Intelligent management and control system for machine room equipment

Abstract

The invention discloses an intelligent management and control system for machine room equipment, and relates to the field of machine room equipment management and control, the system comprises an energy consumption management and control module, a predictive maintenance module and a security situation awareness module, the energy consumption management and control module is based on the synergistic effect of a dynamic load balancing module, a phase change heat dissipation control module and a block chain energy consumption account book module; and the energy consumption cost of machine room equipment is reduced. According to the intelligent management and control system for the machine room equipment, power supply strategies can be switched according to the actual state of the machine room equipment through the dynamic load balancing module, so that the energy consumption of the machine room equipment can be effectively reduced; according to the invention, corresponding measures can be taken at different stages when the equipment has faults, the influence of the equipment faults on services is reduced to the greatest extent, the personnel movement track, the operation instruction and the environmental parameters can be fused through the arranged security situation awareness module, and the illegal intrusion detection accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of equipment control in computer rooms, and particularly to an intelligent control system for computer room equipment. Background Art

[0002] In the IT industry, computer rooms generally refer to places such as telecommunications, China Netcom, China Mobile, dual-line, power, as well as government or enterprises, etc., where servers are stored and IT services are provided for users and employees. Small computer rooms are dozens of square meters, usually with twenty or thirty cabinets placed. Large computer rooms are tens of thousands of square meters with thousands of cabinets placed, or even more. Various servers and minicomputers are usually placed in computer rooms, such as IBM minicomputers, HP minicomputers, SUN minicomputers, and so on.

[0003] When controlling computer room equipment, most rely on fixed strategies and cannot respond in real time to electricity price fluctuations and equipment status changes, resulting in high energy consumption costs. Moreover, the energy consumption data of computer room equipment mostly uses manual meter reading or data from a centralized database, resulting in easy tampering of energy consumption data and inability to meet the requirements of carbon footprint audits.

[0004] Therefore, it is very necessary to propose an intelligent control system for computer room equipment to solve the above problems. Summary of the Invention

[0005] The main purpose of the present invention is to provide an intelligent control system for computer room equipment, which can effectively solve the problems in the background art.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] An intelligent control system for computer room equipment includes an energy consumption control module, a predictive maintenance module, and a security situation awareness module. It is characterized in that: the energy consumption control module realizes the reduction of the energy consumption cost of computer room equipment based on the coordinated action of a dynamic load balancing module, a phase change heat dissipation control module, and a blockchain energy consumption ledger module, and is also used for accurately tracking carbon emissions;

[0008] The predictive maintenance module collects equipment data in real time based on multi-modal sensors, and combines algorithms with digital twin stress analysis to realize early prediction of equipment failures, dynamic assessment of health status, and self-repair control;

[0009] The security situation awareness module constructs an active defense system for the computer room by integrating spatio-temporal behavior analysis and intelligent patrol data, and is used to realize illegal intrusion detection and equipment abnormal replacement protection of computer room equipment.

[0010] Preferably, the dynamic load balancing module includes a multi-dimensional state perception module, an intelligent decision-making strategy module, and a strategy selection module. The multi-dimensional state perception module is used to collect the remaining power of the UPS of the computer room equipment, the power of the backup power supply, and the current load rate; the real-time electricity price and the electricity price prediction for the next 24 hours; the server temperature, the computer room temperature and humidity, and the cumulative operation duration of the equipment; the number of power supply switches, the power grid stability, and the service priority.

[0011] The intelligent decision-making strategy module includes a power supply strategy action library, specifically including: the valley electricity priority strategy, the trigger condition is that the real-time electricity price is in the valley period and it is predicted that the electricity price will rise in the next 2 hours, and the action to be executed is to switch to grid power supply, start charging the backup power supply, and increase the server CPU utilization rate to 80%; the pre-charging switching strategy, the trigger condition is that it is predicted that the load rate will exceed 85% in the next 5 minutes and the remaining power of the UPS < 40%, and the action to be executed is to start pre-charging the backup power supply to 90% 5 minutes in advance, use zero-voltage transition during switching, and automatically detect the bus voltage stability after switching; the life protection strategy, the trigger condition is that the UPS battery health < 40% and the internal resistance growth rate > 15% / month, and the action to be executed is to prohibit the number of single-day switches > 3 times, start the flywheel energy storage system to bear the peak load, and reduce the server operating frequency by 10%; the emergency load shunting strategy, the trigger condition is that the cluster load rate > 90% and lasts for more than 10 minutes, and the action to be executed is to automatically migrate non-core services to the backup server cluster and request temporary grid capacity increase; the green electricity priority strategy, the trigger condition is that the green electricity certificate is available and the green electricity price ≤ 110% of the thermal power price, and the action to be executed is to preferentially use green electricity for power supply, generate a green electricity usage certificate, and preferentially store green electricity; the temperature and load linkage strategy, the trigger condition is that the average server temperature > 35°C and the load rate > 75%, and the action to be executed is to automatically adjust the air supply temperature of the air conditioning system to reduce the temperature by 2°C and trigger the cabinet air damper opening of the computer room equipment to 70%; the power grid stability right-side strategy, the trigger condition is that the power grid voltage fluctuation exceeds ±5% or the frequency deviation > 0.2Hz, and the action to be executed is to cut off the connection with the power grid and switch to UPS power supply, and reduce the power consumption of non-critical equipment by 15%; the service priority strategy, the trigger condition is that the high-priority service load rate > 80%, and the trigger condition is to reserve 20% of the UPS capacity for high-priority services, enable a dedicated heat dissipation channel, and prohibit non-core services from migrating to the high-priority cluster.

[0012] The strategy selection module specifically includes:

[0013] S201: Collection and standardization of core parameters. The core parameters include electricity price, load and equipment health. The electricity price is obtained from the power company API to obtain the real-time electricity price and the forecast for the next 24 hours. The load obtains the average load rate of the cluster through the server management interface. The equipment health is obtained by calculating the remaining life of the UPS battery. The electricity price is normalized to the [0,1] interval according to the peak and valley values. The load rate is set to 1 when it exceeds 85% and 0 when it is less than 50%, with linear mapping in between. The equipment health is set to 1 when the remaining life is >80% and 0 when it is <30%. An exponential function is used to enhance the sensitivity to low health.

