN, but uses GeLU in SwinTransformer, and uses less regularization Normalization, only the Layer Normalization (LN) layer after the Depthwise Convolution is retained. The improved Swin Transformer model introduces the design of large kernel convolution, which performs multiple rounds of 13x13 large kernel convolution before the Shifted Window self-attention, increases the receptive field of the patch-based self-attention in the Swin Transformer, and improves the convergence speed and classification accuracy of the model without increasing the computational complexity.

[0029] The decision layer calculates the laboratory operation risk index by using a decision-level fusion algorithm based on the personnel behavior classification probability matrix and the device state recognition probability matrix. When the laboratory operation risk index is greater than a preset threshold, a warning is triggered.

[0030] The decision-level fusion algorithm is based on the metrology laboratory personnel behavior classification probability matrix and the device state recognition probability matrix, and comprehensively calculates the risk values of all personnel behaviors and device states in the laboratory, and calculates the environmental risk value collected by the environmental perception unit. The personnel behavior risk value, the device state risk value and the environmental risk value are dynamically weighted and fused to obtain the laboratory operation risk index.

[0031] The environmental risk value calculation formula is as follows:

[0032] In the formula: is the environmental risk value; is the temperature risk factor When the temperature T is in the range of 15℃ to 35℃, the temperature risk factor ; is the humidity risk factor, when the humidity H is higher than 85%, the humidity risk factor Take a fixed value of 0.5; when the humidity H is below 40%, the humidity risk factor... According to the formula Calculations show that when humidity H is between 40% and 85%, the humidity risk factor... The value is 0; The gas concentration risk factor primarily considers carbon monoxide and methane gases in the metrology laboratory. When the carbon monoxide concentration (C) exceeds 50 ppm, the gas risk factor is calculated according to the formula... Calculations show that for methane gas, when the concentration C exceeds 1000 ppm, the gas risk factor is calculated according to the formula... calculate; The light intensity risk factor is 0.5 if the light intensity L is less than 100 lux, otherwise it is 0.

[0033] The calculation process for the laboratory operation risk index is as follows: Suppose there are n data sources (such as personnel behavior, device status, and environmental parameters), and each source is divided into m categories (such as normal, warning, and danger).

[0034] The probability matrix for each data source is as follows: ; In the formula: To represent the probability that the i-th data source is classified as category j ( ).

[0035] Assign weight factors Give data source The final class probability is: ; In the formula: The table after weighted merging is judged as type The probability; weighting factor Based on the dynamic adjustment of the laboratory environment, in this embodiment, the weight of personnel behavior is 0.4, the weight of device status is 0.4, and the weight of environmental risk is 0.2. When the gas concentration change rate is >10 ppm / s, the environmental weight is automatically increased to 0.5. When the personnel behavior risk value is >0.7 for three consecutive preset periods, the behavior weight is increased by 0.1. When the device status is abnormal for more than 300 seconds, the device status weight is reduced by 0.15.

[0036] By combining the weighted probabilities of all data sources, a laboratory operation risk index is generated. In this embodiment, the risk index is set to 0.7 when it exceeds a preset threshold, triggering a tiered early warning.

[0037] Example 2: This embodiment provides a method for environmental monitoring and assessment of a smart metrology laboratory, including the following steps: Physical perception data stream is collected through an Internet of Things sensor array; multi-frame image is combined to eliminate occlusion interference; The structure layer is standardized for pretreatment and feature extraction in the virtual platform layer; the intelligent layer generates a personnel behavior and device state probability matrix through an improved SwinTransformer model; The decision layer calculates a laboratory operation risk index based on the probability matrix; and triggers a graded early warning when the risk index exceeds a threshold value.

[0038] The flow updates the environmental information in real time through a database, and compensates for errors of the preliminary identification result.

[0039] Embodiment three The embodiment provides an electronic device, which stores a computer program, and the computer program is executed by a processor to implement a smart measurement laboratory environment monitoring and evaluation method according to any one of the embodiments of the present application.

[0040] Embodiment four The embodiment provides a computer readable medium for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement a smart measurement laboratory environment monitoring and evaluation method according to any one of the embodiments of the present application.

[0041] In the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent the cases of A alone, A and B together, and B alone. Wherein A and B can be singular or plural. The character " / " generally represents an "or" relationship between the front and rear associated objects. "At least one of the following" and the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, wherein a, b, and c can be single or multiple.

