Reinforced computer dust particle prediction and adaptive dust prevention method and system

Abstract

The present application relates to the field of reinforced computer protection technology, in particular to a reinforced computer dust particle prediction and self-adaptive dust prevention method and system. The method comprises: constructing an air duct airflow state boundary model; estimating the motion path of dust particles of different particle sizes based on the model, performing collision detection and stagnation trajectory analysis, and generating a deposition probability distribution structure; extracting a high deposition risk area, constructing an electrostatic flow guide field regulation structure, and calculating flow guide electrode control parameters; generating electrode instructions to drive the flow guide control module to deflect dust particles; configuring a dust collection area according to the dust particle deflection path and performing particle adsorption; combining residual dust particle monitoring data and control parameters to generate a feedback control vector, and realizing self-adaptive update of electrostatic regulation. The present application can accurately predict the behavior of dust particles and dynamically regulate their deposition trend based on CFD and differential model without the need for a sealed structure, realizing high-reliability active dust prevention control in complex operating environments.

Description

Technical Field

[0001] This invention relates to the field of ruggedized computer protection technology, and in particular to a method and system for predicting and adaptively preventing dust particles in ruggedized computers. Background Technology

[0002] This invention relates to the field of ruggedized computer protection technology. With the widespread deployment of high-performance computing devices, their sensitivity to particle deposition in enclosed air ducts, heat dissipation efficiency of air-cooled systems, and long-term stability has significantly increased. Dust particles easily deposit on the walls of air ducts, circuit components, and heat dissipation modules during equipment operation, causing increased thermal resistance, electrical short circuits, and performance degradation. Traditional dust prevention methods mostly rely on physical filters, periodic cleaning, or passive airflow design, which cannot identify dust particle movement trends in real time and are difficult to implement active deflection control for high-risk areas.

[0003] While some existing particle prediction models are based on fluid dynamics simulations, they mostly remain at the level of macroscopic estimation or two-dimensional path trajectories, lacking a hierarchical modeling mechanism for the microscopic behavior of dust particles at different sizes. They also lack comprehensive evaluation methods for particle-boundary interactions, stagnation probability, and spatial deposition tendency. Furthermore, although electrostatic diversion technology can be used for dust deflection, it has not yet formed a real-time feedback linkage mechanism with the particle distribution prediction module. This makes it difficult for the guidance and control strategy to match the actual particle distribution situation, affecting the response efficiency and coverage accuracy of the dust suppression system.

[0004] Therefore, there is an urgent need for a comprehensive particle behavior prediction method that integrates stable airflow boundary modeling, particle size stratification path estimation, collision detection, and deposition probability assessment mechanisms. This method, combined with an electrostatic flow control unit, can be used to construct an adaptive dust control system that can dynamically identify high-risk deposition areas, accurately drive deflection paths, and achieve closed-loop feedback regulation through residual particle monitoring. This will solve the core problems of inaccurate particle prediction, lagging control strategies, and lack of closed-loop regulation in existing technologies. Summary of the Invention

[0005] This invention provides a robust computer-aided dust particle prediction and adaptive dust control method and system to address the problem of how to generate high-precision dust particle distribution prediction results and drive an electrostatic deflection system to complete adaptive closed-loop dust control based on stable airflow boundary modeling and particle size stratification path estimation, fusion of collision detection and path deposition probability functions.

[0006] To address the aforementioned technical problems, this invention provides a method for ruggedized computer dust particle prediction and adaptive dust prevention, comprising:

[0007] Raw wind speed, temperature, and pressure data are collected using a miniature multifunctional environmental sensor array. The collected raw data undergoes data field merging and timestamp alignment to construct a stable set of airflow parameters and generate an airflow state boundary model. The sensor array includes a thermistor wind speed probe, a thermocouple temperature acquisition unit, and a piezoresistive miniature pressure sensor. Wind speed data is recorded in a three-dimensional vector format, temperature data is stored as absolute Celsius values, and pressure data is recorded in Pascals. The airflow state boundary model includes the mapping relationship between the three-dimensional structure of the air duct and the grid points of the control volume, the airflow vector assignment structure of the boundary input surface, a temperature-pressure coupling mapping table, and a pressure gradient map.

[0008] The dust particle motion path is estimated based on the airflow state boundary model, collision detection and stagnation trajectory calibration are performed, candidate deposition paths are generated, and the deposition probability distribution structure is calculated.

[0009] The expression for the deposition probability distribution is:

[0010]

[0011] in, Represents the spatial volume element V m The probability of dust particle deposition within; Z is the normalization factor; N c Indicates the total number of candidate paths; X is the predicted deposition probability value for the j-th path; j This represents the path of the j-th dust particle. It is an indicator function;

[0012] Obtain the deposition probability distribution structure, extract high-risk deposition areas and generate electrode control areas, construct the electrostatic conduction field control structure, and calculate the control parameter matrix of the conduction electrode.

[0013] The control parameter matrix of the flow guiding electrode is obtained, the control parameter matrix is ​​converted into electrode commands and driven to the flow guiding control module to execute the dust particle deflection action and generate dust particle deflection path data.

[0014] The dust collection area is configured based on the dust particle deflection path data, and spatial aggregation and particle adsorption operations are performed to generate residual dust particle monitoring data.

[0015] By combining residual dust particle monitoring data with historical control parameters, a feedback control vector is generated, and the control parameters are updated to complete adaptive adjustment.

[0016] Furthermore, the construction of the stable airflow parameter set includes:

[0017] The system acquires wind speed, temperature, and pressure data within the duct, performs structured analysis and format unification processing, and generates duct airflow characteristic data.

[0018] Normalize and remove disturbances from the airflow characteristic data of the duct to generate a stable set of airflow parameters;

[0019] An airflow state boundary model is constructed based on a set of stable airflow parameters.

[0020] Furthermore, the estimated dust particle movement path includes:

[0021] Obtain the airflow state boundary model, set parameters based on the particle size range, and perform particle motion path estimation.

[0022] Furthermore, the collision detection and stationary trajectory calibration process includes:

[0023] Collision detection and stagnation trajectory calibration are performed on the dust particle movement path to generate candidate deposition paths.

[0024] Furthermore, the calculated deposition probability distribution structure includes:

[0025] The deposition probability distribution structure within the wind tunnel area is calculated based on candidate deposition paths.

[0026] Furthermore, the construction of the electrostatic current-conducting field control structure includes:

[0027] Obtain the deposition probability distribution structure, extract high-risk deposition areas, and construct the spatial control region structure;

[0028] Electric field boundary segmentation and target offset direction planning are performed on the structure of the spatial control area to generate the electrode control area;

[0029] Based on the electrostatic current-conducting field control structure generated in the electrode control region, the control parameter matrix of the current-conducting electrode is calculated.

[0030] Furthermore, the dust particle deflection action includes:

[0031] Obtain the control parameter matrix of the current guiding electrode, and perform drive signal conversion and electrode command generation processing;

[0032] The electrode command is sent to the flow control module to execute the programmable flow electrode control operation;

[0033] The dust particle deflection action is executed according to the programmable flow guide electrode control operation, and dust particle deflection path data is generated.

[0034] Furthermore, the generated residual dust particle monitoring data includes:

[0035] Acquire dust particle deflection path data, configure dust collection target area, and schedule dust collection module;

[0036] The spatial aggregation and particle adsorption operations of the dust collection module are executed to generate dust collection control result data.

[0037] The dust collection control results data are processed to identify and count residual particles, generating residual dust particle monitoring data.

[0038] Furthermore, the adaptive adjustment includes:

[0039] Acquire residual dust particle monitoring data, and generate a feedback control vector by combining the deposition probability distribution structure and the control parameter matrix;

[0040] Update the control parameters in the electrostatic current conduction field control structure according to the feedback control vector;

[0041] The updated control parameters are re-injected into the electrode instruction generation process to complete the adaptive parameter adjustment.

[0042] A ruggedized computer dust particle prediction and adaptive dust prevention system, applied to any of the aforementioned ruggedized computer dust particle prediction and adaptive dust prevention methods, includes:

[0043] The airflow parameter acquisition module is used to acquire wind speed, temperature and pressure data and construct an airflow state boundary model;

[0044] The dust particle path prediction module is used to estimate the dust particle movement path based on the airflow state boundary model and generate a deposition probability distribution structure.

