Data center air quality intelligent early warning method and system based on sensor network

Abstract

The invention relates to the technical field of sensor networks and Internet of Things, in particular to a data center air quality intelligent early warning method and system based on a sensor network. The method comprises the following steps: constructing a self-organizing cooperative network by deploying multiple types of sensor nodes in a key area of a data center, and collecting and correcting multi-dimensional air data in real time; the data reliability is improved through inter-node dynamic weight fusion and local anomaly recognition; establishing an air parameter and machine room structure correlation model by utilizing space mapping and time sequence analysis, and identifying a micro-scale diffusion trend; a self-adaptive dynamic threshold mechanism is constructed, and threshold rolling optimization is realized in combination with historical statistics and real-time feedback; and generating a graded alarm strategy based on multi-stage early warning judgment and triggering conditions, and continuously self-optimizing early warning precision and response efficiency through closed-loop feedback. According to the invention, all-around, intelligent and high-reliability early warning and regulation and control of the air quality of the data center are realized.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of sensor networks and Internet of Things, and in particular to a data center air quality intelligent early warning method and system based on a sensor network. BACKGROUND

[0002] As a key physical facility supporting digital operation, the internal environment stability of a data center directly affects the service life of computing devices and business continuity; in recent years, with the continuous improvement of server power density, the unevenness of airflow and pollutant distribution at the cabinet level microenvironment has become increasingly significant, and higher requirements for the refinement and real-time performance of air quality monitoring have been put forward.

[0003] A real-time intelligent indoor air quality monitoring system is disclosed in Chinese patent application No. CN109596492A, which comprises a sensor perception subsystem, a computer analysis terminal and an intelligent terminal; the sensor perception subsystem is used to collect indoor air quality information, and the sensor perception subsystem comprises a wireless sensor network constructed by a convergence node and a plurality of sensor nodes deployed in the indoor environment; the sensor nodes collect air quality information at the monitoring location, and the convergence node is mainly used to converge the air quality information collected by the sensor nodes and send it to the computer analysis terminal; the computer analysis terminal is connected with a plurality of intelligent terminals to send the received air quality information to the plurality of intelligent terminals.

[0004] At the same time, with the advancement of sensor technology, edge computing and intelligent analysis methods, it has become an important research direction to improve the intelligent level of data center operation and maintenance to build an air quality early warning system that can realize real-time perception, intelligent analysis and autonomous decision-making. SUMMARY

[0005] The purpose of the present application is to solve the problems in the background art, and to propose a data center air quality intelligent early warning method and system based on a sensor network.

[0006] The technical solution of the present application is a data center air quality intelligent early warning method based on a sensor network, comprising the following specific implementation steps: S1, deploying multiple types of sensor nodes in key areas of the data center, constructing a self-organizing cooperative network, collecting multi-dimensional air data in real time, and correcting the collected data at the node end and identifying local anomalies; S2, improving data reliability through dynamic weight fusion and local anomaly identification between nodes, and establishing an air parameter and machine room structure correlation model using spatial mapping and time series analysis to identify micro-scale diffusion trends; S3, constructing an adaptive dynamic threshold mechanism, combining historical statistics and real-time feedback to realize rolling optimization of the threshold; S4, generating a hierarchical alarm strategy based on the multi-level early warning judgment and trigger conditions, and continuously optimizing the early warning accuracy and response efficiency through closed-loop feedback.

[0007] Preferably, step S1 specifically comprises: In the cold aisle, hot aisle, cabinet aisle, air outlet of air conditioner and return air area under the floor of the machine room, temperature and humidity sensors, particulate matter sensors, gas concentration sensors and wind speed and direction sensors are arranged; An air state vector of a single node is constructed, which includes environmental temperature, relative humidity, particulate matter concentration, gas concentration, air flow rate and wind direction; A node self-organizing and cooperating network is constructed, each node forms a self-organizing network through low-power wireless communication, dynamically calculates the cooperation weight according to the distance and health state between nodes, and performs local data fusion and preliminary anomaly identification; The nodes collect air state vectors at an adaptive sampling frequency, and add unique identifiers and time stamps, and high-frequency collection is used in key areas and adjacent nodes are used for redundant collection; The collected data is drift-corrected, filtered, delay-compensated and wind flow-smoothed at the node end, and preliminary anomaly early warning is also supported at the node end.

[0008] Preferably, the node self-organizing and cooperating network is constructed, specifically comprising: Before each sensor node cooperates with adjacent nodes in real time, the cooperation weight is calculated, which comprehensively considers the distance between nodes, the health state of the node and the data reliability; Each node selects an optimal neighbor set according to the cooperation weight, and performs weighted fusion on the data of the neighbor nodes to generate a local cooperative air state vector; The network dynamically maintains the topology according to the node health state and environmental changes, including removing failed nodes, adaptively reconstructing neighbor sets and weights when new nodes are added.

