Machine room multi-dimensional environment intelligent monitoring and early warning system

Abstract

The application discloses a kind of machine room multidimensional environment intelligent monitoring early warning system, it is related to machine room environment monitoring technical field, the present application will be collected data mapping to the multiple path measurement units being set along the pollutant entry path, and in the same collection period, humidity coupling state and pollutant existence state are generated, and when pollutant state variable forms deviation relative to dynamic background state, spatial adjacent difference and historical cumulative difference, erosion trigger state can be generated;Erosion trigger state generates erosion load identification after meeting the continuous condition, and is merged into machine room multidimensional environment early warning event with source path label, so that low amplitude continuous pollutant exposure is output in the form of event associated with machine room area, rack position and source path.

Description

Technical Field

[0001] This invention relates to the field of data center environmental monitoring technology, and in particular to a multi-dimensional intelligent monitoring and early warning system for data center environment. Background Technology

[0002] With the increasing number of coastal data centers, underground communication equipment rooms, rail transit equipment rooms, and edge equipment rooms near fresh air inlets, the environmental monitoring targets for data centers have expanded from temperature, humidity, water leakage, and smoke detection to include particulate matter, salt spray, gaseous corrosive media, fresh air status, filtration status, differential pressure airflow status, and environmental differences at different heights of server racks. Existing data center environmental monitoring systems typically collect data through temperature and humidity sensors, particulate matter sensors, gas sensors, differential pressure sensors, and environmental monitoring platforms, and output anomaly alerts using single-item thresholds, trend analysis, area display, or alarm linkage methods.

[0003] In coastal, underground, or near fresh air inlets, environmental corrosion risks are often caused by the combined effects of contaminant ingress, humidity coupling, and continuous exposure. Particulate matter, salt spray, or gaseous corrosive media may enter the server rack area through fresh air ducts, filter bypasses, access control openings, return airflow, underfloor spaces, or cable trenches, and adhere to the surfaces of terminals, circuit boards, and connectors under conditions of high relative humidity, low dew point margin, or significant local temperature variations. In the early stages, these risks may manifest as low-amplitude, continuous, or path-dependent changes, and the instantaneous concentration at a single measurement point may not necessarily reach the individual alarm threshold.

[0004] Existing threshold alarm methods typically use a single sensor point, a single environmental parameter, or a fixed area as the judgment object, making it difficult to identify the process of pollutant entry by utilizing the sequence of state changes in adjacent areas along the same path, historical background deviations, and differences in cabinet height. When the pollutant state variable is below a single threshold but has undergone abnormal changes relative to the dynamic background state, adjacent path measurement units, or historical cumulative processes, existing alarm results are insufficient to timely characterize the risk of continued exposure. If only the alarm threshold is lowered, frequent alarms are likely to occur during access control maintenance, fresh air switching, or filter differential pressure fluctuations.

[0005] Furthermore, existing data center environmental alarm results often indicate anomalies at a specific measuring point or in a particular area, lacking a unified record of contaminant source paths, humidity coupling status, duration, and rack location. Maintenance and repair typically rely on manual identification of the anomaly source, and the post-repair environmental condition is difficult to correlate with the original alarm event. This leads to similar environmental alarms repeatedly occurring in the same area, increasing the risk of connector malfunctions, circuit board corrosion, and latent equipment failures.

[0006] Therefore, how to generate multi-dimensional environmental early warning events for the computer room by combining the state change sequence of the path measurement unit, the humidity coupling state, and the continuous exposure state when pollutants do not reach a single threshold or only show low-amplitude continuous changes has become a technical problem that needs to be solved. Summary of the Invention

[0007] This application provides a multi-dimensional intelligent environmental monitoring and early warning system for computer rooms, which solves the problem of difficulty in timely path-based early warning of low-level continuous exposure of pollutants in computer rooms.

[0008] This invention provides a multi-dimensional intelligent monitoring and early warning system for computer room environment, which includes a multi-dimensional environment acquisition unit, a path measurement unit construction module, a coupling state generation module, an erosion load generation module, a source path determination module, and an early warning event generation module; The multi-dimensional environmental acquisition unit acquires temperature state variables, humidity state variables, dew point state variables, pollutant state variables, differential pressure airflow state variables, fresh air state variables, filtration state variables, and access control state variables; The path measurement unit construction module maps the collected data to multiple path measurement units set along the pollutant entry path. The path measurement units are associated with the computer room area, cabinet identification, and height layer identification. The coupling state generation module merges the temperature state variables, humidity state variables, dew point state variables and pollutant state variables of the same path measurement unit according to the same acquisition period. It generates a humidity coupling state based on the relative humidity state, dew point margin state and temperature change state. It generates a pollutant existence state based on the deviation of the pollutant state variables from the dynamic background state, spatial adjacency differences and historical cumulative differences. When the humidity coupling state and the pollutant existence state coexist, it generates an erosion trigger state. The erosion load generation module generates an erosion load identifier and an event tracking identifier when the erosion trigger state meets the persistence condition; The source path determination module generates source path labels based on the sequential changes of adjacent path measurement units, differential pressure airflow state variables, fresh air state variables, filtration state variables, and access control state variables. The early warning event generation module merges the erosion load identifier and source path label under the same event tracking identifier, and outputs multi-dimensional environmental early warning events for the data center.

[0009] In some embodiments, the multi-dimensional environment acquisition unit includes an environmental data access interface, a device status access interface, and a computer room status access interface; The environmental data access interface receives temperature measurement data, humidity measurement data, dew point calculation data, particulate matter measurement data, salt spray measurement data, and gaseous corrosive medium measurement data. The pollutant state variables are generated from at least one of the particulate matter measurement data, salt spray measurement data, and gaseous corrosive medium measurement data. The equipment status access interface is connected to at least one of the following: fresh air valve opening, filter differential pressure, air conditioning operation status, exhaust status, and fan operation status; The data center status access interface is connected to at least one of the following: access control open status, maintenance operation status, water leakage status, and rack operation status; The multi-dimensional environmental acquisition unit generates multi-dimensional environmental records based on the acquisition time period, measurement point identification, and computer room area identification.

[0010] In some embodiments, the path measurement unit construction module includes an object table generation submodule and a variable binding submodule; The object table generation submodule creates a data center object table. The data center object table records the correspondence between at least two adjacent areas in the outdoor air intake side, filter side, cold aisle side, cabinet air intake side, cabinet height layer, return air side, underfloor space and cable trench side, as well as the correspondence between data center area, cabinet identifier and height layer identifier. The variable binding submodule binds the measurement point identifiers in the multidimensional environment record to the computer room area, cabinet identifier, or height layer identifier in the computer room object table, and merges the data belonging to the same path measurement unit within the same collection period into the unit status record. The unit status record includes environmental variable record, equipment status record, and computer room status record.

[0011] In some embodiments, the erosion load generation module includes a continuous exposure identification submodule, a humidity modulation submodule, and a load update submodule; The continuous exposure identification submodule identifies a state where the erosion trigger state occurs in multiple consecutive collection periods or repeats within a set observation period as a continuous state. The humidity modulation submodule writes the humidity coupling state into the humidity coupling flag corresponding to the erosion trigger state; The load update submodule generates erosion load identifiers and event tracking identifiers based on pollutant type identifiers, humidity coupling identifiers, continuous status identifiers, and path measurement unit identifiers, and binds and stores the erosion load identifiers and event tracking identifiers.

