Server cabinet fault monitoring system based on big data

Abstract

The invention particularly relates to a server cabinet fault monitoring system based on big data, and relates to the technical field of server cabinet fault monitoring. A data preprocessing and storage module; the core intelligent analysis and diagnosis module is configured to realize sensor fault identification based on preprocessed data and a big data model; and an alarm and visualization interaction module. According to the invention, temperature and light intensity correlation analysis and a multi-dimensional data cross validation mechanism break through the limitation of traditional single temperature threshold judgment; on one hand, temperature data are collected by combining a contact type sensor, a non-contact type sensor and a redundant sensor, parameters such as light intensity and attenuation rate are synchronously obtained through a laser transmitter and a photoelectric detector, a correlation model of temperature and light intensity is constructed, and real temperature abnormity and sensor faults can be accurately distinguished; and on the other hand, associated data such as sensor power supply voltage and communication signal strength are synchronously collected, and a basis is provided for fault traceability.

Description

Technical Field

[0001] The present invention relates to the technical field of server cabinet fault monitoring, and in particular to a server cabinet fault monitoring system based on big data. Background Art

[0002] With the rapid development of the digital economy, the equipment density and computing power requirements of data center server cabinets continue to rise. Temperature stability within the cabinets has become a key factor in ensuring reliable equipment operation. Current mainstream cabinet fault monitoring systems primarily rely on temperature sensors to collect data and determine device status based on a single temperature threshold. However, this model has significant limitations: On the one hand, it is impossible to effectively distinguish between the actual temperature abnormality of the cabinet and the failure of the temperature sensor itself (such as zero point drift, poor line contact, and aging failure), which often leads to waste of operation and maintenance resources due to sensor false alarms, or equipment overheating and shutdown due to sensor omissions; on the other hand, only collecting temperature values ​​and ignoring related parameters such as sensor power supply voltage and communication signal strength makes it difficult to trace the root cause of the fault, increasing the complexity of problem location.

[0003] At the same time, existing analysis models lack dynamic adaptability and are unable to cope with complex scenarios in the long-term operation of cabinets: For example, cabinet load fluctuations with business (such as the difference in computing power between weekday peaks and nighttime lows) will cause dynamic changes in the temperature benchmark, and equipment aging (such as sensor accuracy degradation and decreased laser transmitter light intensity) will gradually make the original threshold invalid. Traditional fixed models cannot adjust the judgment logic in real time, resulting in a continuous decline in long-term monitoring accuracy.

[0004] In addition, alarm information is mostly presented in a single text format, lacking a grading mechanism and visual support. It is difficult for operation and maintenance personnel to quickly locate the fault location and assess the scope of impact, which further extends the fault response time and poses a potential risk to the continuous operation of the data center. Summary of the Invention

[0005] The purpose of the present invention is to solve the above problems and to propose a server cabinet fault monitoring system based on big data.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: A server cabinet fault monitoring system based on big data, comprising: The data acquisition and synchronization module is configured to capture the status data of the temperature sensors in the cabinet through multi-dimensional hardware deployment and precise synchronization mechanism; The data preprocessing and storage module is configured to perform deep cleaning and feature engineering on the raw data, build data sets, and enable storage and fast retrieval; The core intelligent analysis and diagnosis module is configured to realize sensor fault identification based on pre-processed data and big data models; The alarm and visualization interaction module is configured to convert the analysis results into alarm information and visualization charts to support operation and maintenance decisions and realize human-computer interaction.

[0007] Preferably, the data acquisition and synchronization module specifically includes: Deploy sensors in key areas based on cabinet size, including contact sensors, non-contact infrared sensors, and redundant backup sensors; Synchronously collect sensor operating voltage, communication signal strength, and installation coordinates; Red laser and silicon-based photodetector are installed in pairs on both sides of the temperature sensor; Collect relevant parameters, including original light intensity value, light intensity fluctuation frequency, ambient light background value, and laser transmitter operating current; Record the initial laser transmittance of each sensor during installation; Add a timestamp to each frame of data to ensure the time alignment of temperature and light intensity data, and transmit the data; Complete data integrity verification on the gateway side, and cache and prioritize uploading abnormal values ​​exceeding the threshold locally.

[0008] Preferably, the data preprocessing and storage module specifically includes: Clean the data, including noise suppression, missing value processing, and outlier identification; Extract features, including temporal feature extraction and spatial feature extraction; Store data and preprocessing results within a preset time period; store historical data, segmented by time; store extracted feature vectors for model training and trend analysis; A three-level index is constructed based on cabinet ID, sensor ID, and timestamp to support multi-dimensional retrieval.

[0009] Preferably, the core intelligent analysis and diagnosis module specifically includes: A temperature-light intensity benchmark correlation model is constructed using historical normal data, and then the model boundaries and judgment logic are iteratively optimized using abnormal data to achieve accurate identification of temperature sensor faults when the laser is normal.

