Server airflow dynamic adjustment method and system based on thermal imaging data

Abstract

The present application relates to the technical field of dynamic air guiding adjustment, in particular to a server air flow dynamic adjustment method and system based on thermal imaging data, comprising the following steps: based on an infrared temperature sensor array, analyzing multi-angle temperature data, combining neighborhood temperature and air flow state analysis, identifying a space continuous hot spot area, adjusting air deflector deflection and direction, controlling air deflector bending, and obtaining a piece area linkage control result. Through interactive analysis and weight dynamic correction of multi-angle temperature data, the present application realizes automatic identification and replacement of temperature abnormal points, continuity analysis of space point sets promotes directional collection of hot spot aggregation areas, air flow distribution actively adjusts deflection and bending according to hot spot space distribution, adapts to a rapidly changing thermal environment inside a high-density server, strengthens orderly drainage of local heat, responds to variable thermal loads in different areas, and enhances self-adaptive capacity to complex air flow environments, so as to improve server local hot spot dissipation and noise distribution.

Description

Technical Field

[0001] This invention relates to the field of dynamic airflow regulation technology, and in particular to a method and system for dynamic airflow regulation of servers based on thermal imaging data. Background Technology

[0002] Dynamic airflow regulation refers to the technology of using various sensing, control, and execution technologies to adjust and optimize parameters such as airflow direction, flow rate, and velocity in real time according to changes in the environment or operating conditions within a target space, in an intelligent and programmable manner. It is widely used in scenarios such as heat dissipation of electronic equipment, building environment control, industrial process ventilation, and vehicle thermal management. In contrast, traditional server airflow regulation involves using fixed guide vanes or mechanical methods to adjust the posture of guide vanes. By setting the air duct structure and using fixed-point temperature sensors to collect the temperature of key components, when a temperature increase is detected, the mechanical device adjusts the guide vanes or guide vanes at a preset angle to guide part of the airflow to the higher-temperature area, thereby achieving the purpose of passive airflow distribution. This method usually relies on mechanical connections and predetermined strategies for airflow diversion and is difficult to adapt quickly to changes in thermal load under complex operating conditions.

[0003] Traditional adjustment methods are based on mechanical actions and preset rules. Airflow guidance mainly relies on static structures and fixed-point temperature acquisition. They cannot support multi-parameter linkage and coordinated adjustment of spatial partitions. When hot spots are widely distributed, the response is lagging, and local temperature anomalies are difficult to be addressed in a timely manner. This can easily lead to heat dissipation blind spots and noise accumulation. In the event of rapid thermal changes at multiple points, the adjustment of the air guide is not targeted. In scenarios with frequent changes in heat load, the overall operational stability and cooling uniformity of the server are significantly affected. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and propose a method and system for dynamic airflow adjustment of servers based on thermal imaging data.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a server airflow dynamic adjustment method based on thermal imaging data, comprising the following steps:

[0006] S1: Based on the infrared temperature sensor array, analyze multi-angle temperature data, compare the consistency of data in each spatial numbered area, determine the deviation of each temperature reading from the average trend, adjust the weight of abnormal measurements, identify measurement items with sudden trend changes, and obtain the abnormal judgment correction parameters.

[0007] S2: Based on the anomaly determination correction parameters, obtain the temperature of the six neighboring areas of each measurement point, calculate the difference between the spatial number and the temperature of the six neighboring areas, screen the key spatial points of temperature change amplitude, analyze the wind speed and wind direction change patterns between the neighboring areas, determine the airflow state deviation, and obtain the anomaly joint features.

[0008] S3: Based on the aforementioned abnormal joint features, identify areas with drastic temperature fluctuations and decreased or shifted airflow speed, determine the relative positions of three-dimensional disturbance points, analyze the distance and arrangement order of adjacent disturbance points, classify spatial continuous point groups, and obtain the hotspot region sequence distribution.

[0009] S4: Based on the distribution of the hot spot area sequence, determine the spatial vector direction of the aggregation area, compare the angle between the initial angle of the guide vane and the hot spot direction, analyze the area where the deflection angle exceeds the adjustment range, adjust the deflection direction of the guide vane, reset the guiding parameters, and obtain the spatial guiding deflection configuration.

[0010] The present invention is improved in that the anomaly determination correction parameters include correction coefficients, confidence labels, and preferred indexes; the anomaly joint features include temperature distribution patterns, flow field groupings, and feature labels; the hotspot region sequence distribution includes region serial numbers, path indexes, and continuity identifiers; and the spatial guidance deflection configuration includes deflection parameters, mapping relationships, and configuration numbers.

[0011] The present invention is improved in that the step of obtaining the anomaly determination correction parameter is specifically as follows:

[0012] S111: Based on an infrared temperature sensor array, analyze temperature data collected from multiple directions inside the server, compare the temperature performance of measuring points in different directions within the same area, determine the degree of deviation of measuring points in each direction from the average trend of the area, adjust the influence weight of data with abnormal fluctuations, and obtain multi-angle deviation amplitude indicators.

[0013] S112: Based on the multi-angle offset amplitude index, determine the measurement points that have abnormal offsets, and combine the measurement point change trajectories of multiple consecutive time periods to filter the measurement point data with abrupt change characteristics to obtain the trend abrupt change sampling identifier.

[0014] S113: Based on the trend change sampling identifier, analyze the correlation between the corresponding spatial number, time series and six neighboring measurement points, adjust the data reliability of the measurement points and the reference data, and integrate the correlation between the measurement points and the alternative data to obtain the anomaly judgment correction parameters.

[0015] The present invention is improved in that the step of obtaining the abnormal joint features is specifically as follows:

[0016] S211: Based on the anomaly determination correction parameters, analyze the temperature data of each measurement point and its six neighboring points, calculate the temperature difference between each group of data, optimize the average temperature difference between neighboring points, determine whether there is fluctuation in the temperature distribution of each spatial region, filter out abnormal distribution areas, and obtain the average temperature difference of the six neighboring regions.

[0017] S212: Based on the average temperature difference of the six neighboring regions, analyze the distribution of temperature difference in all measuring points in the current cycle, determine whether the temperature change amplitude exceeds the temperature difference baseline, identify the spatial area where the change occurs, and obtain the key measuring point group for temperature difference abrupt change.

[0018] S213: Based on the aforementioned key measurement point group for temperature difference abrupt change, analyze the airflow velocity and direction data of the six neighboring regions, calculate the impact of airflow velocity and direction changes, obtain the joint fluctuation diffusion amount, and obtain the abnormal joint characteristics.

[0019] The present invention is improved in that the step of obtaining the hotspot region sequence distribution is specifically as follows:

[0020] S311: Based on the aforementioned abnormal joint characteristics, analyze the temperature distribution and airflow direction parameters, determine the spatial points in the measurement area where the temperature fluctuates drastically and the airflow speed decreases or the direction deflects, optimize the point numbering, and establish their position mapping in the spatial grid to obtain spatial disturbance positioning data.

[0021] S312: Based on the spatial disturbance positioning data, calculate the three-dimensional distance between adjacent points, filter points that are continuously connected and have the same direction change, compare their arrangement order, identify the direction connection trajectory, and integrate the direction trend relationship by combining the direction angle and spatial number to establish the path direction structure.

[0022] S313: Based on the path direction structure, determine the direction consistency of each path segment, obtain the direction consistency evaluation result, identify the path numbers with continuous arrangement characteristics, and combine their arrangement order and direction characteristics to obtain the hot spot area sequence distribution.

[0023] The present invention is improved in that the step of obtaining the spatial guidance deflection configuration is specifically as follows:

[0024] S411: Based on the hotspot area sequence distribution, compare the consistency between the numbering order of each area and the path direction, determine the direction of the line connecting the center points of adjacent numbers in the spatial path, identify the point series with continuous direction and stable angle change, and obtain the set of spatial direction vector angles.