[0012] In some embodiments, the source path determination module includes a gradient recording submodule and a state association submodule; the gradient recording submodule treats the path measurement unit as a path graph node and the adjacency relationship between adjacent path measurement units as a path graph edge. The gradient recording submodule records the sequential changes of pollutant presence state, humidity coupling state, and pressure difference airflow state variables between adjacent path measurement units. The sequential changes are jointly determined by the formation time of pollutant presence state, the formation time of humidity coupling state, and the path direction represented by the pressure difference airflow state variables of adjacent path measurement units. The state association submodule associates the sequential changes with the fresh air state variables, filter state variables, and access control state variables.

[0013] In some embodiments, the source path determination module further includes a path label generation submodule; Fresh air status variables include fresh air on status and fresh air switching status; filter status variables include filter differential pressure status and filter maintenance status; access control status variables include access control on status, access control off status, and access control location identifier. The path label generation submodule generates one or more source path labels based on the association results of the status association submodule, the path edge set corresponding to the source path label, the running status support value, and the conflict label set. The source path labels include at least one of the following: fresh air introduction label, filter anomaly label, access control disturbance label, return air accumulation label, underfloor diffusion label, and cable trench diffusion label. The one or more source path labels are written into the same event tracking identifier.

[0014] In some embodiments, the path label generation submodule generates a fresh air introduction label when pollutant presence states appear sequentially on the outdoor air intake side, filter side, and cold aisle side, and the fresh air state variable represents the fresh air being turned on; generates a filter abnormality label when the filter pressure difference reaches a filter abnormality state and pollutant presence states appear in the path measurement unit after the filter side; generates an access control disturbance label when pollutant presence states appear in the adjacent computer room area after the access control is turned on; generates a return air accumulation label when the pollutant presence state on the return air side changes earlier than the cold aisle side; and generates an underfloor diffusion label or cable trench diffusion label when the pollutant presence state on the underfloor space or cable trench side changes earlier than the lower height layer of the cabinet.

[0015] In some embodiments, the early warning event generation module includes an event merging submodule, an event classification submodule, and an observation event processing submodule; The event merging submodule merges consecutively occurring erosion load identifiers under the same rack identifier, the same height layer identifier, or the same source path label into a single data center multi-dimensional environmental early warning event. The event classification submodule generates event level identifiers based on pollutant type identifiers, humidity coupling identifiers, persistence status identifiers, and source path labels. The observation event processing submodule records short-lived erosion triggering states that do not meet the conditions for persistence as observation events when access control is open, maintenance work is performed, or fresh air is switched. After the observation event meets the conditions for persistence, the erosion triggering state corresponding to the observation event is output to the erosion load generation module to generate the erosion load identifier corresponding to the persistent state. The erosion load identifier is used by the event merging submodule to generate multi-dimensional environmental early warning events for the computer room.

[0016] In some embodiments, the multi-dimensional environmental early warning event of the data center includes data center area, path measurement unit identifier, cabinet identifier, height layer identifier, contaminant type identifier, humidity coupling identifier, continuous status identifier, source path label, event level identifier, and event tracking identifier; The early warning event generation module records the start time, duration, associated path measurement unit, associated computer room operation status, and early warning output status of the erosion triggering state according to the event tracking identifier.

[0017] In some embodiments, a processing feedback recording module is also included; The handling feedback record module receives handling records corresponding to the event tracking identifier. The handling records include at least one of the following: fresh air adjustment record, filter maintenance record, access control closing record, passage inspection record, cabinet inspection record, underfloor space inspection record, and cable trench inspection record. The handling feedback recording module reads the status of the path measurement unit after the handling record is generated and generates an event closed-loop record. The event closed-loop record includes the erosion load identifier before handling, the erosion load identifier after handling, the source path label before handling, the source path label after handling, and the handling result status.

[0018] Through the above technical solution, the present invention can achieve at least the following beneficial effects: This invention maps collected data to multiple path measurement units set along the pollutant entry path, and generates humidity coupling state and pollutant presence state within the same collection period. When the pollutant state variable deviates from the dynamic background state, spatial adjacency difference, and historical cumulative difference, it can generate erosion trigger state. After the erosion trigger state meets the persistence condition, it generates erosion load identifier, which is merged with the source path label into a multi-dimensional environmental early warning event for the data center, so that low-amplitude continuous pollutant exposure is output in the form of events associated with data center area, rack location, and source path.

[0019] By connecting environmental measurement data, equipment operation status, and computer room operation status through environmental data access interface, equipment status access interface, and computer room status access interface respectively, particulate matter, salt spray, and gaseous corrosive media can be recorded in the same collection period along with fresh air valve opening, filter pressure difference, access control status, and maintenance operation status, providing data input under the same time reference for pollutant presence status, humidity coupling status, and source path label.

[0020] The correspondence between adjacent areas is recorded by the computer room object table, and the measurement point identifier is bound to the computer room area, cabinet identifier or height layer identifier. The pollutant state variables can be merged into the corresponding path measurement unit. The state differences between the cabinet height layer and adjacent areas can be incorporated into the subsequent generation process of erosion trigger state and source path label.

[0021] By using path measurement units as path graph nodes and the adjacency relationship between adjacent path measurement units as path graph edges, source path labels can be generated from the sequential changes of pollutant presence state, humidity coupling state, and pressure difference airflow state variables; fresh air introduction, filtration anomaly, access control disturbance, return air accumulation, underfloor diffusion, and cable trench diffusion can be written into the same event tracking identifier with different source path labels.

[0022] The observation event processing submodule records the erosion trigger state that occurs briefly and does not meet the continuous condition during access control opening, maintenance work, or fresh air switching as an observation event. After the observation event meets the continuous condition, it is converted into the event record corresponding to the continuous state. Short-term operating condition disturbances and continuous exposure events can form different event records.

[0023] The handling feedback recording module maps handling records to event tracking identifiers and records erosion load identifiers, source path labels, and handling result status before and after handling. Multi-dimensional environmental early warning events in the same data center can form a closed-loop record of events corresponding to the status before and after handling. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0025] Figure 1 This is a framework diagram of the multi-dimensional intelligent environmental monitoring and early warning system for the computer room in the embodiment. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0027] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0028] To facilitate understanding, the relevant terms and concepts involved in the embodiments of this application will be introduced below: A path measurement unit refers to a logical data unit set along the path of pollutant entry. It is defined by the room area identifier, cabinet identifier, height layer identifier, measurement point identifier, and adjacent relationship identifier. The lower and upper height layers of the cabinet are determined by the relative positions of the height layer identifiers in the height direction of the cabinet.

[0029] Pollutant state variables refer to environmental variables characterizing the state of particulate matter, salt spray, or gaseous corrosive media during the sampling period. Humidity coupled state refers to the environmental adhesion conditions jointly determined by humidity state variables, dew point state variables, and temperature state variables. Pollutant existence state refers to the state generated based on the deviation of pollutant state variables from the dynamic background state, spatial adjacency differences between adjacent path measurement units, and historical cumulative differences.

[0030] Erosion triggering state refers to the coexistence of humidity coupling state and pollutant presence state within the same data collection period. Erosion load identifier refers to data identifiers recording pollutant type, humidity coupling, persistence state, and path measurement unit. Source path label refers to a label characterizing the source of pollutant entry or diffusion. Multi-dimensional environmental early warning events in the data center refer to event records generated by combining erosion load identifiers and source path labels.