[0014] S202: Dynamic weight allocation, set the basic weight of electricity price to 40%, the basic weight of load to 35%, and the basic weight of equipment health to 25%, and modify the weight based on actual conditions: when the electricity price fluctuates by more than ±10%, the electricity price weight increases by 5-10%, when the load rate is >80% for 30 consecutive minutes, the load weight increases to 45%, and when the health is <40%, the health weight is forcibly increased to 40%; for medical services among high-priority services, the health weight is increased by an additional 10%;

[0015] S203: Construction of strategy scoring model, according to the formula:

[0016] Comprehensive score = (electricity price parameter × electricity price weight) + (load parameter × load weight) + (equipment health parameter × equipment health weight);

[0017] The comprehensive score of each strategy is calculated, and the strategy with the highest comprehensive score is selected. When the score difference between the two strategies with the highest comprehensive scores is <0.05, manual review is performed.

[0018] Preferably, the S203 also includes a strategy adaptation coefficient and a strategy selection logic, wherein the strategy adaptation coefficient is specifically: for the pre-charging switching strategy, when the electricity price is in the valley period, the comprehensive score is +0.15; for the life protection strategy, when the health level is <50%, the comprehensive score is +0.2.

[0019] Preferably, the phase change heat dissipation control module is based on dual closed-loop control, wherein the main loop is a computer room temperature prediction model based on a BP neural network, which is used to predict hot spots 30 minutes in advance; the secondary loop uses a fuzzy PID algorithm to control the working fluid flow of the heat pipe radiator in the computer room. When the predicted temperature exceeds a preset threshold, the heat pipe radiator is started first and the air outlet angle of the air conditioner is adjusted in conjunction.

[0020] Preferably, the blockchain energy consumption ledger module includes a data collection and preprocessing module, a blockchain underlying architecture module, a data structure design module, a carbon footprint calculation and green power traceability module. The data collection and preprocessing module is used to collect data of the power distribution cabinet in the computer room equipment, health parameter data of the used batteries, and environmental data, and clean the collected data;

[0021] The blockchain underlying architecture module is used to design a consortium chain and smart contracts. The consortium chain includes a primary node: the data center operator; verification nodes: power companies, third-party certification agencies; audit nodes: government regulatory departments. At the same time, a consensus mechanism based on practical Byzantine fault tolerance is adopted. The smart contract is an energy consumption data deposit contract written in Solidity, specifically including a data writing interface, a carbon emission calculation function, and a green power matching rule;

[0022] The data structure design module includes generating time series data blocks;

[0023] The carbon footprint calculation and green power traceability module includes a carbon emission calculation module and a green power matching module. The green power matching module automatically scans the REC certificate library based on the smart contract to match the green power usage amount in the same time period. When the green power usage ratio is less than 50%, the green power priority strategy is directly triggered.

[0024] Preferably, the carbon emission calculation module calculates the real-time carbon emissions based on the formula:

[0025]

[0026] Where is the real-time carbon emissions, n is the number of types of equipment in the computer room, P i represents the active power of the i-th type of equipment, Δt is the statistical time interval, CF is the grid carbon intensity factor, and η i is the energy efficiency correction coefficient of the i-th type of equipment;

[0027] It also includes a dynamic grid carbon intensity factor update module, which collects the carbon intensity of the regional grid and the proportion of renewable energy generation released by the power company, and is used for green power matching and dynamically adjusting the carbon intensity factor. The formula is:

[0028] CF new = CF base ×(1 - RER);

[0029] Where CF new is the new grid carbon intensity factor, CF base is the basic carbon intensity factor, and RER is the proportion of renewable energy generation;

[0030] It also includes a carbon footprint generation module. The carbon footprint includes time dimension and space dimension. The time dimension includes generating carbon emission trend reports for daily, weekly, and monthly periods, including year-on-year and month-on-month analyses, to intuitively display the change of carbon emissions over time and statistically show the proportion of carbon emissions during peak periods, helping users understand high-energy-consuming periods. The space dimension includes drawing a carbon emission heat map at the cabinet level, grading different computer room equipment according to the PUE value, intuitively displaying the carbon emission distribution of each cabinet in the computer room, and listing the top 5 equipment with the highest carbon emissions in each area of the data center, facilitating users to focus on and manage high-energy-consuming equipment.

[0031] Preferably, the predictive maintenance module includes a multi-modal data acquisition and preprocessing module, a data fusion model, a health assessment module, a self-repair module, and a knowledge graph construction module. Among them, the multi-modal data acquisition and preprocessing module collects vibration, oil analysis, temperature, and current data of computer room equipment based on sensors and cleans the collected data.

[0032] The data fusion model includes constructing a feature extraction sub-model for extracting the features of the data collected by the multi-modal data acquisition and preprocessing module, cascading the extracted feature vectors, and inputting them into a 3-layer fully connected cascaded neural network to output the failure probability of computer room equipment by learning the association between different modal data features, and training the model at the same time.

[0033] The health assessment module establishes a digital twin of the equipment based on BIM technology, including the geometric structure, material properties, and remaining conditions of the computer room equipment. At the same time, it imports sensor data to drive the digital twin, uses finite element analysis to calculate the stress generated when the cabinet equipment works, and predicts the remaining life of the computer room equipment based on the Miner linear cumulative damage theory.

[0034] The knowledge graph construction module is used to deeply analyze the relationship between the fault phenomena, environmental parameters, and equipment status of computer room equipment, establish a causal relationship network of fault phenomena, environmental parameters, and equipment status, use the fault phenomena, environmental parameters, and equipment status as nodes of the knowledge graph, and their causal relationships as edges, and represent the fault knowledge of the equipment in a graphical way for subsequent reasoning and analysis.