[0010] Preferably, the step 1: establishing a basic model based on historical normal data specifically includes: Select stable operating data within a preset time period after the sensor is installed, and it must include a complete load cycle; Exclude data from the following periods: Abnormal laser light intensity or temperature sensor data deviates from the normal range; abnormal cabinet environment; abnormal data collection; Model composition: Temperature single-dimensional normal baseline: Statistics of temperature distribution in each period, fitting temperature and load correlation curves; output temperature normal range and load response threshold; Light intensity single-dimensional normal baseline: statistics on long-term light intensity attenuation trend, calculation of light intensity fluctuation characteristics; output light intensity normal range and fluctuation threshold; Temperature and light intensity correlation model: The weak correlation law is calculated by Pearson correlation coefficient and the correlation formula is fitted by multiple linear regression: Light intensity prediction value = α·temperature + β·initial light intensity + error term; Output correlation coefficient range, normal fluctuation range of light intensity with temperature.

[0011] Preferably, the step 2: optimizing the model using abnormal data specifically includes: Manually label the following two types of abnormal data to extract fault features that can be used for model optimization: The temperature sensor is normal but the laser is abnormal: the annotation content includes the laser fault type, the start and end time of the abnormal light intensity, and the change trend; the key features extracted are the abnormal light intensity feature and the normal temperature feature; The temperature sensor is abnormal but the laser is normal: the annotation content includes the temperature sensor fault type and the manifestation of the temperature abnormality; the key features extracted are the temperature abnormality feature and the normal light intensity feature; Correct the temperature anomaly judgment boundary: Use the data of abnormal temperature sensor but normal laser to calibrate the temperature anomaly judgment threshold: Strengthen the reliability of determining normal light intensity: Use data showing normal temperature sensors but abnormal lasers to clarify the core characteristics of normal lasers; Construct a feature template for temperature failure when the laser is normal: Extract typical features from data where the temperature sensor is abnormal but the laser is normal, and form a dedicated judgment template.

[0012] Preferably, the step three: after the optimization model determines that the temperature sensor is faulty when the laser is normal, the specific process includes: Through the basic model and optimized feature template, accurate judgment is achieved according to the three-step process of normal light intensity verification, temperature anomaly identification, and fault type matching; A1: Verify whether the laser is working properly: Core verification conditions: Light intensity data is within the single-dimensional baseline range, and the decay rate is less than 0.3% / day; The light intensity fluctuations conform to normal patterns, and the fluctuation period matches the cabinet vibration; The correlation between light intensity and temperature is consistent with the basic model, with no correlation break; A2: Identify whether the temperature sensor is abnormal: Abnormal identification conditions: The temperature continuously deviates from the optimized baseline without synchronous drift of adjacent sensors; Temperature does not respond to load changes; Temperature timing characteristics match fault templates; A3: Matching temperature sensor fault type: By comparing the abnormal temperature characteristics with the temperature fault template when the laser is normal, the specific fault type and confidence level are output.

[0013] Preferably, the step 4: continuous model optimization mechanism specifically includes: Retrain the model weekly using newly generated normal data and confirmed abnormal data to update the correlation coefficient between temperature and light intensity; When manual review finds a misjudgment, it is immediately added to the negative sample library and the judgment weight of the corresponding feature is adjusted; For different cabinet types, sub-models are trained separately to avoid cross-scenario misjudgments.

[0014] Preferably, the alarm and visualization interaction module specifically includes: Convert analysis results into intuitive alarm information and visual charts; match the corresponding warning level based on the abnormal state of sensor monitoring, and issue corresponding level warning; The operation and maintenance personnel enter the processing results of the faulty sensor, adjust the alarm threshold, initiate calibration instructions, and upload the entered information to the cloud. The management personnel download the corresponding faulty sensor information remotely, compile statistics on the historical data of the faulty sensor, and conduct corresponding analysis.

[0015] Preferably, the system management and external integration module is configured to ensure stable operation of the system through device resource management and model parameter management, connect to external systems of dynamic environment and operation and maintenance work orders, and ensure data security and operational compliance through user authority management.

[0016] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. The present invention breaks through the limitations of traditional single temperature threshold judgment through temperature and light intensity correlation analysis and multi-dimensional data cross-validation mechanism. On the one hand, it combines contact, non-contact and redundant sensors to collect temperature data, and simultaneously obtains parameters such as light intensity and attenuation rate through laser emitters and photoelectric detectors to construct a correlation model between temperature and light intensity, which can accurately distinguish between real temperature anomalies and sensor faults themselves. On the other hand, it synchronously collects related data such as sensor power supply voltage and communication signal strength to provide a basis for fault tracing.