[0031] Differential pressure and airflow state variables refer to path direction variables that characterize the direction, amplitude, direction, or amplitude of differential pressure, airflow, or airflow between adjacent path measurement units. They are generated from at least one of differential pressure state variables and airflow state variables. Fresh air state variables refer to data variables characterizing the opening degree of the fresh air valve, the fresh air on / off state, and the fresh air switching state. Filter state variables refer to data variables characterizing the filter differential pressure state and filter maintenance state. Abnormal filter states are formed when the filter differential pressure state reaches the abnormal diameter corresponding to the filtration state variable or the filter maintenance state. The abnormal diameter corresponding to the filtration state variable is determined by the filter differential pressure limit, the filter maintenance state, or historical stable operation samples from the equipment room.

[0032] Access control status variables refer to data variables that characterize the access control system's open state, closed state, and location identifier. Data quality status refers to data markers that characterize whether the measurement data is valid, missing, out of range, frozen, or has an abnormal timestamp.

[0033] The operating status of a computer room refers to a status record formed by at least one of the following: fresh air status variables, filtration status variables, access control status variables, maintenance operation status, air conditioning operation status, exhaust status, and fan operation status.

[0034] Example 1: like Figure 1As shown, this embodiment employs a multi-dimensional intelligent environmental monitoring and early warning system for a computer room, including a multi-dimensional environmental acquisition unit, a path measurement unit construction module, a coupling state generation module, an erosion load generation module, a source path determination module, and an early warning event generation module; The multi-dimensional environmental acquisition unit acquires temperature, humidity, dew point, pollutant, differential pressure airflow, fresh air, filtration, and access control status variables inside and outside the computer room. The path measurement unit construction module maps the collected data to multiple path measurement units set along the pollutant entry path. The path measurement units are associated with the computer room area, cabinet identification, and height layer identification. The coupling state generation module merges the temperature state variables, humidity state variables, dew point state variables, and pollutant state variables of the same path measurement unit within the same acquisition period. It generates a humidity coupling state based on the relative humidity state, dew point margin state, and temperature change state. It generates a pollutant presence state based on the deviation of the pollutant state variables from the dynamic background state, the spatial adjacency difference of adjacent path measurement units, and the historical cumulative difference. When a humidity coupling state and a pollutant presence state exist within the same acquisition period, an erosion trigger state is generated. The erosion load generation module generates an erosion load identifier and an event tracking identifier when the erosion trigger state meets the persistence condition; The source path determination module generates source path labels based on the sequential changes of adjacent path measurement units, differential pressure airflow state variables, fresh air state variables, filtration state variables, and access control state variables. The early warning event generation module merges the erosion load identifier and source path label under the same event tracking identifier, and outputs multi-dimensional environmental early warning events for the data center.

[0035] The state of contaminant presence includes the contaminant presence result, contaminant type identifier, corresponding path measurement unit identifier, and collection period identifier. The erosion triggering state includes humidity coupling state, contaminant presence state, triggering period identifier, and triggering path measurement unit identifier.

[0036] The relative humidity status is generated by comparing the humidity status variable with the relative humidity judgment range. The dew point margin status is generated by comparing the difference between the temperature status variable and the dew point status variable with the dew point margin judgment range. The temperature change status is generated by comparing the change in the temperature status variable of the same path measurement unit within adjacent acquisition periods with the temperature change judgment range. The relative humidity judgment range, dew point margin judgment range, and temperature change judgment range are implementation parameters, determined by stable operation samples in the data center, equipment environmental limits, or validation set adjustments, and are bound to the judgment parameter version identifier. When the humidity status variable or dew point status variable is missing, the humidity coupling status uses the status of the previous acquisition period and is written into the data quality status; if there is no humidity coupling status in the previous acquisition period, no humidity coupling status is generated for this acquisition period.

[0037] In one embodiment, the multi-dimensional environmental acquisition unit includes an environmental data access interface, an equipment status access interface, and a computer room status access interface. The environmental data access interface receives temperature measurement data, humidity measurement data, dew point calculation data, particulate matter measurement data, salt spray measurement data, and gaseous corrosive medium measurement data. The pollutant status variable is generated by at least one of the particulate matter measurement data, salt spray measurement data, and gaseous corrosive medium measurement data. The equipment status access interface receives at least one of the following: fresh air valve opening degree, filter differential pressure, air conditioning operation status, exhaust air status, and fan operation status. The computer room status access interface receives at least one of the following: access control opening status, maintenance operation status, water leakage status, and cabinet operation status. The multi-dimensional environmental acquisition unit forms a multi-dimensional environmental record according to the acquisition time period, measurement point identification, and computer room area identification.

[0038] The multidimensional environmental record includes the data collection period, measurement point identifier, computer room area identifier, cabinet identifier, floor level identifier, temperature status variable, humidity status variable, dew point status variable, pollutant status variable, differential pressure airflow status variable, fresh air status variable, filtration status variable, access control status variable, and data source identifier. The dew point status variable is calculated from temperature and humidity measurement data, or directly output from the dew point measurement equipment. Pollutant status variables are recorded according to pollutant type identifiers, including particulate matter type, salt spray type, and gaseous corrosive medium type.

[0039] A data acquisition period refers to a data time unit formed by merging data within the same time alignment window. The time alignment window is set based on the data acquisition cycle of the data room environment. When data from different measurement points enter the same acquisition period, the merging condition is that the data timestamp falls within the same time alignment window. If multiple valid data entries exist for the same measurement point within the same acquisition period, the data for that acquisition period is determined by the record whose acquisition time is closest to the center of the time alignment window. Data with missing timestamps or data exceeding the time alignment window is written to the data quality status and does not participate in the status merging within the same acquisition period.

[0040] In one implementation, the path measurement unit construction module includes an object table generation submodule and a variable binding submodule; the object table generation submodule establishes a computer room object table, which records the correspondence between at least two adjacent areas among the outdoor air intake side, filter side, cold aisle side, cabinet air intake side, cabinet height layer, return air side, underfloor space and cable trench side, as well as the correspondence between computer room area, cabinet identifier and height layer identifier; The variable binding submodule binds the measurement point identifiers in the multidimensional environment record to the computer room area, cabinet identifier, or height layer identifier in the computer room object table, and merges the data belonging to the same path measurement unit within the same collection period into the unit status record. The unit status record includes environmental variable record, equipment status record, and computer room status record.

[0041] The data center object table includes data center area identifiers, adjacent area identifiers, path direction identifiers, rack identifiers, height layer identifiers, measurement point identifiers, and measurement point type identifiers. Path direction identifiers represent the airflow or diffusion direction between the outdoor air intake side, filter side, cold aisle side, rack air intake side, rack height layer, return air side, underfloor space, and cable trench side. Unit status records include environmental variable records, equipment status records, and data center status records belonging to the same path measurement unit within the same data acquisition period. When multiple measurement points exist for the same path measurement unit, a unit status record for that path measurement unit is generated based on the measurement point type, measurement point location, and data acquisition period.

[0042] In one embodiment, the erosion load generation module includes a continuous exposure identification submodule, a humidity modulation submodule, and a load update submodule; the continuous exposure identification submodule identifies the state in which the erosion trigger state occurs in multiple consecutive collection periods or repeats within a set observation period as a continuous state; the humidity modulation submodule writes the humidity coupling state into the humidity coupling identifier corresponding to the erosion trigger state. The load update submodule generates erosion load identifiers and event tracking identifiers based on pollutant type identifiers, humidity coupling identifiers, continuous status identifiers, and path measurement unit identifiers, and binds and stores the erosion load identifiers and event tracking identifiers.