[0035] Preferably, the self-repair module includes a first-level response module, a second-level response module, and a third-level response module. The first-level response module includes health monitoring and inspection adjustment and spare part procurement recommendation generation. Among them, health monitoring and inspection adjustment: real-time monitor the health indicators of the equipment. When the health is lower than 70%, the system automatically increases the inspection frequency of the equipment from the normal once a day to once an hour; spare part procurement recommendation generation: generate spare part procurement recommendations according to the fault type and historical maintenance records of the equipment, combined with the delivery cycle of the supplier.

[0036] The secondary response module includes fan failure adjustment and server temperature adjustment. Among them, for fan failure adjustment: when an abnormality of the fan is detected, the rotation speed of the redundant fan is dynamically adjusted so that the increased air volume of the redundant fan can compensate for the reduced air volume of the faulty fan, which is used to ensure that the heat dissipation effect of the computer room equipment is not affected; for server temperature adjustment: when the temperature of the server CPU rises abnormally, the operating frequency of the CPU is reduced to reduce the heat generation of the CPU, and at the same time, the heat dissipation efficiency is increased until the CPU temperature returns to the normal range;

[0037] The tertiary response module includes UPS module switching and storage device switching. Among them, for UPS module switching: when a fault occurs in the UPS module of the computer room equipment, the load is switched to the built-in redundant module to ensure uninterrupted power supply for the equipment; for storage device switching: for storage devices, when a fault is detected in a certain hard disk, the hot spare disk is automatically activated and the RAID is rebuilt.

[0038] Preferably, the security situation awareness module includes an abnormal behavior recognition module and an intelligent patrol robot module. The abnormal behavior recognition module includes a behavior data acquisition module, a data modeling and analysis module, an abnormal detection module, and a response module. Among them, the behavior data acquisition module collects the position and movement of personnel in the computer room based on a millimeter-wave radar; an infrared thermal imager monitors the ambient temperature; a sound sensor captures the sound information in the computer room; a camera records the video of personnel behavior; an intelligent access control records the time and position of personnel swiping cards; captures the SSH login logs of servers, database operation instructions, and five-tuple information of network traffic;

[0039] The data modeling and analysis module is used to construct the collected data into a spatio-temporal data cube, covering time stamps, spatial coordinates, personnel IDs, equipment IDs, operation types, and environmental parameters, and extract spatial features, time features, and behavior features from them;

[0040] The abnormal detection module is based on the isolation forest algorithm and introduces a time decay factor to construct a three-dimensional space density tree considering floors, cabinets, and equipment. Among them, the floor considered is the floor of the data center, and the algorithm formula is:

[0041] Score = α × spatial anomaly degree + β × time anomaly degree + γ × behavior anomaly degree;

[0042] Among them, α, β, and γ are the weight coefficients of the spatial anomaly degree, time anomaly degree, and behavior anomaly degree. α is set to 0.4, β is 0.3, γ is 0.3, and Score is the anomaly score;

[0043] The response module is based on an anomaly scoring real-time alarm strategy. The alarm strategy includes a yellow warning with a score of 0.6 - 0.8, automatically pushing risk prompts to the security officer; an orange warning with a score of 0.8 - 0.95, triggering the access control system to lock the current area; a red warning with a score > 0.95, immediately cutting off the device network connection and starting video recording backtracking.

[0044] Preferably, the intelligent inspection robot module is used to control the seven-axis robot to inspect the computer room and the equipment in the computer room.

[0045] Compared with the prior art, the present invention provides an intelligent management and control system for computer room equipment, having the following beneficial effects:

[0046] 1. For this intelligent management and control system of computer room equipment, through the dynamic load balancing module, the power supply strategy can be switched according to the actual state of the computer room equipment. Based on this, the energy consumption of the computer room equipment can be effectively reduced, and at the same time, the loss of battery life is quantified as a decision parameter for subsequent strategy switching. Through the set predictive maintenance module, it can fuse multi-source data such as vibration signals and oil spectra for prediction. It has a high failure prediction rate for computer room equipment, and can dynamically adjust the weights of different data sources to ensure accurate identification of fault symptoms at different operating stages of the computer room equipment, and can take corresponding measures at different stages of equipment failure, minimizing the impact of equipment failure on the business. Through the set security situation awareness module, it can fuse personnel movement trajectories, operation instructions, and environmental parameters to improve the accuracy of illegal intrusion detection.

[0047] 2. For this intelligent management and control system of computer room equipment, by introducing a time decay factor and a three-dimensional space density tree into the improved isolation forest algorithm, the detection ability for recent abnormal behaviors and abnormalities in specific areas can be significantly improved. At the same time, the power grid carbon intensity factor can be updated, combined with the green electricity matching intelligent contract, to automatically adjust the energy consumption strategy to preferentially use green electricity, effectively reducing the carbon intensity factor and achieving an energy-saving effect.

[0048] 3. This intelligent management and control system of computer room equipment has the characteristics of dynamically optimizing energy consumption, maintenance, and security strategies, adapting to complex scenario changes, and can discover and handle faults in advance to reduce losses. Based on this, the service life of computer room equipment can be effectively increased.

[0049] 4. For this intelligent management and control system of computer room equipment, the heat pipe radiator can be linked with the air conditioner, and the refrigeration tasks can be dynamically allocated according to the real-time temperature prediction. The heat pipe radiator can preferentially handle local hot spots, and the air conditioner is responsible for global temperature control. Based on this, the heat island effect can be effectively eliminated.

[0050] 5. When selecting the power supply strategy, the intelligent control system for computer room equipment can automatically adjust according to real-time electricity price fluctuations, load trends, and equipment health status. It incorporates the proportion of renewable energy generation into the carbon intensity calculation model, establishes a mathematical relationship between the green electricity usage ratio and carbon emission calculation, enables the carbon intensity factor to be dynamically adjusted according to the actual energy structure, and takes into account the impact of server temperature and air-conditioning energy efficiency ratio on carbon emissions. The corresponding correction coefficient formulas are respectively derived, which can more accurately reflect the carbon emission situation of the equipment under different operating conditions, and can statistically analyze the carbon emissions from multiple perspectives. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is the system block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] To make the technical means, creative features, achieved purposes, and functions of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0053] Embodiment 1:

[0054] As Figure 1 shown, an intelligent control system for computer room equipment includes an energy consumption control module, a predictive maintenance module, and a security situation awareness module. The energy consumption control module realizes the reduction of the energy consumption cost of computer room equipment based on the coordinated action of a dynamic load balancing module, a phase change heat dissipation control module, and a blockchain energy consumption ledger module, and is also used for accurately tracking carbon emissions.