[0017] 2. The present invention dynamically adapts to long-term scenarios such as cabinet load fluctuations and equipment aging through an iterative optimization mechanism based on basic models and abnormal data: it builds temperature and light intensity baselines based on historical normal data, and continuously corrects judgment thresholds and feature templates based on abnormal data to ensure the model's adaptability to load changes and sensor aging; at the same time, through hierarchical alarms and visual interaction, operation and maintenance personnel can quickly locate the fault location and assess the scope of impact. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Further details, features and advantages of the present application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 This is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0019] Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and for many different purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

[0020] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the relevant art and / or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless expressly defined as such herein. Example 1

[0021] The specific implementation method is combined with the attached Figure 1 Provide detailed explanation.

[0022] Attachment Figure 1 A structural block diagram of a server cabinet fault monitoring system based on big data provided by an embodiment of the present invention shows the connection relationship between the data acquisition and synchronization module and the alarm and visualization interaction module, and marks the main functional interaction process of each module.

[0023] In this embodiment, it includes: The data acquisition and synchronization module is configured to achieve full-scale data capture of the temperature sensor status in the cabinet through multi-dimensional hardware deployment and precise synchronization mechanism, providing the original basis for subsequent analysis; Specifically include: Deploy sensors, including contact sensors, non-contact infrared sensors, and redundant backup sensors, in key areas (such as CPU-intensive areas, power module areas, and heat dissipation blind spots) based on cabinet size (e.g., a 42U standard cabinet). Contact sensor (such as PT100 platinum resistance): directly attached to the surface of the device, collects precise temperature (accuracy ±0.1°C), and the sampling frequency is set to 5Hz (to capture instantaneous temperature fluctuations); Non-contact infrared sensor: installed on the cabinet beam, collects the average temperature of the area (covering a 30cm radius), with a sampling frequency of 1Hz (to reduce redundant data); Redundant backup sensors: Deploy one or two additional sensors of the same type next to core devices (such as the main server) for cross-verification.

[0024] In addition to the temperature value (°C), the sensor's operating voltage (to determine power supply stability), communication signal strength (to determine line connection status), and installation coordinates (accurate to U position + left and right offset in mm) are also collected simultaneously. A 650nm red laser (5mW output power to avoid thermal effects) and a silicon-based photodetector (response time <10μs) are installed in pairs on both sides of the temperature sensor (the angle between the laser path and the normal to the sensor surface is ≤5°); Collect relevant parameters, including the original light intensity value (mW / cm²), light intensity fluctuation frequency (Hz, reflecting whether the sensor is vibrating), ambient light background value (double compensation through the light shield and algorithm), and laser transmitter operating current (to determine whether the device is aging); During installation, record the initial laser transmittance of each sensor (unobstructed and at a standard angle) as a benchmark for subsequent evaluation (±3% is the normal deviation range). Adopting the NTP network time protocol (synchronization accuracy ≤ 1ms), an accurate timestamp is added to each frame of data to ensure the time alignment of temperature and light intensity data (an error exceeding 50ms is marked as synchronization failure), and the data is transmitted; Close-range cabinets (in the same computer room): Directly connected via industrial Ethernet (Gigabit bandwidth), using the MODBUSTCP protocol to encapsulate data frames (including check bits to prevent transmission errors); Long-distance cabinet groups (across floors / campuses): Use LoRaWAN wireless transmission (transmission distance 2-5km), and data is compressed by the edge gateway (compression rate ≥ 30%) before uploading to reduce bandwidth usage; Data integrity verification is completed on the gateway side (data packets with missing key fields are discarded), and abnormal values ​​exceeding the threshold (such as temperature > 80°C) are locally cached and uploaded preferentially to ensure the real-time delivery of urgent data.

[0025] The data preprocessing and storage module is configured to perform in-depth cleaning and feature engineering on raw data, build high-quality data sets, and achieve efficient storage and fast retrieval to support subsequent analysis and modeling; Specifically include: Clean the data, including noise suppression, missing value processing, and outlier identification; Noise suppression: Temperature data: Kalman filtering is used (combined with the cabinet fan speed cycle, the filter coefficient is dynamically adjusted) to eliminate high-frequency fluctuations of ±1°C caused by airflow disturbances. Light intensity data: The 3σ criterion is used to remove impulse noise (such as instantaneous dust obstruction), and the original value is retained for continuous low light intensity (>5s) (which may be real obstruction).

[0026] Missing value handling: Short-term missing data (<10s): linear interpolation is used (based on the 10 valid data points before and after); Long-term absence (>10s): Start filling the backup sensor data and mark the "Rely on backup source" label.

[0027] Outlier identification: The IQR (interquartile range) method is used to define a reasonable range (temperature: cabinet average temperature ±5°C; light intensity: 70% to 130% of the initial value). Out-of-range values ​​are marked as "unverified anomalies" and are not directly eliminated (potential fault clues are retained).

