A clean room pollution source tracing method, system and device with adaptive grid dynamic reconstruction

Abstract

The application discloses a clean room pollution source tracing method, system and equipment of adaptive grid dynamic reconstruction, comprising: acquiring the spatial characteristic value of the clean room, calculating the reference flow field data, and constructing the initial pollution sample library; the clean room space is divided into a primary grid and a secondary grid; real-time acquisition of the concentration time series data of the sensor, calculation of the concentration change gradient information, generation of a tertiary encryption grid; based on the tertiary encryption grid, multiplexing the reference flow field data, locally reconstructing the pollutant transport equation, and extracting the enhanced feature vector; inputting the mapping model, outputting the probability distribution and confidence of the pollution source position; after the tracing is completed, the tertiary encryption grid is released, the optimization parameters of the adaptive threshold are updated, the grid dynamic reconstruction is realized through the spatial distribution unevenness feature triggering mechanism, the problems of insufficient local precision or calculation redundancy of fixed grid division are avoided, the confidence evaluation is provided to support the risk decision, and the timeliness and accuracy of the clean room sudden pollution tracing are improved.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and cleanroom technology, specifically to a method, system, and equipment for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction. Background Technology

[0002] With the rapid development of the semiconductor, biopharmaceutical, and precision manufacturing industries, the environmental safety of cleanrooms has become a critical factor in ensuring product quality and production safety. During cleanroom operation, if sudden contaminant leaks (such as chemical gas leaks or abnormal particle diffusion) cannot be located and traced in a timely manner, it may lead to the scrapping of entire batches of products or even safety accidents. Therefore, achieving rapid and accurate tracing of cleanroom contamination sources has significant engineering application value.

[0003] Existing indoor pollutant tracing technologies primarily rely on a combination of computational fluid dynamics (CFD) simulations and data-driven approaches. Mainstream technical routes typically include numerical inversion-based solution methods, active detection methods based on mobile robots, and matching methods based on pre-built case libraries. Among these, methods based on pre-built case libraries combined with machine learning have become a current research hotspot due to their better balance between timeliness and computational cost. These methods usually pre-build concentration field databases for different pollution source scenarios through CFD simulations, and during actual monitoring, use sensor data to match cases in the database to infer the location of the pollution source.

[0004] However, when applying the aforementioned existing traceability technologies to high-cleanliness cleanroom scenarios, the following significant limitations still exist: 1. Rigid spatial discretization strategies and unreasonable allocation of computational resources: Existing grid-based source tracing methods (such as the common two-level mesh generation technique) typically employ fixed spatial discretization strategies, meaning the grid density and distribution remain constant throughout the source tracing process. However, cleanroom contaminant leaks often exhibit the characteristics of "localized outbreaks followed by global cleanliness." Fixed mesh strategies either sacrifice local source tracing accuracy to maintain global computational efficiency, resulting in the inability to accurately locate minute leak points; or they refine the mesh globally to pursue accuracy, leading to a surge in computational load and failing to meet real-time requirements. There is a lack of a mechanism capable of dynamically adjusting the mesh density based on the contamination diffusion pattern.

[0005] 2. Poor flow field data reusability and high response latency: Traditional methods often fail to fully consider the relative stability of cleanroom airflow organization when constructing sample libraries or performing real-time matching. In the face of sudden situations, recalculating the entire flow field or loading a large database can lead to significant response delays. Existing technologies lack efficient computational strategies for reusing flow fields and reconstructing equations locally for specific regions.

[0006] 3. Lack of confidence assessment in source tracing results, resulting in insufficient decision support: Existing source tracing systems typically output definitive pollution source coordinates without considering uncertainties caused by sensor noise, model errors, and airflow disturbances. In actual cleanroom operation and maintenance, managers not only need to know "where the pollution source is," but also "the reliability of this conclusion," in order to assess whether to immediately shut down the system or strengthen monitoring. The lack of probabilistic output and confidence assessment mechanisms limits the engineering decision support capabilities of source tracing systems.

[0007] 4. Poor scene adaptability and limited threshold settings: Background concentration and noise levels vary greatly between cleanrooms of different cleanliness levels (such as ISO 3 and ISO 5). Existing technologies often use fixed triggering mechanisms or grid division parameters, making it difficult to adapt to different cleanliness levels and easily leading to false alarms or missed alarms in low-concentration environments.

[0008] Therefore, there is an urgent need to develop a cleanroom contamination source tracing method that can dynamically adjust grid density based on real-time monitoring data, reuse benchmark flow field data to reduce computational input, and provide confidence assessment, so as to solve the contradiction between accuracy and efficiency caused by static grid division in existing technologies and improve the level of intelligent cleanroom environmental safety management. Summary of the Invention

[0009] The purpose of this invention is to propose an adaptive grid dynamic reconstruction method, system, and device for tracing cleanroom contamination sources in order to solve the problems mentioned in the background art.

[0010] To achieve the aforementioned objectives, the first technical solution adopted by this invention is: an adaptive grid dynamic reconstruction method for tracing contamination sources in cleanrooms, comprising: S1. Obtain the spatial characteristic values ​​of the cleanroom, calculate the baseline flow field data, and construct the initial contamination sample library; S2, based on the initial contaminated sample library, the cleanroom space is divided into a primary grid and a secondary grid, and sensors are placed at the nodes of the secondary grid; S3, real-time acquisition of concentration time-series data from the sensor, and calculation of concentration change gradient information in each secondary grid region; S4, determine whether the concentration change gradient information is greater than the adaptive threshold. If it is greater than the adaptive threshold, dynamically generate a third-level densified grid in the corresponding second-level grid area. S5. Based on the three-level encrypted grid, reuse the reference flow field data, locally reconstruct the pollutant transport equation, and extract the enhanced feature vector; S6, input the enhanced feature vector into the mapping model, and output the probability distribution and confidence level of the pollution source location; S7. Based on the probability distribution and confidence level of the pollution source location, perform source tracing tasks. After the source tracing is completed, release the three-level encrypted grid and update the optimization parameters of the adaptive threshold.