[0055] The dynamic load balancing module includes a multi-dimensional state perception module, an intelligent decision-making strategy module, and a strategy selection module. The multi-dimensional state perception module is used to collect the remaining power of the UPS of computer room equipment, the power of the standby power supply, and the current load rate; real-time electricity price and electricity price prediction for the next 24 hours; server temperature, computer room temperature and humidity, and cumulative operation duration of equipment; number of power supply switches, power grid stability, and business priority.

[0056] The intelligent decision-making strategy module includes a power supply strategy action library, specifically including: the valley electricity priority strategy, whose trigger condition is that the real-time electricity price is in the valley period and it is predicted that the electricity price will rise in the next 2 hours, and the execution action is to switch to grid power supply, start charging the backup power supply, and increase the server CPU utilization rate to 80%; the pre-charging switching strategy, whose trigger condition is that it is predicted that the load rate will exceed 85% in the next 5 minutes and the remaining power of the UPS is <40%, and the execution action is to start pre-charging the backup power supply to 90% 5 minutes in advance, use zero-voltage transition during switching, and automatically detect the bus voltage stability after switching; the life protection strategy, whose trigger condition is that the UPS battery health is <40% and the internal resistance growth rate >15% / month, and the execution action is to prohibit the number of single-day switches >3 times, start the flywheel energy storage system to bear the peak load, and reduce the server operating main frequency by 10%; the emergency load shunting strategy, whose trigger condition is that the cluster load rate >90% and lasts for more than 10 minutes, and the execution action is to automatically migrate non-core services to the backup server cluster and request temporary grid capacity increase; the green electricity priority strategy, whose trigger condition is that the green electricity certificate is available and the green electricity price ≤ 110% of the thermal power price, and the execution action is to preferentially use green electricity for power supply, generate a green electricity usage certificate, and preferentially store green electricity; the temperature and load linkage strategy, whose trigger condition is that the average server temperature >35°C and the load rate >75%, and the execution action is to automatically adjust the air supply temperature of the air conditioning system to reduce the temperature by 2°C and trigger the cabinet air damper opening of the computer room equipment to 70%; the grid stability right-side strategy, whose trigger condition is that the grid voltage fluctuation exceeds ±5% or the frequency deviation >0.2Hz, and the execution action is to cut off the connection with the grid and switch to UPS power supply, and reduce the power consumption of non-critical equipment by 15%; the service priority strategy, whose trigger condition is that the high-priority service load rate >80%, and the trigger condition is to reserve 20% of the UPS capacity for high-priority services, enable a dedicated heat dissipation channel, and prohibit non-core services from migrating to the high-priority cluster;

[0057] The strategy selection module specifically includes:

[0058] S201: Acquisition and standardization of core parameters. The core parameters include electricity price, load, and equipment health. Among them, the electricity price obtains the real-time electricity price and the predicted electricity price for the next 24 hours from the power company API; the load obtains the cluster average load rate through the server management interface; the equipment health is obtained by calculating the remaining life of the UPS battery. The electricity price is normalized to the [0,1] interval according to peak and valley values; the load rate exceeding 85% is set to 1, and the load rate below 50% is set to 0, with linear mapping in between; the equipment health is set such that the remaining life >80% is 1, and <30% is 0, and an exponential function is used to enhance the sensitivity to low health;

[0059] S202: Dynamic weight allocation, set the basic weight of electricity price to 40%, the basic weight of load to 35%, and the basic weight of equipment health to 25%, and modify the weight based on actual conditions: when the electricity price fluctuates by more than ±10%, the electricity price weight increases by 5-10%, when the load rate is >80% for 30 consecutive minutes, the load weight increases to 45%, and when the health is <40%, the health weight is forcibly increased to 40%; for medical services among high-priority services, the health weight is increased by an additional 10%;

[0060] S203: Construction of strategy scoring model, according to the formula:

[0061] Comprehensive score = (electricity price parameter × electricity price weight) + (load parameter × load weight) + (equipment health parameter × equipment health weight);

[0062] The comprehensive score of each strategy is calculated, and the strategy with the highest comprehensive score is selected. When the score difference between the two strategies with the highest comprehensive scores is <0.05, manual review is adopted. It also includes the strategy adaptation coefficient and strategy selection logic. The strategy adaptation coefficient is specifically as follows: for the pre-charging switching strategy, when the electricity price is in the valley period, the comprehensive score is +0.15; for the life protection strategy, when the health level is <50%, the comprehensive score is +0.2.

[0063] The phase change cooling control module is based on dual closed-loop control, in which the main loop is a computer room temperature prediction model based on BP neural network, which is used to predict hot spots 30 minutes in advance; the secondary loop uses fuzzy PID algorithm to control the working fluid flow of the heat pipe radiator in the computer room. When the predicted temperature exceeds the preset threshold, the heat pipe radiator is started first and the air outlet angle of the air conditioner is adjusted in linkage.

[0064] The blockchain energy consumption ledger module includes a data collection and preprocessing module, a blockchain underlying architecture module, a data structure design module, and a carbon footprint calculation and green electricity traceability module. The data collection and preprocessing module is used to collect data from the power distribution cabinets in the equipment room, health parameter data of the batteries used, and environmental data, and clean the collected data.

[0065] The underlying blockchain architecture module is used to design alliance chains and smart contracts. The alliance chain includes master nodes: data center operators; verification nodes: power companies, third-party certification agencies; audit nodes: government regulatory departments, and a consensus mechanism based on practical Byzantine fault tolerance is adopted. The smart contract is an energy consumption data storage contract written in Solidity, which specifically includes a data writing interface that requires signature authentication from the master node, a carbon emission calculation function based on the ISO14064 standard, and green electricity matching rules that automatically associate REC certificates.

