Modularized data center detection method and system

Abstract

According to the modular data center detection method and system provided by the invention, the temperature sensor is set to be in a three-dimensional space collection mode to collect the historical temperature data, the historical temperature data is processed to generate the three-dimensional temperature field data, the three-dimensional temperature field data is calculated, the hot spot area is identified, and the detection accuracy is improved. And a regional temperature evolution model is constructed according to the regional temperature data of the hot spot region, and prediction and early warning are performed on the temperature of the preset future time point through the regional temperature evolution model and the current temperature data, so that accurate identification and prediction of the hot spot region are realized.

Description

TECHNICAL FIELD

[0001] The embodiment of the present application relates to the technical field of temperature monitoring, in particular to a modular data center detection method and system. BACKGROUND

[0002] As the core of modern information technology infrastructure, the stable operation of the data center is crucial to the digitalization of enterprises and society. With the continuous improvement of computing density, the temperature management problem caused by the concentrated release of heat from server cabinets has become increasingly prominent, becoming a key factor affecting the reliability of data centers.

[0003] The traditional data center temperature monitoring system generally has the problems of uneven distribution of monitoring points and incomplete data collection, and can only obtain local temperature information. Due to the diversity and dispersion of temperature data, it is difficult to establish a unified data model for management, which leads to the inability to effectively correlate and analyze temperature data, further hindering the accurate identification and prediction of hot spots. SUMMARY

[0004] Therefore, the purpose of the embodiment of the present application is to provide a modular data center detection method and system, which can realize temperature prediction of hot spots.

[0005] To achieve the above purpose, the embodiment of the present application provides a modular data center detection method, comprising: acquiring historical temperature data of a data center collected by a temperature sensor, the temperature sensor forming a three-dimensional space temperature collection network when being set at a corresponding equipment monitoring point of the data center; performing data processing on the collected historical temperature data and adding a three-dimensional space coordinate label to obtain a structured temperature data matrix; performing continuous processing on discrete temperature points in the structured temperature data matrix to generate three-dimensional temperature field data; calculating the three-dimensional temperature field data, and identifying a hot spot area according to the calculation result; performing time series analysis on the area temperature data of the hot spot area, and constructing a temperature evolution model according to the time series analysis result; acquiring a current time point and corresponding current temperature data collected by a target temperature sensor, inputting the current time point and the current temperature data into the temperature evolution model to obtain a temperature prediction value of a preset future time point; and generating high-temperature early warning information of a collection area corresponding to the target temperature sensor when the temperature prediction value exceeds a preset dangerous temperature threshold.

[0006] Further, acquiring historical temperature data of a data center collected by a temperature sensor comprises: receiving historical temperature data collected by the temperature sensor, the temperature sensor carrying a unique identification code and three-dimensional coordinate positioning information; if the change amplitude of the historical temperature data exceeds a preset threshold, increasing the sampling frequency of the corresponding temperature sensor, and marking the historical temperature data with a time stamp.

[0007] Further, the discrete temperature points in the structured temperature data matrix are processed to become continuous, generating three-dimensional temperature field data. This includes: acquiring the abnormal temperature values ​​and abnormal sensors corresponding to the discrete temperature points; compensating the abnormal temperature values ​​by using the average temperature of adjacent sensors to obtain a discrete temperature data matrix; processing the discrete temperature data matrix to become spatially continuous, obtaining temperature estimates; generating three-dimensional temperature field data based on the temperature estimates and the structured temperature data matrix, wherein the three-dimensional temperature field data is a gridded data structure; if the temperature of a grid point in the gridded data structure exceeds a preset temperature threshold, the corresponding grid point is marked as a hotspot region, and a three-dimensional visualization graphic is generated based on the three-dimensional temperature field data. Different temperature values ​​are mapped to a color spectrum range using color to obtain a complete three-dimensional temperature distribution map.

[0008] Further, the three-dimensional temperature field data is calculated, and hotspot regions are identified based on the calculation results, including: if the Euclidean distance between the temperature values ​​of any two monitoring points of the device in the three-dimensional temperature field data is less than a preset neighborhood radius and the temperature difference is within a preset range, they are grouped into the same region cluster, resulting in a set of region clusters; the average temperature value and standard deviation of the temperature values ​​within each region cluster are calculated based on the set of region clusters, and whether the region corresponding to the region cluster is a hotspot region is determined based on the average temperature value and the standard deviation; the peak temperature of the hotspot region is obtained, and if the peak temperature exceeds a preset danger threshold, a high priority is triggered, and the specific value of the peak temperature and the corresponding three-dimensional coordinate position are recorded; the duration is calculated based on the time series data of the hotspot region, and if the duration exceeds a preset duration threshold, a complete hotspot region file record is generated.

[0009] Furthermore, a time-series analysis is performed on the regional temperature data of the hotspot area, and a temperature evolution model is constructed based on the time-series analysis results. This includes: acquiring historical temperature records of each monitoring point within the hotspot area; if the time span of the temperature records at the monitoring points exceeds a preset minimum time threshold, arranging the regional temperature data in chronological order to obtain a time-series temperature dataset; calculating the temperature rise rate and acceleration value based on the time-series temperature dataset; and constructing the temperature evolution model based on the temperature rise rate and acceleration value.

[0010] Furthermore, the method further includes: acquiring hotspot temperature data and equipment operating parameters corresponding to the hotspot area collected by the temperature sensor; calculating the temperature gradient of each monitoring point of the equipment based on the hotspot temperature data and the equipment operating parameters; analyzing the hotspot temperature data to obtain the hotspot propagation path corresponding to the hotspot area and marking the corresponding coordinate position and temperature value to obtain heat propagation matrix data; calculating the diffusion rate and influence range of each high-temperature propagation path based on the heat propagation matrix data; acquiring the load history data and temperature anomaly records corresponding to the equipment; analyzing the boundary coordinates corresponding to the influence range and the correlation relationship of the equipment data obtained within the influence range based on the load history data and the temperature anomaly records to generate a correlation data table; constructing a response function based on the correlation data table, the response function being used to calculate the temperature value based on the load value of the equipment; and determining whether to generate a load adjustment command and send it to the corresponding equipment based on the change in the temperature value.

[0011] Furthermore, the method further includes: receiving real-time temperature data of the hotspot area collected by the temperature sensor, the real-time temperature data including temperature values ​​and timestamps; calculating the temperature change gradient and abnormal duration based on the real-time temperature data; calculating the current temperature rise rate based on the temperature change gradient and the abnormal duration, and generating evolution characteristic parameters based on the current temperature rise rate; acquiring device attributes and distance information of devices within a preset range surrounding the hotspot area; calculating a heat propagation attenuation coefficient based on the evolution characteristic parameters and the device attributes; calculating the propagation distance based on the heat propagation attenuation coefficient, and if the propagation distance is less than a safe distance, generating impact assessment data, the safe distance being determined based on the distance information; calculating the impact assessment data through a preset risk rating matrix to obtain a risk value and a corresponding risk level; triggering a graded early warning signal when the risk level exceeds a preset level, and assigning different processing priorities to the hotspot areas with different risk levels.