[0011] Furthermore, in step S2, the division ratio between the primary and secondary grids is the initial configuration parameter.

[0012] Furthermore, in step S4, the density of the third-level encrypted grid is greater than that of the second-level grid, and the ratio of the side length of the third-level encrypted grid to that of the second-level grid is a dynamically configured parameter. The dynamically configured parameter is adjusted in real time according to the cleanliness level of the cleanroom or the diffusion rate of pollutants.

[0013] Furthermore, in step S3, the concentration change gradient information is calculated based on the rate of change of concentration over time and the spatial distribution curvature of the concentration; wherein, the spatial distribution curvature is used to characterize the degree of difference between the current grid node concentration and the average concentration of the surrounding adjacent grid nodes.

[0014] Further, in step S4, the setting logic of the adaptive threshold is to dynamically set the adaptive threshold θ=f(C0, σ, k) based on the background concentration change gradient of the cleanroom and the sensor noise level; where θ is the adaptive threshold, C0 is the background concentration level of the cleanroom, σ is the sensor noise level, k is the cleanroom cleanliness level coefficient; and the function f is a monotonically increasing function.

[0015] Furthermore, in step S5, the conditions for reusing the reference flow field data include: the total area of ​​the region where the three-level densified grid is dynamically generated is less than a preset proportion threshold of the total area of ​​the cleanroom; if it exceeds the preset proportion threshold, the global flow field data is recalculated.

[0016] Furthermore, step S6 also includes iterative processing, specifically including: Obtain the confidence score and compare it with a preset confidence threshold; If the confidence level is less than the preset confidence threshold, then steps S4 to S6 are triggered again on the basis of the current three-level encryption grid to generate a four-level encryption grid until the confidence level is greater than or equal to the preset confidence threshold.

[0017] Furthermore, in step S6, outputting the probability distribution and confidence level of the pollution source location specifically includes: Probability density cloud map generation: The probability sampling point set output by the mapping model is mapped to a three-level encrypted grid coordinate system; the probability sampling point set is spatially smoothed using a kernel density estimation algorithm or a Gaussian mixture model to construct a continuous probability density field; the continuous probability density field is reconstructed by bilinear interpolation based on the topological structure of the three-level encrypted grid; and rendered as a pseudo-color probability density cloud map. Spatial confidence interval extraction: Based on the prediction variance or posterior distribution of the mapping model, the highest posterior density region less than the preset confidence level is extracted, and the highest posterior density region is discretized into a spatial confidence polygon as the spatial confidence interval of the pollution source location. Spatial coupling degree calculation: Calculate the proportion of the spatial intersection area between the spatial confidence polygon and the CAD envelope of each device in the cleanroom equipment layout database to obtain the spatial coupling degree S of each device. i ; Risk scoring and classification: Risk scoring and classification combine equipment process sensitivity weights and pollutant time-series exposure decay factors to calculate risk scores and classify risk levels; Risk Equipment List Output: The output of the risk equipment list includes equipment number, risk level, and spatial coupling degree S. i The system generates a list of risky equipment with confidence coverage ratios and handling strategies; it then maps the probability density cloud map, spatial confidence polygon, and risky equipment list to the cleanroom BIM or digital twin model in real time to generate a visual decision dashboard.

[0018] To achieve the aforementioned objectives, the second technical solution adopted by this invention is: an adaptive grid dynamic reconstruction cleanroom contamination source tracing system, comprising: The basic calculation module is used to obtain the spatial characteristic values ​​of the cleanroom, pre-calculate the baseline flow field data, and construct the initial contamination sample library; The initial grid module is used to divide the cleanroom space into primary and secondary grids, and to place sensors on the secondary grid nodes. The gradient monitoring module is used to collect the concentration time-series data from the sensor in real time and calculate the concentration change gradient in each secondary grid region. The dynamic reconstruction module is used to dynamically generate a third-level densified grid in the corresponding second-level grid region when the concentration change gradient is greater than the adaptive threshold. The local recalculation module is used to reuse the reference flow field data based on the three-level encrypted grid, locally reconstruct the pollutant transport equation, and extract enhanced feature vectors. The probability output module is used to input the enhanced feature vector into the mapping model and output the probability distribution and confidence level of the pollution source location. The resource management module is used to release the three-level encrypted grid and update the optimization parameters of the adaptive threshold after the tracing task is completed.

[0019] To achieve the above-mentioned objectives, the third technical solution adopted by the present invention is: an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to realize a cleanroom pollution source tracing method with adaptive grid dynamic reconstruction.

[0020] Due to the application of the above-mentioned technical solution, the present invention has the following advantages compared with the prior art: Dynamic grid reconstruction is achieved by triggering a mechanism based on spatial distribution non-uniformity characteristics, avoiding local precision deficiencies or computational redundancy caused by fixed grid division. At the same time, confidence assessment is provided to support risk decision-making, significantly improving the timeliness and accuracy of tracing the source of sudden contamination in cleanrooms. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall process of an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the principle of adaptive mesh dynamic reconstruction in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the logic of encryption triggered by concentration change gradient in an embodiment of the present invention. Figure 4 This is a schematic diagram of the probability density cloud map and spatial confidence region according to an embodiment of the present invention; Figure 5 This is an architecture diagram of a cleanroom contamination source tracing system based on adaptive grid dynamic reconstruction, according to an embodiment of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or system that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or systems.