[0028] Extract features, including temporal feature extraction and spatial feature extraction; Time series feature extraction: Temperature characteristics: maximum / minimum value / standard deviation within a 1-minute sliding window, temperature rise rate (°C / min, exceeding 5°C / min is considered abnormal), temperature phase offset within a 24-hour period (compared with the same period in history); Light intensity characteristics: light intensity attenuation rate (daily attenuation / initial value), cross-correlation coefficient between temperature and light intensity (should be weakly correlated under normal installation, |r| < 0.3), number of light intensity mutations (> 3 times within 1 hour is considered abnormal).

[0029] Spatial feature extraction: Calculate the temperature gradient (°C / m) and light intensity difference between adjacent sensors (compared with the theoretical attenuation model) to identify abnormal clustering patterns in local areas (e.g., simultaneous light intensity drops across multiple sensors in a certain area).

[0030] Stores data and preprocessing results within a preset time period (supports millisecond-level queries) to meet real-time monitoring needs; stores historical data (retained for one year), sharded by time (one data block per hour), and supports high compression ratios (compression ratio ≥ 10:1); and stores extracted feature vectors (such as daily temperature fluctuation characteristics) for model training and trend analysis. A three-level index is built based on cabinet ID, sensor ID, and timestamp, supporting multi-dimensional retrieval (e.g., "Query all sensors in cabinet A with a light intensity attenuation rate greater than 2% / day"). The query response time is ≤ 1 second.

[0031] The core intelligent analysis and diagnosis module is configured to realize sensor fault identification based on pre-processed data and big data models; Specifically include: A temperature-light intensity benchmark correlation model is constructed using historical normal data, and then the model boundaries and judgment logic are iteratively optimized using abnormal data to achieve accurate identification of temperature sensor faults when the laser is normal.

[0032] Step 1: Establish a basic model based on historical normal data, including: Select stable operating data within a preset time period after sensor installation (avoiding the initial installation and commissioning period), and include a complete load cycle (such as weekdays / weekends, peak / off-peak periods); Data from the following periods were excluded to avoid contaminating the baseline: Abnormal laser light intensity (e.g., attenuation > 5%, fluctuation > 1%), or temperature sensor data deviating from the normal range (e.g., exceeding the baseline by ±2°C); abnormal cabinet environment (e.g., air conditioner failure, fan stoppage, or obstruction by newly added equipment); or abnormal data collection (e.g., communication interruption, synchronization error > 50ms). Sample size requirement: The normal data volume corresponding to each sensor must be ≥ 100,000 (calculated at a 1Hz sampling frequency, approximately 12 days of data), ensuring coverage of all normal scenarios (different load, temperature, and vibration conditions); The basic model consists of three parts: "temperature single-dimensional normal baseline", "light intensity single-dimensional normal baseline" and "temperature-light intensity correlation model", forming a three-dimensional judgment benchmark; Model composition: Temperature single-dimensional normal baseline: Calculates temperature distribution for each time period (e.g., mean 35°C, standard deviation 2°C from 9:00 AM to 6:00 PM on weekdays), fits temperature and load correlation curves (from CPU usage to temperature predictions), outputs normal temperature range (e.g., 35°C ± 3°C), and load response thresholds (e.g., if the usage increases by 10%, the temperature should rise by 0.5°C). Normal baseline of single-dimensional light intensity: Statistics on long-term light intensity attenuation trend (normal ≤0.1% / day), calculation of light intensity fluctuation characteristics (standard deviation within 1 minute ≤0.2mW / cm²); output light intensity normal range (initial value ±3%), fluctuation threshold (such as fluctuation caused by vibration ≤1%); Temperature-light intensity correlation model: The weak correlation law was calculated using the Pearson correlation coefficient (for every 10°C increase in temperature, the light intensity is allowed to fluctuate by ±0.5%), and the correlation formula was fitted using multiple linear regression: Light intensity prediction value = α·temperature + β·initial light intensity + error term (α and β are coefficients); Output correlation coefficient range (such as 0.1±0.05), normal fluctuation range of light intensity with temperature; Formula Description: α Temperature: This reflects the effect of temperature on light intensity. α is the temperature coefficient, meaning that for every 1 unit change in temperature (e.g., 1°C), the light intensity will change by α units (e.g., if the light intensity unit is mW / cm², the change is mW / cm²). This reflects the linear relationship between temperature and light intensity.

[0033] β Initial Light Intensity: Initial light intensity is the baseline light intensity value initially collected by the system when the sensor and laser are functioning properly and properly installed. β is a weighting factor for initial light intensity, which is used to maintain the basic influence of initial light intensity on the current predicted light intensity, allowing the prediction to take into account the initial state of the device.

[0034] Error term: represents other factors that affect light intensity that the formula cannot accurately cover, such as slight fluctuations in ambient light and extremely subtle instabilities (non-fault-related) in the laser device itself. It supplements the deviation between theoretical calculations and actual conditions.