[0043] Persistence conditions include continuous occurrence conditions and recurring occurrence conditions. Continuous occurrence conditions are determined by the same path measurement unit continuously forming erosion trigger states within adjacent data collection periods; recurring occurrence conditions are determined by the same path measurement unit forming erosion trigger states multiple times within the observation period. Erosion load identifiers include pollutant type identifiers, humidity coupling identifiers, persistence status identifiers, path measurement unit identifiers, start time identifiers, and update time identifiers. Event tracking identifiers are generated from the path measurement unit identifier, pollutant type identifier, and the start time of the erosion trigger state; the source path label is written into the same event tracking identifier after generation.

[0044] In one implementation, the path label generation submodule generates a fresh air introduction label when pollutant presence is sequentially observed on the outdoor air intake side, filter side, and cold aisle side, and the fresh air state variable indicates that the fresh air is turned on; generates a filter abnormality label when the filter pressure difference reaches a filter abnormality state and pollutant presence is observed in the path measurement unit after the filter side; generates an access control disturbance label when pollutant presence is observed in the adjacent computer room area after the access control is turned on; generates a return air accumulation label when the pollutant presence on the return air side changes earlier than the change on the cold aisle side; and generates an underfloor diffusion label or a cable trench diffusion label when the pollutant presence on the underfloor space or cable trench side changes earlier than the change in the lower height layer of the cabinet.

[0045] In one implementation, the early warning event generation module includes an event merging submodule, an event classification submodule, and an observation event processing submodule. The event merging submodule merges consecutively occurring erosion load identifiers under the same cabinet identifier, the same height layer identifier, or the same source path label into a multi-dimensional environmental early warning event for the same computer room. The event classification submodule generates event level identifiers based on pollutant type identifiers, humidity coupling identifiers, continuous status identifiers, and source path labels. The observation event processing submodule records short-lived erosion triggering states that do not meet the persistence conditions as observation events when access control is open, maintenance work is performed, or fresh air is switched. After the observation event meets the persistence conditions, the erosion triggering state corresponding to the observation event is output to the erosion load generation module to generate the erosion load identifier corresponding to the persistence state. The erosion load identifier is used by the event merging submodule to generate multi-dimensional environmental early warning events for the computer room.

[0046] An observation event refers to a short-lived erosion-triggered state that occurs during access control activation, maintenance work, or fresh air switching, and does not meet the conditions for persistence. Observation events include the observation path measurement unit identifier, the observation start time, the observation end time, the associated equipment room operating status, and the observation result status. The observation event processing submodule converts the observation event into an event record corresponding to the persistent state when the observation event meets the conditions for continuous or recurring occurrence; if the observation event does not meet the conditions for persistence, the observation result status is written to the observation event record.

[0047] An event level identifier is a data identifier that characterizes the priority of handling multi-dimensional environmental early warning events in a data center under the same event tracking identifier. It is jointly determined by the pollutant type identifier, humidity coupling identifier, persistence status identifier, and source path label. When the pollutant type identifier, humidity coupling identifier, persistence status identifier, or source path label is missing, the event classification submodule writes the corresponding missing item into the data quality status and generates the event level identifier based on the valid input items; when the valid input items are empty, the data quality status of the event level identifier is written as missing.

[0048] Short-term formation refers to a situation where the duration of the erosion triggering state is less than the number of collection periods corresponding to the continuous occurrence condition, and the recurrence condition is not met within the observation period. Observation result states include observation release state and observation transformation state; the observation release state is formed when the observation event does not meet the continuous condition, and the observation transformation state is formed when the observation event meets the continuous occurrence condition or the recurrence condition.

[0049] In one implementation, the multi-dimensional environmental early warning event of the data center includes data center area, path measurement unit identifier, cabinet identifier, height layer identifier, pollutant type identifier, humidity coupling identifier, continuous status identifier, source path label, event level identifier, and event tracking identifier; the early warning event generation module records the start time, duration, associated path measurement unit, associated data center operating status, and early warning output status of the erosion triggering state according to the event tracking identifier.

[0050] In one embodiment, the system further includes a handling feedback recording module; the handling feedback recording module receives handling records corresponding to event tracking identifiers, the handling records including at least one of fresh air adjustment records, filter maintenance records, access control closing records, passage inspection records, cabinet inspection records, underfloor space inspection records, and cable trench inspection records; the handling feedback recording module reads the status of the path measurement unit after the handling record is generated, and generates an event closed-loop record, the event closed-loop record including a pre-handling erosion load identifier, a post-handling erosion load identifier, a pre-handling source path label, a post-handling source path label, and a handling result status.

[0051] The handling record includes an event tracking identifier, handling type identifier, handling object identifier, handling start time, handling end time, and handling execution status. The handling feedback recording module reads the unit status record of the corresponding path measurement unit during the observation period after the handling ends and generates a post-handling erosion load identifier and a post-handling source path label. The handling result status includes load reduction status, source reduction status, load persistence status, and source transfer status. The event closed-loop record binds the handling record, pre-handling erosion load identifier, post-handling erosion load identifier, pre-handling source path label, post-handling source path label, and handling result status to the same event tracking identifier.

[0052] The load decline status is determined by the removal or downgrading of the erosion load identifier after treatment relative to the erosion load identifier before treatment. The source decline status is determined by the removal of the source path label after treatment relative to the source path label before treatment. The load persistence status is determined by the erosion load identifier after treatment still meeting the persistence condition. The source transfer status is determined by the source path label after treatment being inconsistent with the source path label before treatment, and the source path label after treatment meeting the output condition. If there are abnormal data quality statuses during the observation period after treatment, the treatment feedback recording module generates the treatment result status based on the valid acquisition period and writes the abnormal acquisition period into the event closed-loop record.

[0053] In a preferred embodiment of Example 1, when generating the pollutant presence state, the coupling state generation module establishes a combined judgment relationship of dynamic background state, spatial adjacency difference, and historical cumulative difference for the same path measurement unit and the same pollutant type identifier.

[0054] Pollutant state variables are represented by the measured concentration under the same pollutant type identifier or by the measured value after dimension unification. The dynamic background state is updated according to the background maintenance window to avoid abnormal pollution processes being written into the background state.

[0055] The background maintenance window refers to the set of data collection periods during which the dynamic background status can be updated. It is determined by access control status variables, maintenance operation status, fresh air status variables, filter status variables, and the pollutant presence status of the previous data collection period. When the background maintenance window is active, the dynamic background status is updated according to the current pollutant status variables; when the background maintenance window is inactive, the dynamic background status uses the values ​​from the previous data collection period.

[0056] The dimensionally unified measurement value refers to the value obtained by converting measurement data from different data sources under the same pollutant type identification to the same judgment scale. The dimensional unification process uses pollutant type identification, measurement point type identification, and data source identification as indexes, and determines the value using conversion relationships formed by stable operating samples or equipment calibration relationships. When conversion relationships or equipment calibration relationships are missing, the pollutant state variable is written into the data quality status and is not included in the calculation of new values ​​for dynamic background status, spatial adjacency differences, historical cumulative differences, and evidence values ​​of pollutant presence.

[0057] Before calculating the dynamic background state, spatial adjacency difference, historical cumulative difference, and pollutant presence evidence value, data quality assessment is performed on the pollutant state variable, pressure difference state variable, and airflow state variable. Data with missing, out-of-range, frozen, or abnormal timestamp status will not participate in the dynamic background state update, spatial adjacency difference generation, historical cumulative difference generation, and pollutant presence evidence value generation for the current acquisition period.