[0024] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] like Figures 1-3As shown, this embodiment of the invention provides a method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction, comprising: S1. Obtain the spatial characteristic values ​​of the cleanroom, calculate the baseline flow field data, and construct the initial contamination sample library; S2, based on the initial contaminated sample library, the cleanroom space is divided into a primary grid and a secondary grid, and sensors are placed at the nodes of the secondary grid; Specifically, the primary grid is the basic monitoring area with a coarser computational granularity, used for macroscopic flow field pre-calculation and routine low-frequency inspections; the secondary grid is the key monitoring area with a finer computational granularity, with sensors densely arranged on the secondary grid nodes for real-time concentration time-series data acquisition and gradient monitoring. The division ratio of the primary and secondary grids can be initially configured according to the cleanroom geometry and monitoring requirements.

[0026] S3, real-time acquisition of concentration time-series data from the sensor, and calculation of concentration change gradient information in each secondary grid region; S4, determine whether the concentration change gradient information is greater than the adaptive threshold. If it is greater than the adaptive threshold, dynamically generate a third-level densified grid in the corresponding second-level grid area. S5, based on a three-level densified grid, reuses benchmark flow field data, locally reconstructs the pollutant transport equation, and extracts enhanced feature vectors; S6 will enhance the feature vector input mapping model and output the probability distribution and confidence level of the pollution source location; S7 performs source tracing tasks based on the probability distribution and confidence level of the pollution source location. After the source tracing is completed, the three-level encrypted grid is released, and the optimized parameters of the adaptive threshold are updated.

[0027] According to an embodiment of the present invention, in step S2, the division ratio between the primary grid and the secondary grid is the initial configuration parameter.

[0028] According to an embodiment of the present invention, in step S4, the density of the third-level encrypted grid is greater than that of the second-level grid, and the ratio of the side length of the third-level encrypted grid to that of the second-level grid is a dynamic configuration parameter. The dynamic configuration parameter is adjusted in real time according to the cleanliness level of the cleanroom or the diffusion rate of pollutants.

[0029] Specifically, in step S4, the adaptive threshold is dynamically set based on the background concentration level and background noise level of the cleanroom to avoid false triggering in low-concentration environments. The adaptive threshold judgment logic includes: monitoring the instantaneous change rate of concentration time-series data and combining it with the spatial second-order difference characteristics of adjacent sensor data; when the comprehensive index exceeds the threshold, encryption is triggered.

[0030] Specifically, the ratio of the side length of the third-level encrypted grid to that of the second-level grid is a configurable parameter, and the number of encrypted grids is dynamically adjusted according to the cleanroom cleanliness level.

[0031] According to an embodiment of the present invention, in step S3, the concentration change gradient information is calculated based on the rate of change of concentration over time and the spatial distribution curvature of concentration; wherein, the spatial distribution curvature is used to characterize the degree of difference between the current grid node concentration and the average concentration of surrounding adjacent grid nodes.

[0032] Specifically, in step S3, the calculation of the concentration change gradient takes into account both the rate of change of concentration over time and the spatial second derivative of concentration, and is used to identify the frontier region of pollutant diffusion.

[0033] Specifically, in step S3, the formula for calculating the concentration change gradient information G is: ; The specific calculation methods for each parameter are as follows: (1) Calculation of the concentration-time change rate dC / dt: Approximate calculations can be performed using forward difference or central difference schemes: dC / dt≈[C(t)-C(t-Δt)] / Δt; Where C(t) is the concentration measurement value at the current time t, C(t-Δt) is the concentration measurement value at the previous time, and Δt is the sampling time interval.

[0034] (2) Calculation of spatial distribution curvature ∇²C: ∇²C is the spatial Laplacian operator for concentration, characterizing the spatial curvature of the concentration field and used to identify local extrema and diffusion fronts. It is calculated using a five-point difference scheme in a two-dimensional discrete grid. ∇²C≈[C(i+1,j)+ C(i-1,j)+C(i,j+1)+ C(i,j-1)-4C(i,j)] / Δx²; Where C(i,j) is the concentration value of the current grid node (i,j), C(i±1,j) and C(i,j±1) are the concentration values ​​of the four adjacent grid nodes, and Δx is the grid side length.

[0035] (3) Setting the weighting coefficients α and β: The weighting coefficients α and β have the dual functions of dimension adjustment and feature weight allocation, and they can be set in one of the following ways: Method 1: Coefficient setting based on dimensional balance of physical quantities; ; ; in, and To meet The weighting coefficient is denoted by D, which is the diffusion coefficient of pollutants in cleanroom air (dimensions L² / T). By introducing the diffusion coefficient D, the second term is made more efficient. The dimensions are changed from [C / L²] to [C / T], and the first term... The dimensions are kept consistent, thereby achieving the unification of physical dimensions.

[0036] Method 2: Coefficient setting based on feature normalization; ; ; Where C0 is the background concentration level of the cleanroom, Tc is the characteristic time (such as the cleanroom air exchange cycle or the characteristic diffusion time of pollutants), and Lc is the characteristic length (such as the typical spacing of sensors or the side length of the secondary grid). and To meet The weighting coefficients are then determined. Through this normalization process, both indicators are converted into dimensionless relative rates of change, allowing for direct weighted fusion.

[0037] Method 3: Coefficient setting based on empirical calibration; Based on the cleanroom cleanliness level, sensor density, and contaminant type, the optimal range of values ​​for α and β is determined through historical data regression analysis or numerical simulation experiments. For example, in an ISO Class 5 cleanroom, the typical range of values ​​for α is [0.3, 0.7], and the typical range of values ​​for β is [0.01, 0.1] m² / s.

[0038] (4) Physical meaning of concentration change gradient information G: The G value is a comprehensive monitoring index, and its physical meaning is as follows: When |dC / dt| is large, it indicates that the concentration in the region changes rapidly over time, which may be a source of pollutant release or a diffusion front. When |∇²C| is large, it indicates that the concentration in this region is significantly non-uniform in the spatial dimension, which may be a high gradient region near the pollution source; By weighted fusion of the two indicators, the G value can simultaneously capture the temporal dynamics and spatial distribution characteristics of pollutant diffusion, providing a comprehensive basis for the adaptive threshold judgment in step S4.