[0066] The data structure design module includes generating time - series data blocks, specifically generating one block every 5 minutes, which contains the ID of the computer room equipment, energy consumption data, environmental parameters, and the trigger conditions of the smart contract. At the same time, an indexing mechanism is established: establishing a mapping relationship between the device ID - block hash, supporting fast query of vertical data of a single device, establishing a hierarchical index according to the time range, and improving the audit efficiency;

[0067] The carbon footprint calculation and green electricity traceability module includes a carbon emission calculation module and a green electricity matching module. The green electricity matching module automatically scans the REC certificate library based on the smart contract to match the green electricity usage amount in the same time period. When the green electricity usage ratio is less than 50%, the green electricity priority strategy is directly triggered.

[0068] The carbon emission calculation module calculates the real - time carbon emissions based on the formula:

[0069]

[0070] Where is the real - time carbon emissions, n is the number of types of equipment in the computer room, P i represents the active power of the i - th type of equipment, Δt is the statistical time interval, CF is the grid carbon intensity factor, and η i is the energy efficiency correction coefficient of the i - th type of equipment;

[0071] It also includes a dynamic grid carbon intensity factor update module, which collects the carbon intensity of the regional grid and the proportion of renewable energy generation released by the power company, and is used for green electricity matching and dynamically adjusting the carbon intensity factor. The formula is:

[0072] CF new =CF base ×(1 - REF);

[0073] Where CF new is the new grid carbon intensity factor, CF base is the basic carbon intensity factor, and RER is the proportion of renewable energy generation;

[0074] It also includes a carbon footprint generation module. The carbon footprint includes a time dimension and a space dimension. The time dimension includes generating carbon emission trend reports for the day, week, and month, including year - on - year and month - on - month analysis, to intuitively show the change of carbon emissions over time, and to statistically show the proportion of carbon emissions during peak hours, helping users understand high - energy - consumption periods; The space dimension includes drawing a carbon emission heat map at the cabinet level, grading different computer room equipment according to the PUE value, intuitively showing the carbon emission distribution of each cabinet in the computer room, and listing the top 5 equipment with the highest carbon emissions in each area of the data center, facilitating users to focus on and manage high - energy - consumption equipment.

[0075] Example 2:

[0076] As shown Figure 1 in the figure, an intelligent management and control system for computer room equipment. The predictive maintenance module collects equipment data in real time based on multimodal sensors, combines algorithms with digital twin stress analysis to achieve early prediction of equipment failures, dynamic assessment of health status, and self-repair control;

[0077] The predictive maintenance module includes a multimodal data collection and preprocessing module, a data fusion model, a health status assessment module, a self-repair module, and a knowledge graph construction module. Among them, the multimodal data collection and preprocessing module collects vibration, oil analysis, temperature, and current data of computer room equipment based on sensors and cleans the collected data;

[0078] The data fusion model includes constructing a feature extraction sub-model, which is used to extract the features of the data collected by the multimodal data collection and preprocessing module, cascade the extracted feature vectors, and input them into a three-layer fully connected cascade neural network, which is used to learn the association between different modal data features and output the failure probability of computer room equipment. At the same time, the model is trained;

[0079] The health status assessment module establishes a digital twin of the equipment based on BIM technology, including the geometric structure, material properties, and remainder conditions of the computer room equipment. At the same time, sensor data is imported to drive the digital twin, and finite element analysis is used to calculate the stress generated when the cabinet equipment works. Based on the Miner linear cumulative damage theory, the remaining life of the computer room equipment is predicted;

[0080] The knowledge graph construction module is used to deeply analyze the relationship between the failure phenomena, environmental parameters, and equipment status of computer room equipment, establish a causal relationship network of failure phenomena, environmental parameters, and equipment status, use the failure phenomena, environmental parameters, and equipment status as nodes of the knowledge graph, and the causal relationship between them as edges. Based on this, the failure knowledge of the equipment is represented in a graphical way for subsequent reasoning and analysis.

[0081] The self-repair module includes a first-level response module, a second-level response module, and a third-level response module. The first-level response module includes health status monitoring and inspection adjustment and spare part procurement recommendation generation. Among them, health status monitoring and inspection adjustment: real-time monitor the health status indicators of the equipment. When the health status is lower than 70%, the system automatically increases the inspection frequency of the equipment from the normal once a day to once an hour; Spare part procurement recommendation generation: generate spare part procurement recommendations according to the failure type and historical maintenance records of the equipment, combined with the delivery cycle of the supplier;

[0082] The secondary response module includes fan failure regulation and server temperature regulation. Among them, for fan failure regulation: when an abnormality of the fan is detected, the rotation speed of the redundant fan is dynamically adjusted so that the increased air volume of the redundant fan can compensate for the reduced air volume of the faulty fan, which is used to ensure that the heat dissipation effect of the computer room equipment is not affected; for server temperature regulation: when the temperature of the server CPU rises abnormally, the operating frequency of the CPU is reduced to reduce the heat generation of the CPU, and at the same time, the heat dissipation efficiency is increased until the CPU temperature returns to the normal range.

[0083] The tertiary response module includes UPS module switching and storage device switching. Among them, for UPS module switching: when a failure occurs in the UPS module of the computer room equipment, the load is switched to the built-in redundant module to ensure uninterrupted power supply of the equipment. During the switching process, strict electrical detection is carried out to ensure the stability and safety of the switching; for storage device switching: for storage devices, when a failure is detected in a certain hard disk, the hot spare disk is automatically activated and the RAID is rebuilt.

[0084] Embodiment 3:

[0085] As Figure 1 shown, an intelligent management and control system for computer room equipment. The security situation awareness module constructs an active defense system for the computer room by integrating spatio-temporal behavior analysis and intelligent patrol data, which is used to realize the detection of illegal intrusion of computer room equipment and the protection against abnormal replacement of equipment.