[0009] Preferably, step S2 specifically comprises: Map the correction parameters of each node to the cabinet coordinates, air supply and exhaust paths and cold aisle structure of the data center, and construct a dynamic correspondence framework of multi-dimensional air state and spatial layout; The change rate, change direction and local disturbance persistence of the air parameters are analyzed in a continuous time window sliding manner, and the micro-scale diffusion or aggregation trend is identified in the spatial distribution framework; By constructing a cross-regional factor chain, the cabinet heat load change, air duct resistance, adjacent equipment operation fluctuation external factors and air quality trend are coupled and evaluated to quantify the potential driving force of abnormal formation and diffusion; Based on the disturbance trend and factor risk quantization results, the comprehensive credibility and possible causes of the air quality anomaly are output by the fusion probability model.

[0010] Preferably, the correction parameters of each node are mapped to the rack coordinates, air supply and exhaust paths, and cold aisle structure of the data center to construct a dynamic correspondence framework of multi-dimensional air state and spatial layout, specifically including: The corrected multi-dimensional vector and local anomaly label are taken out from the node data packet, and the node and neighborhood consistency residual is calculated; The initial credibility score is calculated, and the local smoothing is performed by using the inter-node adjacency weight to obtain the final credibility score; Based on the inter-node adjacency weight and credibility, an observation weight matrix is formed, and spatial and temporal analysis is performed by combining graph Laplace smoothing and temporal smoothing.

[0011] Preferably, step S3 specifically includes: According to the historical fusion index sequence and abnormal events, the regional mean and standard deviation are calculated, and the initial threshold value of each region is generated by weighted combination; The contribution weight of each component to the threshold value is corrected in real time by using the regional uncertainty, and the stability is ensured by combining the historical weight; According to the adjusted weight and fusion index, the threshold value is updated rolling, and the sudden abnormal event is amplified and corrected; The false positives and false negatives are counted, the rolling update rate and safety factor are adjusted by the false positive and false negative indicators, and closed-loop optimization is realized.

[0012] Preferably, according to the historical fusion index sequence and abnormal events, the regional mean and standard deviation are calculated, and the initial threshold value of each region is generated by weighted combination, specifically including: The fusion index sequence and corresponding abnormal event label in the historical period are selected, and the statistical distribution including mean and standard deviation is calculated for each region; The initial threshold value is defined based on the weighted combination of mean and standard deviation, and the safety factor is considered; The initial threshold value is generated by weighted combination of multi-dimensional components, and the component weight is based on historical statistics and risk correlation.

[0013] Preferably, step S4 specifically includes: By comparing the regional index with the dynamic threshold value and combining the uncertainty correction rule, the environmental state is divided into normal, mild, moderate, severe warning levels, etc.; By setting the time accumulation coefficient and the neighborhood coordination factor, the persistence and spatial consistency of the warning trigger are judged to avoid false positives caused by short-term fluctuations or local incidents; According to the matching of warning levels and trigger results, different levels of execution actions are matched, and the strategy strength is dynamically adjusted according to the importance of the region and the real-time state; Based on the actual alarm effect, false positive and false negative situation, and air quality recovery situation, the threshold interval, trigger parameter and alarm priority are automatically optimized.

[0014] Preferably, the early warning trigger is continuously and spatially consistent by setting a time accumulation coefficient and a neighborhood synergy factor, specifically including: Introducing a time accumulation factor to calculate the proportion of the past sliding window that meets the threshold condition; Introducing a neighborhood synergy factor to calculate the proportion of the threshold region existing in the neighborhood at the current time; Combining time accumulation and neighborhood synergy, setting the trigger condition, triggering the early warning when one of the continuous judgment or neighborhood resonance meets the trigger condition.

[0015] The technical scheme of the application: an intelligent early warning system for air quality in a data center based on a sensor network, which is used to execute the above-mentioned intelligent early warning method for air quality in a data center based on a sensor network, comprising: Sensor nodes and self-organizing network modules are responsible for actual physical perception and reliable collection, including multiple types of sensor nodes and low-power wireless self-organizing gateways, nodes have local self-checking, health scoring, positioning identification, adjustable sampling rate and node end correction capability, support for inter-node collaboration weight calculation and redundant collection strategy; Edge preprocessing and multi-dimensional fusion module, deployed in the edge server, receives the corrected multi-dimensional state and local anomaly label, completes the credibility evaluation, neighborhood credibility propagation, graph time sequence weighted interpolation and robust fusion, outputs the fusion estimation with uncertainty quantification and regional level index on the continuous space; Adaptive threshold and prediction decision module, using regional index, uncertainty and historical event knowledge base to build and evolve dynamic threshold, including component weight adapter, threshold rolling updater, burst amplification and buffer strategy, and short-term trend prediction engine based on time sequence model, realizing threshold verification feedback and online learning mechanism; Multi-level early warning and intelligent alarm execution module, responsible for converting threshold decision into executable operation and maintenance actions, including early warning level determinator, time accumulation and neighborhood synergy trigger, alarm strategy library, alarm issuing and execution interface, and closed-loop feedback collection unit.