[0025] S412: Based on the set of spatial direction vector angles, compare it with the initial angle of the guide vane deflection component, calculate the angle between the dominant direction of the hot spot path and the current direction of the component, analyze the deviation of the regional direction, determine the correction requirement, and obtain the deflection adjustment range of the guide vane;

[0026] S413: Call the deflection adjustment range of the guide vane, analyze the angle difference between the target direction and the current direction, optimize the control parameters and direction mapping structure of the guide vane, determine the control index parameters, and obtain the spatial guidance deflection configuration.

[0027] The present invention is improved in that the steps further include:

[0028] S5: Based on the spatial guiding deflection configuration, analyze the temperature difference fluctuation and airflow velocity change in the corresponding area, determine the gradient distribution and airflow diffusion characteristics of the hot spot center, calculate the local response trend of the guide vane, execute pulse current, drive the guide vane to bend, and obtain the area linkage control result.

[0029] The results of the area linkage control include linkage commands, deformation information, and feedback indicators.

[0030] The present invention is improved in that the specific steps for obtaining the area linkage control results are as follows:

[0031] S511: Based on the spatial guidance deflection configuration, combined with the main direction of the hot spot area and the current state of the guide vane, determine the change between the airflow velocity and the direction of the guidance path in the area, screen the key features of the guide vane's alignment with the hot spot path, and obtain the airflow path offset amplitude.

[0032] S512: Based on the airflow path offset amplitude, analyze the airflow diffusion and temperature difference on the guide path, and combine the trend of the deflection direction of the guide vane to analyze the synchronous response trend of the guide vane to hot spot disturbance, so as to obtain the degree of guide response offset.

[0033] S513: Based on the degree of guide response offset, combined with the response characteristics and deformation conditions of the guide vane, a pulse current is executed to drive the guide vane to deform, and the deformation characteristics and feedback information of the guide vane in the linkage area are analyzed to obtain the linkage control result of the area.

[0034] A server airflow dynamic adjustment system based on thermal imaging data, the system comprising:

[0035] The anomaly identification module is based on an infrared temperature sensor array. It analyzes the collected multi-angle temperature data, compares the consistency between the angle data in each spatial number area, judges the deviation of each temperature reading from the average trend, adjusts the measurement weight of the identified deviation item, identifies the measurement item with a sudden trend change in the continuous period, and obtains the anomaly identification correction parameters.

[0036] Based on the anomaly determination correction parameters, the heat flow determination module obtains the temperature of the six neighboring areas of each measurement point, calculates the temperature difference between the spatial number and the six neighboring areas, filters the spatial points with key temperature change amplitudes among each measurement point, analyzes the variation law of wind speed and wind direction between the measurement point and the neighboring areas, judges the deviation of the airflow state of each group of measurement points, and obtains the anomaly joint characteristics.

[0037] Based on the aforementioned abnormal joint features, the hotspot tracking module identifies areas with drastic temperature fluctuations and decreased or shifted airflow speed, determines the relative positions of three-dimensional spatial disturbance points, analyzes the distance and arrangement order of adjacent disturbance points in space, classifies point groups with spatial continuity features, and obtains the hotspot region sequence distribution.

[0038] The deflection control module determines the spatial vector direction of the cluster area based on the hot spot area sequence distribution, compares the initial angle of the server guide vane deflection component with the angle of the hot spot area direction, analyzes the guide vane area where the deflection angle exceeds the allowable adjustment range, adjusts the deflection direction of the guide vane, and resets the guidance parameters to obtain the spatial guidance deflection configuration.

[0039] Based on the spatial guide deflection configuration, the linkage drive module analyzes the temperature difference fluctuation and airflow velocity change in the corresponding area, judges the gradient distribution and airflow diffusion characteristics at the center point of the hot spot area, calculates the local response trend of the guide vane, executes pulse current, drives the guide vane to bend, and obtains the linkage control result of the area.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0041] In this invention, through interactive analysis and dynamic weight correction of multi-angle temperature data, the automatic identification and replacement of temperature anomalies are achieved. Temperature and airflow multi-parameter characteristics are cross-correlated, and the continuity analysis of spatial point sets promotes the directional aggregation of hotspot clusters. Airflow distribution actively adjusts deflection and curvature according to the spatial distribution of hotspots to adapt to the rapidly changing thermal environment inside high-density servers, strengthens the orderly dissipation of local heat, responds to varying heat loads in zones, and enhances the adaptive capability to complex airflow environments. This enables real-time, intelligent, and efficient thermal management and adjustment to improve the dissipation of local hotspots and noise distribution in servers. Attached Figure Description

[0042] Figure 1 This is a flowchart of the main steps of the present invention;

[0043] Figure 2 This is a flowchart illustrating the process of obtaining the anomaly detection correction parameters in this invention.

[0044] Figure 3 This is a flowchart illustrating the acquisition of abnormal joint features in this invention;

[0045] Figure 4 This is a flowchart illustrating the process of obtaining the hotspot region sequence distribution in this invention.

[0046] Figure 5 This is a flowchart illustrating the process of obtaining the spatial guidance deflection configuration in this invention.

[0047] Figure 6 This is a flowchart of the process for obtaining the results of the area linkage control in this invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0049] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0050] Example

[0051] Please see Figure 1 This invention provides a technical solution: a method for dynamic airflow adjustment of a server based on thermal imaging data, comprising the following steps:

[0052] S1: Based on the infrared temperature sensor array, analyze the collected multi-angle temperature data, compare the consistency between the angle data in each spatial number area, determine the deviation of each temperature reading from the average trend, adjust the measurement weight of the determined deviation item, identify the measurement item with a sudden trend change in the continuous period, optimize the alternative data of the abnormal item, and obtain the abnormal judgment correction parameters.

[0053] S2: Based on the anomaly judgment correction parameters, obtain the temperature of the six neighborhoods of each measurement point, calculate the temperature difference between the spatial number and the six neighborhoods, screen the key spatial points with the temperature change amplitude in each measurement point, analyze the change law of wind speed and wind direction between the measurement point and the neighborhood, judge the deviation of the airflow state of each group of measurement points, classify the temperature and airflow data of each group of spatial points, and obtain the joint anomaly characteristics.

[0054] S3: Based on the joint characteristics of anomalies, identify areas with drastic temperature fluctuations and decreased or shifted airflow speed, determine the relative positions of three-dimensional disturbance points in space, analyze the distance and arrangement order of adjacent disturbance points in space, classify point groups with spatial continuity characteristics, and integrate them into directional paths to obtain the hot spot region sequence distribution.

[0055] S4: Based on the hotspot area sequence distribution, determine the spatial vector direction of the cluster area, compare the initial angle of the server guide vane deflection component with the angle of the hotspot area direction, analyze the guide vane area where the deflection angle exceeds the allowable adjustment range, adjust the deflection direction of the guide vane, and reset the guidance parameters to obtain the spatial guidance deflection configuration.

[0056] S5: Based on the spatial guide deflection configuration, analyze the temperature difference fluctuation and airflow velocity change in the corresponding area, determine the gradient distribution and airflow diffusion characteristics at the center point of the hot spot area, calculate the local response trend of the guide vane, execute pulse current, drive the guide vane to bend, and obtain the area linkage control result.

[0057] The anomaly judgment correction parameters include correction coefficients, confidence labels, and preferred indexes; the anomaly joint features include temperature distribution patterns, flow field grouping, and feature labels; the hotspot region sequence distribution includes region number, path index, and continuity identifier; the spatial guidance deflection configuration includes deflection parameters, mapping relationships, and configuration numbers; and the area linkage control results include linkage commands, deformation information, and feedback identifiers.