[0058] When the background maintenance window is open, the dynamic background state is updated in the following way: , in, For path measurement unit In pollutant type identification and collection period Dynamic background state below, For path measurement unit In pollutant type identification The upper limit of the background state that is allowed to be written, For path measurement unit In pollutant type identification The background state limit for allowing write access is set below. For path measurement unit In pollutant type identification The background update coefficient is below, and Greater than 0 and less than or equal to 1 For path measurement unit In pollutant type identification Compared to the dynamic background state in the previous data collection period, For path measurement unit In pollutant type identification and collection period The pollutant state variables under the following conditions and Used to restrict the dynamic background state between the background state limit and the background state upper limit.

[0059] The initial value of the dynamic background state is generated from calibration data when the computer room is in a stable operating state. Stable operating states include closed access, no maintenance work, fresh air status not switched, normal filtration status, and no confirmed pollutant presence. The calibration data spans multiple air conditioning operation cycles, and the upper and lower limits of the dynamic background state are set according to the allowable fluctuation range of this calibration data. The background update coefficient is set according to the pollutant type identifier. The background update coefficient for particulate matter pollutants is higher than that for gaseous corrosive media pollutants, allowing short-term particulate matter fluctuations to be quickly incorporated into background estimation, while abnormal increases in gaseous corrosive media pollutants are retained for a longer period of observation.

[0060] When there is no previous collection period during the first collection period, the initial value of historical cumulative difference is set to 0, the initial value of pollutant existence state is set to 0, and the initial value of humidity coupling state is set to 0.

[0061] The coupling state generation module generates spatial adjacency differences based on the differences in the same acquisition time period between adjacent path measurement units: , in, For path measurement unit In pollutant type identification and collection period The spatial adjacency difference below For path measurement unit Adjacent path measurement units, For an effective set of adjacent path measurement units, Measurement units for adjacent paths In pollutant type identification and collection period The pollutant state variables under the following conditions For path measurement unit Measurement units along adjacent paths In pollutant type identification Spatial difference calibration values ​​below, Pollutant type identification The lower limit of concentration normalization is used to avoid excessively small normalization denominators. , , and Following the definition above.

[0062] The effective adjacent path measurement unit set refers to the set of path measurement units that are adjacent to the current path measurement unit within the same acquisition period, have the same pollutant type identifier, and whose pollutant state variables have valid data quality status. When the effective adjacent path measurement unit set is empty, the spatial adjacency difference is set to 0, and this acquisition period is recorded as a conservative period where spatial adjacency difference cannot be triggered by a single factor. Within the conservative period, the pollutant existence status is generated by the combined evidence of single concentration exceeding the limit or historical cumulative differences exceeding the limit.

[0063] Spatial difference calibration values ​​are generated from the differences in the same type of pollutants between adjacent path measurement units under stable operating conditions. When the pollutant state variable is below the individual concentration threshold, if it shows a continuous increase relative to adjacent path measurement units, the spatial adjacent difference still participates in the pollutant presence state determination; when the pollutant state variable is below that of adjacent path measurement units, this difference is counted as 0, so that the determination result mainly responds to the local abnormal rise rather than the overall environmental synchronous decrease.

[0064] The coupled state generation module generates historical cumulative differences based on the continuous deviation between the dynamic background state and the current pollutant state variables: , in, For path measurement unit In pollutant type identification and collection period The historical accumulation of differences, Pollutant type identification The historical memory coefficient below, and Greater than or equal to 0 and less than 1 For path measurement unit In pollutant type identification The historical cumulative difference compared to the previous data collection period, For path measurement unit In pollutant type identification The background deviates from the permitted value. , and Following the definition above.

[0065] The background deviation allowance is generated by the permissible deviation of the pollutant state variable from the dynamic background state under stable operating conditions. The history memory coefficient is used to control the degree of retention of historical deviations; the larger the history memory coefficient, the higher the contribution of continuous low-amplitude deviations to the pollutant's state of existence.

[0066] After obtaining dynamic background deviation, spatial adjacency difference, and historical cumulative difference, the coupled state generation module generates evidence values ​​for the presence of pollutants: , in, For path measurement unit In pollutant type identification and collection period Evidence of the presence of pollutants below. Pollutant type identification The background deviation weight is below. Pollutant type identification Spatial adjacency difference weights, Pollutant type identification The historical cumulative difference weighting, and , and All three are greater than or equal to 0 and less than or equal to 1, and their sum is 1. , , , , and Following the definition above.

[0067] Background deviation weight, spatial adjacency difference weight, and historical cumulative difference weight are determined based on the pollutant type identifier. The spatial adjacency difference weight corresponding to particulate matter measurement data is higher than that corresponding to gaseous corrosive medium measurement data, and the historical cumulative difference weight corresponding to gaseous corrosive medium measurement data is higher than that corresponding to particulate matter measurement data. The background deviation weight and historical cumulative difference weight corresponding to salt spray measurement data are used together in the determination.

[0068] The state of pollutant existence is generated according to the rules of single concentration exceeding the limit, combined evidence exceeding the limit, and the combined significance of spatial differences and historical cumulative differences: like or or ,but, , like and and and ,but, , Other situations ; in, For path measurement unit In pollutant type identification and collection period The state of pollutants is indicated by a value of 1, which represents the presence of generated pollutants, and a value of 0, which represents the presence of no generated pollutants. Pollutant type identification The single concentration threshold below, Pollutant type identification The evidence trigger threshold below, Pollutant type identification The threshold for removing evidence below, and Greater than , Pollutant type identification Spatial difference trigger threshold, Pollutant type identification The historical cumulative trigger threshold is below. For path measurement unit In pollutant type identification The state of the pollutants in the previous collection period. , , and Following the definition above.

[0069] Individual concentration thresholds are determined by equipment limits, industry limits, or historical stable operating samples from the data center corresponding to the pollutant type identifier. Evidence trigger threshold, evidence release threshold, spatial difference trigger threshold, and historical cumulative trigger threshold are implementation parameters, determined by historical sample quantiles or validation set integration results, and the evidence trigger threshold is higher than the evidence release threshold. Background deviation weight, spatial adjacency difference weight, and historical cumulative difference weight are configured according to the same pollutant type identifier; all three participate in the generation of evidence values ​​for the existence of the same pollutant and are bound to the same judgment parameter version identifier.

[0070] When the path measurement unit In pollutant type identification and collection period Pollutant state variables under Dynamic background status when data quality status is missing, out of range, frozen, or timestamp abnormal. Continue Spatial adjacency differences Set to 0, historical cumulative difference Continue Evidence of the presence of pollutants No new values ​​are generated; the pollutant exists in its current state. Continue This data collection period is only used as a data quality record in the event tracking identifier and is not used as a separate trigger for the presence of new pollutants.

[0071] Example 2: Based on Example 1, this example provides a specific method for recording the sequential changes of pollutant presence state, humidity coupling state, and differential pressure airflow state variables between adjacent path measurement units in a multi-dimensional intelligent environmental monitoring and early warning system for computer rooms. The source path determination module includes a gradient recording submodule and a state association submodule. The gradient recording submodule uses the path measurement unit as a path graph node and the adjacency relationship between adjacent path measurement units as path graph edges. The gradient recording submodule records the sequential changes of pollutant presence state, humidity coupling state, and pressure difference airflow state variables between adjacent path measurement units. The sequential changes are jointly determined by the formation time of pollutant presence state, the formation time of humidity coupling state, and the path direction represented by the pressure difference airflow state variables of adjacent path measurement units. The state association submodule associates the sequential changes with the fresh air state variables, filter state variables, and access control state variables.