[0039] (5) Specific laboratory calculation method one example: In an example of an ISO 5 level electronic cleanroom, a certain gaseous molecular contaminant was present.

[0040] Sensor sampling interval Δt 10s Data acquisition time step Secondary grid side length Δx 0.5m Spatial Discrete Scale Background concentration level <![CDATA[C0]]> 20μg / m³ Background concentration of a certain gaseous molecular pollutant in a clean environment Pollutant diffusion coefficient D <![CDATA[2.0×10 -5 m / s]]> Typical diffusion coefficient of a certain gaseous molecular pollutant in air Time weighting coefficient <![CDATA[ω1]]> 0.6 Relative weights of time-varying terms Spatial weight coefficient <![CDATA[ω2]]> 0.4 Relative weights of space curvature terms

[0041] Weighting coefficient calculation (using method 1: physical dimensional balance method): α = ω1 = 0.6 (dimensionless); β=ω2 •D=0.4×2.0×10 -5 =8.0×10 -6 m² / s; The monitoring data of a certain grid node (i,j) at a certain moment is as follows: C(t)= 35μg / m³, C(t-Δt)= 28μg / m³, C(i+1,j)= 32μg / m³, C(i-1,j)= 31μg / m³, C(i,j+1)= 30μg / m³, C(i,j-1)= 33μg / m³; The calculation yields: dC / dt=[C(t)-C(t-Δt)] / Δt=(35-28) / 10=0.7μg / (m 3 ⋅s); ∇²C=(32+31+30+33-4*35) / 0.5²=(126-140) / 0.25=-56μg / m5; G = 0.6 × |0.7| + 8 × 10 -6 ×|-56| = 0.42+0.000448=0.420448; μg / (m 3 ⋅s); The G value will be compared with the adaptive threshold θ in step S4. If G > θ, then the three-level grid encryption of the area will be triggered.

[0042] According to an embodiment of the present invention, in step S4, the setting logic of the adaptive threshold is to dynamically set the adaptive threshold θ=f(C0, σ, k) based on the background concentration change gradient of the cleanroom and the sensor noise level; where θ is the adaptive threshold, C0 is the background concentration level of the cleanroom, σ is the sensor noise level, k is the cleanroom cleanliness level coefficient; and the function f is a monotonically increasing function.

[0043] According to an embodiment of the present invention, in step S5, the conditions for reusing the reference flow field data include: the total area of ​​the region where the three-level densified grid is dynamically generated is less than a preset ratio threshold of the total area of ​​the cleanroom; if the preset ratio threshold is exceeded, the global flow field data is recalculated.

[0044] Specifically, in step S5, when the area of ​​the dynamically generated three-level encrypted grid is less than the preset proportion of the total area of ​​the cleanroom, the reference flow field data in step S1 is directly reused, and only the pollutant transport equation is resolved.

[0045] According to an embodiment of the present invention, step S6 further includes iterative processing, specifically including: Obtain the confidence score and compare it with a preset confidence threshold; If the confidence level is less than the preset confidence threshold, then steps S4 to S6 are triggered again on the basis of the current three-level encryption grid to generate a four-level encryption grid until the confidence level is greater than or equal to the preset confidence threshold.

[0046] According to an embodiment of the present invention, step S6, outputting the probability distribution and confidence level of the pollution source location specifically includes: (1) Generation of probability density cloud map: The probability sampling point set output by the mapping model is mapped to the coordinate system of the three-level encrypted grid; the probability sampling point set is spatially smoothed by the kernel density estimation algorithm or Gaussian mixture model to construct a continuous probability density field; the continuous probability density field is reconstructed by bilinear interpolation based on the topology of the three-level encrypted grid; and rendered as a pseudo-color probability density cloud map. (2) Spatial confidence interval extraction: Based on the prediction variance or posterior distribution of the mapping model, extract the highest posterior density region (HPD) at a preset confidence level (e.g., 95%), and discretize the region into a spatial confidence polygon as the spatial confidence interval of the pollution source location; (3) Spatial coupling degree calculation: Calculate the proportion of the spatial intersection area between the spatial confidence polygon and the CAD envelope of each device in the cleanroom equipment layout database, and obtain the spatial coupling degree S of each device. i The calculation formula is: Where CI is the spatial confidence polygon, and Device i For the CAD envelope projection of the i-th device; (4) Risk scoring and classification: The risk scoring and classification combines the equipment process sensitivity weight and the pollutant time-series exposure decay factor to calculate the risk score and classify the risk level. Risk level score = S i × Equipment process sensitivity weight × Pollutant time-series exposure decay factor; Equipment process sensitivity weight , where T type This is a coefficient for equipment type, such as lithography equipment = 1.0, etching = 0.8, L level This is the process sensitivity level coefficient, such as ISO Class 3 = 1.0, ISO Class 4 = 0.8, ISO Class 5 = 0.6, decreasing with the cleanliness level. value The product value coefficient is normalized to [0,1], and α, β, and γ are weighting coefficients that satisfy α+β+γ=1; the pollutant time-series exposure decay factor F decay The formula is as follows: ; The physical meaning and dimensions of each parameter are explained below: Δt: Exposure duration, refers to the time interval from the moment t0 when the pollutant concentration first exceeds the warning threshold to the current moment t, i.e. Its dimension is [T], and its unit is usually seconds (s) or minutes (min).

[0047] λ: Comprehensive decay coefficient, dimensionless (like ( ), which represents the rate at which the concentration of pollutants decreases per unit time due to the combined effects of multiple physicochemical mechanisms.

