Multi-source electrostatic data integrated ESD intelligent monitoring system and method

Abstract

The invention relates to the technical field of electrostatic protection and industrial safety monitoring, and discloses an ESD intelligent monitoring system and method for multi-source electrostatic data integration, and the system comprises a hardware architecture and a data processing and decision-making unit. The acquisition layer acquires multi-source heterogeneous data, pre-processes the multi-source heterogeneous data, inputs the pre-processed multi-source heterogeneous data into a space-time correlation potential field model construction module, fuses static environment attributes and dynamic disturbance, and calculates an STCPF risk potential field in real time. The risk analysis module identifies a high-risk area through potential field topology analysis and performs early warning; and the regulation and control module solves an electrical safety stability function based on the early warning information, and generates a coordinated regulation and control instruction to close the steady state control loop. And the closed-loop self-calibration module confirms signal backtracking characteristics by using the discharge event and updates model parameters. According to the invention, spatial interpolation and dynamic weight modulation technologies are utilized, global continuous sensing, active adaptive regulation and control and continuous model optimization of electrostatic risks are realized, and monitoring blind areas are effectively eliminated.

Description

Technical Field

[0001] This invention relates to the field of electrostatic discharge (ESD) protection and industrial safety monitoring technology, specifically to an ESD intelligent monitoring system and method that integrates multi-source electrostatic data. Background Technology

[0002] With the development of precision electronic manufacturing processes, electrostatic discharge (ESD) has an increasingly significant impact on product yield and equipment reliability. Existing ESD monitoring technologies mainly employ the deployment of non-contact electrostatic voltage sensor networks at key workstations in the production workshop to assess environmental electrostatic risks by collecting voltage data at specific locations.

[0003] However, this traditional monitoring method is essentially a threshold-triggered mechanism based on discrete observation points. Because the physical distribution of the electrostatic field has spatial continuity, and the dynamic processes involved in manufacturing, such as personnel movement, logistics transportation, and equipment operation, cause real-time distortion of the electrostatic field, relying solely on a limited number of fixed sensor nodes cannot fully cover the complex three-dimensional production space. In areas where sensor monitoring coverage is insufficient, electrostatic accumulation cannot be detected in a timely manner by the existing system, leading to the failure to effectively identify potentially high-risk areas and directly causing missed detections.

[0004] Furthermore, existing monitoring systems primarily focus on data acquisition and over-limit alarm functions. When monitored values ​​exceed preset safety limits, the system typically only triggers audible and visual alarms or sends simple equipment shutdown signals, lacking the ability to close-loop control over the current complex environmental potential state. The system cannot automatically calculate and execute environmental parameter adjustment strategies based on the real-time monitored electrostatic field distribution trend; for example, it cannot coordinate the control of the operating parameters of ion neutralization equipment to smooth potential fluctuations, nor can it actively maintain the dynamic steady state of the electrical environment.

[0005] Meanwhile, existing monitoring models or algorithm parameters are typically set based on ideal environments during the initial system construction and remain static during subsequent operation. Faced with complex changes in the production environment, temperature and humidity fluctuations, and sensor aging, models with fixed parameters cannot adapt to long-term environmental drift. When an electrostatic discharge event actually occurs but the system fails to provide an early warning, existing technologies lack an effective backtracking mechanism to use the failure data to correct and optimize the model parameters in reverse. This results in the monitoring system failing to adaptively improve its monitoring accuracy as operating time increases. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an ESD intelligent monitoring system and method that integrates multi-source electrostatic data, solving the problems of existing electrostatic monitoring technologies relying on single-point threshold triggering, lacking spatial continuity analysis, and being unable to perform closed-loop adaptive control in response to dynamic environmental changes.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of the present invention provides an ESD intelligent monitoring system that integrates multi-source electrostatic data. The system includes a hardware architecture and a data processing and decision-making unit running on the hardware architecture.

[0008] The hardware architecture includes a data acquisition layer and a status display and control execution layer. The data acquisition layer is used to acquire multi-source heterogeneous data related to ESD risks; the status display and control execution layer is connected to the data processing and decision-making unit, and includes an environmental parameter adjustment interface unit, which is used to receive instructions and transmit them to the environmental adjustment equipment to close the steady-state control loop of the test environment.

[0009] The data processing and decision-making unit is connected to the data acquisition layer and is used to receive multi-source heterogeneous data, run core algorithms, and generate analysis results and calibration signals. This unit deploys and runs multiple logical modules, specifically including a data acquisition and preprocessing module, a spatiotemporal correlation potential field model construction and evolution module, a risk analysis and prediction early warning module, an ESD electrical environment dynamic control module, and a closed-loop self-calibration module.

[0010] The data acquisition and preprocessing module is responsible for receiving multi-source heterogeneous data and performing spatiotemporal alignment and standardization on the data to generate standardized spatiotemporal data, thereby eliminating the differences between data of different dimensions.

[0011] The spatiotemporal correlation potential field model construction and evolution module is the core computing unit of the system, used to receive standardized spatiotemporal data. This module integrates static physical environment attributes, continuous electrostatic field strength distribution, and dynamic process disturbances to calculate and update the STCPF risk potential field covering the target area in real time. Its innovative principle lies in using spatial interpolation technology to reconstruct discrete sensor data into a continuous physical field distribution, and combining it with a dynamic weight modulation mechanism to quantify and superimpose process disturbances such as personnel movement and equipment operation into the potential field model, thereby achieving spatial full-domain perception of electrostatic risks.

[0012] The risk analysis and prediction early warning module receives the STCPF risk potential field, performs topological structure analysis on the potential field to identify high-risk areas, predicts the spatiotemporal evolution trend of high-risk areas, and generates high-risk early warning information. This module characterizes the drasticness of potential changes by calculating the potential field gradient magnitude and sets critical conditions based on dielectric breakdown characteristics, thereby accurately locating potential discharge points.