[0012] Furthermore, the method further includes: when the temperature value in the historical temperature data exceeds a preset temperature threshold range, generating temperature anomaly event data and forming an anomaly event database; calculating the support and confidence between temperature anomalies and equipment failure types in the anomaly event data; if the confidence and support meet preset rules, generating a set of association rules between the temperature anomaly and the equipment failure type; extracting temperature anomaly feature fingerprints from the set of association rules, encoding the temperature anomaly feature fingerprints, and constructing a temperature anomaly pattern library based on the feature fingerprint encoding and corresponding processing measures; calculating the similarity between the current temperature anomaly event feature fingerprint and the temperature anomaly feature fingerprints in the temperature anomaly pattern library; if the similarity exceeds a preset similarity value, retrieving the corresponding processing measures from the temperature anomaly pattern library and determining the corresponding processing measure scheme.

[0013] Furthermore, the method also includes: classifying the hotspot area into anomaly levels and generating a temperature change trend map; obtaining the equipment operating parameters and equipment load data corresponding to the equipment in the hotspot area; and drawing a heat migration path map based on the temperature change trend map, the equipment operating parameters, and the equipment load data.

[0014] To achieve the above objectives, this invention also provides a modular data center detection system, comprising: an acquisition module for acquiring historical temperature data of a data center collected by temperature sensors, wherein the temperature sensors form a three-dimensional spatial temperature acquisition network when set at the corresponding equipment monitoring points in the data center; a first data processing module for processing the acquired historical temperature data and adding three-dimensional spatial coordinate labels to obtain a structured temperature data matrix; a second data processing module for performing continuous processing on discrete temperature points in the structured temperature data matrix to generate three-dimensional temperature field data; a calculation module for calculating the three-dimensional temperature field data and identifying hotspot areas based on the calculation results; an analysis module for performing time-series analysis on the regional temperature data of the hotspot areas and constructing a temperature evolution model based on the time-series analysis results, wherein the temperature evolution model is used to predict the temperature value at a specific future time point; a prediction module for acquiring the current time point and corresponding current temperature data collected by a target temperature sensor, inputting the current time point and the current temperature data into the temperature evolution model to obtain a preset temperature prediction value for a future time point; and a generation module for generating high-temperature warning information for the area collected by the target temperature sensor when the predicted temperature value exceeds a preset danger temperature threshold.

[0015] The modular data center detection method and system provided in this invention sets the temperature sensor to a three-dimensional spatial acquisition mode to collect historical temperature data. The historical temperature data is processed to generate three-dimensional temperature field data. The three-dimensional temperature field data is calculated to identify hotspot areas. Then, a regional temperature evolution model is constructed based on the regional temperature data of the hotspot areas. The temperature at future time points is predicted and warned by the regional temperature evolution model and the current temperature data, thus realizing accurate identification and prediction of hotspot areas. Attached Figure Description

[0016] Figure 1 This is a flowchart of an embodiment of the modular data center detection method of the present invention; Figure 2 This is a schematic diagram of the program modules of Embodiment 2 of the modular data center detection system of the present invention. Detailed Implementation

[0017] 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. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0018] Example 1: See Figure 1 This document illustrates a flowchart of the modular data center detection method according to Embodiment 1 of the present invention. It is understood that the flowchart in this embodiment is not intended to limit the order of execution steps. The following description uses computer device 2 as the execution subject. Specifically: Step S100: Obtain historical temperature data of the data center collected by temperature sensors. When the temperature sensors are set at the corresponding equipment monitoring points in the data center, a three-dimensional spatial temperature acquisition network is formed.

[0019] In this embodiment, based on the airflow distribution pattern of the data center server room, equipment monitoring points are marked in the airflow convergence area and at locations with significant temperature gradient changes. Temperature sensors are installed in the temperature difference area of ​​the server rack inlet and outlet, the temperature fluctuation area at the junction of hot and cold aisles, and the area around the air conditioning outlet, forming a three-dimensional spatial temperature acquisition and monitoring network covering the data center. The temperature sensor nodes have unique identification codes and three-dimensional coordinate positioning information, which can more accurately locate when the temperature is abnormal.

[0020] For example, the data center used CFD software (general-purpose computational fluid dynamics software, such as ANSYS Fluent, cfx, STAR-CCM, cossol, OpenFOAM, Phoenics, etc.) to simulate airflow distribution maps, and found that a significant airflow convergence zone was formed at the junction of hot and cold aisles, with a temperature gradient change of 5℃ / meter. Based on this, equipment monitoring points were marked at the air inlets and outlets of the 3rd, 7th, and 12th layers of the rack. The temperature difference at these locations typically exceeds 15℃, allowing for better temperature monitoring. The temperature sensor deployment adopted a layered strategy, with 5 temperature sensors installed at the air inlets of a standard 42U rack (U is a unit of measurement, 1U=44.45mm): bottom, 1 / 4, middle, 3 / 4, and top; and 3 temperature sensors installed at the air outlets: bottom, middle, and top.

[0021] For example, step S100 further includes: receiving historical temperature data collected by a temperature sensor, the temperature sensor carrying a unique identification code and three-dimensional coordinate positioning information; if the change range of the historical temperature data exceeds a preset threshold, increasing the sampling frequency of the corresponding temperature sensor and timestamping the historical temperature data.

[0022] In this embodiment, each temperature sensor is equipped with a unique identification code "TSXX-YY-ZZ", where XX represents the rack number, YY represents the height, ZZ represents the air inlet / outlet indicator, and three-dimensional coordinate information. When a temperature fluctuation of more than 3°C is detected in a certain area within 60 seconds, the sampling frequency is automatically increased from the standard 5 minutes / time to 30 seconds / time. Based on the timestamp, the specific time period in which the temperature change occurred can be located.

[0023] Step S102: Process the collected historical temperature data and add three-dimensional spatial coordinate labels to obtain a structured temperature data matrix.

[0024] In this embodiment, a sliding window filter is used to remove noise interference from historical temperature data. If a data point deviates from the mean of adjacent data points by more than a preset deviation range, the data point is marked as abnormal data and cleaned temperature data is generated. A spatial location index table is constructed based on the three-dimensional coordinate information of the temperature sensor. For the missing data points in the cleaned temperature data, a bilinear interpolation algorithm is used to calculate the estimated temperature value at the missing location, obtaining the completed temperature data. The completed temperature data is standardized using a minimum-maximum normalization method, and a spatial coordinate label corresponding to the temperature sensor is assigned to each temperature data point, forming labeled data including temperature value and location information. The labeled data is arranged into a two-dimensional data matrix according to the spatial coordinate order using a row and column indexing method, resulting in a structured temperature data matrix. The data matrix elements store standardized temperature values ​​and spatial coordinate information.

[0025] For example, the data center uses a sliding window of length 7 to process the raw data collected by the temperature sensor TS01-15-IN. When a temperature value of 32.8℃ is detected at a certain moment, while the average value of the six adjacent data points is 24.2℃, the deviation reaches 8.6℃, exceeding the preset threshold of 3℃. This abnormal point is automatically marked and removed. This processing method can effectively eliminate data noise caused by momentary sensor failure or electromagnetic interference, ensuring the accuracy of subsequent analysis.

[0026] In this embodiment, the data center space is divided into 2m × 2m × 1.5m grid units, and each temperature sensor node is assigned a corresponding grid number based on its X, Y, and Z coordinates. When temperature sensor TS05-08-OUT experiences data loss due to hardware failure, a bilinear interpolation algorithm is invoked to estimate the temperature using temperature data from the four nearest surrounding sensors. The missing location coordinates are (15.2, 8.6, 4.5) meters. The data from the four nearest sensors are selected as 23.1℃, 24.8℃, 22.9℃, and 25.2℃, respectively. A weighted calculation yields an estimated temperature value of 24.0℃.