[0045] The electrostatic flow field generation module is used to obtain the deposition probability distribution structure, generate the electrostatic flow field control structure, and calculate the control parameter matrix.

[0046] The flow guide electrode execution module is used to convert the control parameter matrix into flow guide electrode commands and execute the dust particle deflection operation;

[0047] The dust collection and residue monitoring module is used to configure the dust collection area based on dust particle deflection path data and collect residual dust particle monitoring data.

[0048] An adaptive update module is used to generate a feedback control vector based on the residual dust particle monitoring data and the control parameter matrix, and to complete the control parameter update process.

[0049] The key innovations of this invention include:

[0050] (1) CFD-based particle size stratification motion path modeling and Reynolds number coupled trajectory derivation mechanism. It realizes the dynamic correlation between particle velocity, position, drag and airflow velocity, and introduces "path-level differential estimation and dynamic scoring" closed-loop modeling in dust prevention of air-cooled systems, which has high accuracy and real-time performance.

[0051] (2) Constructing the deposition probability function and three-dimensional mapping structure of spatial candidate paths. An explicit regional statistical mechanism is introduced to form a voxel-level probability heatmap, which overcomes the lack of predictive ability in traditional dust control systems and provides a precise positioning basis for deflection strategies.

[0052] (3) Automatic generation of electrode control area and real-time calculation of electric field deflection direction. The system generates spatial electrode control area and electrostatic conduction parameter matrix through S310–S330, realizing precise guidance of electric field before air inlet and effectively avoiding the problem of "deflection misalignment".

[0053] (4) Construction and execution mechanism of multi-level feedback adaptive control path. The system introduces a mechanism for comparing residual dust particle monitoring results with control parameters, which can adjust the deflection intensity and spatial control area coverage in reverse according to the residual dust particle distribution to achieve periodic self-adjustment.

[0054] The following are its main beneficial effects:

[0055] (1) This invention constructs a stable airflow state boundary model in step S100 and introduces a Lagrange path differential model based on Reynolds number and Stokes drag in steps S210–S230 to achieve a fine estimation of the spatial path of dust particles of different sizes. Furthermore, by integrating collision detection and stagnation assessment mechanisms with spatial deposition probability mapping functions, it can accurately identify dust deposition hotspots in the duct. Compared with traditional particle distribution methods based on empirical rules or macroscopic estimation, its prediction error is lower.

[0056] (2) An electrode control region is constructed in S300–S330, and an electrostatic current conduction control parameter matrix is ​​generated. Combined with the instruction conversion and programmable current conduction execution mechanism in S400, the deflection direction and voltage intensity can be automatically adjusted according to the spatial distribution changes of dust particles. Furthermore, through the feedback path in S600, the residual dust particle monitoring results are fed back to the control matrix generation module to form a closed-loop adaptive adjustment path, which effectively improves the continuous accuracy and real-time performance of the guidance and control strategy in complex environments.

[0057] (3) Traditional dust control solutions typically rely on mechanical blocking structures such as fixed filters, which are prone to clogging and thus affect the system's heat dissipation performance. This invention introduces an adjustable electrostatic flow guiding system, which dynamically applies an electric field based on dust particle path prediction results to achieve "path-guided" dust particle deflection control. An electric field barrier is formed in advance in areas with significant deposition trends, causing dust particles to deviate from the critical heat-sensitive element area and accumulate in the detachable dust collection module. This balances airflow and dust particle isolation, effectively reducing reliance on static dust control structures and improving the system's heat dissipation stability and protection response capabilities.

[0058] (4) The invention modularizes the entire method into airflow modeling, particle path simulation, regional control generation, control execution, deflection effect acquisition, and feedback self-adjustment, forming a self-consistent data flow closed loop and structure mapping closed loop. This system can be reused in other air-cooled equipment scenarios, such as military servers and rail transit computing platforms, and has general adaptability. Attached Figure Description

[0059] Figure 1 A schematic flowchart illustrating the ruggedized computer dust particle prediction and adaptive dust prevention method provided in this application embodiment;

[0060] Figure 2 This is a structural block diagram of a ruggedized computer dust particle prediction and adaptive dust prevention system provided in an embodiment of this application. Detailed Implementation

[0061] Example 1: Refer to Figure 1 This is a flowchart illustrating the ruggedized computer dust particle prediction and adaptive dust prevention method provided in an embodiment of the present invention. The process may include at least steps S100-S600:

[0062] S100: By acquiring wind speed, temperature, and pressure data, a stable set of airflow parameters is constructed, and an airflow state boundary model is generated.

[0063] S200: Based on the airflow state boundary model, estimate the dust particle movement path, perform collision detection and stagnation trajectory calibration, generate candidate deposition paths, and calculate the deposition probability distribution structure.

[0064] S300: Obtain the deposition probability distribution structure, extract high-risk deposition areas and generate electrode control areas, construct the electrostatic flow field control structure, and calculate the control parameter matrix of the flow guiding electrode.

[0065] S400: Obtain the control parameter matrix of the flow guiding electrode, convert the control parameter matrix into electrode commands and drive the flow guiding control module to execute the dust particle deflection action and generate dust particle deflection path data.

[0066] S500: Configure the dust collection area based on the dust particle deflection path data, perform spatial aggregation and particle adsorption operations, and generate residual dust particle monitoring data.

[0067] S600 combines residual dust particle monitoring data with historical control parameters to generate a feedback control vector, and updates the control parameters to complete adaptive adjustment.

[0068] Step S100 includes at least steps S110-S130:

[0069] S110. Obtain wind speed, temperature and pressure data in the duct, perform structured analysis and format unification processing to generate duct airflow characteristic data.

[0070] Specifically, wind speed, temperature, and pressure data are first collected using a miniature, multi-functional environmental sensor array distributed at multiple points along the chassis's heat dissipation duct. The sensor array includes a thermistor wind speed probe, a thermocouple-type temperature acquisition unit, and a piezoresistive miniature pressure sensor, respectively positioned at the duct inlet, middle section, and near the heat dissipation channels of critical components. Its sampling frequency is no less than fifty times per second, meeting the requirements for dynamic flow field capture. Wind speed data is recorded in a three-dimensional vector format, temperature data is stored as absolute Celsius values, and pressure data is recorded in Pascals.

[0071] For the collected multidimensional raw data, the embedded data preprocessing module is invoked to perform structured parsing of the raw wind speed, temperature, and pressure data. The structured parsing process includes data field merging and timestamp alignment. Wind speed data is expanded according to airflow direction (i.e., x-axis, y-axis, z-axis) to construct a wind speed vector list. Temperature and pressure data are uniformly mapped to the corresponding wind speed sampling points according to spatial coordinates, and the sampling time identifier is simultaneously integrated to form a joint sampling data matrix with spatiotemporal consistency.

[0072] Furthermore, the parsed data undergoes a format unification process, specifically including data unit conversion, data missing point imputation, and outlier removal. Unit conversion includes converting wind speed to standard meters per second, temperature data to standard physical degrees, and pressure data to absolute pressure. Data missing point imputation uses linear interpolation or third-order spline interpolation to fill in time-related data gaps caused by occasional sensor failures. Outlier removal employs a sliding window-based z-score statistical method, marking abrupt changes exceeding 3σ as outliers and discarding or smoothing them out. After these processes, a duct airflow characteristic data structure containing wind speed vectors, temperature scalars, and pressure scalars for each sampling point is generated.

[0073] The aforementioned airflow characteristic data will be directly used as the input structure for step S120 to support the generation of a stable airflow parameter set, and will also serve as the initial boundary parameter input for the construction of the dust particle path prediction model in step S200.

[0074] S120. Normalize and remove disturbances from the airflow characteristic data of the duct to generate a stable set of airflow parameters.

[0075] Specifically, the airflow characteristic data of the duct is first normalized. The wind speed vector is normalized using a component normalization method. For each wind speed direction, a maximum and minimum normalized scaling function is constructed to map the three-axis wind speed values ​​one by one. The temperature and pressure data are normalized using a mean-variance normalization method. Based on the historical mean and standard deviation of each parameter, the instantaneous data is dimensionless, thereby achieving amplitude scale alignment between parameters of different dimensions.