[0072] In one implementation, the source path determination module further includes a path label generation submodule; The fresh air status variables include fresh air on status and fresh air switching status; the filter status variables include filter differential pressure status and filter maintenance status; and the access control status variables include access control on status, access control off status, and access control location identifier. The path label generation submodule generates one or more source path labels based on the association results of the status association submodule, the path edge set corresponding to the source path label, the operating status support value, and the conflict label set. The source path labels include at least one of the following: fresh air introduction label, filter anomaly label, access control disturbance label, return air accumulation label, underfloor diffusion label, and cable trench diffusion label. One or more source path labels are written to the same event tracking identifier.

[0073] The operational status support value is a data value that characterizes the degree of support for the corresponding source path label for the fresh air on status, filtration abnormal status, access control on status, maintenance operation status, return air passage in circulation status, and the presence of pollutants in the underfloor space or cable trench.

[0074] In one embodiment, the path diagram edges include the adjacent relationships between the outdoor air intake side and the filter side, the filter side and the cold aisle side, the cold aisle side and the cabinet air intake side, the cabinet low-level height layer and the upper-level height layer, the return air side and the cold aisle side, the underfloor space and the cabinet low-level height layer, and the cable trench side and the cabinet low-level height layer.

[0075] Each path graph edge is bound to the pollutant presence state, humidity coupling state, and pressure difference airflow state variables within the corresponding data collection period, which are used to record the propagation direction and state change sequence between adjacent path measurement units.

[0076] For any adjacent path measurement unit, the gradient recording submodule generates directional constraint values ​​based on the pressure difference state variable and the airflow state variable: , in, For path measurement unit To path measurement unit During the collection period The downward direction constraint value, As the path measurement unit that serves as the starting point of propagation, For path measurement unit Adjacent path measurement units, For the data collection period, For path measurement unit To path measurement unit Pressure difference weight, For path measurement unit During the collection period The pressure difference state variable under the following conditions For path measurement unit During the collection period The pressure difference state variable under the following conditions For path measurement unit To path measurement unit The differential pressure calibrated amplitude between them For path measurement unit To path measurement unit The normalized lower bound of the pressure difference between them For path measurement unit To path measurement unit airflow weight, For path measurement unit To path measurement unit During the collection period The airflow state variable is defined below, and a positive value indicates that the airflow is measured by the path measurement unit. Pointing Path Measurement Unit , For path measurement unit To path measurement unit The airflow calibration amplitude between them For path measurement unit To path measurement unit The lower limit of normalized airflow between them.

[0077] Differential pressure weight and airflow weight Both are greater than or equal to 0 and less than or equal to 1; when both the pressure difference state variable and the airflow state variable are valid. and The sum is 1. When the data quality status of the differential pressure state variable is missing, out of range, frozen, or has an abnormal timestamp, and the data quality status of the airflow state variable is valid, the airflow weight participates in the directional constraint value calculation by 1; when the airflow state variable is missing or marked as an abnormal measurement value, and the differential pressure state variable is valid, the differential pressure weight participates in the directional constraint value calculation by 1; when both the differential pressure state variable and the airflow state variable are missing or both are marked as abnormal measurement values, the directional constraint value does not meet the directional constraint threshold, and the directed path edge cannot be used as a separate trigger for adding a new source path label.

[0078] The differential pressure calibration amplitude and airflow calibration amplitude are generated by the differential pressure fluctuation range and airflow fluctuation range between adjacent path measurement units under stable operating conditions in the equipment room. When the directional constraint value is greater than 0, it indicates that the differential pressure and airflow state variables support the flow of pollutants from the path measurement units. To path measurement unit spread.

[0079] The gradient recording submodule extracts state transition times according to the time window within the event period. The acquisition period when the pollutant's state changes from non-generated to generated is taken as the pollutant transition time, and the acquisition period when the humidity coupling state changes from non-generated to generated is taken as the humidity transition time. , , in, For path measurement unit In pollutant type identification and event cycle The transition moments of pollutants below, For identifying pollutant types, For the event cycle, Used to represent the transition properties of the state of a pollutant. Event cycle The corresponding collection time period set, For path measurement unit In pollutant type identification and collection period The state of the pollutants below, For path measurement unit In pollutant type identification The state of the pollutants in the previous collection period. For path measurement unit In the event cycle The humidity transition moment under the following conditions Transition properties used to represent humidity-coupled states For path measurement unit During the collection period Humidity coupling state under the following conditions For path measurement unit Humidity coupling status during the previous data collection period This means that the earliest collection time period is selected from the collection time periods that meet the conditions.

[0080] If the set of conditions used to extract pollutant transition moments or humidity transition moments is empty, then no corresponding transition moment will be generated; if the event period is... If the first collection period is already in a state of pollutant presence or humidity coupling, then the first collection period is recorded as the continuation start period. The continuation start period is only used to align with the already generated event tracking identifier and is not used as a separate trigger for adding new source path labels; when both ends of a directed path edge only have continuation start periods and there are no new transitions, the sequence of changes of the directed path edge is recorded as 0.

[0081] When no pollutant transition occurs within the event period, the path measurement unit is not considered the starting point for the pollutant propagation sequence within that event period. Similarly, when no humidity transition occurs within the event period and the pollutant transition is not in a humidity-coupled state, the humidity coupling state of the path measurement unit is only recorded as background correlation and not as a basis for humidity matching between sequential change records. The length of the event period is calibrated based on the computer room air exchange cycle, the air conditioning supply and return air circulation cycle, and the data acquisition cycle. The update frequency of the event period is consistent with the data acquisition time period.

[0082] After obtaining the pollutant transition time, humidity transition time, and directional constraint values, the gradient recording submodule first generates the humidity coupling matching state: like Existence, and ,but , Other situations ; in, For path measurement unit In pollutant type identification and event cycle The humidity coupling matching state is set to 1, indicating that the humidity coupling state at the pollutant transition time meets the path recording conditions, and 0, indicating that the humidity coupling state at the pollutant transition time does not meet the path recording conditions. For path measurement unit At the moment of pollutant transition Humidity coupling state under the following conditions , and Following the definition above.

[0083] The gradient recording submodule generates sequential change records for adjacent path measurement units: like and Both exist, and ,and ,and ,but ; Other situations , in, For path measurement unit To path measurement unit In pollutant type identification and event cycle The record of changes in sequence is set, with a value of 1 indicating that the change was recorded by the path measurement unit. Pointing Path Measurement Unit The sequential relationship of changes is represented by a value of 0, indicating that the sequential relationship was not recorded. Pollutant type identification The permissible lag window for the spread of pollutants. For path measurement unit To path measurement unit The directional constraint threshold between them , , , , , , and Following the definition above.

[0084] The allowable lag window for pollutant propagation is calibrated based on the distance between adjacent path measurement units, airflow velocity, and acquisition cycle; the allowable lag window for humidity coupling synchronization is calibrated based on the humidity sensor response time, air conditioning adjustment cycle, and dew point change lag time.