[0027] In this embodiment, minimum and maximum value normalization can eliminate measurement biases between different sensors. In a certain data processing, the original temperature data ranges from 18.5℃ to 31.2℃. The values ​​are mapped to the interval of 0 to 1 using a normalization formula, where the lowest temperature of 18.5℃ corresponds to a normalized value of 0, and the highest temperature of 31.2℃ corresponds to a normalized value of 1. Simultaneously, spatial coordinate labels are added to each temperature data point, forming a complete data record including the normalized temperature value of 0.67 and coordinate information of 12.5, 6.8, and 3.0.

[0028] In this embodiment, the two-dimensional data matrix is ​​constructed using a row-column index mapping method, dividing the data center into a 20×15 grid area, with each grid corresponding to an element position in the matrix. The temperature sensor data located at coordinates (10.4, 2.6, 2.1) meters is mapped to the 5th row and 6th column of the matrix. This element stores the normalized temperature value of 0.42 and the corresponding three-dimensional coordinate information, facilitating subsequent spatial temperature distribution analysis and visualization.

[0029] Specifically, the temperature gradient changes between adjacent elements in a structured temperature data matrix can reflect the distribution characteristics of local hotspots. When the temperature value of the element in the 8th row and 12th column of the matrix is ​​0.85, while the temperature values ​​of adjacent elements are all below 0.60, it indicates the presence of a significant hotspot phenomenon in that area. Matrix operations can quickly identify temperature anomaly regions, i.e., hotspot regions, providing data support for subsequent thermal management optimization.

[0030] Step S104: Perform continuous processing on the discrete temperature points in the structured temperature data matrix to generate three-dimensional temperature field data.

[0031] In this embodiment, if the temperature sensor data transmission is normal, the temperature value at the corresponding spatial coordinate position is recorded as the normal temperature value. If the temperature sensor shows an abnormal reading, the average value of the adjacent temperature sensors is used for data compensation to obtain a structured discrete temperature data matrix. The discrete temperature data matrix is ​​spatially continuous using the Kriging interpolation algorithm. The spatial autocorrelation parameter is determined by calculating the semi-variogram. If the distance is less than the correlation distance, the weighting coefficient is larger. Several temperature values ​​with larger weighting coefficients are obtained for data compensation to obtain the temperature estimate. Three-dimensional temperature field data is generated based on the temperature estimate and the normal temperature value. The three-dimensional temperature field data is a gridded data structure. If the temperature of a grid point exceeds a preset threshold, it is marked as a hotspot region. An isothermal surface extraction algorithm is used to determine the spatial distribution boundary of the same temperature value.

[0032] In this embodiment, 256 temperature sensors are deployed within the data center, evenly distributed throughout the server room at 4-meter intervals. When temperature sensor node S128 experiences an abnormal reading, the temperature values ​​of its four adjacent sensors S127, S129, S130, and S131 are obtained: 24.2℃, 25.1℃, 24.8℃, and 25.3℃, respectively. The average value of 24.85℃ is calculated as compensation data for location S128. This abnormal data processing mechanism ensures the integrity and reliability of the structured temperature data matrix.

[0033] Specifically, the core of the Kriging interpolation algorithm lies in the calculation of the semivariogram and spatial autocorrelation analysis. When the distance between two measuring points is 2 meters, the semivariogram value is 0.15, and the correlation distance is set to 8 meters, at which point the weighting coefficient is 0.82. However, when the distance between the measuring points increases to 12 meters, it exceeds the correlation distance range, and the weighting coefficient decreases to 0.23. Through this distance attenuation weighting mechanism, the temperature value at any spatial location can be accurately estimated, realizing the conversion from discrete point data to a continuous temperature field. The formula for the Kriging interpolation algorithm is: Z (x)= μ (x)+ ε Z(x) is the observation at position x; μ(x) is the deterministic trend, which can be fitted by a polynomial; and ε(x) is the random error, which is spatially autocorrelated.

[0034] For example, step S104 further includes: acquiring the abnormal temperature value and abnormal sensor corresponding to the discrete temperature point; compensating the abnormal temperature value by using the average temperature of the adjacent sensors of the abnormal sensor to obtain a discrete temperature data matrix; performing spatial continuity processing on the discrete temperature data matrix to obtain a temperature estimate; generating three-dimensional temperature field data based on the temperature estimate and the structured temperature data matrix, wherein the three-dimensional temperature field data is a gridded data structure; if the temperature of a grid point in the gridded data structure exceeds a preset temperature threshold, the corresponding grid point is marked as a hotspot area, and a three-dimensional visualization graphic is generated based on the three-dimensional temperature field data, and different temperature values ​​are mapped to a color spectrum range by color to obtain a complete three-dimensional temperature distribution map.

[0035] In this embodiment, the data center uses a 0.5m × 0.5m × 0.3m grid cell for spatial discretization. Each grid cell corresponds to a temperature sensor and a temperature, resulting in three-dimensional temperature field data. Temperature sensors exhibiting abnormal temperature transmission are represented as discrete temperature points. When the estimated temperature of a grid cell reaches 35°C, it is marked as a hotspot and an early warning mechanism is triggered. The isothermal surface extraction algorithm tracks the spatial distribution of identical temperature values ​​to determine the boundary range of the 30°C isothermal surface, forming a closed three-dimensional surface. This processing method can intuitively display temperature gradient changes and areas of heat accumulation.

[0036] In this embodiment, a 3D visualization graphic is generated based on three-dimensional temperature field data. Different temperature values ​​are mapped to a color spectrum range using color mapping to obtain a complete 3D temperature distribution map reflecting the real-time temperature status of various areas within the data center. The complete temperature distribution map uses the HSV color space for temperature mapping, mapping the temperature range of 18℃ to 40℃ to a blue to red color spectrum. Areas with a temperature value of 20℃ are displayed as dark blue, 25℃ as green, 30℃ as yellow, and areas above 35℃ as red. This color coding method allows maintenance personnel to quickly identify the temperature distribution and abnormal hotspot locations within the data center.

[0037] It should be noted that the entire temperature field reconstruction process achieved a transformation from point measurement to surface analysis. The original 256 discrete measurement points were interpolated to generate continuous temperature field data including approximately 500,000 grid points. This densification process significantly improved the spatial resolution of temperature monitoring, enabling the accurate capture and location of minute temperature changes and local hotspots.

[0038] In this embodiment, the three-dimensional temperature field data is updated every 30 seconds to generate a dynamic three-dimensional temperature distribution map. When an abnormal temperature rise is detected around a server rack, the complete three-dimensional temperature distribution map can display the spatial path and range of heat diffusion in real time, providing accurate spatial location information for cooling system adjustment and equipment maintenance.

[0039] Step S106: Calculate the three-dimensional temperature field data and identify hotspot areas based on the calculation results.

[0040] In this embodiment, a density clustering algorithm is used to automatically group spatially adjacent temperature data points with similar temperatures. If the Euclidean distance between two temperature data points is less than a preset neighborhood radius and the temperature difference is within a tolerable range, they are grouped into the same cluster, resulting in a set of region clusters with similar temperature characteristics. The arithmetic mean and standard deviation of the temperature values ​​within each cluster are calculated based on the region cluster set. If the difference between the average temperature of a region cluster and the global mean is greater than twice the global standard deviation, the region is determined to be a hotspot region.