[0076] After normalization, disturbance removal is performed on the normalized results. The disturbance identification operation mainly targets two types of disturbances: first, airflow oscillation disturbances caused by local object obstruction, fan start-up and shutdown, etc.; and second, sampling abrupt disturbances caused by electromagnetic interference or signal mutations. To achieve disturbance identification and removal, the system uses a sliding window method combined with an adaptive trend evaluation algorithm. The sliding window is set to evaluate the rate of change of wind speed, temperature, and pressure over a continuous time period. If the rate of change exceeds three times the standard deviation of the historical steady state, it is marked as a disturbance segment. Further spatial co-analysis is used to confirm whether the disturbance affects the system's stability.

[0077] After disturbance removal, the spatial mean field and principal component vectors are calculated for the remaining stable data to form a set of stable airflow parameters. This set of stable airflow parameters includes the mean wind speed principal vector field, the mean temperature distribution scalar field, and the mean pressure equipotential surface configuration, which serve as the boundary modeling input conditions required for S130.

[0078] The set of stable airflow parameters is not only the basis for constructing the airflow state boundary model, but also the basic fluid input field data structure on which the CFD model is started in the subsequent S200 step. Its steady-state characteristics directly affect the accuracy and convergence of dust particle path estimation.

[0079] S130. Construct an airflow state boundary model based on a stable set of airflow parameters and complete the airflow state initialization process.

[0080] Specifically, the system first calls the wind speed principal vector field data and the spatial distribution structure of temperature and pressure from the stable airflow parameter set and loads them into the airflow state boundary modeling unit. The modeling unit describes the geometric layout of the airflow duct based on a graph structure topology, divides the physical domain of the airflow duct through a three-dimensional mesh partitioning method, and sets the boundary condition region, internal control volume region, and key node positions. The airflow direction principal vector field is mapped to the input surface of the boundary region to form the initial inlet boundary conditions; the temperature field and pressure field are mapped to the boundary nodes and the internal region, respectively, forming the thermal flux boundary constraint and density field difference structure.

[0081] Using a modeling engine, a first-order stable solution field operation is performed on the above boundary conditions to solve the duct flow field structure under initial passive disturbance conditions, thus constructing an airflow state boundary model. This model stores the following structural information: the mapping relationship between the duct's three-dimensional structure and the control volume grid points, the airflow vector assignment structure at the boundary input surface, the temperature-pressure coupling mapping table, and the pressure gradient direction map related to the heat load distribution.

[0082] After the airflow state boundary model is constructed, the model data will serve as the initial boundary input structure for the dust path estimation algorithm in the subsequent step S210. Specifically, the particle trajectory estimation in step S210 is based on the principal direction of the velocity field and the isobaric surface flow information provided by the boundary model, further performing Eulerian-Lagrange path tracking calculations. Therefore, the airflow state boundary model output in S130 not only serves as the boundary driving condition for subsequent path prediction, but its encompassed steady-state parameters are also used in S230 for deposition probability density mapping calculations, ensuring the consistency and closed-loop stability of the overall inference chain.

[0083] Through the continuous execution process from S110 to S130, the system achieves dynamic acquisition, structured processing, and stability modeling of the three parameters of wind speed, temperature, and pressure in the duct environment. This provides a highly timely and consistent physical environment model for subsequent CFD path prediction and electrostatic flow control. In particular, in this invention, the airflow state boundary model output from S130 is directly supplied to step S200, ensuring that the dust particle motion path estimation has accuracy constraints based on the real stable field, and enhancing the overall prediction reliability of the system and the closed-loop consistency of the flow control logic.

[0084] Step S200 includes at least steps S210-S230:

[0085] S210. Obtain the airflow state boundary model, combine it with the particle size range setting parameters, and perform particle motion path estimation.

[0086] Specifically, after completing the construction of the airflow state boundary model in step S130, this step first calls the set of stable airflow parameters output by the model, which includes the average velocity vector field. vorticity tensor field and pressure distribution field This set covers the three-dimensional spatial region of the air duct in a grid format.

[0087] The system combines the user-defined dust particle size range [D] min D max For each typical particle size (Where k = 1, 2, ..., K, a total of K discrete particle sizes are sampled) to perform particle trajectory estimation simulation.

[0088] The particle path estimation is based on the Lagrange particle tracking model, and the following differential expression for particle motion is constructed:

[0089]

[0090] in:

[0091] This represents the spatial position vector of dust particle numbered f at time t in the i-th step;

[0092] This represents the velocity vector of the dust particle at the same time step;

[0093] t: time variable;

[0094] i: Index of the discrete time step;

[0095] f: Particle number (used to distinguish different simulated particles).

[0096] Formula ① describes the time-varying positional evolution of dust particles in three-dimensional space, forming the backbone of the Lagrange particle tracking model. Based on this formula, the system performs path integral simulations for each particle, generating its trajectory. This achieves high-precision position prediction for individual particles, providing a continuous path basis for subsequent collision detection and deposition area determination.

[0097] Furthermore, based on Newton's second law and the mechanical model of a particle of unit mass, the update of the dust particle velocity is constructed as follows:

[0098]

[0099] in:

[0100] The acceleration of the dust particle at the current time step;

[0101] This is the drag acceleration term experienced by the dust particle at step i;

[0102] F G : represents the constant gravitational acceleration term.

[0103] Formula ② establishes a dynamic model of a unit mass particle based on Newton's second law, combining drag and gravity to simulate the evolution of particle velocity. This ensures that the particle motion simulation responds to changes in hydrodynamics while maintaining realistic gravitational influence, thus improving the consistency of the simulation's physics.

[0104] Furthermore, the resistance term is constructed using a modified expression based on the Stokes resistance model as follows:

[0105]

[0106] In the formula:

[0107] μ: air viscosity coefficient;

[0108] ρ f : represents the density of dust particles;

[0109] This is the current particle size sample value;

[0110] The velocity vector of the airflow at the current position of the dust particle is obtained by interpolation of the velocity field generated in S130.

[0111] Formula ③ is used to characterize the viscous drag between particles and airflow under low Reynolds number conditions, reflecting that the larger the particle size, the greater the velocity difference, and the stronger the drag. It enables modeling of the dynamic impact of drag on particles of different sizes, providing a key adjustment factor for path prediction and improving the physical accuracy of simulating the behavior of small-diameter particles.

[0112] Furthermore, to evaluate the intensity of the interaction between particles and the flow field, the Reynolds number is defined:

[0113]

[0114] in:

[0115] Re (i) The Reynolds number of the particle at step i;

[0116] ρ a Air density;

[0117] Formula ④ is used to quantify whether the dynamic behavior of dust particles relative to the airflow belongs to the viscous-dominated or inertial-dominated range. When Re (i) When the value is >1000, it indicates that particle motion is more influenced by inertia, and subsequent paths may deviate from the mainstream region. Therefore, the deflection path needs to be enhanced in S220. This provides a criterion for path offset and deflection prediction, supporting the enhancement of high-inertia particle paths in S220, thereby improving the accuracy of deposition identification.

[0118] The above formulas (①-④) constitute a differential iterative system for dust particle path estimation. The system performs time-domain discrete integration on each initial position point and particle size value to form a path vector sequence. For use in the next step.

[0119] S220. Collision detection and stagnation trajectory calibration are performed on the dust particle movement path to generate candidate deposition paths.

[0120] This step receives the dust particle path sequence output by S210 and establishes a spatial collision detection mechanism for the inner wall of the air duct, component boundaries, and the surface of the heat sink fins. Specifically, the air duct structure set is defined as follows: This represents L boundary surfaces or component surfaces.

[0121] For each path Determine whether its minimum distance to either side of the interface is lower than a set threshold. c ,Right now:

[0122]

[0123] in:

[0124] Dust particle numbered f is on the i-th path at time t. j The corresponding position vector;

[0125] S l : Air duct structure collection The l-th boundary surface in;

[0126] L: Total number of boundary surfaces;

[0127] ||·||: Represents Euclidean distance operation;

[0128] ∈ c : This represents the minimum contact distance threshold for collision detection;

[0129] This represents the minimum distance value across all boundary surfaces;

[0130] This expression is used to identify whether a particle trajectory has made physical contact or come into close proximity (less than ∈) with either side of the interface. c Once the conditions are met, the system records it as a collision event and marks the point segment as a candidate collision location. By introducing a continuous spatial distance judgment mechanism, high-precision particle path and boundary interaction identification is achieved, significantly improving the spatial resolution of deposition hotspots.