[0042] Those of ordinary skill in the art can realize that the units and algorithm steps described in the embodiments disclosed herein can be realized by electronic hardware, computer software and a combination of electronic hardware and computer software. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0043] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the system, device and unit described above can refer to the corresponding processes in the foregoing method embodiments, and will not be described here.

[0044] In several embodiments provided in the present application, any function, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (Read-Only Memory; hereinafter referred to as: ROM), a random access memory (Random Access Memory; hereinafter referred to as: RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0045] The above is only an embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent flow transformation, or direct or indirect application in other related technical fields, is also included in the patent protection scope of the present application.

Claims

1. A smart metrology laboratory environmental monitoring and assessment system, characterized in that, include: The physical entity layer collects physical sensing data streams in real time through an IoT sensor array deployed within the metrology laboratory; The virtual platform layer, based on digital twin technology, constructs a virtual mapping model of the metrology laboratory, including: The structural layer performs data standardization preprocessing and multi-source information feature extraction on the collected physical sensing data stream; The intelligent layer inputs multi-source information features into the improved SwinTransformer model and outputs a probability matrix for classifying the behavior of metrology laboratory personnel and a probability matrix for recognizing the status of equipment. At the decision-making level, a decision-level fusion algorithm is used to calculate the laboratory operation risk index based on the probability matrix of personnel behavior classification and the probability matrix of device status recognition. When the laboratory operation risk index is greater than the preset threshold, an early warning is triggered.

2. The intelligent metrology laboratory environmental monitoring and assessment system according to claim 1, characterized in that: The IoT sensor array includes: Metering device monitoring unit: Real-time acquisition of electrical parameter signals such as voltage, current, and power factor; Personnel behavior capture unit: RGB-D cameras deployed in key areas of the laboratory; Environmental sensing unit: temperature and humidity sensor, light sensor and gas concentration detector.

3. The intelligent metrology laboratory environmental monitoring and assessment system according to claim 1, characterized in that: The human behavior capture unit eliminates occlusion interference through multiple frames of images.

4. The intelligent metrology laboratory environmental monitoring and assessment system according to claim 1, characterized in that: The virtual mapping model of the metrology laboratory constructed based on digital twin technology is a five-dimensional model: Physical objects: Physical entity data acquired through IoT sensor arrays; Twin data: Iterative optimization data of physical objects, virtual models, and service systems; Virtual model: A four-layer digital mapping model integrating geometry, physics, behavior, and rules; Service system: Monitoring services provided based on physical objects and virtual models; Data connection: The communication link that enables real-time data transmission between the above dimensions.

5. The intelligent metrology laboratory environmental monitoring and assessment system according to claim 1, characterized in that: The improved SwingTransformer model adds a Large Kernel Block module before the original SwingTransformer's window multi-head self-attention (W-MSA) module, cross-window attention (SW-MSA) module, and patch merging.

6. The intelligent metrology laboratory environmental monitoring and assessment system according to claim 5, characterized in that: The Large Kernel Block module optimizes feature extraction through a triple mechanism, including: A deep convolutional kernel is used to replace the traditional stacking of small-sized convolutional kernels to expand the receptive field, and multi-level feature fusion is achieved through single-layer convolution. Use the GeLU activation function instead of the ReLU activation function to preserve negative value feature information; Only the Layer Normalization layer after depthwise convolution is retained, simplifying the regularization structure.

7. The intelligent metrology laboratory environmental monitoring and assessment system according to claim 2, characterized in that: The decision-level fusion algorithm is based on the probability matrix of personnel behavior classification and the probability matrix of device status recognition in the metrology laboratory. It comprehensively calculates the risk value of all personnel behavior and device status in the laboratory, and calculates the environmental risk value collected by the environmental sensing unit. The personnel behavior risk value, device status risk value and environmental risk value are dynamically weighted and fused to obtain the laboratory operation risk index.

8. A method for environmental monitoring and assessment of a smart metrology laboratory, characterized in that, Includes the following steps: Collect physical sensing data streams through an IoT sensor array; In the virtual platform layer, standardization preprocessing and feature extraction are performed through the structural layer. In the intelligent layer, the probability matrix of personnel behavior and device state is generated by improving the SwinTransformer model; The decision-making body calculates the laboratory operation risk index based on the probability matrix; A tiered warning is triggered when the risk index exceeds the threshold.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the intelligent metrology laboratory environmental monitoring and evaluation method as described in claim 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent metrology laboratory environmental monitoring and evaluation method as described in claim 8.

Images