[0058] In S1, multi-angle temperature data refers to temperature data collected simultaneously from multiple angles (or orientations) within the same area of ​​the server by an array of infrared temperature sensors deployed in different locations inside the server (such as around core heat-generating components like the CPU, GPU, and memory). This helps overcome blind spots or errors that occur with single-point temperature measurement. Spatial numbered areas refer to the spatial grids or partitions within the server chassis, divided according to physical space, with each acquisition area or monitoring unit assigned a unique number. Each set of temperature data belongs to a unique spatial number, facilitating data aggregation, management, and analysis. Deviation refers to the numerical difference between a measured data point and the average or expected trend of all temperature measurements within that spatially numbered area. Used to measure the outlierness of data; measurement weight refers to the numerical factor assigned based on the reliability and stability of the data when aggregating or judging multiple sets of temperature data in the same area. Data with higher weights have a greater impact on the final judgment result, and the weight of measurement items with large deviations will be reduced or corrected; trend change refers to the phenomenon that the rate or direction of temperature change of the same monitoring point or measurement item suddenly changes within a continuous monitoring period (time series), which usually indicates that the environment has changed rapidly or the measurement is abnormal; alternative data for outliers refers to the original temperature data that has been judged to be abnormal or unreliable, which needs to be replaced and corrected with reliable data from other sources (such as the average of nearby sensors, historical trends) to ensure the continuity and credibility of the data.

[0059] In S2, the six-neighbor temperature refers to the temperature data of six grid points directly adjacent to a measurement point in the three-dimensional coordinate system in six directions (up, down, left, right, front, and back). It is a commonly used neighborhood definition in spatial analysis. Temperature difference refers to the numerical difference between the temperature of a measurement point and its six neighboring points, used to analyze the uniformity and intensity of local temperature distribution. The spatial point with critical temperature change amplitude refers to the spatial location where the temperature change amplitude is significant among all measurement points and has a large impact on the overall distribution. It is usually a hot spot or a cold spot and needs to be monitored closely. The deviation of airflow state refers to the significant change in the direction or velocity of airflow in space. Combined with the analysis of temperature distribution, it can reveal the influence of airflow disturbance on heat distribution.

[0060] In S3, the relative position of the perturbation points refers to the spatial distance and relationship (such as distance and orientation) between points that are identified as having drastic temperature fluctuations or sudden changes in airflow speed, which is used for subsequent spatial clustering and region grouping. The point group with spatial continuity refers to a set of perturbation points that are adjacent in space and have a certain continuity (such as being arranged in patches, lines or clusters), representing the actual distribution of thermal and flow field perturbations in space.

[0061] In S4, the spatial vector direction refers to the dominant direction of the hot spot area or cluster in three-dimensional space. It is usually represented by a vector method through the geometric distribution trend of hot spots in space, which facilitates airflow guidance control. The deflection component specifically refers to the guide vane mechanism in the server used to guide and adjust the airflow direction (such as a dynamic controllable guide vane made of SMA shape memory alloy material), which can deflect or deform according to the control signal.

[0062] In S5, temperature fluctuation refers to the degree of temperature change and its fluctuation amplitude in the hot spot area during the monitoring period, used to judge changes in heat load and local cooling requirements; gradient distribution refers to the rate distribution of temperature change with distance in a spatial region (temperature field gradient), used to analyze the direction of heat diffusion and transfer in space; airflow diffusion characteristics refer to parameters such as the uniformity of airflow distribution, diffusion speed and diffusion range in space, which affect heat removal and heat dissipation efficiency; response trend guides the real-time response tendency and state of the wafer to local heat load based on the detected temperature and airflow changes (such as deflection amplitude, bending speed).

[0063] The flow guide is an SMA (shape memory alloy) flow guide with an ultra-thin wedge structure (thickness 0.3–0.8 mm) that achieves bending deformation under a 5V / 100–500ms pulse current.

[0064] Parameters: Temperature gradient threshold |----|SMA response time |----|Infrared sampling rate;

[0065] Limited range: 10–20℃ / cm |----|100–500ms |----|≥50Hz;

[0066] To avoid ambiguity: Threshold below 10℃ leads to over-adjustment |----| >500ms cannot suppress transient thermal fluctuations |----| <50Hz results in temperature field reconstruction distortion

[0067] Please see Figure 2 The specific steps for obtaining the anomaly detection correction parameters are as follows:

[0068] S111: Based on an infrared temperature sensor array, analyze temperature data collected from multiple directions inside the server, compare the temperature performance of measuring points in different directions within the same area, determine the degree of deviation of measuring points in each direction from the average trend of the area, adjust the influence weight of data with abnormal fluctuations, and obtain multi-angle deviation amplitude indicators.

[0069] First, multiple key location points are set inside the server, such as infrared sensors deployed above the CPU, on the side of the GPU, and near the memory slots. Each sensor is assigned to a corresponding spatial region, such as A1, B3, C2, etc. The sensors collect temperature information at a unified time point and categorize the temperature values ​​collected from different directions within each spatial region. For example, region A1 is covered by measurement points from the front, back, left, and right directions. The collected data is retrieved uniformly within a single time acquisition cycle. After comparing these four data points, the average temperature level within the region is calculated. The difference between the actual temperature at each measurement point and the average level is evaluated. If a certain direction... If the data is significantly higher or lower than the average, for example, by more than 1.5 degrees Celsius, it is considered to have deviated. Then, the weight value of the measurement point is adjusted according to the degree of deviation. If the weight originally used for comprehensive calculation is 1, it is reduced to 0.5 to reduce its influence in the analysis. This adjustment process is performed one by one in all spatial numbered regions. For example, the temperature in the CPU region generally fluctuates around 35 degrees Celsius, while the reading in a certain direction in the GPU region shows an abnormal value of 39 degrees Celsius. In this case, the weight of the data in that direction will be weakened and its difference and azimuth information will be recorded. After this process, each spatial region will generate a set of data structures containing the deviation of data in each direction, azimuth labels, and adjusted weights as the result.

[0070] S112: Based on the multi-angle offset amplitude index, identify the measurement points that have abnormal offsets, and combine the measurement point change trajectories of multiple consecutive time periods to filter the measurement point data with abrupt change characteristics and obtain the trend abrupt change sampling identifier.

[0071] For measurement points marked as having abnormal offsets, time series backtracking is performed. Temperature records from multiple consecutive time periods are retrieved sequentially, and temperature change trajectories are plotted. For a given measurement point, if its temperature fluctuates relatively little over multiple periods but suddenly rises or falls sharply in a certain period, it can be determined that it has a sudden change. In this case, the point is marked as a trend abrupt change point within that time period. For example, in the past 5 periods, the temperature of the measurement point on the left side of the GPU region gradually rose from 36 degrees Celsius to 37 degrees Celsius, a slow change, but suddenly rose to 41 degrees Celsius in the 6th period. This change will be marked as a sudden change. In addition, the temperature change trajectories of its adjacent sensors are also used for verification. If most neighboring measurement points do not show similar changes, the abrupt change determination is more reliable. This judgment process will label the abrupt change according to its degree of change. For example, a measurement point in the 6th period is marked as a sudden change sampling point. At the same time, it will be statistically analyzed whether it has multiple abrupt changes in the entire time series, forming a set of labels for abrupt change points for subsequent processing.