[0131] Furthermore, the system records all instances of contact or speed reduction below a threshold (e.g.) The points and segments constitute a stagnant region. Among them, This indicates that the dust particles are at time step t. j velocity modulus; ∈ v The set velocity threshold is used to determine whether a particle's movement is so slow as to be considered stationary.

[0132] For each path j, the system defines its deposition probability as follows:

[0133]

[0134] in:

[0135] Predicted deposition probability value for path j;

[0136] This represents the percentage of stalled segments in the j-th path, calculated as the ratio of the number of stalled points to the total number of trajectory points; This indicates the number of points in the path that are considered stalled. This represents the total number of time steps for the path;

[0137] It represents the collision frequency density, which is the ratio of the frequency of collisions in the path to the total length of the path; Indicates the number of collisions that occur in the path; The path length is represented and can be calculated by accumulating the points.

[0138] α1, α2: These are weighting coefficients, with a default value of α1 = 0.6 and α2 = 0.4, which can be adjusted based on empirical data. Formula ⑤ is the key innovative expression of this module, used to extract the set of "candidate depositional paths" by combining path dynamic characteristics. Where N c Indicates the total number of candidate paths; Represents the set of candidate depositional paths; X j This represents the path of the j-th dust particle.

[0139] Formula ⑤ combines two core dynamic indicators in a linear fashion: the percentage of stagnation δ (j) With collision frequency density θ (j) This comprehensive measurement, reflecting the depositional tendency of dust particle paths, is the core probabilistic metric in the process of identifying depositional hotspots. It achieves a numerical mapping from dust particle dynamics to depositional trends, overcoming the limitations of traditional methods that rely on static thresholds or empirical rules to judge deposition. It possesses high adjustability and interpretability, providing a precise spatial basis for subsequent electrode-controlled region generation.

[0140] S230. Calculate the deposition probability distribution structure within the wind tunnel area based on candidate deposition paths.

[0141] This step uses the aforementioned candidate path set. Based on this, a three-dimensional spatial deposition probability distribution function is constructed. The deposition tendency of each cell in the duct volume grid is estimated using a statistical region approach.

[0142] Specifically, let the spatial discrete grid be... Each volume element V m The deposition probability is defined as follows:

[0143]

[0144] in:

[0145] Represents the spatial volume element V mThe probability of dust particle deposition within;

[0146] This is an indicator function that indicates whether the path traverses the voxel; Indicates if path X j Passing through or through volume V m The function takes a value of 1 otherwise. It is used to filter paths that traverse the current voxel.

[0147] Z: Normalization factor, equal to the sum of probabilities of all paths.

[0148] The final output is a sedimentation probability volume data structure. Used as input for the subsequent S300 electrostatic control region extraction module.

[0149] Through the complete encapsulation of the S200 module described above, the system can achieve particle size stratification simulation, path estimation, behavior discrimination, and spatial probability mapping based on stable airflow parameters. Specifically:

[0150] Formulas ①–④ construct a differential system for estimating motion paths;

[0151] Formula ⑤ achieves a comprehensive score for path behavior;

[0152] Formula 6 achieves a closed loop for high-dimensional probability mapping.

[0153] The module's technical advantage lies in providing full-path-level particle deposition prediction capabilities based on CFD and multi-factor fusion algorithms. This forms the core reasoning foundation of the entire "hardened computer dust particle prediction and adaptive dust prevention method and system," providing accurate, high-dimensional, and real-time data support for subsequent electrostatic control strategy formulation.

[0154] Step S300 includes at least steps S310-S330:

[0155] S310. Obtain the deposition probability distribution structure, extract high deposition risk areas, and construct the spatial control area structure.

[0156] The system first acquires the deposition probability distribution structure output in the previous step S230. This deposition probability distribution structure is a three-dimensional distribution calculated based on candidate deposition paths, describing the normalized probability density values ​​of dust particle deposition at different locations in the spatial coordinate domain. Specifically, the system calls the three-dimensional deposition probability structure map output by module S230, scans and calculates each voxel unit in the map, and identifies regional units exceeding the deposition risk threshold as candidate hotspots for concentrated dust particle deposition.

[0157] Understandably, the deposition risk threshold is a statically set parameter, determined based on historical operating data and equipment safety tolerance standards, typically taking the percentile coefficient of the corresponding probability density distribution function. When the deposition probability of a spatial unit exceeds the set threshold, the system identifies it as a high-risk deposition point. To enhance the continuity of spatial identification, the system uses a region growing algorithm to cluster adjacent high-deposition-probability units, forming structured deposition risk clusters.

[0158] Furthermore, the system constructs an envelope control volume with each deposition risk accumulation zone as its core boundary. This envelope control volume undergoes processes such as bounding box fitting, boundary expansion, and connection channel extension to form a spatial control region structure for electrostatic current conduction regulation. This spatial control region serves as direct input data for subsequent electrode planning, constituting the geometric boundary basis of the current conduction region.

[0159] During the construction of the spatial control area structure, the system needs to call the airflow state boundary model generated in S130 to screen the overlap between the control area and the stable airflow boundary, retaining only the control volume in the main airflow channel, thereby eliminating non-core spatial units that have limited impact on dust particle movement, and ensuring the effectiveness and coverage efficiency of electrode resource allocation.

[0160] S320. Perform electric field boundary segmentation and target offset direction planning on the spatial control area structure to generate the electrode control area.

[0161] After completing the construction of the spatial control region structure, the system enters the electric field region generation stage. Specifically, the system first performs a spatial segmentation operation on each spatial control region structure, dividing it into several sub-electric field control units based on the distribution position of its boundary in the airflow vector field. The division rules of the sub-electric field control units are mainly based on the following two parameters: one is the angle relationship between the airflow direction vector and the normal of the control region boundary, and the other is the distribution density of the deposition probability gradient field inside the control region.

[0162] For boundary sections with nearly perpendicular angles, the system prioritizes using a high-gradient electric field for forced deflection control; for boundary sections with smaller angles, the system employs a weak electric field for fine-tuning control to reduce disturbance to the main airflow velocity.

[0163] After the electric field boundary segmentation is completed, the system performs target offset direction planning based on the particle size range setting parameters output in step S210. During this process, the system needs to combine the particle size range setting parameters with the airflow state boundary model to calculate the optimal dust particle deflection direction within each electric field control unit. This direction is defined as the direction vector that causes the dust particle movement path to avoid high-risk deposition areas as much as possible and deflect towards the dust collection module deployment area.

[0164] To improve control accuracy, the system incorporates the stagnation trajectory calibration results output in step S220 during the target offset direction planning process. By backfitting the aggregation direction of the stagnation trajectory, the system further determines the locations where dust particles are prone to deposit and their response sensitivity to the electric field direction. This information is integrated into the target offset direction generation function, forming a multi-electrode control direction adapted to dust particles of different sizes.

[0165] The system fuses the target offset direction, boundary attributes, and airflow field information within each control unit to generate an electrode control region. This electrode control region serves as the geometric reference basis for the generation of the electrostatic flow guiding structure. Its data structure includes core fields such as a three-dimensional position index, electric field domain boundary, target offset direction vector, and adaptive particle size parameter range, which are then read and used by subsequent modules.

[0166] S330. Based on the electrode control region, generate an electrostatic current-conducting field control structure and calculate the control parameter matrix of the current-conducting electrode.

[0167] Based on the electrode control region generated in step S320, the system enters the stage of constructing the electrostatic current conduction field control structure and calculating control parameters.

[0168] Specifically, the system first transforms the electrode control region into a spatial discrete mesh model, and then constructs an electrostatic conduction substructure in each control unit. The substructure includes: electrode unit geometric layout, voltage application area allocation, polarity direction constraints, and edge spacing tolerance information. The system initializes the electric field direction in each electrostatic structure based on the offset direction vector defined in the electrode control region.