[0016] Compared with the prior art, the above technical scheme of the application has the following beneficial technical effects: This invention designs an intelligent early warning method and system for data center air quality based on sensor networks. By constructing a self-organizing collaborative sensor network, it achieves high-precision and high-reliability real-time monitoring of air quality in key areas of the data center, effectively overcoming the limitations of single-point sensing being susceptible to interference and data silos. Through dynamic weight fusion between nodes and a local anomaly identification mechanism, the quality and reliability of the raw data are significantly improved, laying a solid foundation for subsequent analysis. A dynamic correlation model is established between air quality parameters and the spatial structure and airflow organization of the computer room, which can accurately analyze the diffusion path and trend of abnormal pollution, achieving a leap from perception to cognition. Utilizing an adaptive dynamic threshold mechanism, it can continuously self-optimize based on environmental changes and historical feedback, effectively balancing the sensitivity and stability of the early warning and significantly reducing false alarms and missed alarms. Through intelligent hierarchical early warning and closed-loop feedback execution, the system transforms monitoring results into operable operation and maintenance instructions, realizing full automation and intelligence from risk perception to control and disposal, significantly improving the initiative and operation and maintenance efficiency of data center environmental management. Attached Figure Description

[0017] Figure 1 This is a flowchart of a data center air quality intelligent early warning method based on sensor networks proposed in this invention. Figure 2 This is a system architecture diagram of a data center air quality intelligent early warning system based on sensor networks proposed in this invention. Detailed Implementation

[0018] Example 1, as Figure 1 As shown, the present invention proposes a data center air quality intelligent early warning method based on sensor networks, which includes the following specific implementation steps: S1. By deploying multiple types of sensor nodes in key areas of the data center, multi-dimensional air data such as temperature, humidity, particulate matter, gas concentration, and wind speed and direction are collected. Combined with a self-organizing collaborative network of nodes, real-time acquisition, node-end calibration, and preliminary anomaly identification, high precision, real-time performance, and reliability of air quality monitoring are achieved, providing a foundation for subsequent data fusion and intelligent early warning. The specific implementation process is as follows: S11. Deploy temperature, humidity, particulate matter, gas, and airflow sensors in the cold aisle, hot aisle, and rack aisles of the computer room. Integrate multi-dimensional information into node state vectors, and achieve refined spatial monitoring through high-density deployment in key areas to improve the integrity of air quality characterization. Specifically: Multiple types of sensor nodes are deployed in key areas of the computer room (cold aisle, hot aisle, server rack aisle, air conditioning vents, and underfloor return air area). Each node collects the following information: Ambient temperature relative humidity , concentration of fine particulate matter (PM2.5, PM10) in the air , monitoring , concentration of VOC and other gases , measuring air flow rate , and wind direction ; Constructing a single-node air state vector : ; Wherein n represents the number of chemical gas species collected; represents the sensor node in the machine room plane coordinate position; t represents the collection time stamp; S12, construct a node self-organizing and cooperative network, each node forms a self-organizing network through low-power wireless communication, dynamically calculates the cooperation weight according to the distance and health state between nodes, carries out local data fusion and preliminary identification of abnormality, improves the monitoring reliability and response speed of key area, specifically: Before each sensor node cooperates with the adjacent node in real time, the cooperation weight is calculated , considering the distance between nodes, node health state and data reliability: ; ; ; ; Among them, represents the state score of sensor j, which is evaluated according to whether each sensor of the node works normally and error drift, and 1 represents that all sensors work normally, ; represents the power supply state score, reflecting the stability and remaining power of the node power supply, 1 represents stable power supply, ; represents the communication state score, reflecting the communication stability of the node with the center or neighbor node, 1 represents stable communication without packet loss, ; 、 and represent weight coefficients, ; N represents the total number of historical data collection; represents the sensor historical factor; represents the cooperation weight of node i to neighbor node j; represents the Euclidean distance between node i and node j; represents the health state score of node j, ; represents the historical data reliability score of node j, which is calculated according to the stability and abnormal rate of past collected data, ; Each node selects the optimal neighbor set according to the cooperation weight : ; The node i performs weighted fusion on the data of its neighbor nodes to generate a local cooperative air state vector: ; Wherein, represents the cooperative neighbor set of node i; represents a set weight threshold for screening high-reliability neighbor nodes; represents the local cooperative air state vector of node i; represents the air state vector of neighbor node j; It should be noted that after the cooperation between nodes is completed, the network performs dynamic topology maintenance according to the node health state and environmental changes: the node can automatically adjust the cooperation relationship according to its own health state and neighbor state, and the failed node is temporarily excluded; when a new node joins or a key node fails, the network self-adapts the neighbor set and weight to ensure data coverage continuity; and periodically evaluates the contribution of each node, optimizes the communication path between nodes, and reduces network delay and energy consumption; S13, the node collects the air state vector at an adaptive sampling frequency, and attaches a unique ID and a time stamp, and the key area collects high frequency and can be redundantly collected by the adjacent nodes, to ensure data continuity and spatiotemporal accuracy, and to provide high-quality raw data for subsequent analysis, specifically: Each node automatically adjusts the collection frequency based on the local cooperative state vector and historical air quality fluctuation characteristics : ; Wherein, represents the node basic collection frequency, ensuring the minimum collection interval; represents the adjustment coefficient, which is used to amplify the influence of air quality fluctuation on the collection frequency; represents the standard deviation of the local cooperative state vector, which is used to measure the amplitude of air quality change; Based on the self-organizing network, when a node continuously collects abnormally or fails to communicate for S times, the neighbor node redundant collection mechanism is automatically started: ; Wherein, represents the redundant collection air state of node i when it is abnormal; The node performs self-monitoring during the collection process to generate a complete collection data packet : ; Wherein, represents the node i at time the collected data packet; represents the unique identification of the node; S14, the collected data is drift-corrected, filtered, delay-compensated and wind flow-smoothed at the node end, and uploaded to the central processing unit, while supporting preliminary abnormal early warning at the node end, improving data reliability and response efficiency, reducing central processing pressure, specifically: collected data complete preliminary processing and correction at the node end: signal filtering (such as temperature and humidity filtering, particulate matter smoothing), data format standardization and timestamp correction, and local anomaly marking (such as local peaks exceeding the threshold); The node uses the local fusion state output by the cooperative network to determine local anomalies with the corrected data: ; wherein, represents the local anomaly flag, 1 indicating abnormality and 0 indicating normality; represents the node's corrected multi-dimensional air state vector; and represents an adaptive threshold value, dynamically determined according to historical statistical characteristics and key area risk levels; The node uploads the corrected multi-dimensional air state and the local anomaly marker and node state information to form a data packet: ; wherein, represents the final data packet uploaded at the node end.