[0072] S113: Based on the trend change sampling identifier, analyze the correlation between the corresponding spatial number, time series and six neighboring measurement points, adjust the data reliability of the measurement points and the reference data, and integrate the correlation between the measurement points and the alternative data to obtain the anomaly judgment correction parameters;

[0073] For each abrupt change point, spatial correlation analysis is performed. First, the spatial number to which the point belongs is extracted, and the temperature measurement information of the six adjacent directions of the point is retrieved. The temperature changes of the neighboring points in the same time period are analyzed. If the temperatures of multiple neighboring points remain stable or fluctuate slightly during the abrupt change period, while the abrupt change point changes drastically, it indicates that there is a problem with the reliability of its measurement results. Then, the reliability label of the abrupt change point is downgraded, for example, its original reliability label is adjusted from high to medium or low, and the neighboring point with the closest historical change trend is found as a reference point. For example, if the measurement point directly below has the most consistent temperature change pattern with the abrupt change point in all cycles, the current cycle temperature of the measurement point is selected as the substitute data value of the abrupt change point, and the mapping relationship between the two is recorded in the data structure. At the same time, the original abrupt change point number, the adjusted reliability level, the selected substitute temperature value, and the associated reference point label are retained. A dataset is output for subsequent analysis and adjustment of control actions to support higher quality airflow regulation decisions.

[0074] Please see Figure 3 The specific steps for obtaining joint anomaly features are as follows:

[0075] S211: Based on the anomaly detection correction parameters, analyze the temperature data of each measurement point and its six neighboring points, calculate the temperature difference between each set of data, optimize the average temperature difference between neighboring points, determine whether there is fluctuation in the temperature distribution of each spatial region, filter out abnormal distribution areas, and obtain the average temperature difference of the six neighboring regions.

[0076] The system acquires the current temperature data from each measuring point inside the server chassis and simultaneously reads the temperature values ​​of its six spatially adjacent measuring points. For example, measuring point P0 is located at the center of the CPU area, and its six neighboring points are located above, below, left, right, front, and back, numbered P1 to P6. If the current temperature of P0 is 42.1℃, and its neighboring points are 41.6℃, 42.5℃, 43.2℃, 41.0℃, 40.9℃, and 42.3℃ respectively, the temperature difference between P0 and each neighboring point is calculated, resulting in six sets of difference values. These six difference values ​​are then summed and divided by six to obtain the average temperature difference in the neighborhood of that measuring point. If this average value exceeds 2.0℃, it is considered that there is a local temperature fluctuation at that measuring point. The benchmark value of 2.0℃ is set based on the regional temperature difference distribution pattern under stable operating conditions in long-term sampling data. If the temperature difference among all measuring points exceeds 2.0℃, the system will determine the average temperature difference. If the average temperature difference is below 1.5℃ for more than 80% of the measurements, then 2.0℃ is set as the threshold for judging the deviation. This difference calculation is performed on each measuring point sequentially and compared with the average temperature difference benchmark. If the average temperature difference of two or more consecutive measuring points in any spatial number region is higher than 3.0℃, then the region is marked as a local thermal disturbance region. The historical average temperature values ​​of the neighboring measuring points in the marked region are further collected and compared with the current period. If the deviation exceeds the set temperature change trigger threshold of 4.0℃, then the region is identified as an abnormal distribution and recorded as an abnormal point set. In this process, the data call, difference calculation and averaging between each measuring point and its neighboring region are all based on the real-time temperature sampling data stream. The processed result is a set of average temperature differences for each spatial region, which is used to characterize the temperature uniformity fluctuation characteristics.

[0077] S212: Based on the average temperature difference of the six neighboring regions, analyze the distribution of temperature difference among all measuring points in the current cycle, determine whether the temperature change exceeds the temperature difference baseline, identify the spatial area where the change occurs, and obtain the key measuring point group for temperature difference abrupt change.

[0078] Based on the average temperature difference across the six neighboring regions obtained above, a global analysis is performed on all server monitoring points within the current time period. The average difference value of each monitoring point is compared with a set temperature difference baseline value. The baseline value is set at 1.8℃ based on temperature distribution statistics, as this value falls within the upper limit of the stable fluctuation zone in historical data. If the average temperature difference of a monitoring point exceeds 3.0℃ in the current period, and the temperature difference increase in the area where that monitoring point is located exceeds 1.2℃ compared to the previous period, then that point is considered a critical change point. This process is repeated for each monitoring point, and all points meeting the above two conditions are considered critical change points. The measurement points of the device are included in the mutation candidate set. Then, the spatial numbers of each measurement point in the mutation candidate set are aggregated to determine whether they constitute a continuous regional distribution. For example, if measurement points numbered A3, A4, A5, B3, and B4 in the CPU region all meet the mutation criteria and are arranged in an adjacent structure, then the region is treated as a key mutation region group, and its measurement point number, location, number of neighbors, and mutation magnitude level are recorded. Through this process, the key measurement point group of temperature difference mutation is screened and output, forming a set of measurement points that are spatially continuous and significantly deviate from the baseline in terms of temperature change, which is used for subsequent wind direction and wind speed field analysis and adjustment triggering.

[0079] S213: Based on a key measurement point group for sudden temperature changes, analyze the airflow velocity and direction data in a six-neighbor area, calculate the impact of changes in airflow velocity and direction, and use the following formula:

[0080]

[0081] By obtaining the joint wave diffusion, the joint anomaly characteristics are obtained, where, v represents the joint wave diffusion at the i-th measurement point. i v represents the airflow velocity at the i-th measuring point. j θ represents the airflow velocity at point j, which is one of the six adjacent points of the i-th measuring point. i θ represents the airflow direction at the i-th measuring point. j T represents the airflow direction at point j, which is one of the six adjacent points of the i-th measuring point. i This represents the temperature at the i-th measuring point. This represents the average temperature of the six adjacent points of the i-th measuring point.

[0082] The joint wave diffusion is a multivariate spatial wave intensity that each measurement point exhibits relative to its six neighboring points in the server's spatial grid, across multiple physical quantities such as airflow velocity, airflow direction, and temperature. This quantity reflects the overall intensity of change of the measurement point and its surrounding neighborhood in three key indicators (airflow velocity, airflow direction, and temperature). By statistically analyzing the joint wave diffusion of each measurement point, the most representative anomaly clusters in the space can be located.

[0083] The system retrieves airflow velocity, direction, and temperature data from six neighboring points within a spatially numbered grid for each measuring point. It then calculates the changes in these three physical dimensions between the measuring point and its six neighbors. When analyzing airflow velocity differences, the system extracts the velocity v at the measuring point. i =1.2 m / s, the six-neighbor velocity data is:

[0084] [1.5,1.3,1.1,1.4,1.2,1.3]m / s;

[0085] The corresponding normalized result is:

[0086] [0.75, 0.81, 0.69, 0.88, 0.75, 0.81];

[0087] The wind speed difference is averaged after processing the square of the deviation for each group, and then the wind direction angle difference is processed. Let the direction of the measuring point be θ. i =45°, neighborhood direction is:

[0088] [50,55,60,65,70,75]°;

[0089] After normalization, it becomes:

[0090] [0.67, 0.73, 0.80, 0.87, 0.93, 1.00];

[0091] Calculate the squared mean of the directional differences, then extract the temperature parameter, with the temperature at the measuring point being T. i =35.2℃, neighborhood temperature is:

[0092] [36.0, 35.5, 34.9, 36.2, 35.8, 35.3]℃;

[0093] Neighborhood average temperature After normalization, they are respectively T i =0.62, Substituting the above data into the calculation formula, the sum of squares of wind speed difference term is:

[0094]

[0095] The sum of squares of wind direction difference is:

[0096]

[0097] The square term of the temperature difference is:

[0098]

[0099] Substituting into the formula, we get:

[0100]

[0101] The joint wave diffusion at this measuring point is obtained as follows: This value is compared with the set joint fluctuation baseline (e.g., 0.25). Since it is higher than the reference range, it indicates that the point is in a region where multiple physical quantity disturbances are concentrated, which meets the criteria for abnormal measurement points. Therefore, the dataset of this point is classified into the category of abnormal joint features for subsequent spatial disturbance trend identification and guidance control calculation. The formula achieves the superposition and quantification of cross-dimensional multi-field data by introducing the equalization processing of three normalized parameters: wind speed, wind direction, and temperature. This allows the local comprehensive fluctuation effect to be accurately captured under the condition of no unit interference.