[0169] Furthermore, the system combines the deposition probability distribution structure generated in S230 to perform voltage gradient optimization distribution processing within each electrode control region. This processing iteratively optimizes the electrode voltage values, ensuring that the resulting electric force vector field can effectively act on dust particles, achieving real-time deflection of their motion paths. The system employs a multi-dimensional parameter optimization model based on gradient descent to iteratively update the objective function. This objective function consists of the following parts: first, minimizing the error of dust particles deviating from the deposition hotspot path; second, the disturbance effect of the overall electric field intensity on airflow stability; and third, the constraint of electrode layout compactness.

[0170] The output of the above optimization process is the control parameter matrix of the current-conducting electrode. This matrix structure contains the spatial coordinates, voltage intensity value, polarity, and timing number of each electrode node. To improve the response efficiency and reliability of the control system, the system further converts the voltage scheduling values ​​in the control parameter matrix into a drive signal format, which can be directly used by the electrode command generation and module drive stages in subsequent steps.

[0171] Throughout the S330 process, the system needs to simultaneously read the airflow state boundary model output by S130 to ensure that the electric field formed by the flow guide electrode control does not cause critical disturbance to the main airflow flux, thereby maintaining the stability of the overall heat dissipation performance of the system.

[0172] Through the continuous implementation of the three steps S310 to S330 described above, this invention achieves a closed-loop path from deposition risk perception and spatial electrode planning to electrostatic current field generation. Specifically:

[0173] Based on the deposition probability distribution structure, S310 realizes spatial identification and structural calibration of high-risk deposition areas;

[0174] S320 constructs a highly adaptable electrode control region by constraining the airflow field and reasoning about the target direction;

[0175] Based on this, S330 generates a refined, programmable electrostatic flow field control matrix to ensure that dust particles are effectively deflected before entering the air duct, guiding them away from the deposition hotspot area.

[0176] The above processing results will be directly used as the input basis for the S400 module electrode execution process, ensuring a closed loop from sensing to regulation, forming a dynamic adaptive structure of "prediction-planning-control".

[0177] Step S400 includes at least steps S410-S430:

[0178] S410: Obtain the control parameter matrix of the current guiding electrode, and perform drive signal conversion and current guiding electrode command generation processing.

[0179] After generating the control parameter matrix for the current-conducting electrode in step S330, the system enters the conversion stage from control parameters to execution instructions. Specifically, the system first calls the control parameter matrix in the electrostatic current-conducting field control structure generated in step S330. The control parameter matrix is ​​a set of control structures describing the spatial configuration, voltage intensity, and polarity direction of the electrode units. In this embodiment, the matrix data structure is presented as a multi-dimensional discrete array and has time synchronization and spatial grid matching characteristics.

[0180] Furthermore, the system invokes the electrode parameter mapping module to convert the electrode space indices and deflection intensity values ​​in the control parameter matrix into low-order voltage codes compatible with the drive circuit. This mapping module has a voltage quantization conversion interface, and its processing includes:

[0181] 1. Analyze the electrode index identifier field in the control parameter matrix and extract its spatial coordinate attributes and their correspondence with the electric field boundary;

[0182] 2. Based on the control weight and target offset direction setting value of each electrode unit, the continuous control parameters are discretized into digital drive signals and encapsulated using a programmable grid register structure format.

[0183] Third, based on the system's scheduling time slice structure, the above electrode drive signal sequence is time-coded to generate an electrode instruction set with timeliness identifiers.

[0184] The electrode instruction set is encapsulated and output in byte stream format, containing information fields such as the drive address, deflection direction instruction, voltage level, and control duration for each current-conducting electrode unit. In this step, the system performs redundancy checks and electrical tolerance verification on this instruction set structure to ensure that it can be stably parsed and executed by the current-conducting electrode control module in subsequent steps.

[0185] The electrode instruction generation process also includes an error tolerance mechanism. When the system detects cross-boundary electrode overlap or polarity conflict fields during the parsing of the current-conducting electrode control parameter matrix, the system will automatically execute the vector field rollback mechanism, calling the verified results from the previous control parameter buffer as a temporary replacement to ensure the integrity of the drive logic.

[0186] S420: Send the current guiding electrode command to the current guiding control module to execute the programmable current guiding electrode control operation.

[0187] After the instruction set for the current guiding electrode is encapsulated and generated, the system will enter the stage of issuing current guiding instructions and driving the current guiding control module. Specifically, the system calls the central control bus interface to establish a low-latency communication link with the current guiding electrode execution module. This communication link is implemented using a CAN bus protocol or an SPI high-speed bus protocol, and its structure has clock synchronization and multi-channel fault tolerance characteristics to ensure the stability and accuracy of instruction transmission.

[0188] During the transmission preparation phase, the system performs multi-channel notification and scheduling flag setting on the electrode instruction set to ensure that the corresponding electrode execution unit in the current diversion control module can receive control signals according to the preset timing sequence. Subsequently, the system sends each group of control byte streams containing position index, deflection voltage, and polarity direction to the input buffer of the current diversion control module according to the electrode block sequence.

[0189] The flow control module has an internal parsing component for receiving electrode control commands, and its functions include:

[0190] 1. Receive the control byte stream structure, perform address mapping and CRC check to ensure instruction integrity;

[0191] 2. Based on the time-series tag reordering module, instructions are queued and their execution cycles are scheduled to achieve clock synchronization control;

[0192] Third, based on the analysis results, the unit modules in the current-guiding electrode array are driven to perform dynamic deflection control operations. Each electrode unit can be programmed to set the voltage amplitude, electric field direction and control duration.

[0193] In this embodiment, the programmable current guiding electrode unit is constructed based on a MEMS electrode array structure, which has the characteristics of small size, high response speed and low power consumption. It can respond to the electric field change settings in the control command in real time and stably maintain the deflection action.

[0194] Furthermore, the current diversion control module includes a state feedback path. This feedback path monitors the actual deflection voltage, current response, and electric field formation of the electrodes, and transmits the detected values ​​back to the central control module for subsequent feedback vector generation and adaptive parameter update processing.

[0195] To prevent command failures in complex airflow disturbances or high dust concentration environments, the system is designed with an electrode redundancy zone and a skip-step control strategy. When the critical flow guiding electrode fails to execute or responds with a delay, the system can automatically call upon the redundant electrode unit to intervene in the control, forming an electric field topology adjustment and deflection path migration operation to ensure that the dust particle deflection mechanism is not interrupted.

[0196] S430: Performs dust particle deflection action according to the programmable current guide electrode control operation, and generates dust particle deflection path data.

[0197] After the programmable flow guide electrode successfully receives the control command issued by the flow guide control module and completes the electric field reconstruction, the electrostatic flow guide field in the duct space will form an adjustable non-uniform potential distribution. This potential distribution structure is consistent with the target offset direction preset by the electrostatic flow guide field control structure in step S330, and has the ability to spatially deflect the particle group carried by the incoming airflow.

[0198] Specifically, the guiding electrode array establishes a gradient-guided electric field structure inside the air duct based on the spatial arrangement and voltage excitation values ​​set by the control parameter matrix. This structure, using the boundary of the high-risk deposition probability area as a reference, forms a stable particle deflection field towards the dust collection module, away from the sensitive area. This particle deflection field selectively deflects dust particles of different sizes, charge states, and inertial parameters by applying voltage differences of varying intensity and polarity at different locations.

[0199] After dust particles enter the guiding field area, their trajectories deviate relative to the original airflow path due to the combined effects of electrostatic and aerodynamic forces. The system records these changes in real time using a multi-channel capacitive particle trajectory sensing array mounted on the duct wall or behind the guiding electrodes. This sensing array employs sub-millisecond response devices, enabling it to stably sense particle movement paths under complex airflow disturbance conditions and archive the degree of deflection.

[0200] To ensure the reliability of the particle deflection path data, the system performs the following steps on the initial offset data output by the induction array:

[0201] First, perform noise reduction and segment registration on the particle response values ​​at each acquisition point to eliminate spurious electrical signal responses caused by background airflow disturbances.

[0202] 2. Based on the known wind speed vector field and the inertial characteristic model of dust particles, calculate the instantaneous velocity vector and deflection acceleration of the particle motion;

[0203] 3. Based on the motion trajectory of the particles from the guide inlet to the dust collection target area, construct a deflection trajectory dataset in three-dimensional space.

[0204] The system encapsulates the above particle trajectory data into a dust particle deflection path data structure, which includes, but is not limited to, fields such as particle number, particle size class, initial position coordinates, final deflection position, average offset vector per unit time, and whether it has entered the preset dust collection area.