[0013] The ESD (Electrical Safety and Security) dynamic control module is used to generate coordinated control commands to maintain environmental potential stability after receiving high-risk early warning information, based on a preset electrical safety and stability function. This module establishes a control device model, evaluates the ability of different devices to regulate environmental potential, and searches for the parameter set that optimizes the comprehensive evaluation index among multiple candidate schemes, thus realizing a shift from passive monitoring to active intervention.

[0014] The closed-loop self-calibration module receives discharge event confirmation signals, backtracks to extract high-risk event features, and then updates the model parameters in the spatiotemporal correlation potential field model construction and evolution module. This module constructs high-dimensional feature vectors to characterize the spatiotemporal statistical features of the discharge moment and uses the actual results to reverse-correct the model's weight parameters, enabling the system to have an adaptive capability that continuously optimizes as the running time increases.

[0015] In the above system, the data acquisition layer specifically includes a non-contact electrostatic voltage sensor network for acquiring intrinsic electrostatic data, an environmental sensor network for acquiring environmental impact data of temperature and humidity, an industrial IoT gateway for accessing the manufacturing execution system or equipment control system to obtain process disturbance data, and an ESD event confirmation sensor network for capturing electrostatic discharge event confirmation signals.

[0016] Furthermore, when processing numerical data, the data acquisition and preprocessing module employs a standardization method for dimensionless calculations, outputting the calculation results as standardized spatiotemporal data. The spatiotemporal correlated potential field model construction and evolution module, when constructing the potential field, on the one hand, calls upon the standardized spatiotemporal data, uses an interpolation algorithm to calculate standardized estimates and converts them into physical voltage values ​​to construct a continuous electrostatic field strength distribution; on the other hand, it uses a dynamic weighted modulation function to calculate the dynamic influence factors of the grid cells to quantify dynamic process disturbances; finally, it combines static physical environment attributes to perform multi-factor weighted summation, obtaining a dimensionless comprehensive risk index as the STCPF risk potential field.

[0017] When performing topology analysis, the risk analysis and prediction early warning module calculates the potential field gradient magnitude for each grid cell in the STCPF risk potential field using the central difference method. This magnitude is then compared with a gradient threshold set based on the breakdown voltage characteristics of air or ambient media to identify cells with high discharge risk. When predicting evolution trends, the module extracts features of high-risk regions to construct a risk state vector, and updates it over time and by measurement using a state prediction algorithm, outputting a risk state estimate for the next time step.

[0018] When performing control, the ESD electrical environment dynamic control module establishes a control equipment model containing neutralization efficiency definitions, and generates multiple candidate correction schemes in response to early warning information. By substituting into the electrical safety stability function for calculation and traversing the search, it extracts the set of equipment operating parameters corresponding to the optimal scheme.

[0019] When performing self-calibration, the closed-loop self-calibration module responds to the discharge event confirmation signal, traces back historical data, calculates the spatiotemporal statistical characteristics of the local region, and encapsulates them into a high-dimensional feature vector. When updating parameters, it constructs an objective function based on the deviation between this vector and the actual results, calculates the comprehensive gradient vector, and uses a gradient descent algorithm to iteratively correct the model parameters.

[0020] A second aspect of this invention provides an intelligent ESD monitoring method integrating multi-source electrostatic data. This method, based on the aforementioned system, includes the following steps: Step 1: Perform system initialization. Load the 3D mesh model of the target area, the preset electrostatic affinity coefficient values ​​of each mesh cell, and the location and parameter information of ESD intervention resources. Use the above information to construct the basic static environmental data.

[0021] Step 2: Collect and preprocess data. Periodically acquire multi-source heterogeneous data, and perform spatiotemporal alignment and standardization on the multi-source heterogeneous data to output standardized spatiotemporal data.

[0022] Step 3: Construct and update the potential field. Map the standardized spatiotemporal data onto the static environmental base data, and generate and update the STCPF risk potential field covering the target area at the current moment through co-kriging interpolation, dynamic weight modulation, and multi-factor fusion calculation.

[0023] Step 4: Analyze the potential field and issue an early warning. Perform gradient and divergence analysis on the STCPF risk potential field to identify high-risk areas and calculate the risk metric. If the risk metric is higher than the safety threshold, a high-risk early warning message is generated.

[0024] Step 5: Perform electrical environment control. Upon receiving a high-risk warning, based on the topological characteristics of the high-risk area and the status of the control equipment, solve for a preset electrical safety stability function to generate a coordinated control command. Then, convert the coordinated control command into a calibration signal and send it to the environmental parameter adjustment interface unit for execution.

[0025] Step six: Perform model closed-loop self-calibration. This step is independent of the main monitoring process. When a real electrostatic discharge event confirmation signal is detected, the high-risk feature signature corresponding to the electrostatic discharge event confirmation signal is traced back and extracted. Based on the high-risk feature signature, the model parameters required to construct the potential field are adjusted and optimized.

[0026] This invention provides an intelligent ESD monitoring system and method that integrates multi-source electrostatic data. It has the following beneficial effects: 1. This invention constructs a spatiotemporally correlated potential field model (STCPF) and utilizes a cooperative kriging interpolation algorithm and dynamic weight modulation technology to process multi-source heterogeneous data, solving the monitoring blind zone problem caused by existing technologies that rely solely on single-point threshold triggering from discrete sensors. This scheme can reconstruct a continuous electrostatic field strength distribution covering the entire area from limited observation point data and quantify the dynamic disturbances generated by personnel movement or equipment operation, thereby achieving spatial continuity analysis of electrostatic risks in the target area and improving the accuracy of identifying potentially high-risk areas in the gaps of sensor network coverage.