[0019] S2, by constructing a hierarchical air state analysis mechanism oriented to the operation characteristics of the data center, realizing the transformation from the original local monitoring results to the interpretable and inferable indoor air dynamic structure, establishing a dynamic correlation framework between air quality parameters and cabinet distribution, air duct structure in a spatial mapping manner, and then extracting the spreading trend of local anomalies in micro-scale space in a time-continuous change mode; based on the influence chain of cross-regional factors, an interference level evaluation model is constructed to structure and quantify the abnormal diffusion risk; the air disturbance trend and possible sources are comprehensively judged through probability fusion, and the specific implementation process is as follows: S21, by mapping the correction parameters of each node to the cabinet coordinates, air supply and exhaust paths, and cold aisle structure of the data center, a dynamic correspondence framework of multi-dimensional air state and spatial layout is constructed, realizing the structured positioning of air quality parameters, specifically: extract the corrected multi-dimensional vector and the local anomaly marker from the node packet, and calculate the node and neighborhood consistency residual : ; calculate the initial credibility score : ; To avoid extreme fluctuations of isolated nodes, a local smoothing is performed on the adjacency weights between nodes : ; wherein, denotes the local consistency residual; denotes a stabilization constant to prevent the denominator from being zero or too small, ensuring numerical stability; denotes a region typical amplitude for scale normalization (in this embodiment, the historical mean of the region is taken); denotes the initial (unsmoothed by neighborhood) raw trust score of node i; , , and denote the weight coefficients; denotes the health status score of node i; denotes the data reliability score of node i; denotes the confidence coefficient, controlling the mixing proportion of the node self-score and the neighborhood average trust score; denotes the normalized cooperation weight, i.e., the adjacency weight between nodes is normalized to obtain; S22, in a continuous time window sliding manner, analyzes the change rate, change direction and local disturbance persistence of the air parameters, and identifies whether there is a micro-scale diffusion or aggregation trend in the spatial distribution framework, specifically: based on the adjacency weight between nodes , and forms an observation weight matrix based on the trustworthiness: ; ; define the graph Laplacian , solve the spatial smoothing minimization problem (for each component): ; introduce an exponentially weighted history term for each grid point for time series smoothing: ; wherein, denotes the observation weight matrix, i.e., the diagonal matrix; L denotes the graph Laplacian matrix; D denotes the degree matrix, which is a diagonal matrix, and its elements are defined as , i.e., the sum of the connection weights of node i and all neighbors; W denotes the weighted adjacency matrix between nodes, and each element is ; denotes the spatial smoothing strength coefficient, controlling the trade-off between the observation fitting term and the graph smoothing term; represents vectorized observation data (concatenating all nodes of the same component into a column vector); represents a spatially smoothed estimate solution vector (corresponding to a smoothed version of ); represents temporal smoothing coefficients (exponential weights) for combining current time estimate with historical estimates ; S23, by constructing a cross-regional factor chain, the cabinet heat load change, air duct resistance, adjacent area equipment operation fluctuation and other external factors are coupled with the air quality trend to evaluate the potential driving force of abnormal formation and diffusion, specifically: According to the credibility and original label, the nodes are divided into three categories: trusted set : ; suspicious set : ; low credibility set : ; For each sensor , the difference between the neighborhood smoothed estimate and the local estimate is calculated: ; If (adaptive threshold, determined by the historical fluctuation standard deviation), it is determined as a real burst anomaly, otherwise it is considered as a sensor anomaly or noise; For each spatial point, the fusion value z is calculated using the weighted Huber loss: ; ; The IRLS (iterative reweighted least squares) is used to solve the above, and the initial weight is ; If a certain i is determined to be a real burst, the fusion result at this point allows higher sensitivity (local reduction, or weight temporary promotion), to ensure that the real scene mutation will not be smoothed out too much; wherein and represent the credibility threshold set for stratification, and in the embodiment , ; represents the difference between the node observation and the fusion estimate; represents an adaptive consistency threshold for comparing whether it is significant; denoted by Huber loss function, used for robust handling of residuals; z represents fusion estimate, which is the robust fusion value obtained by minimizing Huber loss on observations (weighted) at a certain spatial point or grid point; The switching point (threshold) is when the residual Use quadratic loss; otherwise, it grows linearly. S24. Based on the disturbance trend and factor risk quantification results, a fusion probability model is used to output the comprehensive credibility and possible causes of air quality anomalies, providing verifiable input for subsequent early warning strategies. Specifically: Define an interpretable indicator that transitions from multidimensional to scalar (regional AQI-like index): ; First, principal component analysis (PCA) is used to obtain the variance contribution rate of each component, and then its correlation with equipment risk is scored. Fusion, generating weights : ; Clustering nodes into risk level zones (based on) With uncertainty U), it is used for differentiated threshold management and alarm strategies, and uncertainty is quantified: ; Output triples for each region / grid point And return the node-level anomaly label (whether it is determined to be a real outbreak); in, This represents the normalized value of the m-th component after fusion. The comprehensive air quality index for region p; This represents the dynamic weight of the m-th component in the comprehensive index (summing to 1); M represents the total number of components. This represents the score of the m-th component in terms of its contribution to the principal components or variance (a data-driven importance measure). The equipment risk sensitivity score (engineering prior) represents the m-th component, characterizing the strength of the variable's impact on equipment health or failure risk; and Indicates the adjustment coefficient; This represents the quantified value of the uncertainty in region p; This represents the fusion center value of region p.