[0102] Please see Figure 4 The specific steps for obtaining the hotspot region sequence distribution are as follows:

[0103] S311: Based on the joint characteristics of anomalies, analyze the temperature distribution and airflow direction parameters, identify spatial points in the measurement area where the temperature fluctuates drastically and the airflow speed decreases or the direction deflects, optimize the point numbering, and establish their position mapping in the spatial grid to obtain spatial disturbance location data.

[0104] First, the temperature distribution data within each spatially numbered area is compared point by point. The temperature fluctuation range of each measuring point in the current cycle is retrieved and compared with the temperature change in the previous cycle. Measuring points with a change exceeding 3℃ are marked as high fluctuation points. At the same time, the wind speed and wind direction data of the same measuring point in the corresponding cycle are retrieved. If the wind speed decreases by more than 20%, or the wind direction deflects by more than 15 degrees, the measuring point is labeled as an airflow disturbance. For example, measuring point C2 had a wind speed of 2.5 m / s in the previous cycle and 1.9 m / s in the current cycle, a decrease of 24%. At the same time, its wind direction changed from directly in front to 45 degrees to the right front. It meets the conditions of decreasing airflow speed and changing direction and is marked as a disturbance point. Its number is an item in the set of disturbance point numbers. Then, the position of this type of disturbance point in the server's three-dimensional spatial grid is mapped. After extracting the X, Y, and Z coordinates of each measurement point, a one-to-one correspondence is made with the spatial number to construct a spatial index table. Further checks are made for duplicate numbers, discontinuous numbers, or skips. If the numbering is found to be disordered, for example, the numbering of the same area skips the number B3.5 between B3 and B4, it is optimized to be consecutive numbered, and the spatial index information of the point is updated to form spatial disturbance location data. This data records the positional relationship, disturbance feature label, and number mapping of all spatial points marked as having severe temperature fluctuations and significant airflow disturbances.

[0105] S312: Based on spatial disturbance positioning data, calculate the three-dimensional distance between adjacent points, filter points that are continuously connected and have the same direction change, compare their arrangement order, identify the direction connection trajectory, and integrate the direction trend relationship by combining the direction angle and spatial number to establish the path direction structure.

[0106] Based on spatial disturbance location data, the coordinate information of all marked disturbance points in three-dimensional space is extracted. Three-dimensional distance calculation is performed on each pair of adjacent disturbance points. By calculating the point-to-point distance between all disturbance points one by one, it is determined whether the distance between point pairs is less than a spatial connection threshold of 5 cm. This value is set based on the minimum spacing standard of the server's internal component layout. Among point pairs that meet the connection threshold, connection paths with consistent directional change trends are selected. The judgment method is to compare the spatial displacement direction between every two points. If the directional angle change between multiple connection points is within 10 degrees, it is determined to be a path in the same direction. Continuous paths are further... Points are arranged from the starting point to the ending point according to their spatial location. For example, if the points are numbered A1-A2-A3 consecutively and have the same spatial direction, they are recorded as a valid trajectory. Then, the direction angle information between each point in each trajectory is extracted and combined with its spatial number to generate a direction angle-number key-value structure. For example, if a path is from number B2 to B3 with a direction angle of 30 degrees, it is recorded as B2_30, B3_30, and so on. Finally, the key-value sets of multiple trajectories with the same direction are merged to generate a path direction structure dataset. Each data item in the structure includes the starting point number, ending point number, direction angle label, and number of consecutive points.

[0107] S313: Based on the path direction structure, determine the consistency of direction among each path segment using the following formula:

[0108]

[0109] Obtain the directional consistency assessment result RD s By identifying path numbers with consecutive arrangement characteristics and combining their arrangement order and direction characteristics, the hotspot region sequence distribution is obtained, where n RD Dθ represents the total number of disturbance points in the current disturbance path. z This represents the directional angle at the z-th disturbance point, used to describe the dominant direction of airflow or heat propagation at that point. Qd represents the average of the directional angles of all disturbance points along the disturbance path, used to construct a directional deviation benchmark. z Ds represents the three-dimensional spatial distance between the z-th perturbation point and its predecessor, used to reflect the influence of path segment length on directional changes. z Dv represents the perturbation intensity of the z-th perturbation point in the six-neighbor direction, used to measure the local temperature fluctuation at that point. z This represents the change in airflow direction at the z-th disturbance point, used to measure the instability of the airflow near that point.

[0110] The directional consistency assessment result refers to the degree of deviation between the dominant directional angle of each disturbance point and the overall average directional angle in a spatial path formed by multiple disturbance points. It is a result of comprehensive calculation based on the distance between points, disturbance intensity, and airflow changes. The result reflects the degree of consistency of directional changes within the path. It is used to identify which path segments in space have strong directional coherence, thereby assisting in locating hotspot area sequences and guiding the dynamic adjustment of server airflow.

[0111] First, extract the direction angle Dθ of the disturbance point. z 3D distance Qd z Disturbance intensity Ds z and airflow change Dv z For each disturbance point along the path, calculate the average angle between the direction of each point and the direction of the path. The absolute deviation between points is multiplied by the three-dimensional spatial distance from the point to the previous disturbance point to form the cumulative term in the numerator. Simultaneously, the sum of the squares of the disturbance intensity and the airflow change at each disturbance point is calculated to form the sum of squares in the denominator. Five disturbance points are selected for calculation, with directional angles of 35°, 38°, 40°, 36°, and 37°, and three-dimensional distances of 4.5cm, 4.8cm, 5.0cm, 4.3cm, and 4.6cm, corresponding to disturbance intensities of 12℃, 14℃, 13℃, 15℃, and 14℃, and airflow changes of 0.7m / s, 0.8m / s, 0.6m / s, 0.9m / s, and 0.7m / s. The directional angles are averaged to obtain... Perform for each disturbance point The calculations yielded the following results:

[0112] The first point is 2.2 × 4.5 = 9.9;

[0113] The second point is 0.8 × 4.8 = 3.84;

[0114] The third point is 2.8 × 5.0 = 14.0;

[0115] Point 4 is 1.2 × 4.3 = 5.16;

[0116] The fifth point is 0.2 × 4.6 = 0.92;

[0117] The summation of the numerators gives 33.82;

[0118] Subsequently, after normalizing the disturbance intensity and airflow change, the corresponding normalized data are as follows:

[0119] Point 1 Ds z =0.48, Dv z =0.28;

[0120] Point 2 Dsz =0.56, Dv z =0.32;

[0121] Point 3 Ds z =0.52, Dv z =0.24;

[0122] Point 4 Ds z =0.60, Dv z =0.36;

[0123] Point 5 Ds z =0.56, Dv z =0.28;

[0124] Substituting the denominators into the formula and performing the sum of squares, we get:

[0125] The first point is 0.48. 2 +0.28 2 =0.3088;

[0126] The second point is 0.56. 2 +0.32 2 =0.4160;

[0127] Point 3 is 0.52 2 +0.24 2 =0.3280;

[0128] Point 4 is 0.60 2 +0.36 2 =0.4896;

[0129] Point 5 is 0.56 2 +0.28 2 =0.3920;

[0130] The sum is 1.9344. Taking the square root of this value gives the denominator as... Substitute into the formula:

[0131]

[0132] Directional Consistency Assessment Result RD s The calculated result is 24.32, which measures the degree of consistency in the orientation of disturbance points within the current path, where RD s A larger value indicates a significant deviation in the direction of disturbance within the path. This result, combined with the path number sequence and spatial orientation information, records path segment numbers as items that do not meet the direction consistency criterion and excludes them from the subsequent construction of the hotspot region sequence. The calculated RD... sPath segments with values ​​below 10 are categorized into a set of paths with consistent direction, and their numbers and order of arrangement are extracted to generate a hotspot region sequence distribution. The formula introduces the dual influence factors of disturbance intensity and airflow disturbance in the calculation of directional deviation, which can accurately reflect the degree of flow-thermal coupling under the background of directional fluctuation, thereby realizing the application of joint control of directional characteristics and disturbance characteristics in path distribution identification.