[0027] 2. This invention introduces an ESD electrical environment dynamic control module. By establishing a control device model and solving the electrical safety stability function, it solves the problem of traditional systems lacking closed-loop adaptive control for dynamic environmental changes. After generating high-risk warning information, the system can automatically calculate the coordinated control command to keep the environmental potential balanced based on the optimization algorithm, and drive the environmental control device to perform the operation. This realizes the transformation from passive monitoring and alarm to actively maintaining the steady state of the electrical environment, effectively reducing the probability of discharge caused by electrostatic accumulation.

[0028] 3. This invention solves the problem that static model parameters cannot adapt to the long-term evolution of complex production environments by using a closed-loop self-calibration mechanism and extracting high-risk feature signatures through discharge event confirmation signals. By constructing high-dimensional feature vectors and using gradient descent algorithm to perform inverse iterative correction of model parameters, the system can use actual abnormal events as truth feedback to continuously optimize the potential field construction parameters, giving the monitoring system the adaptive ability to continuously improve monitoring accuracy as operating time increases. Attached Figure Description

[0029] Figure 1 This is a functional structure diagram of an ESD intelligent monitoring system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an embodiment of the ESD intelligent monitoring method integrating multi-source electrostatic data according to an embodiment of the present invention.

[0030] The system comprises: 11. Non-contact electrostatic voltage sensor network; 12. Environmental sensor network; 13. Industrial IoT gateway; 14. ESD event confirmation sensor network; 20. Data processing and decision-making unit; 21. Data acquisition and preprocessing module; 22. Spatiotemporal correlation potential field model construction and evolution module; 23. Risk analysis and prediction early warning module; 24. ESD electrical environment dynamic control module; 25. Closed-loop self-calibration module; 31. Central monitoring and display terminal; 32. Local audible and visual alarm device; and 33. Environmental parameter adjustment interface unit. Detailed Implementation

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] See attached document Figure 1 The present invention provides an ESD intelligent monitoring system that integrates multi-source electrostatic data. The system includes a hardware architecture and a logic module running on the architecture.

[0033] The system's hardware architecture includes a data acquisition layer, a data processing and decision-making unit 20, and a status display and control execution layer.

[0034] The data acquisition layer is used to acquire raw data related to ESD risks. This layer specifically includes: a non-contact electrostatic voltage sensor network 11 for acquiring intrinsic electrostatic data; an environmental sensor network 12 for acquiring environmental impact data such as temperature and humidity; an industrial IoT gateway 13 for connecting to the manufacturing execution system and equipment control system to acquire process disturbance data; and an ESD event confirmation sensor network 14 for capturing confirmation signals of electrostatic discharge events.

[0035] The data processing and decision-making unit 20 is connected to the various networks and gateways of the data acquisition layer. This unit receives and processes various types of acquired data, runs core algorithms, and generates analysis results and control commands. The data processing and decision-making unit 20 can be a server, industrial control computer, or edge computing device.

[0036] The status display and control execution layer is connected to the data processing and decision-making unit 20. This layer includes a central monitoring and display terminal 31 for visually presenting the risk potential field and early warning information; a local audible and visual alarm device 32 for executing immediate alarms; and an environmental parameter adjustment interface unit 33 for converting the calibration signal generated by the decision-making unit into a standard level signal and transmitting it to the environmental adjustment equipment to close the steady-state control loop of the test environment.

[0037] The data processing and decision-making unit 20 has multiple logical modules deployed and running within it. These modules work together to implement the method of this invention. Specifically, they include: A data acquisition and preprocessing module 21 receives data from a non-contact electrostatic voltage sensor network 11, an environmental sensor network 12, and an industrial IoT gateway 13, and performs spatiotemporal alignment and standardization processing on the data.

[0038] A spatiotemporal correlated potential field (STCPF) model construction and evolution module 22, based on the data output by the data acquisition and preprocessing module 21, integrates static physical environment attributes, continuous electrostatic field strength distribution and dynamic process disturbances, and calculates and updates the STCPF risk potential field covering the target area in real time.

[0039] A risk analysis and prediction early warning module 23 performs topological structure analysis on the potential field generated by the STCPF model construction and evolution module 22, identifies high-risk areas, predicts their spatiotemporal evolution trend, and generates risk early warning information accordingly.

[0040] An ESD electrical environment dynamic control module 24, after receiving high-risk warning information generated by the risk analysis and prediction warning module 23, performs optimization solution according to the preset electrical safety stability function to generate a set of coordinated control instructions for maintaining the stability of the environmental potential, thereby ensuring the electrical safety test benchmark of the object under test.

[0041] A closed-loop self-calibration module 25 receives discharge event confirmation signals from the ESD event confirmation sensor network 14, and extracts high-risk event features based on the signals, thereby updating the model parameters in the STCPF model construction and evolution module 22.

[0042] See attached document Figure 2 This invention provides an intelligent ESD monitoring method that integrates multi-source electrostatic data, the method comprising the following steps: S100, System initialization: This step loads the three-dimensional mesh model of the target area, the preset electrostatic affinity coefficient values ​​of each mesh unit, and the location and parameter information of ESD intervention resources.

[0043] S200, Acquisition and Preprocessing of Multi-Source Heterogeneous Data: This step periodically acquires intrinsic electrostatic data, environmental impact data, and process disturbance data, and performs spatiotemporal alignment and standardization on the data.

[0044] S300, Construct and update the Spatiotemporal Correlation Potential Field (STCPF): This step generates and updates the global STCPF risk potential field covering the target area at the current moment based on the preprocessed data through co-kriging interpolation, dynamic weight modulation, and multi-factor fusion calculation.

[0045] S400, Analyze the STCPF risk potential field and make a risk assessment: This step performs gradient and divergence analysis on the current potential field to identify high-risk areas and calculates the risk metric. If the risk metric is lower than the preset safety threshold, the process returns to step S200; if the risk metric is higher than the safety threshold, an early warning message is generated and step S500 is executed.