[0035] The parameters α and β are determined as follows: Data fitting: Collect a large amount of historical data (temperature, corresponding light intensity, and initial light intensity) from when both the temperature sensor and laser were operating normally. Using a linear regression algorithm, we ensure that the formula (α·temperature + β·initial light intensity) is as close as possible to the actual light intensity. By minimizing the error, we determine the optimal values ​​for α and β. For example, using the Python scikit-learn library, we input temperature and initial light intensity as features and actual light intensity as labels, and train a linear regression model to obtain α and β.

[0036] Scenario Adaptation: Temperature can have varying effects on light intensity in different cabinets (such as server cabinets and network cabinets) and installation environments (such as room temperature stability and vibration). For example, in a high-temperature room, temperature fluctuations may have a more significant impact on light intensity. After a basic fit, you can adjust α and β within a small range based on the actual scenario to make the formula more suitable for the application.

[0037] The role of the error term: It acknowledges the gap between the model and reality, ensuring that the formula is not absolutely idealized, retaining tolerance for complex and random influencing factors, and avoiding ignoring interference that cannot be fully modeled in practice due to the pursuit of precise formulas. This ensures that the light intensity prediction is more in line with the actual situation. On the one hand, during model training, the error term can help us evaluate the model fitting effect (such as indicators such as mean square error) and guide the optimization of α and β. On the other hand, in actual applications, when the deviation between the actual light intensity and the predicted value is within a reasonable range of the error term, the device status can be considered normal. If the deviation is too large, it is determined that there may be a fault. The output form of the basic model is: Numerical baseline: such as normal temperature range [30℃, 40℃], normal light intensity range; Curve baselines: such as the "load vs. temperature" fitting curve (used to determine whether the temperature changes normally with the load) and the "temperature vs. light intensity" correlation curve (used to determine whether the light intensity fluctuates normally with the temperature). Feature vector library: Extracts the time series features of normal data (such as temperature fluctuation within 10 minutes ≤ 2°C, light intensity attenuation rate ≤ 0.05% / hour) to form a normal feature template.

[0038] Step 2: Optimize the model using abnormal data, including: Manually label the following two types of abnormal data (marking fault type, abnormal period, and key features) to extract fault features that can be used for model optimization: The temperature sensor is normal but the laser is abnormal: The annotation content includes the laser fault type (such as transmitter attenuation, power supply failure), the start and end time of the abnormal light intensity, and the change trend. The key features extracted are the abnormal light intensity characteristics (such as attenuation rate > 0.5% / day, no load-related sudden drop) and the normal temperature characteristics (still meeting the single-dimensional baseline). The temperature sensor is abnormal but the laser is normal: The annotation content includes the temperature sensor fault type (such as drift, line fault) and the manifestation of the temperature anomaly (such as fixed offset, fluctuation). The key features extracted are temperature anomaly characteristics (such as deviation from the baseline >3°C, load unresponsiveness) and normal light intensity characteristics (still consistent with the single-dimensional baseline and correlation model). Based on the characteristics of abnormal data, the basic model is corrected in terms of boundaries, weights, and fault feature library, to improve the recognition capability of the target scenario (temperature failure when the laser is normal).

[0039] Correct the temperature anomaly judgment boundary (for temperature sensor failure): Use the data of the temperature sensor anomaly but the laser normal to calibrate the temperature anomaly judgment threshold: For example, in the basic model, a temperature deviation of 3°C from the baseline is considered abnormal, but actual data shows that drift faults usually deviate by more than 4°C before they stabilize, so the judgment threshold is adjusted from 3°C to 4°C (to reduce false positives). High-frequency fluctuation characteristics of poor line contact (such as standard deviation > 5°C within 1 minute) are marked, and the judgment condition that the fluctuation frequency is > 2 Hz and lasts > 10 seconds is added to the model.

[0040] Enhance the reliability of determining normal light intensity (eliminating interference from abnormal lasers): Using data from cases where the temperature sensor is normal but the laser is abnormal, clarify the core characteristics of normal lasers and avoid misjudging laser abnormalities as temperature faults. For example, when a laser is abnormal, the light intensity may drop suddenly but the temperature is normal. This type of data can be used to mark the counter-signals of normal light intensity (such as no sudden drop, normal attenuation rate, and fluctuations that conform to vibration laws). Added secondary verification logic for normal light intensity in the model: If the light intensity data meets the single-dimensional baseline and is normally correlated with the temperature, it is confirmed that the laser is normal (eliminating its interference with temperature determination).

[0041] Construct a feature template for temperature failure when the laser is normal: Extract typical features from data showing abnormal temperature sensors but normal lasers to form a unique judgment template; Measurement drift (zero drift): Light intensity is within the normal range (e.g., ±3%), combined with a temperature deviation from the baseline of >4°C and constant (e.g., a continuous increase of 5°C), combined with a normal correlation between temperature and light intensity (light intensity fluctuates with temperature); Poor line contact: light intensity fluctuation <1% combined with high-frequency temperature fluctuation (standard deviation >5°C) and communication signal strength fluctuation >20dBm; Hardware failure: Light intensity is stable (fluctuation < 0.5%), temperature is locked to a fixed value (e.g. 125°C), and power supply voltage is abnormal (< 3V).