[0048] Where: λ is the linear superposition of the coefficients of each sub-attenuation mechanism: ; The ventilation attenuation coefficient is positively correlated with the cleanroom air change rate (ACH). (ACH unit is times / hour); : Surface deposition / adsorption attenuation coefficient, which is related to pollutant particle size, surface material and electrostatic properties; The degradation attenuation coefficient due to chemical reaction is related to the chemical activity of pollutants and the ambient temperature and humidity. The turbulent dilution-diffusion attenuation coefficient is related to the uniformity of airflow organization and the intensity of local vortices. Specifically, ; Where: (1) ε is the turbulent diffusion coefficient, with dimensions of , in m² / s, characterizes the equivalent diffusion capacity caused by turbulent fluctuations.

[0049] Engineering estimation can be achieved using ,in U is the turbulent fluctuation velocity (I is the turbulence intensity, U is the turbulence intensity). mean (meaning average wind speed), L t The integral scale is for turbulence (usually 0.1-0.3 times the grid side length).

[0050] In a certain ISO Class 5 electronic cleanroom, the typical calibration value of ε under an ionized ammonia scenario is 4.5 × 10⁻⁶. -4 m² / s.

[0051] (2) L c The characteristic mixing length, with dimensions [L] and units of m, represents the characteristic spatial scale of pollutant concentration homogenization.

[0052] For the secondary grid region, L cThe grid side length Δx can be taken as 1 / 2; when considering the equipment blocking effect, a correction factor can be introduced. In a 10m×10m×3m cleanroom, L c The typical value is 0.25 m.

[0053] (3) Calculation example: λ dil =(4.5×10 -4 (m² / s) / (0.25m)² = 0.0072s -1 ; After considering the local vortex enhancement effect, we take λ. dil ≈0.010s -1 .

[0054] Dimensional verification: [m² / s] / [m²] = [s -1 It meets the dimensional requirements of the attenuation coefficient.

[0055] λ filter High efficiency filtration attenuation coefficient, which is related to the efficiency of HEPA / ULPA filters and the airflow velocity.

[0056] Specifically, λ chem_filter = η filter ·Q eff / V; Where: (1) η filter For filtration efficiency, Q eff V represents the effective circulating air volume, and V is the cleanroom volume. (2) In an ISO Class 5 electronic cleanroom, with ULPA + chemical composite filtration, the ammonia ion removal efficiency is ≥95%, and the effective circulating air volume dimension is [L]. 3 / T], such as the unit m 3 / s (3) Calculation example: In a certain ISO 5 cleanroom environment, Q eff =f recirc ⋅f chem ⋅Q total =0.90×0.60×25=13.5m 3 / s; C(t): The measured value of pollutant concentration at the current time t, with the dimension of [concentration], such as μg / m³; C peak The peak concentration observed in the monitored area during this pollution incident has the same dimensions as C(t). Ratio C(t) / C peak This is a dimensionless relative concentration index, with a value range of [0,1].

[0057] The time decay term is a dimensionless exponential function, representing the proportion of pollutants remaining after exposure time Δt under the influence of the comprehensive decay coefficient λ. When When this value approaches 0, it indicates that the pollutants have sufficiently decayed; when When the value approaches 1, it indicates that the pollutants have not yet significantly decreased.

[0058] F decay Physical meaning of F decay This is a dimensionless risk attenuation factor, with a value range of [0,1], used for time-series weighting of risk scores. When pollutants just exceed the limit ( When F decay ≈1, the risk score is not affected by decay; When pollutants have been continuously decaying for a long time ( When F decay ≪1, the risk score is significantly reduced, reflecting the engineering logic that the longer the exposure time and the lower the current concentration, the smaller the immediate risk.

[0059] Calibration method of λ: The total attenuation coefficient λ can be calibrated in any of the following ways:

Method 1

Method 2

Method 3

[0060] Typical value examples (molecular contaminants below 0.1 μm): In an ISO 5 level electronic cleanroom, molecular contaminants (primarily ionic ammonia NH4) were targeted. + (For example), the relevant parameter settings are as follows: Basic parameter settings: Types of pollutants - <![CDATA[Ammonium ion (NH4 + )]]> Typical molecular pollutants, exhibiting adsorption properties and weak reactivity. air exchange rate ACH 300 times / hour ISO Level 5 Typical Values Background concentration level <![CDATA[C0]]> 20μg / m³ Background concentration limits of ionized ammonia in clean environments Sensor sampling interval Δt 1s High-frequency monitoring Secondary grid side length Δx 0.5m Spatial Discrete Scale

[0061] Calculation of the overall attenuation coefficient λ (specifically for molecular pollutants): <![CDATA[λ vent (Ventilation) Clean air replacement dilution ACH / 3600 = 300 / 3600 <![CDATA[0.0833s -1 ]]> <![CDATA[λ ads (Surface adsorption) Physical / chemical adsorption of molecules on wall / equipment surfaces <![CDATA[λ ads =V d ▪ The A / V ratio is based on a cleanroom with a ceiling height of 3m, where the effective surface area to volume ratio is approximately 2.5m². -1 (Considering walls, ceilings, floors, and equipment surfaces) Regarding molecular contaminants (ionic ammonia), V d V is contributed by both physical adsorption and chemisorption, and can be estimated using the following model. d =1 / (1 / V diff +1 / V react Consider a value of 0.0012 m / s. <![CDATA[0.0030s -1 <!-- 9 -->]]> <![CDATA[λ react (Chemical reaction) <![CDATA[Neutralization reaction with acidic gases (such as HCl, HNO3) in the environment]]> <![CDATA[Based on second-order reaction kinetics, [H⁺] ≈ 10 -8 mol / m³, k react ≈ 2×10 5 m³ / (mol·s)]]> <![CDATA[0.0020s -1 ]]> <![CDATA[λ dil (Turbulent dilution) Concentration homogenization caused by local eddies CFD simulation of local mixing efficiency shows that the value taken for laminar flow cleanrooms is too low. <![CDATA[0.0100s -1 ]]> <![CDATA[λ chem_fi lter (Chemical filtration) Activated carbon / chemical filter screen for retaining molecular pollutants <![CDATA[η filter • Q / V ULPA+ chemical composite filter, with an ammonia ion removal efficiency ≥95% and a circulating air velocity of 0.3m / s. <![CDATA[0.0300s -1 ]]> λ (sum) Multi-mechanism coupling attenuation Linear superposition <![CDATA[0.1283≈0.128s -1 ]]>