[0020] S3, build adaptive and dynamic evolution of air quality threshold, realize intelligent early warning of multi-region and multi-dimensional air state of data center, through initial threshold generation, component weight dynamic adjustment, threshold rolling evolution and feedback closed loop optimization, the threshold can consider real-time response, historical statistics and abnormal events, ensure that the early warning system is long-term stable, robust and intelligent, the specific implementation process is as follows: S31, according to the history of fusion index sequence and abnormal event, calculate the mean and standard deviation of the region, generate the initial threshold of each region by weighted combination, provide the basis for subsequent dynamic adjustment, at the same time consider the influence of multi-dimensional component, realize the targeted initial value setting, specific as follows: Select the fusion index sequence in the history period H And the corresponding abnormal event label (Such as alarm time point), calculate the statistical distribution of each region: ; ; Define initial threshold ; For multi-dimensional component, weighted combination can be used: ; Where, H represents the set of historical time period, used to calculate the historical distribution of regional fusion index; Represents the average value of the historical fusion index of region p; Represents the standard deviation of the historical fusion index of region p; Represents the average value function; Represents the standard deviation function; Represents the safety factor, which controls the degree of initial threshold higher than the historical average value; Represents the weight of multi-dimensional component, which is used to fuse the contribution proportion of multi-dimensional air state to the threshold; S32, use regional uncertainty to real-time correct the contribution weight of each component to the threshold, reduce the current measurement weight under high uncertainty, combine with the historical weight to ensure stability, realize the adaptive sensitivity of threshold to different regions and different states, specific as follows: Calculate the uncertainty adjustment coefficient: ; Update the component weight: ; Where, Represents the uncertainty adjustment coefficient; Represents the uncertainty sensitive adjustment parameter; Represents the reference uncertainty; Represents the historical average weight, used to compensate the current weight under high uncertainty; S33, according to the adjusted weight and fusion index, the threshold is updated rolling, and the sudden abnormal event is amplified and corrected, so as to realize the rapid response and long-term stable balance of the threshold to real-time changes, specifically: The threshold is updated rolling: ; If step S23 marks that the current area has a real sudden abnormal event , the threshold is temporarily lowered by a certain percentage to ensure that high-risk events trigger alarm in time, and the update strategy is: ; Wherein, represents the adaptive threshold of area p at time ; represents the rolling update rate, which controls the response speed of the threshold to new data; represents the abnormal amplification correction coefficient, which lowers the threshold proportion when a real abnormal event occurs; represents the mth component fusion index of area p at time ; represents the adaptive weight of the mth component at time ; S34, statistics of historical false positives and false negatives, adjustment of rolling update rate and safety coefficient through FP and FN indicators, closed-loop optimization; the threshold parameters evolve adaptively over time to improve the accuracy of early warning and the long-term stability of the system, forming a complete adaptive threshold management system, specifically: Select the historical window H as the verification interval to count false positives and false negatives: ; ; Feedback adjustment parameters: ; Update the safety coefficient : ; Wherein, represents the number of false negatives of area p (real abnormal events not triggered by the threshold); represents the number of false positives of area p (threshold triggered but no actual abnormal event); and represent the coefficient of feedback adjustment of rolling update rate; represents the weight parameter of feedback adjustment of safety coefficient ; represents the indicator function, which is used to count whether the event occurs (1 for occurrence, 0 for non-occurrence).