[0042] Step 3: After the optimization, the model determines that the temperature sensor is faulty when the laser is normal. The specific process includes: Through the basic model and optimized feature template, accurate judgment is achieved according to the three-step process of normal light intensity verification, temperature anomaly identification, and fault type matching; A1: Verify whether the laser is normal (eliminate abnormal laser interference): Based on the optimized light intensity normal judgment logic, it is confirmed that there is no abnormality in the laser, eliminating interference for subsequent temperature fault judgment: Core verification conditions (must be met simultaneously): The light intensity data is within the single-dimensional baseline range (e.g., initial value ±3%), and the decay rate is less than 0.3% / day (no laser attenuation failure); The light intensity fluctuation conforms to normal patterns (standard deviation within 1 minute is less than 1%), and the fluctuation period matches the cabinet vibration (such as fan speed). This excludes loose lasers or power supply failures. The correlation between light intensity and temperature conforms to the basic model (for example, for every 10°C increase in temperature, the light intensity fluctuates within ±0.5%), and there is no correlation break (excluding faults such as laser wavelength offset); A2: Identify whether the temperature sensor is abnormal (based on optimized temperature judgment logic): After confirming that the laser is normal, identify anomalies by comparing the temperature data with the optimized baseline: Abnormal identification conditions (any one of them can be met): The temperature continuously deviates from the optimized baseline (e.g., > ±4°C for 30 consecutive minutes), and there is no synchronization offset between adjacent sensors (excluding environmental anomalies). The temperature does not respond to load changes (for example, if the CPU usage increases from 20% to 80%, the temperature change is less than 1°C, which is below the normal response threshold); Temperature timing characteristics match fault templates (such as high-frequency fluctuations and fixed value locks); A3: Match the temperature sensor fault type (accurately locate the problem): By comparing the abnormal temperature characteristics with the temperature fault template when the laser is normal, the specific fault type and confidence level are output: For example, a sensor displays a temperature of 39°C (baseline 34°C, deviation 5°C), the laser intensity is normal (10mW / cm²±0.2), and the correlation between temperature and intensity is normal (intensity increases slightly by 0.3% when temperature rises) to match the measurement drift (zero drift) template with a confidence level of 98%; If the feature matching degree is less than 60% (such as a new type of fault), the manual review process is triggered and the case is added to the feature library for subsequent model iterations.

[0043] Step 4: Model continuous optimization mechanism, including: Retrain the model weekly using newly generated normal data and confirmed abnormal data (after annotation) to update the correlation coefficient between temperature and light intensity (to adapt to slight sensor aging); When manual review reveals a misjudgment (e.g., misjudging normal temperature fluctuations as faults), the case is immediately added to the negative sample library and the judgment weight of the corresponding feature is adjusted (e.g., reducing the abnormal weight of short-term fluctuations); For different cabinet types (such as high-density server cabinets and network cabinets), sub-models are trained separately (due to different loads and vibration environments, the normal baselines are different) to avoid cross-scenario misjudgments.

[0044] The alarm and visualization interaction module is configured to convert analysis results into intuitive alarm information and visualization charts to support operation and maintenance decisions and achieve efficient human-computer interaction; Convert analysis results into intuitive alarm information and visual charts to support operation and maintenance decisions and achieve efficient human-computer interaction.

[0045] Emergency Alarm (P1): A sensor fails completely (e.g., due to hardware damage), making it impossible to monitor the temperature in a critical area. Or a loose installation may cause the equipment to overheat, triggering the following conditions: an immediate SMS, phone call, or audio and visual alarm in the equipment room. A response must be made within 15 minutes. Important alarm (P2): Sensor drift / mild obstruction (light intensity attenuation 5% to 10%) does not affect basic monitoring but needs to be processed within 30 minutes and pushed to the operation and maintenance app, displayed at the top of the system homepage; Prompt alarm (P3 level): Slight angular deviation (light intensity attenuation <5%) is recorded in the maintenance checklist and a summary report is generated every week.

[0046] Alarm details: Includes fault sensor ID, cabinet location (with 3D cabinet diagram), fault type (with data waveform), impact assessment (such as "may cause temperature monitoring in this area to deviate by 2°C"), and handling suggestions (such as "recalibrate the sensor and use a laser level to adjust the angle").

[0047] Cabinet heat map: Displays the temperature of each area in color coding (blue < 25°C, red > 40°C), and superimposes abnormal light intensity marks (yellow flashing dots); Sensor status list: sorted by fault level, showing key parameters such as temperature, light intensity attenuation rate, and the time of the last calibration; Trend comparison chart: Compare the current temperature / light intensity curve of the same sensor with the historical baseline for the same period, visually showing the deviation.