[0133] Please see Figure 5 The specific steps for obtaining the spatial guidance deflection configuration are as follows:

[0134] S411: Based on the hotspot area sequence distribution, compare the consistency between the numbering order of each area and the path direction, determine the direction of the line connecting the center points of adjacent numbers in the spatial path, identify the point series with continuous direction and stable angle change, and obtain the set of spatial direction vector angles.

[0135] Extract the spatial ID and path direction identifier corresponding to each hotspot region in the sequence. Sort the IDs according to the order of their appearance in the sequence. For example, if a hotspot sequence is IDs B1, B2, B3, B5, B6, then determine if there is a skipped number between B3 and B5, record the skipped number, and check if it is caused by a non-continuous path. Next, extract the three-dimensional position coordinate data of the center point coordinates between consecutive IDs. Calculate the direction vector for any two adjacent center point coordinates and determine if their spatial directions are consistent. By comparing the angle difference between the directions of all center point lines in the current path, classify the angle changes. The set angle variation range is 0 to 15 degrees. This setting is based on the allowable range of physical angle changes in the duct guidance response. For example, if the angle between B1 and B2 is 5 degrees, between B2 and B3 is 8 degrees, between B3 and B5 is 18 degrees, and between B5 and B6 is 10 degrees, then B1 to B3 can be considered to form a continuous and stable change path. B3 to B5 is excluded because it exceeds the set angle variation limit. The same processing is performed on each continuous point series. All point series number segments that meet the angle stability requirements are recorded, and point series combinations are constructed according to their numbering order. The direction vector angles of the point series are extracted and classified, and summarized into a set of spatial direction vector angles.

[0136] S412: Based on the set of spatial direction vector angles, compare it with the initial angle of the guide vane deflection component, calculate the angle between the dominant direction of the hotspot path and the current direction of the component, analyze the regional direction deviation, and use the formula:

[0137]

[0138] Determine the correction requirement and obtain the deflection adjustment range ΔAθ′ of the guide vane, where ΔAθ represents the angle between the dominant direction of the hotspot path and the initial direction of the guide vane, used to measure the deviation between the target direction and the current direction. aσ represents the deviation of the dominant direction in the hotspot region, reflecting the intensity of the change in the dominant direction within the hotspot region. σ represents the density of disturbance points in three-dimensional space within the hotspot region; higher density indicates more concentrated disturbances. m The average index of a path represents the arithmetic mean of the index numbers of spatial points on a hot path, reflecting the complexity of the path. (AC) m This represents the average level of path continuity, and the average level of hotspot path continuity. The higher the value, the more coherent the path.

[0139] The deflection adjustment range of each guide vane is a set of adjustment angles that each guide vane needs to perform based on the deviation between the current airflow direction and the target airflow direction, taking into account various spatial disturbances and path characteristics. This is used to dynamically optimize the airflow distribution of the server.

[0140] Based on the set of spatial directional vector angles, and compared with the initial angle of the deflector assembly, it is necessary to calculate the angle between the dominant direction vector of each hotspot path and the current direction angle of the deflector. The angle is defined as the difference between the vectors, denoted as ΔAθ. For example, if the dominant direction of the hotspot path is 63.4° and the current initial direction of the deflector is 37.0°, then the angle ΔAθ = 26.4° is used to measure the degree of deviation of the airflow direction. Subsequently, it is necessary to analyze the fluctuation amplitude of the dominant direction in the path structure, which is characterized by the standard deviation of the change in the angle between multiple continuous vectors, and the deviation amplitude AD is set. a =12.7, then call the perturbation point density σ, which is obtained by converting the number of perturbation points per unit space. In this embodiment, it is 9.0. After normalization, σ = 0.78 is adopted. Then, call the path point number index sequence and take its arithmetic mean as AP. m =5.8, and simultaneously extract the average length of the longest consecutively numbered segment in this path segment as the average level of path continuity AC. m =3.4, after both are normalized to the same dimension, let AP m =0.66, AC m =0.59, substitute into the following formula for calculation:

[0141]

[0142] The calculation result is ΔAθ′≈17.75, which means that the corresponding guide vane needs to be deflected by 17.75° in the current state. This fully considers the directional angle, disturbance degree and path continuity. As the angle input item for the deflection adjustment command, it can be called in subsequent steps. By introducing the coordinated correction of disturbance density and path parameters, the formula can dynamically respond to the disturbance characteristics of different regions, strengthen the response logic of angle adjustment, and thus provide a calculable basis for spatial guidance deflection configuration.

[0143] S413: Call the deflection adjustment amplitude of the guide vane, analyze the angle difference between the target direction and the current direction, optimize the control parameters and direction mapping structure of the guide vane, determine the control index parameters, and obtain the spatial guidance deflection configuration.

[0144] First, extract the current deflection angle value from the current guide vane state and compare it with the target direction angle recorded in the direction to be adjusted to calculate the angle difference. For example, if a guide vane is currently facing forward at 0 degrees and the target direction is deflected to the right at 30 degrees, the angle difference is 30 degrees. Set the allowable deflection adjustment range to between 0 and 40 degrees. This range is determined based on the controllable deformation characteristics and response time curve of the SMA guide vane material. If the angle difference is within the allowable adjustment range, proceed to the control parameter optimization process. Extract historical adjustment data corresponding to the currently numbered guide vane from the configured control parameters and analyze the current pulse amplitude and duration used for similar angle adjustments in the past. For example... Historical records show that the guide vane uses 5V current for 300 milliseconds when deflected at 30 degrees. Therefore, the current configuration uses the same control parameters by default and constructs a mapping structure by combining the number and the direction angle. The form D12_30 indicates that the guide vane numbered D12 faces the 30-degree direction. Uniqueness verification is performed on this mapping structure to avoid conflicts caused by multiple guide vanes using the same target direction. If a conflict occurs, the mapping structure is reordered according to the priority of hot spot areas. Control index parameters are established for each guide vane. The index content includes the guide vane number, deflection angle, target direction, control voltage and duration. The output is used as the space guidance deflection configuration.

[0145] Please see Figure 6 The specific steps for obtaining the results of the area linkage control are as follows:

[0146] S511: Based on the spatial guidance deflection configuration, combined with the main direction of the hot spot area and the current state of the guide vane, the change between the airflow velocity and the direction of the guidance path in the area is determined, the key features of the guide vane's alignment with the hot spot path are selected, and the airflow path offset amplitude is obtained.