[0205] Furthermore, the dust particle deflection path data will serve as the fundamental input for the spatial aggregation and particle adsorption processes performed in the subsequent S500 stage, providing spatial layout support for the configuration and scheduling of the dust collection area. Simultaneously, the number and position of particles that were not successfully deflected along the path will be used to generate the feedback control vector for the S600 stage, providing a closed-loop basis for the dynamic adjustment of the control parameter matrix.

[0206] The system also incorporates a dust particle deviation failure alarm mechanism in this step. When particles fail to deflect effectively within the guide zone, or when the deflection direction deviates too much from the target dust collection area, the system will record the abnormal particle event and generate a field identifier indicating a failure to map control parameters. This information will be used by the adaptive module to perform parameter correction and readjustment of the guide electric field structure.

[0207] By implementing steps S410 to S430, the system completes the mapping from the control parameters of the guiding electrode to the physical deflection action, constructing a closed-loop structure from electrostatic field control command generation and command execution to deflection path perception. This ensures that dust particles are effectively deflected before they occur in high-risk deposition areas and achieves a highly efficient and dynamically adaptive electrostatic guiding control mechanism. The dust particle deflection path data output by this module not only provides accurate input for subsequent dust collection strategies but also forms a key foundation for constructing an adaptive update mechanism for control parameters, ensuring that the entire system has long-term stable dust prevention effects and self-optimization capabilities.

[0208] Step S500 includes at least steps S510-S530:

[0209] S510: Obtain dust particle deflection path data, configure the dust collection target area, and schedule the dust collection module.

[0210] Specifically, this step first acquires the dust particle deflection path data generated in the preceding step S430. This dust particle deflection path data records the actual trajectory of dust particles from the guiding initiation surface to the duct terminal under the control of the programmable guiding electrode, including a sequence of dust particle positions, velocity vectors, timestamps, and guiding offset coordinate system information. This deflection path data serves as a key input parameter for dust collection area configuration and dust collection module scheduling, and is organized as a multi-dimensional tensor at the data structure level and transmitted to the data preprocessing unit of this module.

[0211] The data preprocessing unit performs path clustering and density assessment analysis on the dust particle deflection path data. By analyzing the spatial clustering degree of multiple deflection paths, spatial clustering regions with high dust particle trajectory overlap can be extracted, and these regions are then marked as primary dust collection candidate regions. Furthermore, combined with the deposition probability distribution structure generated in step S230, the system further compares the intersection degree between the deflected path trajectory and the high deposition risk region, and filters and marks the final dust collection target region boundary.

[0212] To improve dust collection accuracy and module scheduling efficiency, the system performs geometric projection mapping on the dust collection target area based on the electric field conduction endpoint distribution map and aerodynamic distribution model, forming a matching dust collection area structure adapted to the air inlet and filter media arrangement of the dust collection module. This structure uses a matrix-style three-dimensional mesh to describe the boundary information of the dust collection space and the particle guidance inlet position, and forms input parameters for subsequent use by the dust collection module.

[0213] The system calls the dust collection control module interface and assigns initialization control commands to the dust collection equipment through the interface scheduling strategy mapping function. These control commands include, but are not limited to: adsorption medium activation commands, air pressure adjustment commands, electrostatic adsorption plate opening commands, and particle guide valve synchronization commands. All commands are generated in real-time based on the spatial layout and air field stability model of the dust collection target area and are executed through the dust collection module drive channel.

[0214] S520: Perform the spatial aggregation and particle adsorption operation of the dust collection module to generate dust collection control result data.

[0215] Furthermore, after completing the dust collection module scheduling operation, this step begins the spatial aggregation and particle adsorption process of the dust collection module. The dust collection module includes at least an electrostatic adsorption plate structure, programmable airflow guide vanes, a particle guiding channel, a filter material module, and a residual dust collection chamber. These components work together to complete the spatial aggregation, trajectory convergence, and particle capture of dust particles within the target area.

[0216] Specifically, the electrostatic adsorption plate structure establishes a dynamic electric field through voltage field control commands injected into the control module. The direction of this electric field is synchronously adjusted according to the control parameter matrix of the current guiding electrode in S330, causing particles in the dust collection area to shift towards the filter surface under the guidance of the electric field. The filter material module uses a high-efficiency, low-resistance electrostatic fiber medium, which can perform mechanically barrier-free adsorption treatment on various types of dust particles with particle sizes ranging from submicron to tens of micron, thus achieving both high-precision adsorption and maintenance of airflow.

[0217] To prevent secondary diffusion or unexpected trajectory deviation of particles under air disturbance, the system further utilizes programmable airflow guide vane modules to establish a reverse airflow pressure differential layer in the dust collection channel inlet area. This airflow structure is automatically generated and updated by the system based on dust particle deflection path data, achieving aerodynamic capture of dust particles during the accumulation process and improving particle adsorption efficiency.

[0218] During dust collection, each submodule synchronously records operating parameters, including key indicators such as changes in adsorption intensity, particle density, electric field response amplitude, and filter flux. The system encapsulates this data to form dust collection control result data, which is output in a structured data format for subsequent residue identification. The dust collection control result data includes at least: the dust collection target area identification number, the control time period identifier, the dust collection efficiency of the filter unit, the remaining charge capacity of the adsorption plate, the stability index of the collecting airflow, and the count of abnormal adsorption events.

[0219] S530: Perform residual particle identification and counting processing on the dust collection control result data to generate residual dust particle monitoring data.

[0220] After the dust collection operation is completed and the dust collection control results data are obtained, the system enters the residual particle identification and counting stage. The core of this stage is to identify residual particles that have not been adsorbed after dust collection, monitor the actual working efficiency of the dust collection system, and provide a quantitative basis for subsequent feedback control.

[0221] Specifically, the system first invokes the high-resolution imaging module for the dust collection area, which is located on the inner wall of the dust collection structure or the side wall of the downdraft duct. Immediately after the dust collection task is completed, the module acquires an image sequence of the target area. This image sequence undergoes background modeling and differential extraction by the image preprocessing module to form an intermediate image structure containing dust particle contour information.

[0222] The system invokes a particle recognition neural network model to perform multi-scale particle recognition operations on the aforementioned intermediate image structure, calibrating the position, size, reflectivity, and morphological parameters of various unadsorbed particles. The recognition model constructs residual feature templates based on the training set and eliminates misidentified items by incorporating spatial distribution constraints. Confirmed residual particle information is uniformly stored in a particle residual dataset for subsequent counting and statistical analysis.

[0223] The system performs quantitative analysis based on the identified residual particle set data. Using a particle category density function and a regional activity rate model, it calculates the residual particle density per unit area and per unit time, and compares it with historical data or preset thresholds to determine if residual levels exceed standards. All statistical parameters and analysis results are organized into structured residual dust particle monitoring data for use by the S600 module when updating feedback control parameters.

[0224] Through the continuous processing flow from S510 to S530 in this step, the present invention achieves a dynamic mapping from dust particle deflection path data to the physical dust collection structure's actions. This not only establishes a closed-loop link between dust particle motion path prediction and physical adsorption response, but also achieves modular decoupling between spatial partitioning control of the dust collection area and particle adsorption execution. Finally, the formation of residual dust particle monitoring data provides a reliable data input basis for subsequent control parameter updates, ensuring that the adaptive control strategy has a quantifiable feedback path, thus forming a closed-loop control structure of "prediction-control-collection-evaluation-feedback".

[0225] Step S600 includes at least steps S610-S630:

[0226] S610. Acquire residual dust particle monitoring data, and generate a feedback control vector by combining the deposition probability distribution structure and the control parameter matrix.

[0227] Specifically, the system first calls the residual dust particle monitoring data generated in step S530. This residual dust particle monitoring data is a set of structured residual particle indicators output by the dust collection control module after performing particle adsorption and identification. This set is obtained by statistically processing the number, particle size distribution, spatial distribution density, and residual heat trap density of dust particles not yet adsorbed in the dust collection area. The residual particle identification method includes a weighted fusion result of three methods: particle image reflectance analysis, capacitance anomaly identification, and dust collection surface adsorption redundancy verification.