[0021] S4, through four levels of early warning level judgment, trigger condition optimization, alarm strategy generation and feedback self-learning, realize multi-level early warning trigger driven by dynamic threshold, and combine uncertainty, time accumulation and space neighborhood information to build stable and sensitive alarm process, and adjust threshold, trigger parameter and strategy priority according to alarm execution feedback, so that the system has the self-evolution ability to adapt to environmental changes, and the specific implementation process is as follows: S41, by comparing the regional index with the dynamic threshold and combining the uncertainty correction rule, the environment state is divided into normal, mild, moderate and severe early warning, so that different degrees of air quality deviation can be accurately classified; This process not only uses real-time index, but also includes threshold offset interval and credibility in judgment, so that the early warning level is sensitive and stable, and provides clear hierarchical basis for subsequent triggering and strategy generation, which is specific as follows: Define multi-level threshold interval: ; And introduce the uncertainty correction factor to correct the level: ; ; Get the modified continuous score and map it back to the discrete level, and round up to the nearest integer, and map the text level to an ordered score (normal = 0, mild = 1, moderate = 2, severe = 3): ; Among them, represents the early warning level of region p at time ; max represents the reference upper bound for normalizing uncertainty to [0,1], in this embodiment, the 95% quantile of historical uncertainty is selected to avoid being pulled up by extreme anomalies; represents the normalized uncertainty; represents the correction strength coefficient; represents the sign function, indicating the deviation direction of the current index relative to the threshold; represents the continuous level correction amount; represents the continuous score limited in the allowed level range (0 to 3); represents the modified continuous level score; represents the threshold lower bound buffer; represents the threshold upper bound buffer; S42, by setting time accumulation coefficient and neighborhood synergy factor, the persistence and spatial consistency of early warning trigger are judged to avoid false alarm caused by short-term fluctuations or local sporadic; the final trigger condition is established by one of the continuous determination or neighborhood resonance, which makes the early warning trigger more consistent with the air diffusion law and the actual machine room scene characteristics, and provides a reliable trigger signal for strategy execution, specifically: The time accumulation factor is introduced: ; The neighborhood synergy factor is introduced: ; The trigger condition is set by combining time accumulation and neighborhood synergy: ; Among them, represents the time accumulation factor, that is, the proportion that satisfies in the past sliding window ; represents the neighborhood synergy factor, that is, the proportion of the existence of the threshold region in the neighborhood at the current time ; represents the binary trigger identifier, which is 1 if the trigger condition is met, otherwise it is 0; represents the time accumulation trigger threshold; represents the neighborhood synergy trigger threshold; S43, according to the early warning level and the trigger result, the execution action of different levels is matched, and the strategy strength can be dynamically adjusted according to the importance of the region and the real-time state, so that the alarm not only stays in information prompt, but also is converted into executable operation and maintenance instruction, realizing the coherent closed loop from detection to disposal, specifically: The alarm strategy is defined according to the level and trigger identifier: ; The historical feedback mechanism is used to fine-tune the strategy response level, and the node end priority is introduced, for example, the key area triggers more stringent measures, and the alarm instruction is output accordingly; Among them, represents the alarm action or strategy to be executed for region p at time (for example, "prompt monitoring", "limit ventilation / start air purification", "comprehensive emergency response"); S44, based on the actual alarm effect, false alarm and air quality recovery, the threshold interval, trigger parameter and alarm priority are automatically optimized to make the early warning system have continuous evolution ability; by regularly counting the accuracy and adjusting the update rate and strategy sensitivity, the system always keeps synchronized with the change of the machine room environment, and can maintain high reliability and low false alarm rate in long-term operation.

[0022] Embodiment two, as Figure 2As shown, the application proposes a sensor network-based data center air quality intelligent early warning system, which is used to implement the sensor network-based data center air quality intelligent early warning method proposed in embodiment one, and includes a sensor node and self-organizing network module, an edge preprocessing and multi-dimensional fusion module, an adaptive threshold and prediction decision module, and a multi-level early warning and intelligent alarm execution module.