[0041] In this embodiment, the density clustering algorithm identifies similar temperature regions by setting a neighborhood radius of 1.5 meters and a temperature tolerance range of 2°C. When the coordinates of grid cell A are (10, 15, 2.5) and the temperature is 28.5°C, and the coordinates of grid cell B are (11, 15.5, 2.8) and the temperature is 29.2°C, the Euclidean distance between the two points is 1.3 meters and the temperature difference is 0.7°C. Both meet the clustering conditions and are therefore grouped into the same cluster. The 500,000 grid points are automatically grouped into approximately 800 temperature region clusters, with each cluster containing an average of 600 adjacent data points.

[0042] For example, when the global average temperature is 24.8℃ and the global standard deviation is 3.2℃, if the average temperature of a certain region cluster reaches 31.2℃, the difference between it and the global average temperature is 6.4℃, exceeding the critical value of 6.4℃ (twice the global standard deviation), then this region is immediately marked as a hotspot region. This method of determining hotspot regions based on statistical distribution can adapt to changes in temperature benchmarks under different seasons and load conditions.

[0043] For example, step S106 further includes: if the Euclidean distance between the temperature values ​​of any two device monitoring points in the three-dimensional temperature field data is less than a preset neighborhood radius and the temperature difference is within a preset range, they are classified into the same region cluster to obtain a set of region clusters; the average temperature value and standard deviation of the temperature values ​​within each region cluster are calculated based on the set of region clusters, and the region corresponding to the region cluster is determined to be a hotspot region based on the average temperature value and standard deviation; the peak temperature of the hotspot region is obtained, and if the peak temperature exceeds a preset danger threshold, a high priority is triggered, and the specific value of the peak temperature and the corresponding three-dimensional coordinate position are recorded; the duration is calculated based on the time series data of the hotspot region, and if the duration exceeds a preset duration threshold, a complete hotspot region file record is generated, which includes the specific value, three-dimensional coordinate position, and duration.

[0044] In this embodiment, a maximum value search algorithm is used to find peak temperature points within hotspot areas. If the peak temperature exceeds a preset danger threshold, a high-priority marker is triggered, and the specific value of the peak temperature and its corresponding three-dimensional coordinates are recorded. The duration parameter is calculated based on the time-series data of the hotspot area. If the duration exceeds a preset duration threshold, a complete hotspot area record is generated, including spatial coordinates, peak temperature value, and duration. The maximum value search algorithm uses a sliding window mechanism to locate peak temperature points within identified abnormal areas. In an abnormal area comprising 1200 grid points, a point-by-point scan revealed that the temperature value at coordinates (25.5, 18.0, 1.8) was 38.7℃, the highest value in the area. Since this peak temperature exceeds the preset danger threshold of 35℃, a high-priority marker is automatically triggered, and the precise coordinates and temperature value of this location are stored in the abnormal event database.

[0045] In this embodiment, time series analysis calculates the duration parameter by continuously monitoring temperature changes at the same spatial location. When a hotspot area is first detected to have a temperature anomaly, its temperature status is recorded every 30 seconds. If the area remains in an abnormal state for more than 15 minutes (30 monitoring cycles), it is determined to be a persistent hotspot, and a complete hotspot area record is generated. This complete hotspot area record includes multi-dimensional information to support subsequent analysis and decision-making. The spatial range coordinates recorded in the record are from the starting point (22.0, 16.5, 1.2) to the ending point (28.0, 20.5, 2.4), covering a volume of approximately 126 cubic meters. The peak temperature of 38.7℃ and its corresponding coordinates (25.5, 18.0, 1.8) identify the core location of the hotspot, and the duration of 18 minutes reflects the severity of the anomaly. This automated hotspot detection mechanism significantly improves the intelligence level of data center temperature monitoring. Traditional inspection methods require 2 hours to complete a full-room temperature anomaly investigation, while based on the above steps, anomaly identification and location of all 500,000 grid points can be completed within 5 minutes. The generated hotspot profiles provide maintenance personnel with accurate fault location information, reducing the response time for equipment repair and cooling adjustment from an average of 45 minutes to 8 minutes.

[0046] Step S108: Perform time-series analysis on the regional temperature data of the hotspot area, and construct a temperature evolution model based on the time-series analysis results. The temperature evolution model is used to predict the temperature value at a specific point in the future.

[0047] In this embodiment, historical temperature records of each monitoring point within the hotspot area are acquired. If the time span of the temperature records at a monitoring point exceeds a preset minimum time threshold, the temperature data is arranged in chronological order to obtain a complete time-series temperature dataset. Based on the time-series temperature dataset, the ratio of the temperature difference between adjacent time points to the time interval is calculated, i.e., the rate of temperature increase.v i =Δ tT i+1 - T i Δt=t i+1 -t i Δ t = t i+1 - t i constant time interval v i >0 indicates a temperature rise, i>0. If the rate of temperature rise is positive for three consecutive time periods, the acceleration of the temperature change can be calculated using second-order difference. a i = ( v i+2 - v i ） / 2Δ t Unit: °C / s², i>0. A temperature evolution function is constructed based on the temperature rise rate and acceleration values, for... T(t) = T i + v(t- t i )+0.5a i (t-t i ) 2 If i > 0, the least squares method is used to fit the variation curve of historical temperature data. If the goodness-of-fit coefficient is greater than 0.85, it indicates that the established temperature evolution function can well describe the temperature change pattern of the region, thus obtaining a temperature evolution model. This high goodness of fit ensures the accuracy and reliability of the prediction results.

[0048] For example, step S108 further includes: obtaining historical temperature records of each monitoring point within the hotspot area; if the time span of the temperature records of the monitoring points exceeds a preset minimum time threshold, arranging the regional temperature data in chronological order to obtain a time series temperature dataset; calculating the temperature rise rate and acceleration values ​​based on the time series temperature dataset; and constructing a temperature evolution model based on the temperature rise rate and acceleration values.

[0049] In this embodiment, the acquisition of historical temperature records for monitoring points within the hotspot area needs to ensure data continuity and integrity. When the temperature record time span of a certain monitoring point reaches more than 72 hours, these temperature data will be automatically rearranged according to the chronological order of collection time to form a complete time-series temperature dataset. The time-series temperature dataset can truly reflect the evolution of the temperature field over time, providing a reliable data foundation for subsequent predictive analysis. When calculating the rate of temperature change between adjacent time points, the temperature difference within each time interval is extracted and then divided by the corresponding time interval length. When the rate of temperature increase is positive for three consecutive time intervals, it indicates that the area is in a state of continuous warming. At this time, the acceleration value of temperature change can be obtained through second-order difference calculation, which reflects whether the temperature rise trend is further intensifying.

[0050] Step S110: Obtain the current time point and corresponding current temperature data collected by the target temperature sensor, input the current time point and current temperature data into the temperature evolution model, and obtain the temperature prediction value of the preset future time point.

[0051] In this embodiment, the temperature prediction value for a preset future time point is calculated based on the current time point and the current temperature data input into the temperature evolution function. If the predicted temperature value exceeds the preset danger temperature threshold, a high temperature warning message for the corresponding time point is generated. During the temperature prediction process, the current temperature data and time parameters are input into the determined temperature evolution function to calculate the temperature prediction value for the preset future time point.