[0228] Furthermore, the system retrieves the deposition probability distribution structure already constructed in step S230. This distribution structure is constructed through dust particle movement path simulation, candidate deposition path generation, and multi-region probability estimation function calculation. It possesses dual identification capabilities of spatial coordinate resolution and probability density, and covers the entire heat dissipation channel area. Combining the above two types of structural data, the system compares the predicted deposition probability value with the actual residual dust particle monitoring value for each high-risk deposition area, performs difference mapping operations, and performs temporal alignment to form a structured residual parameter set.

[0229] The system further incorporates the control parameter matrix output in step S330. This matrix records the voltage parameters, electrode activation sequence, deflection angle control factor, and polarity adjustment strategy parameters of each guiding electrode in the electrostatic current guiding field control structure. To construct the feedback vector, the system generates a feedback control vector based on the nonlinear mapping relationship between the control parameter matrix and the residual parameter set, using a high-dimensional parameter compression and encoding function. This feedback control vector is a set of multi-dimensional control deviation indices used to characterize the degree of deviation between the current guiding control strategy and the actual dust particle behavior, supporting element-wise adjustment of subsequent control parameters.

[0230] The process of generating the feedback control vector is executed within the adaptive update module 60 of the system of the present invention. It calls the three input structures mentioned above: residual dust particle monitoring data, deposition probability distribution structure, and control parameter matrix. After completing the structural standardization and dimensional normalization processing, it performs encoding, alignment, and combination operations to form a multidimensional feedback control vector, which is then used by the subsequent update module.

[0231] S620. Update the control parameters in the electrostatic current conduction field control structure according to the feedback control vector.

[0232] In this step, after receiving the feedback control vector, the system automatically calls the electrostatic current conduction field control structure constructed in step S330, locks its internal control parameter submodule, and starts the control parameter update process.

[0233] Specifically, the electrostatic current-conducting field control structure includes multiple current-conducting electrode nodes, boundary deflection units, and electric field gradient equalization units. The aforementioned control parameter submodule contains five key control attributes: the spatial arrangement order of the current-conducting electrodes, the driving voltage amplitude, the polarity configuration direction, the response delay, and the stability coefficient of the controlled electric field. During the control parameter update process, the system performs parameter-by-parameter correction operations based on the residual direction, adjustment amplitude, and control offset factor carried in the feedback control vector.

[0234] During the correction process, the system employs a distributed adaptive filtering algorithm to perform stepwise gradient updates for each control parameter. The system sequentially calls the control parameter values ​​from the previous cycle, the corresponding elements of the current feedback control vector, the target deflection direction of the flow field, and the regional priority strategy factor to form a parameter offset calculator. This calculator is then mapped to the corresponding positions of each module within the electrostatic flow field control structure to complete the parameter update.

[0235] Furthermore, to ensure the stability and continuity of control parameter updates, the system performs three stability verification operations after each update: First, it performs a physical consistency verification on the updated control parameter matrix to ensure that the voltage value, polarity direction, and response delay do not violate the electric field safety domain; second, it performs an electrode cascade response verification to verify that the deflection operation will not cause electric field interference conflicts between regions; and third, it performs a thermal conduction influence constraint assessment to avoid the generation of local thermal conduction bottlenecks in the high-heat region by the electrode deflection control.

[0236] After completing the above multiple verifications, the system officially injects the updated control parameters into the control structure, completing the refresh operation of the current control parameter matrix.

[0237] S630: Re-inject the updated control parameters into the current guide electrode command generation process to complete the adaptive parameter adjustment process.

[0238] After updating the control parameter matrix, the system enters the instruction update process reconstruction stage. First, the adaptive update module reads the updated control parameter matrix and maps it to the current-guiding electrode instruction generation process defined in step S410.

[0239] Specifically, the system sequentially feeds the voltage control vector, polarity configuration vector, response time sequence vector, and spatial positioning matrix from the new control parameter matrix into the instruction conversion module. This module then executes the electrode instruction construction process based on the previously set drive signal model and instruction encoding specifications. This process includes four stages: signal standardization processing, electrode address mapping, control mode matching, and output sequence synthesis, ultimately generating a new set of current-conducting electrode instructions.

[0240] The system sends the set of current-guiding electrode instructions to the current-guiding control module, which then interfaces with the programmable current-guiding electrode control operation in step S420. To ensure the atomicity and consistency of instruction updates, the system sets up a dual-channel instruction caching mechanism. This means that new instructions are loaded in parallel within the old instruction execution cycle, and the contents of the current-guiding electrode instruction cache are synchronously replaced by a switching trigger flag before the current control cycle ends.

[0241] It is worth noting that the deflection path generation logic in the flow guide electrode command generation process also needs to be updated synchronously at this stage. The system combines the command set with the current deposition probability distribution structure to perform a path mapping backtracking operation to ensure that the new control strategy can effectively cover all high-risk deposition areas and achieve real-time synchronous control of dust particle deflection path data.

[0242] At this point, the adaptive parameter adjustment process is complete, and the system achieves closed-loop correction of the flow guidance control logic.

[0243] Through the coordinated execution of steps S610, S620, and S630, the dynamic updating and real-time write-back of dust particle deflection control parameters are achieved, ensuring that the system can continuously optimize the flow guidance control strategy based on the offset trend between the current residual dust particle behavior and the historical deflection control strategy. Compared to the traditional fixed control parameter mode, the adaptive parameter adjustment mechanism described in this embodiment can enhance the closed-loop control performance without external intervention, significantly improve the accuracy and reliability of dust particle deflection, and ensure that the system maintains the optimal heat dissipation-dust prevention balance during long-term operation.

[0244] The key innovations of this invention include:

[0245] (1) CFD-based particle size stratification motion path modeling and Reynolds number coupled trajectory derivation mechanism. It realizes the dynamic correlation between particle velocity, position, drag and airflow velocity, and introduces "path-level differential estimation and dynamic scoring" closed-loop modeling for the first time in dust prevention of air-cooled systems, which has high accuracy and real-time performance.

[0246] (2) Constructing the deposition probability function and three-dimensional mapping structure of spatial candidate paths. An explicit regional statistical mechanism is introduced to form a voxel-level probability heatmap, which overcomes the lack of predictive ability in traditional dust control systems and provides a precise positioning basis for deflection strategies.

[0247] (3) Automatic generation of electrode control area and real-time calculation of electric field deflection direction. The system generates spatial electrode control area and electrostatic conduction parameter matrix through S310–S330, realizing precise guidance of electric field before air inlet and effectively avoiding the problem of "deflection misalignment".

[0248] (4) Construction and execution mechanism of multi-level feedback adaptive control path. The system introduces a mechanism for comparing residual dust particle monitoring results with control parameters, which can adjust the deflection intensity and spatial control area coverage in reverse according to the residual dust particle distribution to achieve periodic self-adjustment.

[0249] The following are its main beneficial effects:

[0250] (1) Significantly improves the spatial resolution and physical consistency of dust particle distribution prediction. This invention constructs a stable airflow state boundary model in step S100 and introduces a Lagrange path differential model based on Reynolds number and Stokes drag in steps S210–S230 to achieve fine estimation of the spatial path of dust particles of different sizes; and integrates collision detection and stagnation assessment mechanisms with spatial deposition probability mapping functions to accurately identify dust particle deposition hotspots in the duct. Compared with traditional particle distribution methods based on empirical rules or macroscopic estimation, its prediction error is lower.

[0251] (2) Achieving dynamic adaptive adjustment of dust particle deflection control. The system constructs an electrode control region and generates an electrostatic current conduction control parameter matrix in S300–S330. Combined with the instruction conversion and programmable current conduction execution mechanism in S400, the deflection direction and voltage intensity can be automatically adjusted according to changes in the spatial distribution of dust particles. Furthermore, through the feedback path in S600, the residual dust particle monitoring results are fed back to the control matrix generation module, forming a closed-loop adaptive adjustment path, which effectively improves the continuous accuracy and real-time performance of the guidance and control strategy in complex environments.

[0252] (3) Reduce reliance on fixed dustproof components and improve heat dissipation efficiency. Traditional dustproof solutions typically rely on mechanical blocking structures such as fixed filters, which are prone to clogging and thus affect the system's heat dissipation performance. This invention introduces an adjustable electrostatic flow guiding system, which dynamically applies an electric field based on dust particle path prediction results to achieve "path-guided" dust particle deflection control. An electric field barrier is formed in advance in areas with significant deposition trends, causing dust particles to deviate from the critical heat-sensitive element area and accumulate in the detachable dust collection module, balancing airflow and dust particle isolation. This effectively reduces reliance on static dustproof structures and improves the system's heat dissipation stability and protection response capabilities.