[0046] S500, Decision and Execution of Electrical Environment Parameter Correction: This step is initiated after receiving the early warning information. Based on the topological characteristics of the high-risk area and the status of the control equipment, a preset electrical safety stability function is solved to obtain the optimal parameter correction strategy. Cooperative control commands are generated and sent to the control execution layer for execution to eliminate electrostatic field anomalies that cause electrical faults.

[0047] S600, Perform closed-loop self-calibration of the model: This step is independent of the main processes of S200 to S500 and is event-driven. When a real electrostatic discharge event confirmation signal is detected, the high-risk feature signature of the event is traced back and extracted, and the relevant parameters of the STCPF model are adjusted and optimized based on the signature.

[0048] To further clarify the present invention, the technical details involved in each of the above steps will be described in detail below.

[0049] In a specific embodiment of the present invention, the data acquisition and preprocessing module 21 is responsible for providing high-quality data input with a uniform format for subsequent potential field construction.

[0050] The data acquisition and preprocessing module 21 collects multi-source heterogeneous data. For intrinsic electrostatic data, environmental impact data, and ESD event confirmation data, data is collected by deploying corresponding non-contact electrostatic voltage sensor networks 11, environmental sensor networks 12, and ESD event confirmation sensor networks 14 within the target area. These sensor networks can connect to the data processing and decision-making unit 20 via wired or wireless methods such as Ethernet, Wi-Fi, or LoRaWAN, and transmit data using protocols such as MQTT or Modbus TCP. For process disturbance data, it is obtained by establishing an interface with the factory's Manufacturing Execution System (MES), Production Control System (PLC), or automated equipment scheduling system through the Industrial Internet of Things gateway 13. Specific implementation methods of the interface include: subscribing to relevant data nodes using the OPC Unified Architecture (OPCUA) protocol; obtaining information such as work order number, product ESD sensitivity level, and material batch from the production database by executing SQL queries; or obtaining real-time information such as the AGV's real-time coordinates, speed, and task status by calling the WebAPI interface.

[0051] The collected raw data undergoes spatiotemporal alignment and standardization processing by the data acquisition and preprocessing module 21. The specific steps are as follows: S201, Time Standardization. All data acquisition devices and data processing and decision-making units 20 within the system are synchronized with a unified time server via Network Time Protocol (NTP) to ensure that all data points are assigned consistent and high-precision timestamps, providing a reliable time standard for subsequent analysis.

[0052] S202, Perform spatial coordinate unification. Map the coordinate information of all data with location attributes to a preset unified three-dimensional grid coordinate system. In one implementation, for two-dimensional coordinates obtained from the AGV scheduling system... It can be converted into coordinates in a three-dimensional mesh coordinate system using a preset linear transformation matrix. In another implementation, the system can pre-build a lookup table (LUT) that stores the correspondence between two-dimensional coordinates and three-dimensional grid coordinates for a specific area. This table allows for rapid coordinate mapping. This step ensures that spatial data from different sources can be fused and calculated within the same coordinate system.

[0053] S203, Perform data standardization. To eliminate the interference of different physical dimensions on subsequent model calculations, the data acquisition and preprocessing module 21 standardizes numerical data such as electrostatic intrinsic data and environmental impact data, making them dimensionless values. This processing uses the Z-score standardization formula for calculation. The Z-score standardization formula is: ; in, The standardized value is a dimensionless value. Raw measurement value, i.e., the original reading collected by the sensor at a specific moment; The mean of the sample dataset; : Standard deviation of the sample dataset.

[0054] In one implementation, the sample dataset may consist of historical data within a fixed-size sliding time window, for example, data from before the current time. The data from each sampling point is used as a sample dataset. In another implementation, it can consist of all relevant measurement data within a specific production batch or work order cycle.

[0055] Through the above steps, the data output by the data acquisition and preprocessing module 21 has a unified time base, unified spatial coordinates, and dimensionless numerical characteristics, which can be directly used for the accurate construction of the subsequent STCPF risk potential field.

[0056] The STCPF model construction and evolution module 22 receives the data output by the data acquisition and preprocessing module 21 and performs static modeling of the physical environment and generation of continuous electrostatic field strength distribution.

[0057] This module first performs static modeling of the physical environment. In one implementation, this modeling includes the following steps: S301, Perform spatial meshing. Based on the Building Information Model (BIM) or Computer-Aided Design (CAD) drawings of the target area, discretize the three-dimensional physical space into a set of multiple cubic mesh units. Each grid cell The size is set according to the monitoring accuracy requirements.

[0058] S302, assign electrostatic affinity (EAC) values ​​to each mesh cell. Assign a static electrostatic affinity coefficient This coefficient has the physical unit of the reciprocal of the volt (1 / V) and is used to characterize the ability of an object or surface within a grid cell to accumulate static charge.

[0059] Specifically, the system pre-constructs a material-EAC mapping table based on surface resistivity. During assignment, the mesh elements are obtained. The surface resistivity parameters of the main internal materials are determined by referring to a table. The value of . The mapping logic is: for surface resistivity greater than . The insulating material imparts a high EAC value; for surfaces with resistivity at... to Electrostatic dissipative materials between these conditions are given a moderate EAC value; for materials with surface resistivity less than The conductive material is given an EAC value close to zero.

[0060] After completing the physical environment modeling, the STCPF model construction and evolution module 22 generates a continuous electrostatic field strength distribution using measured data. This process employs co-kriging interpolation, utilizing the correlation between electrostatic voltage and ambient temperature for estimation.