[0052] In this embodiment, time series analysis has significant advantages in temperature prediction. By deeply mining historical temperature data, it can discover the inherent patterns and trends in temperature changes, providing a more forward-looking perspective than simple real-time monitoring and issuing early warning signals before temperature anomalies occur. When the rate of temperature increase in a certain hotspot area is 2.5 degrees Celsius per hour, 3.1 degrees Celsius per hour, and 3.8 degrees Celsius per hour in three consecutive time periods, a positive acceleration value is calculated, indicating that the temperature rise trend is accelerating. The temperature evolution function constructed by combining these parameters can accurately predict the highest temperature that the area may reach in the next 6 hours, thus providing a scientific basis for risk management.

[0053] Step S112: When the predicted temperature value exceeds the preset dangerous temperature threshold, generate a high temperature warning information for the target temperature sensor's corresponding acquisition area.

[0054] In this embodiment, when the predicted temperature value exceeds the set dangerous temperature threshold, a corresponding high temperature warning will be generated immediately, which can identify potential high temperature risk areas in advance and provide sufficient time window for the timely implementation of safety protection measures.

[0055] For example, the method further includes: acquiring hotspot temperature data and equipment operating parameters corresponding to the hotspot areas collected by temperature sensors; calculating the temperature gradient of each monitoring point of the equipment based on the hotspot temperature data and equipment operating parameters; analyzing the hotspot temperature data to obtain the hotspot propagation path corresponding to the hotspot area and marking the corresponding coordinate position and temperature value to obtain heat propagation matrix data; calculating the diffusion rate and influence range of each high-temperature propagation path based on the heat propagation matrix data; acquiring the load history data and temperature anomaly records corresponding to the equipment; analyzing the boundary coordinates corresponding to the influence range and the correlation relationship of the equipment data within the influence range based on the load history data and temperature anomaly records to generate a correlation data table; constructing a response function based on the correlation data table, the response function being used to calculate the temperature value based on the load value of the equipment; and determining whether to generate a load adjustment command and send it to the corresponding equipment based on the change in temperature value.

[0056] In this embodiment, temperature data and device operating parameters collected by a temperature sensor network are received. The temperature gradient distribution of each node within the hotspot area is calculated using the finite difference method. If the temperature gradient exceeds a preset threshold, it is marked as a high-temperature propagation path, obtaining heat propagation matrix data including coordinate positions and temperature values. Based on the heat propagation matrix data, the diffusion rate of each path is calculated using the Fourier heat conduction equation. The diffusion rate calculation formula is v = k multiplied by the temperature gradient divided by the density multiplied by the specific heat capacity, determining the boundary coordinates of the heat influence range; where v represents the diffusion rate, unit: W / m 2 : Watts per square meter; k represents the thermal conductivity coefficient, unit: W / (m*K): Watts per meter per Kelvin, characterizing the material's ability to conduct heat, such as copper's thermal conductivity being approximately 401 (W / (m*K)) and air's being approximately 0.026 (W / (m*K)); T represents the temperature gradient vector, unit: K / m: Kelvin per meter. Historical load data and temperature anomaly records of the equipment are acquired. Equipment data within the boundary coordinate range is processed using an association rule mining algorithm. If the correlation coefficient between the load change rate and the temperature change rate is greater than 0.83, a strong association is established, resulting in an association data table of equipment identification, load value, and temperature response time. A load-temperature response function is constructed based on the association data table, and the least squares method is used to fit the temperature change curves under different load conditions. If the current load value exceeds the preset range, a temperature warning mechanism is triggered, generating a load adjustment command.

[0057] In this embodiment, the temperature data acquisition of the sensor network needs to ensure the synchronization and accuracy of data from each monitoring node. When temperature sensor data distributed throughout the equipment room is received, operating parameters such as server CPU load and memory usage are simultaneously acquired. When calculating the temperature gradient between adjacent nodes using the finite difference method, the monitoring area is divided into a regular grid, with each grid point corresponding to a temperature value. When the gradient value obtained by dividing the temperature difference between a grid point and its adjacent points by the distance exceeds 5 degrees per meter, this path is automatically marked as a high-temperature propagation path.

[0058] In this embodiment, the application of the Fourier heat conduction equation can accurately calculate the heat diffusion characteristics in different media. When the thermal conductivity of air is 0.026 W / m / Kelvin, its density is 1.2 kg / m³, and its specific heat capacity is 1005 J / kg / Kelvin, the heat diffusion rate is calculated based on the measured temperature gradient. This calculation method can determine the specific boundaries of the heat's influence, providing accurate spatial positioning information for subsequent equipment correlation analysis.

[0059] In this embodiment, the load-temperature response function is constructed using the least squares method to fit and analyze historical data. Temperature response data under different load conditions are collected, and a mathematical model is established to describe the quantitative relationship between the two. When the current load value of a server exceeds the upper limit of its normal operating range, the potential temperature rise is predicted based on the response function, and a load adjustment command is automatically generated before the temperature reaches a dangerous threshold. This early warning mechanism based on load-temperature correlation analysis enables proactive temperature control. By identifying load patterns that may lead to overheating in advance, corresponding load distribution adjustment measures are taken before temperature anomalies occur, effectively avoiding the risk of performance degradation or failure due to overheating.

[0060] For example, the method further includes: receiving real-time temperature data of the hotspot area collected by a temperature sensor, the real-time temperature data including temperature values ​​and timestamps; calculating the temperature change gradient and abnormal duration based on the real-time temperature data; calculating the current temperature rise rate based on the temperature change gradient and abnormal duration, and generating evolution characteristic parameters based on the current temperature rise rate; obtaining the device attributes and distance information of devices within a preset range around the hotspot area; calculating the heat propagation attenuation coefficient based on the evolution characteristic parameters and device attributes; calculating the propagation distance based on the heat propagation attenuation coefficient, and if the propagation distance is less than the safe distance, generating impact assessment data, the safe distance being determined based on the distance information; calculating the impact assessment data through a preset risk rating matrix to obtain the risk value and the corresponding risk level; triggering a graded early warning signal when the risk level exceeds a preset level, and assigning different processing priorities to hotspot areas with different risk levels.

[0061] In this embodiment, if the temperature rise rate exceeds a preset threshold, evolution characteristic parameters are generated, including peak temperature, duration, and gradient. If the propagation distance is less than a safe distance threshold, impact assessment data is generated, including predicted equipment temperature rise and potential loss. A risk score is calculated based on the impact assessment data using a risk rating matrix. If the abnormal duration exceeds a critical duration and the predicted equipment temperature rise reaches a damage threshold, it is determined to be a high-risk level. Thresholds are compared based on the risk level. If the risk score exceeds a warning threshold, a corresponding level of warning signal is triggered, generating a tiered warning instruction including warning type, response time, and handling plan.