[0062] Where, λ chem_filter = η filter ·Q eff / V; Cleanroom dimensions V 10m×10m×3m Volume V = 300 m³ air exchange rate ACH 300 times / hour ISO Level 5 Typical Values Total air volume <![CDATA[Q total ]]> 25m³ / s <![CDATA[Q total = 300×300 / 3600]]> Recirculated air ratio <![CDATA[f recirc ]]> 0.90 90% air is recycled Chemical filtration section ratio <![CDATA[f chem ]]> 0.60 60% of the recirculated air passes through a chemical filter. Filtration efficiency <![CDATA[η filter ]]> 0.95 Impregnated phosphate activated carbon with ammonium ions

[0063] ① Effective circulating air volume: ; ② Chemical filtration attenuation coefficient: ; Engineering correction: Considering factors such as the influence of filter media moisture and local airflow short-circuiting, a conservative coefficient of 0.7 is adopted. ; The monitoring parameters for a certain pollution incident are as follows: Exposure duration Δt = 45s (from the first time the limit was exceeded to the current assessment time); Peak concentration C peak =50μg / m³ (the maximum monitored value during the event in this area); The current concentration C(t) = 35 μg / m³ (real-time monitoring value at the current moment).

[0064] Pollutant time-series exposure decay factor F decay The calculation process is as follows: ①Time decay term: ; ②Relative concentration term: ; ③Time-series exposure decay factor: .

[0065] F decay = 0.00221 is a dimensionless risk attenuation factor, indicating that the immediate risk in this region has attenuated to 0.221% of the peak risk. Combined with the spatial coupling degree S... i Weight W for equipment process sensitivity process Risk score (Risk=S) can be calculated. i ×W process ×F decay And classify risk levels according to preset thresholds.

[0066] Risk scoring calculation example: Assume the parameters of a cleanroom process-sensitive equipment are as follows: Spatial Coupling S i = 0.45 (confidence polygon covers 45% of the device area); Equipment process sensitivity weight W process = 0.85 (etching equipment, highly sensitive); Pollutant time-series exposure decay factor F decay =0.00221, then the risk score is: Risk=S i ×W process×F decay =0.45×0.85×0.00221≈0.000845.

[0067] Based on risk level classification (reference threshold): Risk ≥ 0.1 High risk Shut down immediately and activate emergency response. 0.01≤Risk<0.1 Medium risk Strengthen monitoring and prepare to switch to backup equipment. Risk<0.01 Low risk Maintain production and record event logs.

[0068] Based on this calculated risk score, cleanroom maintenance personnel can receive decision support for "maintaining production and recording event logs".

[0069] (5) Output of risk equipment list: The output of the risk equipment list includes equipment number, risk level, and spatial coupling degree S. i The system generates a list of risky equipment with confidence coverage ratios and handling strategies; it then maps the probability density cloud map, spatial confidence polygon, and list of equipment under management to the cleanroom BIM or digital twin model in real time to generate a visual decision dashboard.

[0070] like Figure 4 As shown, the probability density cloud map presents the posterior probability distribution of pollution sources in two-dimensional space in pseudo-color: the color gradient from blue to red indicates that the probability density increases from low to high; three black dashed closed contour lines correspond to the highest posterior density (HPD) regions at 68%, 95%, and 99% confidence levels, respectively, with the innermost 95% CI region discretized to generate a spatial confidence polygon; red pentagrams mark the peak positions of probability density, serving as initial source tracing reference points. This cloud map can be mapped to the cleanroom BIM model in real time, supporting the dynamic generation of a list of risky equipment.

[0071] According to an embodiment of the present invention, in step S7, updating the optimization parameters of the adaptive threshold includes: recording the gradient data that triggered grid encryption and the finally confirmed pollution source location in this source tracing task; using the gradient data and the finally confirmed pollution source location as new samples to incrementally learn the mapping model or threshold setting logic.

[0072] In one specific embodiment of the present invention, this embodiment is applied to an ISO 5 level electronic cleanroom with an area of ​​100 square meters and a floor height of 3 meters.

[0073] Step S1: Baseline model construction; A cleanroom geometric model was established, and spatial characteristic values ​​(including the locations of air supply vents, return air vents, and temperature and humidity settings) were defined. The PISO algorithm was used to pre-calculate baseline flow field data. Several virtual pollution source locations and release intensities were defined, and the pollutant transport equation was calculated based on the baseline flow field to construct an initial pollution sample library.

[0074] The flow field calculation here is a one-time pre-calculation, which can be reused in subsequent steps.

[0075] Step S2: Initial mesh generation; The cleanroom space is divided into several primary grids (e.g., 1 meter on each side). Each primary grid is further divided into several secondary grids (e.g., 0.5 meters on each side). Sensors are placed at the central nodes or key locations of the secondary grids.

[0076] The ratio of primary to secondary grids is only in the initial state. Unlike the fixed limitations of existing technologies, this embodiment allows for adjustment according to the scenario.

[0077] Step S3: Real-time monitoring and gradient calculation; The system collects concentration data from each sensor in real time and calculates the concentration change gradient for each secondary grid region.

[0078] In this embodiment, the gradient calculation logic combines the rate of change in the time dimension with the difference value in the spatial dimension. Specifically, the system reads the concentration values ​​at the current moment and the previous moment, and combines them with the concentration difference between adjacent sensors to form a comprehensive and dynamically configurable gradient index.