[0061] First, the target mesh element is calculated using the co-kriging voltage interpolation formula. center point The standardized electrostatic voltage estimate, and the co-Kriging voltage interpolation formula are as follows: ; in, Standardized estimate of continuous electrostatic voltage. It is a dimensionless value; The summation operator. It indicates that the expression following it is summed. Voltage sensor index; : The number of voltage sensors involved in the calculation; : Voltage sensor weighting coefficient; Standardized measured electrostatic voltage value. That is, the dimensionless voltage value output by the data acquisition and preprocessing module 21; : No. The spatial coordinates of a voltage sensor; Covariate sensor index; The number of temperature sensors involved in the calculation; Covariate sensor weighting coefficients; Standardized measured temperature value. That is, the dimensionless temperature value output by the data acquisition and preprocessing module 21; : No. The spatial coordinates of a temperature sensor; Spatial coordinates of the target mesh cell; : Time variable.

[0062] Weighting coefficient and The equations were derived by solving the co-Kriging linear equations, which are based on fitted models of the voltage variogram, temperature variogram, and their cross-variogram. The variogram model is selected from one or a combination of the spherical model, the exponential model, or the Gaussian model.

[0063] Finally, the standardized estimates are denormalized to obtain continuous electrostatic voltage values ​​with physical units. The final continuous electrostatic voltage value is calculated using the denormalization formula. The inverse standardization formula is: ; in, Continuous electrostatic voltage value. The unit is volts (V); The standardized continuous electrostatic voltage estimate obtained from the previous step; Standard deviation of the voltage sample dataset. This value is derived from the statistical results of the data acquisition and preprocessing module 21; : The mean of the voltage sample dataset. This value comes from the statistical results of the data acquisition and preprocessing module 21.

[0064] While generating a continuous electrostatic field strength distribution, the STCPF model construction and evolution module 22 processes process disturbance data from the data acquisition and preprocessing module 21 to generate a dynamic modulation weight field.

[0065] The STCPF model construction and evolution module 22 performs the calculation of dynamic modulation weights. The system receives a set of perturbation data. This set contains real-time state tuples for multiple disturbance sources. For each disturbance source in the dataset, the system calls the corresponding weight mapping function according to preset rules.

[0066] S303, Calculate the dynamic modulation weight values. The dynamic weight modulation function is used to calculate the grid cell values. At any moment The weights, and the dynamic weight modulation function are defined as follows: ; in, : Dynamic modulation weight value. This value is a dimensionless coefficient; : No. One grid cell; Time variable; Weight mapping function; : Process disturbance data set; : Weighted model parameter set; : Spatial coordinates of the target mesh cell.

[0067] In one specific implementation, when the process disturbs the data When a moving electrostatic interference source (such as an Automated Guided Vehicle, AGV) is present, the system uses a Gaussian distribution function to construct the weight field around the interference source. In this case, the weights are calculated using a dynamic weighted Gaussian function formula, which is: ; in, : Benchmark weight value; Weighted amplitude parameter. This parameter is determined by the physical properties and motion state of the interference source. Specifically, it is calculated using the following formula. : ; in, This is a preset material base coefficient (related to the AGV surface insulation material). The preset speed gain coefficient; This refers to the real-time moving speed of the AGV. Exponentiation operator; Euclidean norm operator. Represents coordinate points and The square of the Euclidean distance between them; Real-time spatial coordinates of the AGV; : Weighted influence range parameter. Represents the standard deviation of the Gaussian distribution, used to define the spatial extent of the area affected by the interference source. This parameter is independent of the statistical standard deviation in the aforementioned data standardization step. .

[0068] In another implementation, when the process disturbs the data When dealing with electrostatic-sensitive work orders within a specific area, the system employs a step function. Construct a weight field. The system identifies the workstation area corresponding to the work order and sets the weight values ​​of all grid cells within that area to a preset constant. (in ), grid cells outside the region maintain the baseline weight. .

[0069] S304, calculates the final STCPF risk potential field. This step integrates static physical properties, continuous field strength distribution, and dynamic process perturbations. The system performs calculations on each grid cell. The fusion calculation is performed using the STCPF risk potential field calculation formula, which is as follows: ; in, STCPF risk potential value. This value is a dimensionless comprehensive risk index; Static electrostatic affinity coefficient. The unit is the reciprocal of the volt. ; Continuous electrostatic voltage value. The unit is volts. ; Dynamically modulated weight values. Dimensionless; Multiplication operator.

[0070] Calculated using this formula, the physical dimensions are eliminated, and the generated... As a unified, dimensionless risk metric, it is transmitted to the risk analysis and prediction early warning module 23.

[0071] The risk analysis and prediction early warning module 23 receives STCPF risk potential field data from the STCPF model construction and evolution module 22, and identifies and predicts risks based on field theory analysis methods.

[0072] The risk analysis and prediction early warning module 23 first performs an analysis of the potential field topology, identifying high-risk areas by calculating the gradient characteristics of the potential field. The specific implementation steps are as follows: S401, calculate the potential field gradient magnitude. For each mesh element in the three-dimensional mesh space. The gradient vector of the STCPF potential field is calculated using the central difference method, and the magnitude of this gradient is obtained. The formula for calculating the magnitude of the potential field gradient is as follows: ; in, : Potential field gradient magnitude. This value characterizes the grid cell. The magnitude of the rate of change of the risk potential field space; STCPF risk potential value. For grid cells The center coordinates are given, and the parameters within the function represent the coordinate positions. : Mesh step size. Represents the 3D mesh size at which the mesh moves within a certain range. Dimensions of the unit along the axial direction; Square root operator.

[0073] After the calculation is completed, the calculated result will be... With the preset gradient threshold Compare. Gradient threshold The breakdown voltage characteristics are set based on the air or ambient medium. If the gradient magnitude of a grid cell exceeds this threshold, it is marked as a "high discharge risk cell".