[0062] In this embodiment, a temperature sensor collects temperature data every second within the data center server room. When the temperature in a certain area rises from 25 degrees Celsius to 45 degrees Celsius within 10 minutes, the calculated temperature rise rate is 2 degrees Celsius per minute, exceeding the preset threshold of 1.5 degrees Celsius per minute. The generated evolution characteristic parameters show that the temperature peak reaches 45 degrees Celsius, the abnormal duration is 10 minutes, and the change gradient is 2 degrees Celsius per minute. This precise feature parameter extraction provides a reliable data foundation for subsequent risk assessment. Information on server equipment within a 3-meter radius of the hotspot is obtained, including the thermal conductivity coefficient of the aluminum alloy chassis material (200 watts per meter Kelvin) and the thermal conductivity coefficient of the plastic cable sheath (0.3 watts per meter Kelvin). Based on different material properties and distance attenuation patterns, the heat propagation attenuation coefficient for aluminum alloy equipment is calculated to be 0.8, while that for plastic equipment is 0.2. When the propagation distance is 1.5 meters, less than the safe distance threshold of 2 meters, the predicted temperature rise of the aluminum alloy chassis will reach 15 degrees Celsius, with a potential loss assessment of a 30% decrease in equipment performance. The formula for calculating the heat transfer attenuation coefficient is: T(x,y)=a*x+b*y+c, where a, b, and c are real constants; x and y are two variables of temperature change, such as the thermal conductivity coefficient corresponding to the material properties and the normalized distance; T(x,y) is the heat transfer attenuation coefficient, which represents the rate of temperature change.

[0063] In this embodiment, the risk rating matrix employs a multi-dimensional scoring mechanism, comprehensively considering factors such as the duration of the anomaly, the predicted temperature rise of the equipment, and material sensitivity. When the duration of the anomaly reaches 15 minutes, exceeding the critical duration of 10 minutes, and the predicted temperature rise of the critical server reaches 20 degrees Celsius, approaching the damage threshold of 25 degrees Celsius, a risk score of 85 is calculated, classifying it as a high-risk level. Quantitative temperature assessment accurately identifies potential equipment damage risks. When the risk score of 85 exceeds the warning threshold of 80, a Level 1 warning signal is immediately triggered, generating a tiered warning instruction that includes a "equipment overheating warning" type, a "response within 5 minutes" time requirement, and a "immediately activate the backup cooling system and transfer the load" handling plan. For medium-risk situations, a Level 2 warning is generated, requiring a response within 15 minutes and the implementation of "enhanced ventilation monitoring" measures. This tiered warning mechanism ensures accurate responses at different risk levels, avoiding resource waste and insufficient response, and significantly improving equipment protection effectiveness and operational efficiency.

[0064] This embodiment also automatically generates cooling resource allocation plans for high-risk hotspot areas, calculates the additional cooling capacity required for cooling, adjusts the air supply volume and air supply temperature of the relevant areas, monitors the temperature change effect after adjustment, and if the temperature drop rate is lower than expected, further increases the investment in cooling resources or reduces the load on relevant equipment.

[0065] Specifically, real-time temperature data and spatial location information of high-risk hotspot areas are acquired. The cooling range is calculated based on the temperature data and area. If the temperature exceeds a preset safe range, a target cooling value is determined. The total cooling capacity required for the hotspot area to reach a safe temperature is calculated using thermodynamic formulas. Based on the total cooling capacity requirement and the cooling capacity parameters of existing air conditioning equipment, a load allocation algorithm is used to calculate the airflow adjustment value and airflow temperature setpoint for each air conditioning unit. If the cooling capacity of a single unit is insufficient, multiple units are called upon to work collaboratively, generating a dispatch command including the unit number and airflow parameters. The dispatch command is sent to the target air conditioning unit through the building automation control interface. Real-time temperature sensor data within the adjusted area is collected, the temperature drop rate is calculated, and compared with a preset ideal drop curve. If the actual drop rate is lower than the expected threshold, the current cooling measures are deemed insufficient. Feedback information indicating unsatisfactory temperature reduction is received. The operating power of non-critical electrical equipment within the area is reduced through the equipment load management module, while the amount of backup cooling equipment is increased.

[0066] In this embodiment, when the temperature in a certain area of ​​the data center server room reaches 35 degrees Celsius and the area is 50 square meters, it is calculated based on thermodynamic principles that a temperature reduction of 10 degrees Celsius is needed to reach the safe range of 25 degrees Celsius. At this point, the total cooling capacity requirement is determined by parameters such as area volume, air density, and specific heat capacity. If this area requires 150 kilowatts of cooling capacity to complete the cooling task within 30 minutes, the load allocation algorithm analyzes the cooling capacity parameters of the existing 6 precision air conditioning units. Each unit has a rated cooling capacity of 30 kilowatts, and a single unit cannot meet the total demand of 150 kilowatts. It is calculated that 5 units need to work collaboratively, with 3 units having their airflow adjusted to the maximum setting of 8000 cubic meters per hour and the airflow temperature set to 16 degrees Celsius, and the other 2 units having their airflow set to 6000 cubic meters per hour and the airflow temperature set to 18 degrees Celsius. The generated allocation command includes the unit numbers AC-01 to AC-05, along with the corresponding airflow parameters and temperature settings.

[0067] In this embodiment, the building automation control interface uses the BACnet protocol to send dispatch commands to the target air conditioning equipment. After the command is executed, data from 12 temperature sensors in the area is collected every 2 minutes. The ideal temperature drop curve is set to a 3-degree Celsius drop every 10 minutes, but actual monitoring shows that the temperature only drops by 2 degrees Celsius in the first 10 minutes, which is lower than the expected threshold, indicating that the current cooling measures are insufficient. When the temperature drop effect is not up to standard, the equipment load management module will automatically reduce the power of non-critical lighting equipment in the area to 50%, and at the same time adjust the UPS (Uninterruptible Power Supply) charging power from 100% to 70% to release more power resources for the cooling system. Two backup cooling devices will also be activated, each with a rated cooling capacity of 25 kilowatts, increasing the total cooling capacity to 200 kilowatts. This dynamic adjustment mechanism can maximize the cooling effect while ensuring the normal operation of critical equipment. Through real-time monitoring and feedback adjustment, the cooling system can reduce the area temperature from 35 degrees Celsius to the target 25 degrees Celsius within 15 minutes, avoiding the risk of server downtime due to excessive temperature. Meanwhile, the multi-device collaborative working mode has better energy consumption control compared to a single high-power device. Once the target temperature is reached, the number of operating devices can be gradually reduced to achieve precise temperature control and energy optimization management.

[0068] For example, the method further includes: when the temperature value in the historical temperature data exceeds a preset temperature threshold range, generating temperature anomaly event data and forming an anomaly event database; calculating the support and confidence between temperature anomalies and equipment failure types in the anomaly event data; if the confidence and support meet preset rules, generating a set of association rules between temperature anomalies and equipment failure types; extracting temperature anomaly feature fingerprints from the set of association rules, encoding the temperature anomaly feature fingerprints, and constructing a temperature anomaly pattern library based on the feature fingerprint encoding and corresponding processing measures; calculating the similarity between the current temperature anomaly event feature fingerprint and the temperature anomaly feature fingerprints in the temperature anomaly pattern library; if the similarity exceeds a preset similarity value, retrieving the corresponding processing measures from the temperature anomaly pattern library and determining the response processing measure scheme.