[0023] The sensor node and self-organizing network module is responsible for actual physical sensing and reliable collection, and includes multiple types of sensor nodes (temperature and humidity, particulate matter, gas, wind speed and direction, etc.) and low-power wireless self-organizing gateways. The nodes have local self-checking, health scoring, positioning identification, adjustable sampling rate, and node end correction capability, support inter-node collaboration weight calculation and redundant collection strategy, and provide edge time synchronization and secure communication channel, and simultaneously include an on-site operation subsystem for automatic calibration, filter replacement reminder, and physical protection mechanism. The edge preprocessing and multi-dimensional fusion module is deployed on an edge server, receives the corrected multi-dimensional state and local anomaly label, completes credibility evaluation, neighborhood credibility propagation, graph-time sequence weighted interpolation, and robust fusion, outputs spatially continuous and uncertainty-quantized fusion estimates and regional-level indicators, and generates a fused data packet for use by the upper layer. The adaptive threshold and prediction decision module uses regional indicators, uncertainty, and a historical event knowledge base to construct and evolve dynamic thresholds, including a component weight adapter, a threshold rolling updater, a burst amplification and buffering strategy, and a short-term trend prediction engine based on a time series model. The module simultaneously implements threshold verification feedback and online learning mechanism, uses false / missed report statistics for parameter self-optimization, and thus outputs adaptive thresholds and confidence information for each region. The multi-level early warning and intelligent alarm execution module is responsible for converting threshold decisions into executable operation actions, including but not limited to an early warning level determiner, a time accumulation and neighborhood coordination trigger, an alarm strategy library (including hierarchical response action sets such as prompt monitoring, local ventilation adjustment, air purifier start / stop, cabinet load limiting, etc.), an alarm issuing and execution interface (facing management terminals, mobile push, and on-site controllers), and a closed-loop feedback collection unit for recording processing effects and feedback learning.

[0024] The embodiments of the application are described in detail above in combination with the drawings, but the application is not limited thereto, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the purpose of the application.

Claims

1. A data center air quality intelligent early warning method based on sensor networks, characterized in that, The specific implementation steps include the following: S1. Deploy multiple types of sensor nodes in key areas of the data center to build a self-organizing collaborative network, collect multi-dimensional air data in real time, and perform node-end correction and local anomaly identification on the collected data. S2. Improve data reliability by dynamic weight fusion between nodes and local anomaly identification, and establish a correlation model between air parameters and computer room structure by using spatial mapping and time series analysis to identify microscale diffusion trends. S3. Construct an adaptive dynamic threshold mechanism, combining historical statistics and real-time feedback to achieve threshold rolling optimization; S4. Generates a graded alarm strategy based on multi-level early warning judgment and triggering conditions, and continuously optimizes the early warning accuracy and response efficiency through closed-loop feedback.

2. The intelligent early warning method for data center air quality based on sensor networks according to claim 1, characterized in that, Step S1 specifically includes: Temperature and humidity sensors, particulate matter sensors, gas concentration sensors, and wind speed and direction sensors are installed in the cold aisle, hot aisle, server rack aisle, air conditioning outlet, and underfloor return air area of ​​the computer room. Construct an air state vector for a single node, which includes ambient temperature, relative humidity, particulate matter concentration, gas concentration, air velocity, and wind direction; Construct a node self-organizing and cooperative network. Each node forms a self-organizing network through low-power wireless communication. The cooperative weight is dynamically calculated based on the distance between nodes and their health status. Local data fusion and preliminary anomaly identification are then performed. The nodes collect air state vectors at an adaptive sampling frequency and attach a unique identifier and timestamp. Key areas are collected at high frequency and redundantly supplemented by neighboring nodes. The collected data undergoes drift correction, filtering, delay compensation, and airflow smoothing at the node end, and also supports preliminary anomaly warning at the node end.

3. The intelligent early warning method for data center air quality based on sensor networks according to claim 2, characterized in that, Constructing a self-organizing and collaborative network of nodes specifically includes: Before each sensor node collaborates with its neighboring nodes in real time, a collaboration weight is calculated. This weight takes into account the distance between nodes, the health status of nodes, and the reliability of data. Each node selects the optimal set of neighbors based on the cooperation weight, and performs weighted fusion of the data of the neighboring nodes to generate a local cooperative air state vector; The network performs dynamic topology maintenance based on node health status and environmental changes, including removing failed nodes, adding new nodes, and adaptively reconstructing the neighbor set and weights.

4. The intelligent early warning method for data center air quality based on sensor networks according to claim 3, characterized in that, Step S2 specifically includes: The calibration parameters of each node are mapped to the rack coordinates, air supply and exhaust paths and cold aisle structure of the data center, thus constructing a dynamic correspondence framework between multi-dimensional air conditions and spatial layout. By using a continuous time window sliding method to analyze the rate of change, direction of change, and persistence of local disturbances of air parameters, microscale diffusion or aggregation trends can be identified within the spatial distribution framework. By constructing a cross-regional factor chain, external factors such as cabinet heat load changes, air duct resistance, and fluctuations in the operation of neighboring equipment are coupled with air quality trends for evaluation, thereby quantifying the potential driving forces for the formation and spread of anomalies. Based on the perturbation trend and factor risk quantification results, a fusion probability model is used to output the comprehensive credibility and possible causes of air quality anomalies.