[0079] Step S4: Dynamic triggering and mesh encryption; An adaptive threshold is set. When the concentration change gradient information in a certain region exceeds the adaptive threshold, the region is determined to be a suspected source region. Within the primary grid area corresponding to the suspected source region, the system dynamically generates a secondary-level densified grid. The density of the secondary-level grid is higher than that of the secondary-level grid (the secondary-level densified grid can be dynamically generated according to the cleanliness level).

[0080] The threshold setting logic is "background concentration + noise coefficient". The specific coefficients are calibrated through experiments, and the specific values ​​are not disclosed here to protect the core algorithm parameters.

[0081] Step S5: Local reconstruction and feature extraction; Since the airflow in a cleanroom is usually relatively stable, when the encrypted area is small (e.g., less than 30% of the entire area), the system directly calls the baseline flow field data in step S1 and resolves the contaminant transport equation only within the three-level encrypted grid.

[0082] Extract sensor enhancement feature vectors within the three-level grid, which contain concentration distribution features at higher spatial resolution.

[0083] Step S6: Probability output; The enhanced feature vectors are input into a machine learning model (such as a Bayesian neural network). The model outputs a probability distribution map of the pollution source location and a 95% confidence interval.

[0084] If the confidence level is lower than the preset value (e.g., 80%), the system can automatically trigger grid encryption in adjacent areas for iterative tracing.

[0085] Step S7: Resource recycling; Once the source tracing task is completed or the pollutant concentration returns to normal, the system releases the computing resources occupied by the level 3 encrypted grid and reverts to the level 2 grid state. Simultaneously, the gradient data triggered in this instance is recorded for optimizing the next adaptive threshold.

[0086] Example 2: The difference between this embodiment and Embodiment 1 is that it is applicable to ISO Class 3 cleanroom scenarios.

[0087] In this scenario, the background concentration is extremely low, and sensor noise has a significant impact. The adaptive threshold in step S4 is automatically increased to prevent false encryption due to noise fluctuations. Simultaneously, the flow field reuse decision ratio in step S5 is adjusted to 20% to ensure flow field accuracy at a higher precision.

[0088] like Figure 5 As shown, to achieve the above-mentioned objectives, the second technical solution adopted by this invention is: an adaptive grid dynamic reconstruction cleanroom contamination source tracing system, comprising: The basic calculation module is used to obtain the spatial characteristic values ​​of the cleanroom, pre-calculate the baseline flow field data, and construct the initial contamination sample library; The initial grid module is used to divide the cleanroom space into primary and secondary grids, and to place sensors on the secondary grid nodes. The gradient monitoring module is used to collect the concentration time-series data from the sensor in real time and calculate the concentration change gradient in each secondary grid region. The dynamic reconstruction module is used to dynamically generate a third-level densified grid within the corresponding second-level grid region when the concentration change gradient is greater than the adaptive threshold. The local recalculation module is used to reconstruct the pollutant transport equation locally and extract enhanced feature vectors based on a three-level densified grid, reuse the baseline flow field data, and extract the enhanced feature vectors. The probability output module is used to input the enhanced feature vector into the mapping model and output the probability distribution and confidence level of the pollution source location. The resource management module is used to release the three-level encrypted grid and update the optimized parameters of the adaptive threshold after the traceability task is completed.

[0089] According to an embodiment of the present invention, a visual interactive terminal is also included, which is used to display the probability density cloud map and the list of risky devices, and to provide a manual configuration interface for the confidence threshold.

[0090] To achieve the above-mentioned objectives, the third technical solution adopted by the present invention is: an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the program to realize a cleanroom pollution source tracing method for adaptive grid dynamic reconstruction.

[0091] In summary, this invention achieves dynamic grid reconstruction through a spatial distribution non-uniformity feature triggering mechanism, avoiding the problems of insufficient local accuracy or computational redundancy caused by fixed grid division. At the same time, it provides confidence assessment to support risk decision-making, significantly improving the timeliness and accuracy of tracing the source of sudden pollution in cleanrooms.

[0092] Those skilled in the art will understand that, for ease of explanation, the example is provided with one memory and one processor. In actual terminals or servers, multiple processors and memories may exist. Memory can also be referred to as storage medium or storage device, etc., and the embodiments of this application do not limit this.

[0093] It should be understood that in the embodiments of this application, the processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor may also be a general-purpose microprocessor, graphics processing unit (GPU), or one or more integrated circuits to execute relevant programs to achieve the functions required by the embodiments of this application.

[0094] The processor can also be an integrated circuit chip with signal processing capabilities. In implementation, each step of this application can be completed through integrated logic circuits in the processor hardware or instructions in software form. The aforementioned processor can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the functions required by the units included in the methods, systems, and storage media of the embodiments of this application.

[0095] It should also be understood that the memory mentioned in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache.

[0096] By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0097] The memory can also be a Compact Disc Read-Only Memory (CD-ROM) or other optical disc storage, optical disk storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. The memory can exist independently and be connected to the processor via a bus. The memory can also be integrated with the processor. The memory can store programs, and when the program stored in the memory is executed by the processor, the processor performs the various steps of the method determined in the above embodiments of this application.

[0098] It should be noted that when the processor is a general-purpose processor, DSP, ASIC, FPGA, or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, the memory (storage module) is integrated into the processor. It should be noted that the memory described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0099] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0100] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules within the processor. The software modules can reside in mature storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. Since this storage medium is located in memory, the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method; to avoid repetition, these will not be described in detail here.

[0101] Those skilled in the art will recognize that the various illustrative logical blocks (ILBs) and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented 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 implementations should not be considered beyond the scope of this application.

[0102] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer-programmed program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a processor, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a computer network, or other programmable device.

[0103] This embodiment also provides a computer-readable storage medium storing a computer program that causes a computer to execute in order to implement the above-described method based on multi-stage vortex and intelligent feedforward.

[0104] It should be noted that computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic) or wireless (e.g., infrared, wireless, microwave, etc.) means, or from one website, computer, server, or data center to a mobile phone processor via a wired means. A computer-readable storage medium can be any usable medium that a computer can access, or a data storage system such as a server or data center that integrates one or more usable media. Usable media can be magnetic media (e.g., floppy disks, hard disks), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives), etc.