[0147] First, the direction vector of each segment in the hotspot area path is extracted and compared with the current deflection angle value of the corresponding guide vane to determine whether the guide vane positively covers the main direction of the hotspot path. The determination process is as follows: first, calculate the dominant direction angle between the start and end points in the hotspot path; then, extract the current deflection angle value of the guide vane corresponding to the spatial number of the hotspot area. For example, if the hotspot path direction is 30 degrees and the current deflection direction of the guide vane is 10 degrees, then the angle between the two is 20 degrees. Next, obtain the airflow velocity measured at the current measuring point on the path. If the angle between the airflow direction and the path direction exceeds 20 degrees, and the airflow velocity corresponding to the measuring point is less than 1.5 meters per second, then the measuring point is determined to be a guide mismatch point, and further screening is performed. All measuring points along the path with an angle greater than a set threshold are numbered and their numbers are collected. The consistency between the guide vane and the hot spot direction is evaluated for the measuring points. The alignment degree of each guide vane facing the hot spot path is calculated. An angle less than 10 degrees is considered highly aligned, an angle between 10 and 30 degrees is considered moderately aligned, and an angle greater than 30 degrees is marked as an offset point. At the same time, the airflow velocity value corresponding to each measuring point and the angle value with the hot spot direction are recorded to form a key feature vector group. By statistically analyzing the distribution of the angle values ​​and the trend of airflow velocity changes in the feature group, the airflow path offset amplitude of each hot spot area is generated. If the offset amplitude exceeds 30 degrees and is accompanied by an airflow velocity lower than the benchmark wind speed value of 1.8 meters per second, it is recorded as a severely offset area.

[0148] S512: Based on the airflow path offset amplitude, analyze the airflow diffusion and temperature difference on the guide path, and combine the trend of the guide vane deflection direction change to analyze the synchronous response trend of the guide vane to hot spot disturbance, and obtain the degree of guide response offset.

[0149] The current deflection angle, previous cycle adjustment trend direction, current cycle airflow velocity value, and temperature change trend along the path of each guide vane in all severely offset areas are extracted and correlated. First, the temperature change values ​​recorded at each measuring point along the guide path are statistically analyzed. The temperature difference between two adjacent cycles is extracted. If the number of points with a change above 3.5℃ accounts for more than 40% of the total path length, it is identified as a temperature fluctuation path. Simultaneously, the consistency of the deflection direction of the guide vanes along this path is analyzed. If more than 60% of the guide vanes have inconsistent deflection directions (i.e., the difference in direction angle between adjacent guide vanes exceeds 15 degrees), it is identified as inconsistent response direction. This is then combined with the correlation between the deflection angle of each point along the path and the temperature change trend along the path. Based on the change in airflow velocity, calculate whether the overall wind speed change along the path exceeds 0.8 meters per second. If the airflow velocity fluctuates greatly and the deflection direction is inconsistent, it indicates that there is a steering response delay along the path. By recording the numbers of all guide vanes along the path, the difference between their deflection direction and the deflection trend of the previous cycle, and the change in airflow velocity, a steering response offset record table is formed. By comparing the historical cycle data, it is determined whether the guide vanes can respond synchronously when hot spot disturbances occur. If the response lag time is greater than 500 milliseconds or the direction adjustment angle is less than 20 degrees, the guide vanes along the path are marked as having a lag response. The set of guide vane response degrees in each path is output to constitute the steering response offset degree.

[0150] S513: Based on the degree of guide response offset, combined with the response characteristics and deformation conditions of the guide vane, a pulse current is executed to drive the guide vane to deform, and the deformation characteristics and feedback information of the guide vane in the linkage area are analyzed to obtain the linkage control result of the area.

[0151] The system retrieves the target deflection angle value to be adjusted for each guide vane in the current cycle and its current angle difference to confirm whether deformation conditions are triggered. Deformation conditions are set as an angle difference exceeding 15 degrees and a current airflow velocity below 1.5 meters per second. For guide vanes meeting these conditions, a pulsed current drive is executed with parameters of 5 volts and a pulse duration of 300 milliseconds. The pulse signal is loaded onto the guide vane drive pin via the SMA control module, triggering a bending deformation of the guide vane structure and creating a new deflection state. After deformation, a temperature sensor and an infrared thermal imager monitor whether the temperature change in the hotspot area shows a decreasing trend. Simultaneously, the deflection angle and response of the guide vane before and after deformation are recorded. The deformation of the guide vane is considered effective if the temperature drops by more than 2°C and the wind speed increases by more than 0.5 m / s after deformation, based on the time taken and the local airflow velocity change value. The deformation results of multiple guide vanes in the linkage area are summarized, and the guide vane groups with a deformation time difference within 100 milliseconds are marked as linkage subsets. The number, deformation angle, airflow change and temperature drop trend of the subset are extracted to construct a deformation feature record table. At the same time, the drive completion status code and abnormal alarm value are read from the feedback signal. The area linkage control result is generated by combining the deformation completion status and feedback information. The result includes the linkage area number, deformation synchronization label, response time level and control success indicator.

[0152] A server airflow dynamic adjustment system based on thermal imaging data, the system includes:

[0153] The anomaly identification module is based on an infrared temperature sensor array. It analyzes the collected multi-angle temperature data, compares the consistency between the angle data in each spatial number area, judges the deviation of each temperature reading from the average trend, adjusts the measurement weight of the identified deviation item, identifies the measurement item with a sudden trend change in the continuous period, and obtains the anomaly identification correction parameters.

[0154] The heat flow determination module obtains the temperature of the six neighboring areas of each measurement point based on the anomaly determination correction parameters, calculates the temperature difference between the spatial number and the six neighboring areas, filters the spatial points with key temperature change amplitudes in each measurement point, analyzes the variation law of wind speed and wind direction between the measurement point and the neighboring areas, judges the deviation of the airflow state of each group of measurement points, and obtains the anomaly joint characteristics.

[0155] Based on the joint features of anomalies, the hotspot tracking module identifies areas with drastic temperature fluctuations and decreased or shifted airflow speed, determines the relative positions of three-dimensional disturbance points in space, analyzes the distance and arrangement order of adjacent disturbance points in space, classifies point groups with spatial continuity features, and obtains the hotspot region sequence distribution.

[0156] The deflection control module determines the spatial vector direction of the cluster area based on the hot spot area sequence distribution, compares the initial angle of the server guide vane deflection component with the angle of the hot spot area direction, analyzes the guide vane area where the deflection angle exceeds the allowable adjustment range, adjusts the deflection direction of the guide vane, and resets the guidance parameters to obtain the spatial guidance deflection configuration.

[0157] The linkage drive module is based on the spatial guide deflection configuration. It analyzes the temperature difference fluctuation and airflow speed change in the corresponding area, judges the gradient distribution and airflow diffusion characteristics at the center point of the hot spot area, calculates the local response trend of the guide vane, executes pulse current, drives the guide vane to bend, and obtains the linkage control result of the area.

[0158] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for server airflow dynamic adjustment based on thermographic data, characterized in that, The method comprises the following steps: S1: based on the infrared temperature sensor array, analyzing multi-angle temperature data, comparing the consistency of each spatial number region data, judging the deviation of each temperature reading from the average trend, adjusting the abnormal measurement weight, identifying the trend mutation measurement item, and obtaining the abnormal judgment correction parameter; S2: based on the abnormal judgment correction parameter, obtaining the six-neighborhood temperature of each measurement point, calculating the difference between the spatial number and the six-neighborhood temperature, screening the key spatial points of the temperature change amplitude, analyzing the change rule of the wind speed and direction between the neighborhoods, judging the airflow state deviation, and obtaining the abnormal joint feature; S3: according to the abnormal joint feature, identifying the region where the temperature fluctuates sharply and the airflow speed decreases or the direction deviates, judging the relative position of the three-dimensional disturbance point, analyzing the distance and arrangement order of the adjacent disturbance points, classifying the spatial continuous point group, and obtaining the hotspot region sequence distribution; S4: according to the hotspot region sequence distribution, judging the direction of the spatial vector of the aggregation area, comparing the included angle between the initial angle of the guide vane and the hotspot direction, analyzing the region where the deflection angle exceeds the adjustment range, adjusting the deflection direction of the guide vane, resetting the guide parameter, and obtaining the spatial guide deflection configuration; S5: based on the spatial guide deflection configuration, analyzing the temperature difference fluctuation and airflow speed change of the corresponding region, judging the gradient distribution of the hotspot center and the airflow diffusion characteristics, calculating the local response trend of the guide vane, executing the pulse current, driving the guide vane to bend, and obtaining the piece area linkage control result; The piece area linkage control result comprises linkage instructions, deformation information and feedback marks.