[0253] (4) A generalized and modular dust prevention logic system is formed. The invention breaks down the entire method into airflow modeling, particle path simulation, regional control generation, control execution, deflection effect acquisition and feedback self-adjustment in a modular manner, forming a self-consistent data flow closed loop and structure mapping closed loop. This system can be reused in other air-cooled equipment scenarios, such as military servers and rail transit computing platforms, and has general adaptability.

[0254] Example 2: Figure 2 A structural block diagram of a ruggedized computer-aided dust particle prediction and adaptive dust control system according to an embodiment of the present invention is shown. Figure 2 As shown, the structure may include:

[0255] The airflow parameter acquisition module 10 is used to acquire wind speed, temperature and pressure data from the heat dissipation duct, perform structured parsing and format unification processing on the raw data, and complete normalization and disturbance removal operations to output a stable airflow parameter set and an airflow state boundary model.

[0256] The dust particle path prediction module 20 is used to call the airflow state boundary model and combine the parameters set by the particle size range to perform dust particle motion path estimation; perform collision detection and stagnation trajectory calibration on the estimation results to form candidate deposition paths; and calculate the deposition probability distribution structure based on the candidate deposition paths.

[0257] The electrostatic flow field generation module 30 is used to read the deposition probability distribution structure, extract high-risk deposition areas and construct a spatial control area structure; on this basis, it completes the electric field boundary segmentation and target offset direction planning to generate the electrode control area; and constructs the electrostatic flow field control structure based on the electrode control area and calculates the control parameter matrix of the flow guiding electrode.

[0258] The flow guide electrode execution module 40 is used to receive the control parameter matrix, convert it into flow guide electrode instructions and send them to the flow guide control module; drive the programmable flow guide electrode to form a dynamic electric field, guide the dust particles to deflect and output the dust particle deflection path data.

[0259] The dust collection and residue monitoring module 50 is used to configure the dust collection target area based on the dust particle deflection path data, schedule the dust collection module to perform spatial aggregation and particle adsorption operations, and generate dust collection control result data; then, the dust collection control result data is used to identify and count residual particles to generate residual dust particle monitoring data.

[0260] The adaptive update module 60 is used to acquire residual dust particle monitoring data, combine the deposition probability distribution structure and control parameter matrix to generate a feedback control vector; update the control parameters in the electrostatic conduction field regulation structure according to the feedback control vector, and write the update result back to the conduction electrode execution module to form a closed-loop regulation.

[0261] This system employs a closed-loop architecture of "real-time airflow parameter acquisition—dust particle path prediction—electrostatic flow field generation—flow guide electrode execution—dust collection and residue monitoring—adaptive updating" to achieve dynamic prediction and active deflection control of dust particle deposition risk while maintaining the heat dissipation airflow, significantly improving the reliability and lifespan of ruggedized computers in high-dust environments.

[0262] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application 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 specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A method for ruggedized computer particle prediction and adaptive dust prevention, characterized in that, Includes the following steps: Raw wind speed, temperature, and pressure data are collected using a miniature multifunctional environmental sensor array. The collected raw data undergoes data field merging and timestamp alignment to construct a stable set of airflow parameters and generate an airflow state boundary model. The sensor array includes a thermistor wind speed probe, a thermocouple temperature acquisition unit, and a piezoresistive miniature pressure sensor. Wind speed data is recorded in a three-dimensional vector format, temperature data is stored as absolute Celsius values, and pressure data is recorded in Pascals. The airflow state boundary model includes the mapping relationship between the three-dimensional structure of the air duct and the grid points of the control volume, the airflow vector assignment structure of the boundary input surface, a temperature-pressure coupling mapping table, and a pressure gradient map. The dust particle motion path is estimated based on the airflow state boundary model, collision detection and stagnation trajectory calibration are performed, candidate deposition paths are generated, and the deposition probability distribution structure is calculated. The expression for the deposition probability distribution is: in, Represents the spatial volume element V m The probability of dust particle deposition within; Z is the normalization factor; N c Indicates the total number of candidate paths; X is the predicted deposition probability value for the j-th path; j This represents the path of the j-th dust particle. It is an indicator function; Obtain the deposition probability distribution structure, extract high-risk deposition areas and generate electrode control areas, construct the electrostatic conduction field control structure, and calculate the control parameter matrix of the conduction electrode. The control parameter matrix of the flow guiding electrode is obtained, the control parameter matrix is ​​converted into electrode commands and driven to the flow guiding control module to execute the dust particle deflection action and generate dust particle deflection path data. The dust collection area is configured based on the dust particle deflection path data, and spatial aggregation and particle adsorption operations are performed to generate residual dust particle monitoring data. By combining residual dust particle monitoring data with historical control parameters, a feedback control vector is generated, and the control parameters are updated to complete adaptive adjustment.

2. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The set of stable airflow parameters includes: The system acquires wind speed, temperature, and pressure data within the duct, performs structured analysis and format unification processing, and generates duct airflow characteristic data. Normalize and remove disturbances from the airflow characteristic data of the duct to generate a stable set of airflow parameters; An airflow state boundary model is constructed based on a set of stable airflow parameters.

3. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The estimated dust particle movement path includes: Obtain the airflow state boundary model, set parameters based on the particle size range, and perform particle motion path estimation.

4. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The collision detection and stationary trajectory calibration process includes: Collision detection and stagnation trajectory calibration are performed on the dust particle movement path to generate candidate deposition paths.

5. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The calculated deposition probability distribution structure includes: The deposition probability distribution structure within the wind tunnel area is calculated based on candidate deposition paths.

6. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The electrostatic current conduction field control structure includes: Obtain the deposition probability distribution structure, extract high-risk deposition areas, and construct the spatial control region structure; Electric field boundary segmentation and target offset direction planning are performed on the structure of the spatial control area to generate the electrode control area; Based on the electrostatic current-conducting field control structure generated in the electrode control region, the control parameter matrix of the current-conducting electrode is calculated.

7. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The dust particle deflection action includes: Obtain the control parameter matrix of the current guiding electrode, and perform drive signal conversion and electrode command generation processing; The electrode command is sent to the flow control module to execute the programmable flow electrode control operation; The dust particle deflection action is executed according to the programmable flow guide electrode control operation, and dust particle deflection path data is generated.

8. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The generated residual dust particle monitoring data includes: Acquire dust particle deflection path data, configure dust collection target area, and schedule dust collection module; The spatial aggregation and particle adsorption operations of the dust collection module are executed to generate dust collection control result data. The dust collection control results data are processed to identify and count residual particles, generating residual dust particle monitoring data.

9. The ruggedized computer dust particle prediction and adaptive dust prevention method according to claim 1, characterized in that, The completion of adaptive adjustment includes: Acquire residual dust particle monitoring data, and generate a feedback control vector by combining the deposition probability distribution structure and the control parameter matrix; Update the control parameters in the electrostatic current conduction field control structure according to the feedback control vector; The updated control parameters are re-injected into the electrode instruction generation process to complete the adaptive parameter adjustment.

10. A ruggedized computer dust particle prediction and adaptive dust prevention system, applied to the ruggedized computer dust particle prediction and adaptive dust prevention method according to any one of claims 1-9, characterized in that, include: The airflow parameter acquisition module is used to acquire wind speed, temperature and pressure data and construct an airflow state boundary model; The dust particle path prediction module is used to estimate the dust particle movement path based on the airflow state boundary model and generate a deposition probability distribution structure. The electrostatic flow field generation module is used to obtain the deposition probability distribution structure, generate the electrostatic flow field control structure, and calculate the control parameter matrix. The flow guide electrode execution module is used to convert the control parameter matrix into flow guide electrode commands and execute the dust particle deflection operation; The dust collection and residue monitoring module is used to configure the dust collection area based on dust particle deflection path data and collect residual dust particle monitoring data. An adaptive update module is used to generate a feedback control vector based on the residual dust particle monitoring data and the control parameter matrix, and to complete the control parameter update process.

Images