[0069] In this embodiment, real-time temperature data is acquired from a temperature monitoring sensor via a database connector. If the temperature data exceeds a preset threshold range, an abnormal event recording mechanism is triggered to generate abnormal event data including a timestamp and device number. Based on the abnormal event data, an association rule mining algorithm is used to calculate the support and confidence between temperature anomalies and device fault types. If the support is greater than a preset minimum support threshold and the confidence is greater than a preset minimum confidence threshold, a set of association rules between temperature anomalies and fault types is obtained. A feature extractor extracts the temperature change amplitude, duration, and rate of change as temperature anomaly feature fingerprints from the association rule set. A hash function is used to encode the feature fingerprints, establishing a temperature anomaly pattern library including feature fingerprint encoding and corresponding handling measures. A similarity calculation function is used to match and compare the feature fingerprints of newly occurring temperature anomalies with those stored in the pattern library. If the similarity exceeds a preset recognition threshold, the corresponding handling measure record is retrieved from the pattern library, and the appropriate handling measure scheme is determined.

[0070] In this embodiment, the temperature monitoring sensor transmits temperature data to the database connector every second via an RS485 bus. When the temperature in a certain area of ​​the computer room suddenly rises from the normal 22 degrees Celsius to 35 degrees Celsius, exceeding the preset threshold of 30 degrees Celsius, an abnormal event record is immediately generated, including the specific timestamp of January 15, 2024, 14:32:15 and key information such as the equipment number AC-Zone-A03. Historical abnormal data is analyzed using an association rule mining algorithm to discover the intrinsic relationship between temperature anomalies and equipment failures. When the calculated support for temperature exceeding the limit and air conditioning compressor failure is 0.65 and the confidence level is 0.82, both exceeding the preset minimum support threshold of 0.6 and minimum confidence level of 0.8, a strong correlation between the two is confirmed. This association rule can reveal the equipment failure mode corresponding to temperature anomalies, providing data support for subsequent fault prediction and handling.

[0071] For example, the feature extractor extracts feature fingerprints from association rules in three core dimensions. The magnitude of temperature change reflects the severity of the anomaly, such as a temperature rise of 13 degrees Celsius; the duration indicates how long the anomaly lasts, such as 8 minutes; and the rate of change reflects the speed at which the anomaly occurs, such as a rise of 1.6 degrees Celsius per minute. These feature fingerprints are encoded into 32-bit strings using an MD5 hash function to ensure that each anomaly pattern has a unique digital identifier. The temperature anomaly pattern library stores feature fingerprint codes for various typical fault scenarios. When air conditioning refrigerant leaks, the characteristic is a slow temperature rise over a long period; when the fan fails, the temperature rises rapidly but by a relatively small margin; when the compressor completely stops, the temperature rises sharply and rapidly. Each pattern corresponds to specific handling measures, such as adding backup refrigeration equipment, adjusting air supply parameters, or activating an emergency cooling program.

[0072] In this embodiment, the similarity calculation uses the Euclidean distance algorithm to match new abnormal events with a pattern library. When the similarity between the feature fingerprint of a newly occurring temperature anomaly and a known pattern in the pattern library reaches 0.89, exceeding the preset recognition threshold of 0.85, the corresponding processing measure record is retrieved. This intelligent matching mechanism can quickly identify the fault type in the early stages of an anomaly and provide targeted solutions, significantly shortening the fault response time and improving the reliability and automation level of the building temperature control system.

[0073] For example, the method further includes: classifying hotspot areas into anomaly levels and generating temperature change trend maps; obtaining equipment operating parameters and equipment load data corresponding to the equipment in the hotspot areas; and drawing heat migration path maps based on the temperature change trend maps, equipment operating parameters, and equipment load data.

[0074] In this embodiment, the hotspot area is divided into a high-temperature zone (>28℃), a low-temperature zone (<18℃), and a temperature fluctuation zone (fluctuation >2℃ within 5 minutes), and a temperature change trend graph is generated to visually display the temperature change curve over 24 hours. When an abnormal temperature increase was detected in a certain server rack group, rising from 24℃ to 29℃, data from surrounding temperature sensors was retrieved. Combined with air conditioning airflow (12000m³ / h) and rack power consumption data (8.5kW / rack), a heat migration path diagram was drawn. The root cause was ultimately located at an airflow obstruction point formed by dense cable management behind the rack, preventing effective heat dissipation. Optimizing the cabling using the heat migration path diagram successfully reduced the temperature in this area by 3.5℃, improving heat dissipation efficiency.

[0075] Example 2: Please continue reading Figure 2This diagram illustrates a program module schematic of a second embodiment of the modular data center detection system of the present invention. In this embodiment, the modular data center detection system 20 may include or be divided into one or more program modules. One or more program modules are stored in a storage medium and executed by one or more processors to complete the present invention and implement the above-described modular data center detection method. The program module referred to in this embodiment of the present invention refers to a series of computer program instruction segments capable of performing specific functions, which are more suitable than the program itself for describing the execution process of the modular data center detection system 20 in the storage medium. The following description will specifically introduce the functions of each program module in this embodiment: Acquisition module 200 is used to acquire historical temperature data of the data center collected by temperature sensors. When the temperature sensors are set at the corresponding equipment monitoring points in the data center, a three-dimensional spatial temperature acquisition network is formed.

[0076] The first data processing module 202 is used to process the collected historical temperature data and add three-dimensional spatial coordinate labels to obtain a structured temperature data matrix.

[0077] The second data processing module 204 is used to perform continuous processing on discrete temperature points in the structured temperature data matrix to generate three-dimensional temperature field data.

[0078] The calculation module 206 is used to calculate the three-dimensional temperature field data and identify hotspot areas based on the calculation results.

[0079] Analysis module 208 is used to perform time series analysis on regional temperature data of hotspot areas, and to construct a temperature evolution model based on the time series analysis results. The temperature evolution model is used to predict the temperature value at a specific point in the future.

[0080] The prediction module 210 is used to acquire the current time point and the corresponding current temperature data collected by the target temperature sensor, input the current time point and the current temperature data into the temperature evolution model, and obtain the temperature prediction value for the preset future time point.

[0081] The generation module 212 is used to generate high temperature warning information for the target temperature sensor's corresponding acquisition area when the predicted temperature value exceeds the preset dangerous temperature threshold.

[0082] The modular data center detection system provided in this invention sets the temperature sensor to a three-dimensional spatial acquisition mode to collect historical temperature data. By processing the historical temperature data, three-dimensional temperature field data is generated. The three-dimensional temperature field data is calculated to identify hotspot areas. Then, a regional temperature evolution model is constructed based on the regional temperature data of the hotspot areas. By using the regional temperature evolution model and the current temperature data, the temperature at the next time point is predicted and warned, thus achieving accurate identification and prediction of hotspot areas.

[0083] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0084] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0085] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A modular data center detection method, characterized in that, include: Historical temperature data of the data center is acquired by temperature sensors, which form a three-dimensional spatial temperature acquisition network when the temperature sensors are set at the corresponding equipment monitoring points in the data center. The collected historical temperature data is processed and three-dimensional spatial coordinate labels are added to obtain a structured temperature data matrix; the discrete temperature points in the structured temperature data matrix are processed to generate three-dimensional temperature field data. The three-dimensional temperature field data is calculated, and hotspot regions are identified based on the calculation results. The regional temperature data of the hotspot regions are subjected to time series analysis, and a temperature evolution model is constructed based on the time series analysis results. The temperature evolution model is used to predict the temperature value at a specific future time point. The system acquires the current time point and corresponding current temperature data collected by the target temperature sensor, inputs the current time point and the current temperature data into the temperature evolution model, and obtains the temperature prediction value for a preset future time point. When the temperature prediction value exceeds the preset danger temperature threshold, a high temperature warning information is generated for the area collected by the target temperature sensor.