[0105] Finally, it should be noted that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction, characterized in that, include: S1. Obtain the spatial characteristic values ​​of the cleanroom, calculate the baseline flow field data, and construct the initial contamination sample library; S2, based on the initial contaminated sample library, the cleanroom space is divided into a primary grid and a secondary grid, and sensors are placed at the nodes of the secondary grid; S3, real-time acquisition of concentration time-series data from the sensor, and calculation of concentration change gradient information in each secondary grid region; S4, determine whether the concentration change gradient information is greater than the adaptive threshold. If it is greater than the adaptive threshold, dynamically generate a third-level densified grid in the corresponding second-level grid area. S5. Based on the three-level encrypted grid, reuse the reference flow field data, locally reconstruct the pollutant transport equation, and extract the enhanced feature vector; S6, input the enhanced feature vector into the mapping model, and output the probability distribution and confidence level of the pollution source location; S7. Based on the probability distribution and confidence level of the pollution source location, perform source tracing tasks. After the source tracing is completed, release the three-level encrypted grid and update the optimization parameters of the adaptive threshold.

2. The method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction as described in claim 1, characterized in that, In step S2, the ratio of the primary grid to the secondary grid is the initial configuration parameter.

3. The method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction as described in claim 1, characterized in that, In step S4, the density of the third-level encrypted grid is greater than that of the second-level grid, and the ratio of the side length of the third-level encrypted grid to that of the second-level grid is a dynamic configuration parameter. The dynamic configuration parameter is adjusted in real time according to the cleanliness level of the cleanroom or the diffusion rate of pollutants.

4. The method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction as described in claim 3, characterized in that, In step S3, the concentration change gradient information is calculated based on the rate of change of concentration over time and the spatial distribution curvature of concentration; wherein, the spatial distribution curvature is used to characterize the degree of difference between the current grid node concentration and the average concentration of surrounding adjacent grid nodes.

5. A cleanroom contamination source tracing method based on adaptive grid dynamic reconstruction as described in SS, characterized in that, In step S4, the adaptive threshold setting logic is to dynamically set the adaptive threshold θ based on the background concentration change gradient of the cleanroom and the sensor noise level; θ=f(C0,σ,k), where C0 is the background concentration level of the cleanroom, σ is the sensor noise level, and k is the cleanroom cleanliness level coefficient; the function f is a monotonically increasing function.

6. The method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction as described in claim 1, characterized in that, In step S5, the conditions for reusing the reference flow field data include: the total area of ​​the region where the three-level densified grid is dynamically generated is less than a preset ratio threshold of the total area of ​​the cleanroom; if it exceeds the preset ratio threshold, the global flow field data is recalculated.

7. The method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction as described in claim 6, characterized in that, Step S6 further includes iterative processing, specifically including: Obtain the confidence score and compare it with a preset confidence threshold; If the confidence level is less than the preset confidence threshold, then steps S4 to S6 are triggered again on the basis of the current three-level encryption grid to generate a four-level encryption grid until the confidence level is greater than or equal to the preset confidence threshold.

8. The method for tracing cleanroom contamination sources using adaptive grid dynamic reconstruction as described in claim 1, characterized in that, In step S6, outputting the probability distribution and confidence level of the pollution source location specifically includes: Probability density cloud map generation: The probability sampling point set output by the mapping model is mapped to a three-level encrypted grid coordinate system; the probability sampling point set is spatially smoothed using a kernel density estimation algorithm or a Gaussian mixture model to construct a continuous probability density field; the continuous probability density field is reconstructed by bilinear interpolation based on the topological structure of the three-level encrypted grid; and rendered as a pseudo-color probability density cloud map. Spatial confidence interval extraction: Based on the prediction variance or posterior distribution of the mapping model, the highest posterior density region less than the preset confidence level is extracted, and the highest posterior density region is discretized into a spatial confidence polygon as the spatial confidence interval of the pollution source location. Spatial coupling degree calculation: Calculate the proportion of the spatial intersection area between the spatial confidence polygon and the CAD envelope of each device in the cleanroom equipment layout database to obtain the spatial coupling degree S of each device. i ; Risk scoring and classification: Risk scoring and classification combine equipment process sensitivity weights and contaminant time-series exposure decay factors to calculate risk scores and classify risk levels. Risk Equipment List Output: The output of the risk equipment list includes equipment number, risk level, and spatial coupling degree S. i The system generates a list of risky equipment with confidence coverage ratios and handling strategies; it then maps the probability density cloud map, spatial confidence polygon, and risky equipment list to the cleanroom BIM or digital twin model in real time to generate a visual decision dashboard.

9. A cleanroom contamination source tracing system with adaptive grid dynamic reconstruction, characterized in that, include: The basic calculation module is used to obtain the spatial characteristic values ​​of the cleanroom, pre-calculate the baseline flow field data, and construct the initial contamination sample library; The initial grid module is used to divide the cleanroom space into primary and secondary grids, and to place sensors on the secondary grid nodes. The gradient monitoring module is used to collect the concentration time-series data from the sensor in real time and calculate the concentration change gradient in each secondary grid region. The dynamic reconstruction module is used to dynamically generate a third-level densified grid in the corresponding second-level grid region when the concentration change gradient is greater than the adaptive threshold. The local recalculation module is used to reuse the reference flow field data based on the three-level encrypted grid, locally reconstruct the pollutant transport equation, and extract enhanced feature vectors. The probability output module is used to input the enhanced feature vector into the mapping model and output the probability distribution and confidence level of the pollution source location. The resource management module is used to release the three-level encrypted grid and update the optimization parameters of the adaptive threshold after the tracing task is completed.

10. 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 method as described in any one of claims 1 to 8.

Images