[0085] The path label generation submodule reads the sequential change records and generates label evidence values ​​according to the set of path edges corresponding to the candidate source path labels: , in, Source path tags In pollutant type identification and event cycle The label evidence value below, For source path tags, Source path tags The corresponding set of path edges, For path measurement unit To path measurement unit The resulting directed path edges, Source path tags The corresponding path attenuation coefficient, and Greater than 0 and less than or equal to 1 For path edges Source path label The order value corresponds to the position value in the path edge set, which increases sequentially from the source side as 0, 1, 2, etc. For path measurement unit To path measurement unit In pollutant type identification and event cycle The record of changes in sequence Source path tags The corresponding running state supports weights, and Greater than or equal to 0, Source path tags In the event cycle The running status supports values, and Greater than or equal to 0 and less than or equal to 1.

[0086] Candidate source path labels refer to source path labels whose label evidence values ​​reach the label evidence threshold within the same event period but have not yet completed conflict constraint determination. Candidate source path label records include the source path label, pollutant type identifier, event period, associated path measurement unit, associated path graph edges, and operational status support values. After conflict constraint determination, candidate source path labels are generated into final source path labels written to the event tracking identifier; candidate source path labels that do not generate final source path labels are written to the event tracking identifier as candidate association records.

[0087] The path edge set corresponding to the fresh air introduction label includes directed path edges from the outdoor air intake side to the filter side and from the filter side to the cold aisle side. The operating status support value is generated based on the fresh air initiation status. The path edge set corresponding to the filtration anomaly label includes directed path edges from the filter side to the cold aisle side and from the cold aisle side to the cabinet air intake side. The operating status support value is generated when the filter pressure difference reaches the filtration anomaly state. The path edge set corresponding to the return air accumulation label includes directed path edges from the return air side to the cold aisle side. The operating status support value is generated when the pollutant presence state on the return air side changes earlier than the cold aisle side and the return air path is in a recirculation state. The path edge set corresponding to the underfloor diffusion label includes directed path edges from the underfloor space to the lower height layer of the cabinet. The operating status support value is generated when the pollutant presence state in the underfloor space changes earlier than the lower height layer of the cabinet. The path edge set corresponding to the cable trench diffusion label includes directed path edges from the cable trench side to the lower height layer of the cabinet. The operating status support value is generated when the pollutant presence state on the cable trench side changes earlier than the lower height layer of the cabinet.

[0088] The state of the return air passage is determined by the fan operation status and the exhaust status, and is used to characterize the existence of return air circulation between the return air side and the cold aisle side.

[0089] The set of path edges corresponding to the access control disturbance label includes directed path edges from the path measurement unit associated with the access control opening position to the adjacent computer room area. The operating status support value is generated by the access control opening status and maintenance operation status. The path measurement unit associated with the access control opening position is determined by the access control position identifier, computer room area identifier and adjacent relationship identifier corresponding to the access control status variable.

[0090] Access control location identifiers refer to the location records of access control devices corresponding to access control status variables, and include at least the access control device identifier and the computer room area identifier. When access control status variables are missing, access control disturbance tags are not output as new source path tags; when event tracking identifiers already exist, the missing access control status variables for that collection period are written into the data quality status corresponding to the event tracking identifier.

[0091] When multiple candidate source path labels simultaneously meet the label evidence conditions, the path label generation submodule generates the final source path label according to the conflict constraints: like and ,but , Other situations ; in, Source path tags In pollutant type identification and event cycle The output status below, with a value of 1 indicating the output source path label. A value of 0 indicates that the source path tag is not output. , Source path tags The corresponding label evidence threshold, For source path tags Other source path tags that have conflicting relationships Source path tags The corresponding set of conflicting tags, Source path tags Relative to source path tags The threshold for evidence difference Source path tags In pollutant type identification and event cycle The label evidence value below, , , and Following the definition above.

[0092] The conflict label set is set based on the degree of path edge overlap, directional constraint direction, and runtime status support values. Only mutually exclusive tags that cannot be simultaneously output as the final source path tag within the same event period are written; source path tags that can simultaneously interpret the same event are not written into each other's conflicting tag sets. When the fresh air introduction tag and the filter anomaly tag meet the condition for simultaneous output, they are not treated as mutually exclusive tags.

[0093] Candidate association records refer to data records that are associated with the same event tracking identifier but are not output as the final source path label. These include candidate source path labels, candidate evidence sources, association path graph edges, association collection time periods, and reasons for not being output. Reasons for not being output include at least one of the following: evidence difference not meeting the condition, directional constraint not meeting the condition, missing operational status support values, or conflict label priority not being met.

[0094] Fresh air introduction tags and filter anomaly tags can be output simultaneously within the same event cycle, provided that the sequential changes from the outdoor intake side to the filter side, the sequential changes from the filter side to the cold aisle side, and the filter anomaly status are all met. If only the path measurement unit after the filter side shows a lagging change, and the outdoor intake side does not form an earlier pollutant presence state, then a filter anomaly tag will be output instead of a fresh air introduction tag. When there is a directional conflict between the return air accumulation tag and the fresh air introduction tag, the directional constraint value and the earliest pollutant transition time will be used for joint determination. If the pollutant transition time on the return air side is earlier than that on the cold aisle side, and the directional constraint value supports the propagation from the return air side to the cold aisle side, then a return air accumulation tag will be output. When both the underfloor diffusion label and the cable trench diffusion label point to the lower level of the cabinet, if the changes in the underfloor space and the cable trench side occur earlier than those in the lower level of the cabinet within the allowable lag window for contaminant propagation, the evidence values ​​of the corresponding labels are compared. If the evidence difference reaches the evidence difference threshold, the label with the higher evidence value is output. If the evidence difference does not reach the evidence difference threshold, both are written as candidate association records into the event tracking identifier, and the corresponding source path label is output only after the directional constraint value or state transition time in the subsequent event cycle forms a difference.

[0095] Dynamic background status, spatial difference calibration value, background deviation allowable value, directional constraint threshold, label evidence threshold, evidence difference threshold, weight, and window parameters are all bound to the judgment parameter version identifier. The judgment parameter version identifier is only used to record the calibration and threshold calibers used within the same event period; if an event tracking identifier has already been generated within the same event period, the judgment parameter version identifier at the beginning of the event will be used for judgment. The new judgment parameter version will take effect in the next event period or during collection periods that are not in a continuous state, to avoid inconsistencies in source path labels or event level identifiers before and after the handling of the same early warning event due to version switching.

[0096] The judgment parameter version identifier includes version identifier number, effective collection period, applicable data center area identifier, applicable pollutant type identifier, threshold caliber identifier, weight caliber identifier, and window caliber identifier. The threshold caliber identifier corresponds to the single concentration threshold, evidence trigger threshold, evidence release threshold, spatial difference trigger threshold, historical cumulative trigger threshold, directional constraint threshold, and label evidence threshold; the weight caliber identifier corresponds to the background deviation weight, spatial adjacency difference weight, historical cumulative difference weight, and operational status support weight; the window caliber identifier corresponds to the collection period, event cycle, pollutant propagation allowable lag window, and humidity coupling synchronization allowable window.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0098] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A multi-dimensional intelligent monitoring and early warning system for computer room environment, characterized in that, It includes a multi-dimensional environment acquisition unit, a path measurement unit construction module, a coupling state generation module, an erosion load generation module, a source path determination module, and an early warning event generation module; The multi-dimensional environmental acquisition unit acquires temperature state variables, humidity state variables, dew point state variables, pollutant state variables, differential pressure airflow state variables, fresh air state variables, filtration state variables, and access control state variables; The path measurement unit construction module maps the collected data to multiple path measurement units set along the pollutant entry path. The path measurement units are associated with the computer room area, cabinet identification, and height layer identification. The coupling state generation module merges the temperature state variables, humidity state variables, dew point state variables and pollutant state variables of the same path measurement unit according to the same acquisition period. It generates a humidity coupling state based on the relative humidity state, dew point margin state and temperature change state. It generates a pollutant existence state based on the deviation of the pollutant state variables from the dynamic background state, spatial adjacency differences and historical cumulative differences. When the humidity coupling state and the pollutant existence state coexist, it generates an erosion trigger state. The erosion load generation module generates an erosion load identifier and an event tracking identifier when the erosion trigger state meets the persistence condition; The source path determination module generates source path labels based on the sequential changes of adjacent path measurement units, differential pressure airflow state variables, fresh air state variables, filtration state variables, and access control state variables. The early warning event generation module merges the erosion load identifier and source path label under the same event tracking identifier, and outputs multi-dimensional environmental early warning events for the data center.