[0086] The security situation awareness module includes an abnormal behavior recognition module and an intelligent patrol robot module. The abnormal behavior recognition module includes a behavior data collection module, a data modeling and analysis module, an abnormal detection module, and a response module. Among them, the behavior data collection module collects the position and movement of personnel in the computer room based on a millimeter-wave radar; an infrared thermal imager monitors the environmental temperature; a sound sensor captures the sound information in the computer room; a camera records the video of personnel behavior; an intelligent access control records the time and position of personnel swiping cards; captures the SSH login logs of the server, database operation instructions, and five-tuple information of network traffic.

[0087] The data modeling and analysis module is used to construct a spatio-temporal data cube from the collected data, covering time stamps, spatial coordinates, personnel IDs, equipment IDs, operation types, and environmental parameters, and extracts spatial features, time features, and behavior features from it.

[0088] The abnormal detection module is based on the isolation forest algorithm and introduces a time decay factor to construct a three-dimensional spatial density tree considering floors, cabinets, and equipment. Among them, the floor considered is the floor of the data center, and the algorithm formula is:

[0089] Score = α × spatial anomaly degree + β × time anomaly degree + γ × behavior anomaly degree;

[0090] Among them, α, β, and γ are the weight coefficients of spatial anomaly degree, time anomaly degree, and behavior anomaly degree. α is set to 0.4, β is 0.3, γ is 0.3, and Score is the anomaly score;

[0091] The response module is based on the real-time alarm strategy of the anomaly score. The alarm strategy includes a yellow warning for Score 0.6 - 0.8, automatically pushing risk prompts to the security officer; an orange warning for Score 0.8 - 0.95, triggering the access control system to lock the current area; a red warning for Score > 0.95, immediately cutting off the device network connection and starting video recording and backtracking;

[0092] The intelligent inspection robot module is used to control the seven-axis robot to conduct inspections on the computer room and the equipment in the computer room. The seven-axis robot is equipped with an omnidirectional wheel chassis, with a load capacity of up to 150 kg, equipped with a laser SLAM navigation system, a mapping accuracy of ±0.05 m, and ultrasonic obstacle avoidance sensors with an effective distance of 5 m. The operation module includes a seven-axis robotic arm with a repeat positioning accuracy of ±0.02 mm, a high-definition camera supporting 12x optical zoom, an infrared thermal imager with a temperature measurement range of -20°C to 600°C. The seven-axis robot plans the optimal inspection path based on the A* algorithm and has the ability to dynamically adjust. When encountering people, it actively avoids and maintains a safe distance of ≥1.5 m. It executes operation tasks at a fixed time interval, cleans the cabinet surface every 2 hours using a 0.3 MPa air flow; tightens the power interface every 4 hours, with the torque controlled at 1.5 ± 0.1 N·m; takes real-time pictures of the equipment labels. When abnormal situations such as loose screws are found, it marks the position and generates a maintenance work order for temporary reinforcement, and at the same time pushes the warning information. In addition, this robot is also linked with the environmental control system of the computer room. When the surface temperature of the equipment is detected to be > 55°C, it automatically turns on the corresponding cabinet fan; when the color of the cable insulation layer changes, it is linked to cut off the power supply of this circuit.

[0093] It should be noted that the present invention is an intelligent management and control system for computer room equipment. When in use, it includes:

[0094] One: Energy consumption cost reduction and carbon footprint management, real-time analysis of electricity price, load rate, and equipment health, dynamically adjusting the power supply strategy, predicting hot spots 30 minutes in advance, and optimizing the linkage between heat pipe radiators and air conditioners;

[0095] Write the energy consumption data into the alliance chain every 5 minutes, generate a carbon emission report in the time / space dimension, and adjust the grid carbon intensity factor based on the proportion of renewable energy generation;

[0096] Calculate the optimal power supply strategy according to the calculation formula, and select the power supply strategy for execution in combination with the adjusted strategy adaptation coefficient and execution logic;

[0097] II: Improve the reliability of the equipment. Multimodal sensors collect vibration, oil, and temperature data in real time, analyze faults through a cascaded neural network, perform self-repair of the equipment through a three-level response module, and conduct hierarchical management of the health of the equipment in the computer room. When the health of the UPS battery < 40%, prohibit switching more than 3 times in a single day and enable flywheel energy storage to bear the peak load of the equipment;

[0098] III: Strengthen safety protection. Integrate millimeter-wave radar, infrared thermal imaging, and server log data to identify abnormal behaviors, conduct inspections of the equipment in the computer room through intelligent inspection robots, calculate the abnormal score of the current behavior according to the abnormal score formula, and select corresponding response strategies based on the scored data;

[0099] IV: Collaborative control mechanism, including the triggering of the green electricity priority strategy: when the green electricity certificate is available and the price ≤ 110% of thermal power, give priority to using green electricity and generate a usage certificate. When the proportion of green electricity usage < 50%, automatically trigger the adjustment of the load balancing strategy; Grid stability response: when the voltage fluctuation is ±5% or the frequency deviation > 0.2Hz, switch to UPS power supply and reduce the power consumption of non-critical equipment by 15%; High-priority service guarantee: when the high-priority service load > 80%, reserve 20% of the UPS capacity and enable a dedicated heat dissipation channel at the same time.

[0100] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent control system for equipment in a computer room, including an energy consumption control module, a predictive maintenance module, and a security situation awareness module, characterized in that: The energy consumption control module is based on the synergy of the dynamic load balancing module, the phase change heat dissipation control module and the blockchain energy consumption ledger module to reduce the energy consumption cost of the equipment in the computer room and is also used to accurately track carbon emissions; The predictive maintenance module collects equipment data in real time based on multimodal sensors, combines algorithms with digital twin stress analysis, and achieves early prediction of equipment failures, dynamic health assessment, and self-repair control; The security situation awareness module builds an active defense system for the computer room by integrating spatiotemporal behavior analysis and intelligent inspection data, which is used to detect illegal intrusions into the computer room equipment and protect against abnormal equipment replacement.