2. The method of claim 1, wherein the server airflow dynamic adjustment based on thermographic data is characterized by, The abnormal judgment correction parameter comprises correction coefficients, credibility labels and preferred indexes, the abnormal joint feature comprises temperature distribution patterns, flow field groups and feature labels, the hotspot region sequence distribution comprises region serial numbers, path indexes and continuity marks, and the spatial guide deflection configuration comprises deflection parameters, mapping relationships and configuration numbers.

3. The server airflow dynamic adjustment method based on thermal imaging data according to claim 1, wherein, The acquisition step of the abnormal judgment correction parameter is specifically: S111: based on the infrared temperature sensor array, analyzing the temperature data collected by multiple directions inside the server, comparing the temperature performance of the temperature difference direction measurement points in the same region, judging the deviation degree of each direction measurement point from the regional average trend, adjusting the data influence weight of the abnormal fluctuation, and obtaining the multi-angle deviation amplitude index; S112: based on the multi-angle deviation amplitude index, judging the measurement points with abnormal deviation, screening the measurement point data with mutation characteristics combined with the measurement point change trajectory in continuous multiple time periods, and obtaining the trend mutation sampling mark; S113: based on the trend mutation sampling mark, analyzing the correlation of the corresponding spatial number, time sequence and six-neighborhood measurement points, adjusting the data credibility and reference data of the measurement points, and integrating the correlation of the measurement points and the substitute data, and obtaining the abnormal judgment correction parameter.

4. The server airflow dynamic adjustment method based on thermal imaging data according to claim 1, wherein, The acquisition step of the abnormal joint feature is specifically: S211: based on the abnormal judgment correction parameter, analyzing the temperature data of each measurement point and its six adjacent points collected, calculating the temperature difference between each group of data, optimizing the average temperature difference between adjacent points, judging whether the temperature distribution of each spatial region exists fluctuation, screening the abnormal distribution region, and obtaining the six-neighborhood temperature difference average amount; S212: According to the average amount of temperature difference of the six-neighborhood, analyze the distribution of temperature difference in all measuring points in the current period, judge whether the temperature change amplitude exceeds the temperature difference baseline, identify the spatial area where the change occurs, and obtain the key measuring point group of temperature difference mutation; S213: Based on the key measuring point group of temperature difference mutation, analyze the airflow velocity and direction data of the six-neighborhood, calculate the influence of airflow velocity change and direction change, obtain the joint fluctuation diffusion amount, and obtain the abnormal joint feature.

5. The server airflow dynamic adjustment method based on thermal imaging data according to claim 1, wherein, The acquisition step of the hot area sequence distribution is specifically: S311: According to the abnormal joint feature, analyze the temperature distribution and airflow direction parameters, judge the spatial points in the measurement area where the temperature fluctuates violently and the airflow velocity decreases or the direction deflects, optimize the point number, and establish the position mapping in the spatial grid, to obtain the spatial disturbance positioning data; S312: According to the spatial disturbance positioning data, calculate the three-dimensional distance between adjacent points, filter the points that are continuously connected and have consistent direction changes, compare their arrangement order, identify the direction connection track, integrate the direction trend relationship through the combination of direction angle and spatial number, and establish the path direction structure; S313: According to the path direction structure, judge the direction consistency of each path segment, obtain the direction consistency evaluation result, identify the path number with continuous arrangement characteristics, and combine the arrangement order and strike characteristics to obtain the hot area sequence distribution.

6. The server airflow dynamic adjustment method based on thermal imaging data according to claim 1, wherein, The acquisition step of the space guide deflection configuration is specifically: S411: According to the hot area sequence distribution, compare the consistency of the region number order and the path direction, judge the direction of the center point connecting line of adjacent numbers in the spatial path, identify the point list with continuous direction and stable angle change, and obtain the spatial direction vector angle set; S412: Based on the spatial direction vector angle set, compare with the initial angle of the deflector deflection component, calculate the angle between the main direction of the hot path and the current direction of the component, analyze the region direction deviation amplitude, judge the correction requirement, and obtain the deflector deflection adjustment amplitude; S413: Call the deflector deflection adjustment amplitude, analyze the angle difference between the target direction and the current direction, optimize the deflector control parameter and direction mapping structure, determine the control index parameter, and obtain the space guide deflection configuration.

7. The server airflow dynamic adjustment method based on thermal imaging data according to claim 1, wherein, The acquisition step of the slice area linkage control result is specifically: S511: Based on the space guide deflection configuration, combine the main direction of the hot area and the current state of the deflector, judge the change between the airflow velocity and the guide path direction in the region, filter the key features of the deflector alignment to the hot path, and obtain the airflow path offset amplitude; S512: Based on the airflow path offset amplitude, analyze the airflow diffusion and temperature difference on the guide path, combine the deflector deflection direction change trend, analyze the synchronous response trend of the deflector to the hot disturbance, and obtain the guide response offset degree; S513: Based on the guide response offset degree, combine the response characteristics and deformation conditions of the deflector, execute pulse current to drive the deflector to deform, analyze the deformation characteristics and feedback information of the deflector in the linkage area, and obtain the slice area linkage control result.

8. Server airflow dynamic adjustment system based on thermographic data, characterized in that, The system is used for realizing the server airflow dynamic adjustment method based on thermal imaging data in any one of claims 1-7, and the system comprises: The anomaly identification module analyzes the collected multi-angle temperature data based on the infrared temperature sensor array, compares the consistency between the angle data in each spatially numbered region, judges the deviation amplitude between each temperature reading and the average trend, adjusts the measurement weight of the judged deviation item, identifies the measurement item with trend mutation in the continuous period, and obtains an anomaly judgment correction parameter; The heat flow judgment module obtains the six-neighborhood temperature of each measurement point based on the anomaly judgment correction parameter, calculates the temperature difference between the spatial number and the six-neighborhood, screens the spatial points with critical temperature change amplitude among the measurement points, analyzes the change law of the wind speed and direction between the measurement points and the neighborhood, judges the deviation of the airflow state of each group of measurement points, and obtains an anomaly joint feature; The hotspot tracking module identifies the region with violent temperature fluctuation and airflow speed drop or direction deviation according to the anomaly joint feature, judges the relative position of the spatial three-dimensional disturbance point, analyzes the distance and arrangement order of the adjacent disturbance points in the space, classifies the point group with spatial continuous feature, and obtains a hotspot region sequence distribution; The deflection control module judges the spatial vector direction of the aggregation area according to the hotspot region sequence distribution, compares the initial angle of the server deflector deflection component with the included angle of the hotspot region direction, analyzes the deflector area with deflection angle exceeding the allowed adjustment range, adjusts the deflection direction of the deflector, and resets the guiding parameter, to obtain a spatial guiding deflection configuration; The linkage driving module analyzes the temperature difference fluctuation and airflow speed change of the corresponding region based on the spatial guiding deflection configuration, judges the gradient distribution and airflow diffusion characteristics at the center point of the hotspot region, calculates the response trend of the local deflector, executes the pulse current, drives the deflector to bend, and obtains a sheet area linkage control result.

Images