2. The intelligent monitoring and early warning system for multi-dimensional environment of the computer room according to claim 1, characterized in that, The multi-dimensional environmental acquisition unit includes an environmental data access interface, an equipment status access interface, and a computer room status access interface; The environmental data access interface receives temperature measurement data, humidity measurement data, dew point calculation data, particulate matter measurement data, salt spray measurement data, and gaseous corrosive medium measurement data. The pollutant state variables are generated from at least one of the particulate matter measurement data, salt spray measurement data, and gaseous corrosive medium measurement data. The equipment status access interface is connected to at least one of the following: fresh air valve opening, filter differential pressure, air conditioning operation status, exhaust status, and fan operation status; The data center status access interface is connected to at least one of the following: access control open status, maintenance operation status, water leakage status, and rack operation status; The multi-dimensional environmental acquisition unit generates multi-dimensional environmental records based on the acquisition time period, measurement point identification, and computer room area identification.

3. The intelligent monitoring and early warning system for multi-dimensional environment of the computer room according to claim 2, characterized in that, The path measurement unit building module includes an object table generation submodule and a variable binding submodule; The object table generation submodule creates a data center object table. The data center object table records the correspondence between at least two adjacent areas in the outdoor air intake side, filter side, cold aisle side, cabinet air intake side, cabinet height layer, return air side, underfloor space and cable trench side, as well as the correspondence between data center area, cabinet identifier and height layer identifier. The variable binding submodule binds the measurement point identifiers in the multidimensional environment record to the computer room area, cabinet identifier, or height layer identifier in the computer room object table, and merges the data belonging to the same path measurement unit within the same collection period into the unit status record. The unit status record includes environmental variable record, equipment status record, and computer room status record.

4. The intelligent monitoring and early warning system for multi-dimensional environment of the computer room according to claim 1, characterized in that, The erosion load generation module includes a continuous exposure identification submodule, a humidity modulation submodule, and a load update submodule; The continuous exposure identification submodule identifies a state where the erosion trigger state occurs in multiple consecutive collection periods or repeats within a set observation period as a continuous state. The humidity modulation submodule writes the humidity coupling state into the humidity coupling flag corresponding to the erosion trigger state; The load update submodule generates erosion load identifiers and event tracking identifiers based on pollutant type identifiers, humidity coupling identifiers, continuous status identifiers, and path measurement unit identifiers, and binds and stores the erosion load identifiers and event tracking identifiers.

5. The intelligent monitoring and early warning system for multi-dimensional environment of a computer room according to claim 1, characterized in that, The source path determination module includes a gradient recording submodule and a state association submodule; the gradient recording submodule uses the path measurement unit as a path graph node and the adjacency relationship between adjacent path measurement units as a path graph edge. The gradient recording submodule records the sequential changes of pollutant presence state, humidity coupling state, and pressure difference airflow state variables between adjacent path measurement units. The sequential changes are jointly determined by the formation time of pollutant presence state, the formation time of humidity coupling state, and the path direction represented by the pressure difference airflow state variables of adjacent path measurement units. The state association submodule associates the sequential changes with the fresh air state variables, filter state variables, and access control state variables.

6. The intelligent monitoring and early warning system for multi-dimensional environment of the computer room according to claim 5, characterized in that, The source path determination module also includes a path label generation submodule; Fresh air status variables include fresh air on status and fresh air switching status; filter status variables include filter differential pressure status and filter maintenance status; access control status variables include access control on status, access control off status, and access control location identifier. The path label generation submodule generates one or more source path labels based on the association results of the status association submodule, the path edge set corresponding to the source path label, the running status support value, and the conflict label set. The source path labels include at least one of the following: fresh air introduction label, filter anomaly label, access control disturbance label, return air accumulation label, underfloor diffusion label, and cable trench diffusion label. The one or more source path labels are written into the same event tracking identifier.

7. The intelligent monitoring and early warning system for multi-dimensional environment of a computer room according to claim 6, characterized in that, The path label generation submodule generates a fresh air introduction label when pollutant presence is sequentially observed on the outdoor intake side, filter side, and cold aisle side, and the fresh air state variable represents the start of fresh air. It generates a filter abnormality label when the filter pressure difference reaches an abnormal state and pollutant presence is observed in the path measurement unit after the filter side. It generates an access control disturbance label when pollutant presence is observed in the adjacent computer room area after the access control is open. It generates a return air accumulation label when the pollutant presence on the return air side changes earlier than the change on the cold aisle side. It generates an underfloor diffusion label or cable trench diffusion label when the pollutant presence in the underfloor space or cable trench side changes earlier than the change in the lower height layer of the cabinet.

8. The intelligent monitoring and early warning system for multi-dimensional environment of a computer room according to claim 1, characterized in that, The early warning event generation module includes an event merging submodule, an event classification submodule, and an observation event processing submodule; The event merging submodule merges consecutively occurring erosion load identifiers under the same rack identifier, the same height layer identifier, or the same source path label into a single data center multi-dimensional environmental early warning event. The event classification submodule generates event level identifiers based on pollutant type identifiers, humidity coupling identifiers, persistence status identifiers, and source path labels. The observation event processing submodule records short-lived erosion triggering states that do not meet the conditions for persistence as observation events when access control is open, maintenance work is performed, or fresh air is switched. After the observation event meets the conditions for persistence, the erosion triggering state corresponding to the observation event is output to the erosion load generation module to generate the erosion load identifier corresponding to the persistent state. The erosion load identifier is used by the event merging submodule to generate multi-dimensional environmental early warning events for the computer room.

9. The intelligent monitoring and early warning system for multi-dimensional environment of a computer room according to claim 1, characterized in that, The multi-dimensional environmental early warning events for the data center include data center area, path measurement unit identifier, rack identifier, height layer identifier, contaminant type identifier, humidity coupling identifier, continuous status identifier, source path label, event level identifier, and event tracking identifier; The early warning event generation module records the start time, duration, associated path measurement unit, associated computer room operation status, and early warning output status of the erosion triggering state according to the event tracking identifier.

10. The intelligent monitoring and early warning system for multi-dimensional environment of a computer room according to claim 9, characterized in that, It also includes a handling feedback recording module; The handling feedback record module receives handling records corresponding to the event tracking identifier. The handling records include at least one of the following: fresh air adjustment record, filter maintenance record, access control closing record, passage inspection record, cabinet inspection record, underfloor space inspection record, and cable trench inspection record. The handling feedback recording module reads the status of the path measurement unit after the handling record is generated and generates an event closed-loop record. The event closed-loop record includes the erosion load identifier before handling, the erosion load identifier after handling, the source path label before handling, the source path label after handling, and the handling result status.

Images