2. According to claim 1, the intelligent control system for computer room equipment is characterized by: The dynamic load balancing module includes a multi-dimensional state perception module, an intelligent decision-making strategy module, and a strategy selection module, wherein the multi-dimensional state perception module is used to collect the UPS remaining power, backup power supply power, current load rate, real-time electricity price, and electricity price forecast for the next 24 hours of the equipment in the computer room; Server temperature, room temperature and humidity, and accumulated equipment operating time; Power switching times, grid stability, and business priorities; The intelligent decision-making strategy module includes a power supply strategy action library, specifically including: valley power priority strategy, the trigger condition is that the real-time electricity price is in the valley period and the electricity price is predicted to rise in the next 2 hours, and the execution action is to switch to the grid power supply, start the backup power supply charging, and increase the server CPU utilization rate to 80%; pre-charge switching strategy, the trigger condition is that the load rate in the next 5 minutes is predicted to exceed 85% and the remaining power of the UPS is <40%, and the execution action is to start the backup power supply to pre-charge to 90% 5 minutes in advance, use zero voltage transition during switching, and automatically detect the bus voltage stability after switching; life protection strategy, the trigger condition is that the UPS battery health is <40%, the internal resistance growth rate is >15% / month, the execution action is to prohibit the number of switching times per day >3 times, start the flywheel energy storage system to bear the peak load, and reduce the server operating frequency by 10%; emergency load diversion strategy, the trigger condition is that the cluster load rate is >90% and lasts for more than 10 minutes, and the execution action is to automatically migrate non Core business to the backup server cluster, request temporary capacity increase of the power grid; green electricity priority strategy, the trigger condition is that the green electricity certificate is available and the green electricity price is ≤110% of the thermal power price, the execution action is to give priority to green electricity supply, generate green electricity use certificate, and give priority to storing green electricity; temperature and load linkage strategy, the trigger condition is that the average server temperature is >35℃ and the load rate is >75%, the execution action is to automatically adjust the air supply temperature of the air conditioning system to reduce the temperature by 2℃ and trigger the cabinet damper opening of the equipment in the computer room to 70%; grid stability right side strategy, the trigger condition is that the grid voltage fluctuation exceeds ±5% or the frequency deviation is >0.2Hz, the execution action is to cut off the connection with the grid and switch to UPS power supply, and reduce the power consumption of non-critical equipment by 15%; business priority strategy, the trigger condition is that the high priority business load rate is >80%, the trigger condition is to reserve 20% UPS capacity for high priority business, enable dedicated heat dissipation channel, and prohibit non-core business from migrating to high priority cluster; The strategy selection module specifically includes: S201: Collection and standardization of core parameters. The core parameters include electricity price, load and equipment health. The electricity price is obtained from the power company API to obtain the real-time electricity price and the forecast for the next 24 hours. The load obtains the average load rate of the cluster through the server management interface. The equipment health is obtained by calculating the remaining life of the UPS battery. The electricity price is normalized to the [0,1] interval according to the peak and valley values. The load rate is set to 1 when it exceeds 85% and 0 when it is less than 50%, with linear mapping in between. The equipment health is set to 1 when the remaining life is >80% and 0 when it is <30%. An exponential function is used to enhance the sensitivity to low health. S202: Dynamic weight allocation, set the basic weight of electricity price to 40%, the basic weight of load to 35%, and the basic weight of equipment health to 25%, and modify the weight based on actual conditions: when the electricity price fluctuates by more than ±10%, the electricity price weight increases by 5-10%, when the load rate is >80% for 30 consecutive minutes, the load weight increases to 45%, and when the health is <40%, the health weight is forcibly increased to 40%; for medical services among high-priority services, the health weight is increased by an additional 10%; S203: Construction of strategy scoring model, according to the formula: Comprehensive score = (electricity price parameter × electricity price weight) + (load parameter × load weight) + (equipment health parameter × equipment health weight); The comprehensive score of each strategy is calculated, and the strategy with the highest comprehensive score is selected. When the score difference between the two strategies with the highest comprehensive scores is <0.05, manual review is performed.

3. According to claim 2, the intelligent control system for computer room equipment is characterized in that: The S203 also includes a strategy adaptation coefficient and a strategy selection logic, wherein the strategy adaptation coefficient is specifically: for the pre-charging switching strategy, when the electricity price is in the valley period, the comprehensive score is +0.15; for the life protection strategy, when the health level is <50%, the comprehensive score is +0.

2.

4. The intelligent control system for computer room equipment according to claim 1 is characterized in that: The phase change heat dissipation control module is based on dual closed-loop control, where the main loop is a computer room temperature prediction model based on BP neural network, which is used to predict hot spots 30 minutes in advance; The secondary loop uses fuzzy PID algorithm to control the working fluid flow of the heat pipe radiator in the computer room. When the predicted temperature exceeds the preset threshold, the heat pipe radiator is started first and the air outlet angle of the air conditioner is adjusted in linkage.

5. The intelligent control system for computer room equipment according to claim 1 is characterized in that: The blockchain energy consumption ledger module includes a data collection and preprocessing module, a blockchain underlying architecture module, a data structure design module, and a carbon footprint calculation and green electricity traceability module. The data collection and preprocessing module is used to collect data from the power distribution cabinet in the computer room equipment, health parameter data of the batteries used, and environmental data, and clean the collected data; The blockchain underlying architecture module is used to design alliance chains and smart contracts, where the alliance chain includes master nodes: data center operators; Verification nodes: power companies, third-party certification agencies; Audit nodes: government regulatory departments, while using a consensus mechanism based on practical Byzantine fault tolerance; Smart contracts are energy consumption data storage contracts written in Solidity, specifically including data writing interfaces, carbon emissions calculation functions, and green electricity matching rules; The data structure design module includes generating a time series data block; The carbon footprint calculation and green electricity traceability module includes a carbon emission calculation module and a green electricity matching module. The green electricity matching module automatically scans the REC certificate library based on the smart contract to match the green electricity usage in the same time period. When the green electricity usage ratio is lower than 50%, the green electricity priority strategy is directly triggered.