2. The modular data center detection method according to claim 1, characterized in that, Acquiring historical temperature data of a data center collected by a temperature sensor includes: receiving historical temperature data collected by the temperature sensor, wherein the temperature sensor carries a unique identification code and three-dimensional coordinate positioning information; if the variation of the historical temperature data exceeds a preset threshold, increasing the sampling frequency of the corresponding temperature sensor and timestamping the historical temperature data.

3. The modular data center detection method according to claim 1, characterized in that, The discrete temperature points in the structured temperature data matrix are processed to become continuous, generating three-dimensional temperature field data, including: abnormal temperature values ​​and abnormal sensors corresponding to the discrete temperature points; temperature compensation is performed on the abnormal temperature values ​​using the average temperature of adjacent sensors to obtain a discrete temperature data matrix; spatial continuity processing is performed on the discrete temperature data matrix to obtain temperature estimates; three-dimensional temperature field data is generated based on the temperature estimates and the structured temperature data matrix, wherein the three-dimensional temperature field data is a gridded data structure; if the temperature of a grid point in the gridded data structure exceeds a preset temperature threshold, the corresponding grid point is marked as a hotspot region, and a three-dimensional visualization graphic is generated based on the three-dimensional temperature field data, mapping different temperature values ​​to a color spectrum range to obtain a complete three-dimensional temperature distribution map.

4. The modular data center detection method according to claim 1, characterized in that, The three-dimensional temperature field data is calculated, and hotspot regions are identified based on the calculation results. This includes: if the Euclidean distance between the temperature values ​​of any two monitoring points of the device in the three-dimensional temperature field data is less than a preset neighborhood radius and the temperature difference is within a preset range, they are grouped into the same region cluster, resulting in a set of region clusters; the average temperature value and standard deviation of the temperature values ​​within each region cluster are calculated based on the set of region clusters, and whether the region corresponding to the region cluster is a hotspot region is determined based on the average temperature value and the standard deviation; the peak temperature of the hotspot region is obtained, and if the peak temperature exceeds a preset danger threshold, a high-priority marker is triggered, and the specific value of the peak temperature and the corresponding three-dimensional coordinate position are recorded; the duration is calculated based on the time series data of the hotspot region, and if the duration exceeds a preset duration threshold, a complete hotspot region record is generated.

5. The modular data center detection method according to claim 1, characterized in that, A time-series analysis is performed on the regional temperature data of the hotspot area, and a temperature evolution model is constructed based on the time-series analysis results. This includes: acquiring historical temperature records of each monitoring point within the hotspot area; if the time span of the temperature records of the monitoring points exceeds a preset minimum time threshold, arranging the regional temperature data in chronological order to obtain a time-series temperature dataset; calculating the temperature rise rate and acceleration value based on the time-series temperature dataset; and constructing the temperature evolution model based on the temperature rise rate and acceleration value.

6. The modular data center detection method according to claim 1, characterized in that, The method further includes: acquiring hotspot temperature data and equipment operating parameters corresponding to the hotspot area collected by the temperature sensor; calculating the temperature gradient of each monitoring point of the equipment based on the hotspot temperature data and the equipment operating parameters; analyzing the hotspot temperature data to obtain the hotspot propagation path corresponding to the hotspot area and marking the corresponding coordinate position and temperature value to obtain heat propagation matrix data; calculating the diffusion rate and influence range of each high-temperature propagation path based on the heat propagation matrix data; acquiring the load history data and temperature anomaly records corresponding to the equipment; analyzing the boundary coordinates corresponding to the influence range and the correlation relationship of the equipment data obtained within the influence range based on the load history data and temperature anomaly records to generate a correlation data table; constructing a response function based on the correlation data table, the response function being used to calculate the temperature value based on the load value of the equipment; and determining whether to generate a load adjustment command and send it to the corresponding equipment based on the change in the temperature value.

7. The modular data center detection method according to claim 1, characterized in that, The method further includes: receiving real-time temperature data of the hotspot area collected by the temperature sensor, the real-time temperature data including temperature values ​​and timestamps; calculating the temperature change gradient and abnormal duration based on the real-time temperature data; calculating the current temperature rise rate based on the temperature change gradient and the abnormal duration, and generating evolution characteristic parameters based on the current temperature rise rate; acquiring device attributes and distance information of devices within a preset range surrounding the hotspot area; calculating a heat propagation attenuation coefficient based on the evolution characteristic parameters and the device attributes; calculating the propagation distance based on the heat propagation attenuation coefficient, and if the propagation distance is less than a safe distance, generating impact assessment data, the safe distance being determined based on the distance information; calculating the impact assessment data through a preset risk rating matrix to obtain a risk value and a corresponding risk level; triggering a graded early warning signal when the risk level exceeds a preset level, and assigning different processing priorities to the hotspot areas with different risk levels.

8. The modular data center detection method according to claim 1, characterized in that, The method further includes: when the temperature value in the historical temperature data exceeds a preset temperature threshold range, generating temperature anomaly event data and forming an anomaly event database; calculating the support and confidence between temperature anomalies and equipment failure types in the anomaly event data; if the confidence and support meet preset rules, generating a set of association rules between the temperature anomaly and the equipment failure type; extracting temperature anomaly feature fingerprints from the set of association rules, encoding the temperature anomaly feature fingerprints, and constructing a temperature anomaly pattern library based on the feature fingerprint encoding and corresponding processing measures; calculating the similarity between the current temperature anomaly event feature fingerprint and the temperature anomaly feature fingerprints in the temperature anomaly pattern library; if the similarity exceeds a preset similarity value, retrieving the corresponding processing measures from the temperature anomaly pattern library and determining the corresponding processing measure scheme.

9. The modular data center detection method according to claim 1, characterized in that, The method further includes: classifying the hotspot area into anomaly levels and generating a temperature change trend map; obtaining the equipment operating parameters and equipment load data corresponding to the equipment in the hotspot area; and drawing a heat migration path map based on the temperature change trend map, the equipment operating parameters, and the equipment load data.

10. A modular data center detection system, characterized in that, include: The acquisition module is used to acquire historical temperature data of the data center collected by temperature sensors. When the temperature sensors are set at the device monitoring points corresponding to the data center, they form a three-dimensional spatial temperature acquisition network. The first data processing module is used to process the acquired historical temperature data and add three-dimensional spatial coordinate labels to obtain a structured temperature data matrix. The second data processing module is used to perform continuous processing on the discrete temperature points in the structured temperature data matrix to generate three-dimensional temperature field data. The calculation module is used to calculate the three-dimensional temperature field data and identify hotspot areas based on the calculation results; The analysis module is used to perform time-series analysis on the regional temperature data of the hotspot area, and to construct a temperature evolution model based on the time-series analysis results. The temperature evolution model is used to predict the temperature value at a specific point in the future. The prediction module is used to acquire the current time point and the corresponding current temperature data collected by the target temperature sensor, and input the current time point and the current temperature data into the temperature evolution model to obtain the temperature prediction value at a preset future time point. The generation module is used to generate high temperature warning information for the target temperature sensor's corresponding acquisition area when the predicted temperature value exceeds a preset dangerous temperature threshold.

Images