[0074] Subsequently, spatial clustering is performed on all labeled cells. Specifically, Connected Component Labeling (CCL) is used to scan the 3D mesh space and, based on neighborhood connectivity rules, group adjacent "high-discharge-risk cells" into the same connected component. Each connected component is defined as an independent "high-risk region object." And calculate the arithmetic mean of the coordinates of all the grid cells contained in the object, which is used as the geometric center coordinates of the object. .

[0075] After identifying high-risk areas, the risk analysis and prediction early warning module 23 uses the Kalman filter algorithm to predict the movement trajectory of the risk areas.

[0076] S402, Construct the risk state vector. For each identified high-risk area object... Its motion state is defined as a six-dimensional vector. The risk state vector definition is adopted, which is: ; in, :time The risk state vector; Discrete-time step index. Corresponds to the time frame sequence of the system analysis; Spatial coordinates of the geometric center of the risk area. That is, the coordinates calculated in the preceding steps. The amount; : The velocity components of the risk area along each axis; : Transpose operator.

[0077] S403, perform state prediction. The state transition matrix is ​​used to estimate the state at the next time step. The Kalman state prediction formula is used for calculation. The Kalman state prediction formula is as follows: ; in, The prior estimate of the state at the next moment; : State transition matrix. This matrix is ​​constructed based on a uniform linear motion model, specifically as follows: ; in, Discrete time step and The time interval between; The posterior estimate of the current state; Matrix multiplication operator.

[0078] Through this prediction step, the system obtains the predicted location of high-risk areas at future times, and generates early warning information that includes the risk level and predicted trajectory.

[0079] After receiving the early warning information sent by the risk analysis and prediction early warning module 23, the ESD electrical environment dynamic control module 24 is responsible for formulating and executing the optimal environmental potential correction strategy. Through digital modeling and optimization solution technology, the ESD electrical environment dynamic control module 24 achieves fine control of electrical stress in the area to prevent electrical faults caused by electrostatic discharge.

[0080] The ESD dynamic control module 24 first performs digital modeling of the control equipment. The system maintains a database of equipment parameters containing all active ESD control devices in the factory (such as intelligent ion fans, humidifiers, and automatic grounding switches).

[0081] S501, Construct a model of the control equipment. For each physical control device... The system maps this to a logical object containing a set of attributes. The attribute set includes: the device's spatial coordinates. (These coordinates have been mapped to a unified 3D mesh coordinate system), device type, maximum power. Effective radius of action And its current operating status.

[0082] In addition, each device model is associated with a potential neutralization efficiency function. This is used to quantitatively describe the suppression rate of the electrostatic voltage of the surrounding spatial grid cells by the device when it is in the on state. The neutralization efficiency function formula is used to calculate the device's... In the target grid The rate of change of electric potential at the point is given by the neutralization efficiency function formula: ; in, Rate of change of electric potential. Indicates the device's... Set power Below, make the grid cells The value of the electrostatic voltage decrease per second (unit: V / s). The negative sign indicates the decrease in voltage; : Static electricity elimination efficiency coefficient of the equipment. This constant is determined by the equipment's factory specifications; The set operating power of the equipment; Exponentiation operator; Target mesh cell The center coordinates; Intervention equipment Installation coordinates; The effective radius of action of the equipment.

[0083] When receiving a request for a specific high-risk area When an early warning message is received, the ESD electrical environment dynamic control module 24 executes the optimal correction decision.

[0084] S502, Generate a set of candidate correction schemes. The system determines the risk level based on high-risk areas. Geometric center coordinates Filter out all devices covering the area in the device parameter database (i.e., And the selected intervention devices are those that are in a usable state. Based on the selected devices, the system generates multiple candidate intervention plans. Each candidate correction scheme A set of control parameter vectors is defined, including the IDs of the participating devices and the set operating power. and the planned duration of operation .

[0085] S503 is a system that solves for the optimal strategy based on virtual simulation. The system uses the STCPF model as its foundational environment to analyze each candidate correction scheme. Numerical simulation evaluation is performed. Specifically, the system uses a time-stepping method to simulate the evolution of the potential field after the intervention is applied. For each simulation time step... The voltage value within the region is updated using the potential field update formula, which is: ; in, Simulation time The grid voltage value. Initial time. Equal to the currently measured continuous electrostatic voltage value ; : Summation operator. Represents the summation scheme. Neutralizes the performance of all activated devices.

[0086] The system continues to perform the above iterative calculations until the STCPF potential field value of all grids in the high-risk area decreases below the preset safety threshold. The cumulative simulation time at this point is the expected potential recovery time of this scheme. .

[0087] Based on this, the comprehensive evaluation index of each candidate scheme is calculated using the electrical safety stability function, which is: ; in, Electrical safety and stability evaluation index (the smaller the value, the faster the system recovers to a safe state and the smaller the disturbance to the system). Candidate correction schemes; Sensitivity coefficient for risk duration; : Expected potential recovery time. This value is calculated from the aforementioned virtual simulation steps and characterizes the duration of the electrical risk. Equipment operating load factor; Summation operator; Device index; Equipment operating power; : Duration of equipment operation. This product represents the total energy injection into the electrical system during the control process.

[0088] S504 executes optimal control. The system searches the candidate solution set using an optimization algorithm (such as a genetic algorithm) to find the solution that maximizes the electrical safety stability evaluation index. Minimum solution After determining the optimal solution, the ESD electrical environment dynamic control module 24 encapsulates the corresponding environmental potential compensation parameters into a calibration signal and sends it to the test environment adjustment interface via the industrial bus, instructing the external system to reset and calibrate the environmental electrical parameters in order to maintain the zero-point stability of the monitoring system.

[0089] The closed-loop self-calibration module 25 achieves adaptive optimization of system model parameters by analyzing the deviation between actual ESD events and historical risk potential field prediction data. The closed-loop self-calibration module 25 uses real electrostatic discharge signals as a reference value to perform reverse correction on the calculation model of the preceding modules.