[0048] Historical analysis function: Fault statistics report: Statistics on fault type distribution, processing time, and repeated fault sensors by time period (day / week / month); Installation status trend: Displays the sensor's light intensity attenuation rate over time, predicting when maintenance is required (e.g., "The light intensity attenuation is expected to exceed 10% in 15 days"); Model Accuracy Analysis: Displays the recognition accuracy of various fault types and details of misjudgment cases (supports manual labeling and correction).

[0049] Interactive operations: Supports O&M personnel to manually enter processing results (such as "sensor replaced"), adjust alarm thresholds (for special load cabinets), initiate calibration commands (triggering re-collection of laser initial values), and upload the entered information to the cloud. Management personnel can remotely download information about the corresponding faulty sensors and analyze the historical data of the faulty sensors.

[0050] The system management and external integration modules are configured to ensure stable system operation through device resource management and model parameter management, connect to external systems such as dynamic environment and operation and maintenance work orders, and ensure data security and operational compliance through user authority management.

[0051] Real-time monitoring of the operating status of temperature sensors, laser transmitters, and transmission gateways (e.g., automatic alarm when the laser operating temperature is > 60°C) and recording of equipment lifespan (e.g., prompt when a sensor has been in operation for 18 months and is approaching its 24-month replacement cycle); Dynamic allocation of computing resources (e.g., automatically increasing the CPU / memory usage of the analysis module during peak failure periods) and storage capacity warning (automatically expanding capacity when remaining space is less than 20%). Key nodes (such as data transmission gateways) use a master-slave switching mechanism (switching time < 10 seconds) to avoid system interruption caused by single point failure.

[0052] Save historical model versions (at least 5), support rollback (when the accuracy of the new model decreases), and record model training logs (training data volume, number of iterations, accuracy); Parameter configuration: Allows administrators to adjust core parameters (such as temperature drift judgment threshold and light intensity attenuation alarm threshold), and supports custom configuration templates by cabinet type (such as network cabinet / server cabinet).

[0053] Push sensor fault information to the computer room air conditioning system through the API interface, triggering temporary adjustments (such as increasing the air conditioning speed in the faulty area); Operation and maintenance work order system: automatically creates fault work orders (including processing priority and a list of required tools) and synchronizes work order status (such as dispatched or completed) to the monitoring system; Log audit system: records all operations (such as parameter modifications and manual marking) and system events (such as alarm triggering and model updates). The log is retained for ≥ 1 year and supports compliance auditing.

[0054] Permission roles are divided into system administrators (configure parameters, manage users), operation and maintenance engineers (handle alarms, perform maintenance), and read-only users (view reports, monitor status). Record all key user operations (such as modifying alarm thresholds and marking faults as resolved) to support traceability and auditing; Sensitive data (such as cabinet location coordinates) is stored in encrypted form and transmitted using the HTTPS protocol to prevent information leakage.

[0055] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0056] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0057] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0058] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0059] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0060] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0061] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0062] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0063] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0064] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. A server cabinet fault monitoring system based on big data, characterized in that: include: The data acquisition and synchronization module is configured to capture the status data of the temperature sensors in the cabinet through multi-dimensional hardware deployment and precise synchronization mechanism; The data preprocessing and storage module is configured to perform deep cleaning and feature engineering on the raw data, build data sets, and enable storage and fast retrieval; The core intelligent analysis and diagnosis module is configured to realize sensor fault identification based on pre-processed data and big data models; The alarm and visualization interaction module is configured to convert the analysis results into alarm information and visualization charts to support operation and maintenance decisions and realize human-computer interaction.

2. A server cabinet fault monitoring system based on big data according to claim 1, characterized in that: Data acquisition and synchronization module, specifically including: Deploy sensors in key areas based on cabinet size, including contact sensors, non-contact infrared sensors, and redundant backup sensors; Synchronously collect sensor operating voltage, communication signal strength, and installation coordinates; Red laser and silicon-based photodetector are installed in pairs on both sides of the temperature sensor; Collect relevant parameters, including original light intensity value, light intensity fluctuation frequency, ambient light background value, and laser transmitter operating current; Record the initial laser transmittance of each sensor during installation; Add a timestamp to each frame of data to ensure the time alignment of temperature and light intensity data, and transmit the data; Complete data integrity verification on the gateway side, and cache and prioritize uploading abnormal values ​​exceeding the threshold locally.

3. The server cabinet fault monitoring system based on big data according to claim 1, characterized in that: Data preprocessing and storage module, specifically including: Clean the data, including noise suppression, missing value processing, and outlier identification; Extract features, including temporal feature extraction and spatial feature extraction; Store data and preprocessing results within a preset time period; store historical data, segmented by time; store extracted feature vectors for model training and trend analysis; A three-level index is constructed based on cabinet ID, sensor ID, and timestamp to support multi-dimensional retrieval.