6. The intelligent control system for computer room equipment according to claim 5 is characterized in that: The carbon emission calculation module calculates the real-time carbon emission based on the formula: in is the real-time carbon emissions, n is the number of equipment types in the computer room, P i represents the active power of the i-th type of equipment, Δt is the statistical time interval, CF is the carbon intensity factor of the power grid, η i is the energy efficiency correction factor for the i-th type of equipment; It also includes a dynamic grid carbon intensity factor update module, which collects the carbon intensity of the regional grid and the proportion of renewable energy generation published by the power company, and is used for green electricity matching and dynamic adjustment of the carbon intensity factor. The formula is: CF new =CF base ×(1-RER); where CF new is the new grid carbon intensity factor, CF base is the basic carbon intensity factor, RER is the proportion of electricity generated by renewable energy; It also includes a carbon footprint generation module. The carbon footprint includes time dimension and space dimension. The time dimension includes generating daily, weekly and monthly carbon emission trend reports, including year-on-year and month-on-month analysis, to intuitively display the changes in carbon emissions over time, and count the proportion of carbon emissions during peak periods, to help users understand high energy consumption periods; the space dimension includes drawing cabinet-level carbon emission heat maps, grading different computer room equipment according to PUE values, intuitively displaying the carbon emission distribution of each cabinet in the computer room, and listing the top 5 equipment in terms of carbon emissions in each area of ​​the data center, so that users can focus on and manage high-energy consumption equipment.

7. An intelligent control system for computer room equipment according to claim 1, characterized in that: The predictive maintenance module includes a multimodal data collection and preprocessing module, a data fusion model, a health assessment module, a self-repair module, and a knowledge graph construction module, wherein the multimodal data collection and preprocessing module collects vibration, oil analysis, temperature and current data of the equipment in the computer room based on this sensor, and cleans the collected data; The data fusion model includes constructing a feature extraction sub-model for extracting features of data collected by the multi-modal data collection and preprocessing module, cascading the extracted feature vectors, and inputting them into a 3-layer fully connected cascade neural network, for outputting the failure probability of the equipment in the computer room by learning the association between the features of different modal data, and training the model at the same time; The health assessment module establishes a digital twin of equipment based on BIM technology, including the geometric structure, material properties and remainder conditions of the equipment in the computer room. At the same time, it imports sensor data to drive the digital twin, uses finite element analysis to analyze the stress generated when the cabinet equipment is working, and predicts the remaining life of the equipment in the computer room based on Miner's linear cumulative damage theory. The knowledge graph construction module is used to conduct in-depth analysis of the relationship between the fault phenomenon, environmental parameters and equipment status of the computer room equipment, establish a causal relationship network of fault phenomena, environmental parameters and equipment status, use the fault phenomenon, environmental parameters and equipment status as nodes of the knowledge graph, and the causal relationship between them as edges. Based on this, the equipment fault knowledge is represented in a graphical manner for subsequent reasoning and analysis.

8. An intelligent control system for computer room equipment according to claim 7, characterized in that: The self-repair module includes a primary response module, a secondary response module and a tertiary response module, wherein the primary response module includes health monitoring and inspection adjustment and spare parts procurement suggestion generation, wherein health monitoring and inspection adjustment: real-time monitoring of the health index of the equipment, when the health is lower than 70%, the system automatically increases the inspection frequency of the equipment from the normal once a day to once an hour; spare parts procurement suggestion generation: based on the equipment's fault type and historical maintenance records, combined with the supplier's delivery cycle, generate spare parts procurement suggestions; The secondary response module includes fan fault adjustment and server temperature adjustment, wherein fan fault adjustment: when a fan abnormality is detected, the speed of the redundant fan is dynamically adjusted so that the increased air volume of the redundant fan can compensate for the reduced air volume of the faulty fan, so as to ensure that the heat dissipation effect of the equipment in the computer room is not affected; server temperature adjustment: when the temperature of the server CPU rises abnormally, the operating frequency of the CPU is reduced to reduce the heat generated by the CPU and increase the heat dissipation efficiency until the CPU temperature returns to the normal range; The three-level response module includes UPS module switching and storage device switching, wherein UPS module switching: when the UPS module of the computer room equipment fails, the load is switched to the built-in redundant module to ensure uninterrupted power supply to the equipment; storage device switching: for storage devices, when a hard disk failure is detected, the hot spare disk is automatically activated and the RAID is rebuilt.

9. The intelligent control system for computer room equipment according to claim 1, characterized in that: The security situation awareness module includes an abnormal behavior recognition module and an intelligent inspection robot module. The abnormal behavior recognition module includes a behavior data acquisition module, a data modeling and analysis module, an abnormal detection module, and a response module. The behavior data acquisition module collects the position and movement of personnel in the computer room based on millimeter wave radar; the infrared thermal imager monitors the ambient temperature; The sound sensor captures the sound information in the computer room; the camera records the video of personnel behavior; the smart access control records the time and location of personnel swiping the card; the server's SSH login log, database operation instructions and five-tuple information of network traffic are captured; The data modeling and analysis module is used to construct the collected data into a spatiotemporal data cube, covering timestamps, spatial coordinates, personnel IDs, equipment IDs, operation types and environmental parameters, and extract spatial features, temporal features and behavioral features therefrom; The anomaly detection module is based on the isolation forest algorithm and introduces a time decay factor to construct a three-dimensional spatial density tree that considers floors, cabinets, and equipment, where the floors considered are floors of a data center. The algorithm formula is: Score = α × spatial abnormality + β × temporal abnormality + y × behavioral abnormality; Among them, α, β, and γ are the weight coefficients of spatial abnormality, temporal abnormality, and behavioral abnormality, α is set to 0.4, β is 0.3, γ is 0.3, and Score is the abnormality score; The response module is based on a real-time alarm strategy based on abnormal scores. The alarm strategy includes a yellow warning for scores of 0.6-0.8, which automatically pushes risk warnings to safety officers. Orange warning with score 0.8-0.95 triggers the access control system to lock the current area; If Score>0.95 is a red warning, the device network connection will be immediately disconnected and video playback will be started.

10. The intelligent control system for computer room equipment according to claim 1, characterized in that: The intelligent inspection robot module is used to control the seven-axis robot to inspect the machine room and the equipment in the machine room.

Images