[0090] The closed-loop self-calibration module 25 first performs the extraction of high-risk signature features. When the ESD event confirmation sensor network 14 deployed in the field detects a real electrostatic discharge event, the sensor sends a message containing the event occurrence time to the closed-loop self-calibration module 25. Coordinates of the location where the event occurred and discharge intensity The trigger signal.

[0091] S601, Retrospective spatiotemporal data. After receiving the trigger signal, the closed-loop self-calibration module 25 accesses the historical database and retrieves the time window. STCPF potential field data within. System lock based on event coordinates. Centered on, with radius Extract the potential field distribution sequence within the local grid region.

[0092] S602, Constructing the Feature Signature Vector. To quantify the field strength morphological characteristics that caused this ESD event, the system calculates the spatiotemporal statistical features of this local region and encapsulates them into a high-dimensional feature vector. The feature signature vector is constructed using the following formula: ; in, Spatiotemporal feature signature vector; : Transpose operator; Local potential mean. Represents the arithmetic mean of the STCPF values ​​of all grid cells within a local region; : Local potential space variance. Used to characterize the non-uniformity of risk distribution, its calculation formula is: ; in, For local areas The total number of grid cells within; The summation operator adds up the terms that meet a given condition. : Set membership operator. Represents a grid cell. Belongs to a local region set ; Local maximum gradient magnitude. That is, the magnitude of the gradient within this region. The maximum value; : Rate of change of potential field over time. Used to capture voltage transient characteristics before an event occurs, calculated using the backward difference formula: ; in, : The rate of change of the potential field over time; : Potential field value at the time of the event and at the event coordinates; The system's time sampling interval; A sampling point before the event occurs. The potential field value at the moment before the event occurs; : Potential field increment term.

[0093] After feature extraction is complete, the closed-loop self-calibration module 25 performs inverse self-calibration of the model parameters. Its purpose is to adjust the variable parameter set in the STCPF model construction and evolution module 22. (Specifically including the weight amplitude parameter in the dynamic weighting function) and weight influence range parameter This allows the risk assessment value output by the model to approximate the actual physical laws.

[0094] S603, Construct the target loss function. The system compares the risk potential field value predicted by the model with the actual discharge intensity to construct a loss function that measures the prediction deviation. The calibration loss function formula is used, and the calibration loss function formula is as follows: ; in, Calibration loss value; The set of model parameters to be calibrated; Intensity mapping coefficient; Measured discharge intensity; Model predictions, representing values ​​based on the parameter set. Calculated data containing event coordinates The STCPF potential value at the grid cell; Regularization coefficient; Initial values ​​for parameters; :L2 norm operator.

[0095] S604, Perform parameter update. Use gradient descent algorithm to update the parameter set. Perform iterative adjustments to minimize the loss function. The calculation is performed using a parameter inverse update formula, which is as follows: ; in, : The updated model parameter set; The set of model parameters for the current iteration step; Learning rate; The gradient vector of the loss function; Indicates the parameter set Perform gradient calculations.

[0096] The gradient vector is formed by superimposing the gradient of the data fitting term and the gradient of the regularization term.

[0097] Specifically, based on the formula relationships defined in the preceding module: ; The gradient calculation path for the data fitting term is based on the chain rule: ; Combining the derivative of the regularization term, the final gradient vector calculation formula is: ; in, : Partial derivative operator. The system uses this to calculate the total gradient value, thereby determining the direction of parameter adjustment; STCPF risk potential field value; : Dynamically modulated weight values. The system uses these to calculate the total gradient value, thereby determining the direction of parameter adjustment.

[0098] The system calculates the total gradient value accordingly, thereby determining the direction of parameter adjustment.

[0099] Through the above steps, the system uses real ESD event data to correct the physical model parameters, enabling the STCPF field strength distribution model to continuously approximate the characteristics of the real physical environment.

Claims

1. A multi-source electrostatic data integration ESD intelligent monitoring system, characterized in that, include: Hardware architecture and the data processing and decision-making units running on that hardware architecture; The hardware architecture includes: The data acquisition layer is used to acquire multi-source heterogeneous data related to ESD risks; The status display and control execution layer is connected to the data processing and decision-making unit. The data processing and decision-making unit is connected to the data acquisition layer and is used to receive the multi-source heterogeneous data, run the core algorithm, and generate analysis results and calibration signals. The data processing and decision-making unit has multiple logical modules deployed and running, specifically including: The data acquisition and preprocessing module is used to receive the multi-source heterogeneous data and perform spatiotemporal alignment and standardization processing on the data to generate standardized spatiotemporal data. The spatiotemporal correlation potential field model construction and evolution module is used to receive the standardized spatiotemporal data, integrate static physical environment attributes, continuous electrostatic field strength distribution and dynamic process disturbances, and calculate and update the STCPF risk potential field covering the target area in real time. The risk analysis and prediction early warning module is used to receive the STCPF risk potential field, perform topological structure analysis on the STCPF risk potential field to identify high-risk areas, predict the spatiotemporal evolution trend of the high-risk areas, and generate high-risk early warning information. The ESD electrical environment dynamic control module is used to perform optimization calculations based on a preset electrical safety stability function after receiving the high-risk warning information, and generate a coordinated control command to maintain the stability of the environmental potential. The closed-loop self-calibration module is used to receive discharge event confirmation signals, backtrack to extract high-risk event features, and then update the model parameters in the spatiotemporal correlation potential field model construction and evolution module. The status display and control execution layer includes an environmental parameter adjustment interface unit, which receives the coordinated control command, encapsulates the coordinated control command into a calibration signal and converts it into a standard level signal, and transmits it to the environmental control equipment to close the steady-state control loop of the test environment.

2. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, The data acquisition layer specifically includes: Non-contact electrostatic voltage sensor network for collecting intrinsic electrostatic data; An environmental sensor network is used to collect data on the environmental effects of temperature and humidity. Industrial IoT gateways are used to connect to manufacturing execution systems or equipment control systems to acquire process disturbance data; An ESD event confirmation sensor network is used to capture confirmation signals for electrostatic discharge events; The aforementioned intrinsic electrostatic data, environmental impact data, process disturbance data, and confirmation signals of electrostatic discharge events together constitute the multi-source heterogeneous data.

3. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, The data acquisition and preprocessing module performs the following processing: Receive numerical data from the multi-source heterogeneous data; The numerical data were calculated in a dimensionless manner using the Z-score standardization formula. The output calculation results are used as the standardized spatiotemporal data.

4. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, The spatiotemporal correlation potential field model construction and evolution module performs the following processing: The standardized spatiotemporal data is called, and the standardized estimate is calculated using the co-kriging voltage interpolation formula. The standardized estimate is then converted into a physical voltage value using the inverse standardization formula. The continuous electrostatic field strength distribution is then constructed using the physical voltage value. The standardized spatiotemporal data is called, and the dynamic influence factor of the grid cell is calculated using the dynamic weight modulation function definition. The dynamic influence factor is then used to quantify the dynamic process disturbance. The static physical environment attributes are read, and combined with the continuous electrostatic field strength distribution and the dynamic process disturbance, the STCPF risk potential field calculation formula is used to perform multi-factor weighted summation. The calculated dimensionless comprehensive risk index is output as the STCPF risk potential field.

5. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, When performing topology analysis, the risk analysis and prediction early warning module: For each grid cell in the STCPF risk potential field, the potential field gradient magnitude is calculated using the central difference method and the potential field gradient magnitude calculation formula. The potential field gradient magnitude is numerically compared with a preset gradient threshold. The grid cells whose potential field gradient magnitude is greater than the gradient threshold are marked as high discharge risk cells, and the high discharge risk cells are used as the basis for generating the high risk warning information. The gradient threshold is a critical value set based on the breakdown voltage characteristics of air or the ambient medium.

6. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, The risk analysis and prediction early warning module, when predicting evolution trends: Extract the high-risk area features at the current moment, and construct a risk state vector representing the current state of the system using the risk state vector definition. The risk state vector is input into the state transition matrix, and the Kalman state prediction formula is used for time update and measurement update. Output the risk state estimate for the next time step, and use the risk state estimate as the basis for generating the high-risk early warning information.

7. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, The ESD electrical environment dynamic control module performs the following processing: A control device model is established, and the ability of the control device model to regulate the ambient potential is defined by the neutralization efficiency function formula; In response to the high-risk warning information, multiple candidate correction schemes are generated; Substitute each of the candidate correction schemes into the electrical safety stability function to calculate the comprehensive evaluation index, and then iterate through the search to find the optimal scheme that minimizes the comprehensive evaluation index. Extract the set of equipment operating parameters corresponding to the optimal solution, and encapsulate the set of equipment operating parameters into the coordinated control command.

8. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 1, characterized in that, The closed-loop self-calibration module performs the following processing: In response to the discharge event confirmation signal, historical data of the time and location surrounding the discharge event confirmation signal are retrieved. Calculate the spatiotemporal statistical characteristics of a local region; The spatiotemporal statistical features are encapsulated into high-dimensional feature vectors using a feature signature vector construction formula, and these high-dimensional feature vectors are then used for subsequent model calibration.

9. The ESD intelligent monitoring system integrating multi-source electrostatic data according to claim 8, characterized in that, When the closed-loop self-calibration module updates parameters: Based on the deviation between the high-dimensional feature vector and the actual discharge result, the objective function is constructed using the calibration loss function formula; Calculate the combined gradient vector, which is the superposition of the gradient of the data fitting term and the gradient of the regularization term; The model parameters are iteratively corrected using the gradient descent algorithm and the inverse parameter update formula. The corrected parameter values ​​are then updated into the model parameters.

10. A multi-source electrostatic data integration method for intelligent ESD monitoring, characterized in that, An ESD intelligent monitoring system integrating multi-source electrostatic data as described in any one of claims 1-9 includes the following steps: S1. Perform system initialization: Load the three-dimensional mesh model of the target area, the preset electrostatic affinity coefficient values ​​of each mesh unit, and the location and parameter information of ESD intervention resources, and use the above information to construct the basic data of the static environment; S2. Data acquisition and preprocessing: Periodically acquire multi-source heterogeneous data, perform spatiotemporal alignment and standardization processing on the multi-source heterogeneous data, and output standardized spatiotemporal data; S3. Construct and update the potential field: Map the standardized spatiotemporal data output in step S2 to the static environmental base data loaded in step S1, and generate and update the STCPF risk potential field covering the target area at the current time through collaborative kriging interpolation, dynamic weight modulation and multi-factor fusion calculation. S4. Analyze the potential field and issue an early warning: Perform gradient and divergence analysis on the STCPF risk potential field generated in step S3 to identify high-risk areas and calculate the risk metric. If the risk metric is higher than the safety threshold, generate a high-risk early warning message. S5. Perform electrical environment control: After receiving the high-risk warning information generated in step S4, solve the preset electrical safety stability function according to the topological characteristics of the high-risk area and the status of the control equipment to generate a coordinated control command, and convert the coordinated control command into a calibration signal and send it to the environmental parameter adjustment interface unit for execution. S6. Perform model closed-loop self-calibration: Step S6 is independent of the main process of steps S2 to S5. When a real electrostatic discharge event confirmation signal is detected, backtrack and extract the high-risk feature signature corresponding to the electrostatic discharge event confirmation signal, and adjust and optimize the model parameters used to construct the potential field in step S3 based on the high-risk feature signature.

Images