4. The server cabinet fault monitoring system based on big data according to claim 1, characterized in that: Core intelligent analysis and diagnosis modules include: A temperature-light intensity benchmark correlation model is constructed using historical normal data, and then the model boundaries and judgment logic are iteratively optimized using abnormal data to achieve accurate identification of temperature sensor faults when the laser is normal.

5. The server cabinet fault monitoring system based on big data according to claim 4, characterized in that: Step 1: Establish a basic model based on historical normal data, including: Select stable operating data within a preset time period after the sensor is installed, and it must include a complete load cycle; Exclude data from the following periods: Abnormal laser light intensity or temperature sensor data deviates from the normal range; abnormal cabinet environment; abnormal data collection; Model composition: Temperature single-dimensional normal baseline: Statistics of temperature distribution in each period, fitting temperature and load correlation curves; output temperature normal range and load response threshold; Light intensity single-dimensional normal baseline: statistics on long-term light intensity attenuation trend, calculation of light intensity fluctuation characteristics; output light intensity normal range and fluctuation threshold; Temperature and light intensity correlation model: The weak correlation law is calculated by Pearson correlation coefficient and the correlation formula is fitted by multiple linear regression: Light intensity prediction value = α·temperature + β·initial light intensity + error term; Output correlation coefficient range, normal fluctuation range of light intensity with temperature.

6. The server cabinet fault monitoring system based on big data according to claim 5, characterized in that: Step 2: Optimize the model using abnormal data, including: Manually label the following two types of abnormal data to extract fault features that can be used for model optimization: The temperature sensor is normal but the laser is abnormal: the annotation content includes the laser fault type, the start and end time of the abnormal light intensity, and the change trend; the key features extracted are the abnormal light intensity feature and the normal temperature feature; The temperature sensor is abnormal but the laser is normal: the annotation content includes the temperature sensor fault type and the manifestation of the temperature abnormality; the key features extracted are the temperature abnormality feature and the normal light intensity feature; Correct the temperature anomaly judgment boundary: Use the data of abnormal temperature sensor but normal laser to calibrate the temperature anomaly judgment threshold: Strengthen the reliability of determining normal light intensity: Use data showing normal temperature sensors but abnormal lasers to clarify the core characteristics of normal lasers; Construct a feature template for temperature failure when the laser is normal: Extract typical features from data where the temperature sensor is abnormal but the laser is normal, and form a dedicated judgment template.

7. The server cabinet fault monitoring system based on big data according to claim 6, characterized in that: Step 3: After the optimization, the model determines that the temperature sensor is faulty when the laser is normal. The specific process includes: Through the basic model and optimized feature template, accurate judgment is achieved according to the three-step process of normal light intensity verification, temperature anomaly identification, and fault type matching; A1: Verify whether the laser is working properly: Core verification conditions: Light intensity data is within the single-dimensional baseline range, and the decay rate is less than 0.3% / day; The light intensity fluctuations conform to normal patterns, and the fluctuation period matches the cabinet vibration; The correlation between light intensity and temperature is consistent with the basic model, with no correlation break; A2: Identify whether the temperature sensor is abnormal: Abnormal identification conditions: The temperature continuously deviates from the optimized baseline without synchronous drift of adjacent sensors; Temperature does not respond to load changes; Temperature timing characteristics match fault templates; A3: Matching temperature sensor fault type: By comparing the abnormal temperature characteristics with the temperature fault template when the laser is normal, the specific fault type and confidence level are output.

8. The server cabinet fault monitoring system based on big data according to claim 1, characterized in that: Step 4: Model continuous optimization mechanism, including: Retrain the model weekly using newly generated normal data and confirmed abnormal data to update the correlation coefficient between temperature and light intensity; When manual review finds a misjudgment, it is immediately added to the negative sample library and the judgment weight of the corresponding feature is adjusted; For different cabinet types, sub-models are trained separately to avoid cross-scenario misjudgments.

9. The server cabinet fault monitoring system based on big data according to claim 1, characterized in that: Alarm and visualization interaction module, including: Convert analysis results into intuitive alarm information and visual charts; match the corresponding warning level based on the abnormal state of sensor monitoring, and issue corresponding level warning; The operation and maintenance personnel enter the processing results of the faulty sensor, adjust the alarm threshold, initiate calibration instructions, and upload the entered information to the cloud. The management personnel download the corresponding faulty sensor information remotely, compile statistics on the historical data of the faulty sensor, and conduct corresponding analysis.

10. The server cabinet fault monitoring system based on big data according to claim 1, characterized in that: The system management and external integration modules are configured to ensure stable system operation through device resource management and model parameter management, connect to external systems such as dynamic environment and operation and maintenance work orders, and ensure data security and operational compliance through user authority management.

Images