5. The intelligent early warning method for data center air quality based on sensor networks according to claim 4, characterized in that, The calibration parameters of each node are mapped to the rack coordinates, air supply and exhaust paths, and cold aisle structure of the data center, constructing a dynamic correspondence framework between multi-dimensional air conditions and spatial layout, specifically including: Extract the corrected multidimensional vector and local anomaly marker from the node data packet, and calculate the node-neighborhood consistency residual; Calculate the initial confidence score and perform local smoothing using the adjacency weights between nodes to obtain the final confidence score; An observation weight matrix is ​​formed based on the adjacency weights and credibility between nodes. Spatial and temporal analysis is then performed by combining graph Laplace smoothing and temporal smoothing.

6. The intelligent early warning method for data center air quality based on sensor networks according to claim 5, characterized in that, Step S3 specifically includes: Based on historical fusion index sequences and outliers, calculate the regional mean and standard deviation, and generate the initial threshold for each region through weighted combination. The contribution weights of each component to the threshold are adjusted in real time using regional uncertainty, and stability is ensured by combining historical weights. The thresholds are updated on a rolling basis based on the adjusted weights and fusion index, while amplification corrections are made for sudden abnormal events. By statistically analyzing historical false alarms and missed alarms, and adjusting the rolling update rate and safety factor based on the false alarm and missed alarm indicators, closed-loop optimization can be achieved.

7. The intelligent early warning method for data center air quality based on sensor networks according to claim 6, characterized in that, Based on historical fusion index sequences and outliers, the regional mean and standard deviation are calculated, and an initial threshold for each region is generated through weighted combination, specifically including: Select the fusion index sequence and corresponding abnormal event labels within the historical time period, and calculate the statistical distribution for each region, including the mean and standard deviation; The initial threshold is defined based on a weighted combination of the mean and standard deviation, with a safety factor taken into account. An initial threshold is generated by weighting the multidimensional components, with the component weights based on historical statistics and risk correlation.

8. The intelligent early warning method for data center air quality based on sensor networks according to claim 7, characterized in that, Step S4 specifically includes: By comparing regional indices with dynamic thresholds and combining uncertainty correction rules, environmental conditions are classified into normal, mild, moderate, and severe warning levels. By setting a time accumulation coefficient and a neighborhood coordination factor, the continuity and spatial consistency of the early warning trigger are judged to avoid false alarms caused by short-term fluctuations or local sporadic events. Different levels of execution actions are matched according to the warning level and the triggering result, and the strategy intensity is dynamically adjusted according to the importance of the area and the real-time status. Based on the actual alarm effect, false alarms and missed alarms, and air quality recovery, the threshold range, trigger parameters, and alarm priority are automatically optimized.

9. The intelligent early warning method for data center air quality based on sensor networks according to claim 8, characterized in that, By setting a time accumulation coefficient and a neighborhood coordination factor, the persistence and spatial consistency of early warning triggering are judged, specifically including: Introduce a time accumulation factor to calculate the proportion of times the threshold condition is met within the past sliding window; Introduce a neighborhood collaboration factor to calculate the proportion of regions exceeding the threshold within the neighborhood at the current time; By combining time accumulation and neighborhood collaboration, trigger conditions are set, and an early warning is triggered when either continuous judgment or neighborhood resonance is satisfied.

10. A sensor network-based intelligent early warning system for data center air quality, used to execute the sensor network-based intelligent early warning method for data center air quality as described in any one of claims 1 to 9, characterized in that, include: The sensor node and self-organizing network module is responsible for actual physical sensing and reliable data acquisition. It includes multiple types of sensor nodes and low-power wireless self-organizing gateways. The nodes have local self-testing, health scoring, positioning identification, adjustable sampling rate and node-end calibration capabilities, and support inter-node collaborative weight calculation and redundant acquisition strategies. The edge preprocessing and multidimensional fusion module is deployed on the edge server. It receives the corrected multidimensional state and local anomaly markers, completes credibility assessment, neighborhood credibility propagation, graph time-series weighted interpolation and robust fusion, and outputs a spatially continuous fusion estimate with uncertainty quantification and a region-level index. The adaptive threshold and prediction decision module uses regional indicators, uncertainty and historical event knowledge base to build and evolve dynamic thresholds. It includes component weight adaptor, threshold rolling updater, burst amplification and buffering strategy, and short-term trend prediction engine based on time series model to realize threshold verification feedback and online learning mechanism. The multi-level early warning and intelligent alarm execution module is responsible for transforming threshold decisions into executable operation and maintenance actions. It includes an early warning level determiner, time accumulation and neighborhood collaboration triggers, alarm policy library, alarm issuance and execution interface, and closed-loop feedback acquisition unit.

Images