Energy-saving and consumption-reducing efficient machine room management system

Abstract

The invention discloses an energy-saving and consumption-reducing efficient machine room management system, and relates to the technical field of data center energy management, the management system comprises a cloud edge-end three-level collaborative architecture: an end layer is composed of a multi-mode sensing network and an intelligent execution system, and an edge layer is an intelligent edge controller connected with the end layer through an LoRa / industrial Ethernet; the cloud layer is a dynamic optimization platform connected with the edge layer through a 5G / optical fiber. The method has the advantages that a cloud edge end three-level collaborative architecture sinks a core decision to an edge node, response delay is compressed to be within 50 ms, an edge controller executes emergency control in real time based on a local flow processing engine, monitoring blind areas are dynamically eliminated in combination with a reconfigurable sensing network, and the real-time performance of the edge controller is improved. The spatial resolution reaches 0.1 m < 3 >, compared with a traditional centralized system, the hot spot detection accuracy is improved, the starting and stopping frequency of a compressor is reduced, and the service life of equipment is prevented from being shortened from the source.

Description

Technical Field

[0001] This invention relates to the field of data center energy management technology, specifically to a high-efficiency data center management system that saves energy and reduces consumption. Background Technology

[0002] The existing data center energy management system has significant deficiencies:

[0003] First, the collaborative control lacks real-time performance. The decision-making delay of the centralized architecture exceeds 500ms, making it difficult to respond to dynamic changes in equipment. The distributed solution lacks cross-level coordination, resulting in conflicts between local optimization and global goals.

[0004] Secondly, the energy efficiency optimization is too narrow. Most systems focus only on temperature control, ignoring the coupled effects of multiple parameters such as humidity and airflow organization. Furthermore, the use of static dehumidification materials and threshold control strategies leads to frequent compressor start-stops and accelerates equipment aging.

[0005] Third, the data sensing capability is weak, fixed sensors have monitoring blind spots, and millisecond-level energy efficiency fluctuations cannot be captured. The data silo phenomenon further causes distortion in energy efficiency analysis.

[0006] To address the aforementioned issues, there is an urgent need for a solution that integrates multimodal sensing and dynamic optimization. This solution should ensure precise temperature and humidity control while overcoming real-time limitations, and collaboratively optimize multiple objectives such as energy consumption and equipment lifespan, thereby achieving comprehensive benefits such as reducing the PUE value to below 1.25 and increasing the energy efficiency of the air conditioning system by over 40%. Summary of the Invention

[0007] The purpose of this invention is to provide an energy-saving and consumption-reducing high-efficiency data center management system.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an energy-saving and consumption-reducing high-efficiency data center management system, comprising a management system, wherein the management system includes a three-level collaborative architecture of cloud, edge, and endpoint.

[0009] End layer: Composed of multimodal sensor networks and intelligent execution systems;

[0010] Edge layer: Intelligent edge controllers that connect to the end layer via LoRa / Industrial Ethernet;

[0011] Cloud Layer: A dynamically optimized platform that connects to the edge layer via 5G / fiber optics;

[0012] The multimodal sensing network includes an environmental sensing unit, an equipment status monitoring unit, and an energy consumption metering unit, which are used to collect environmental parameters, equipment status, and energy consumption data in real time.

[0013] The intelligent edge controller includes a lightweight AI inference module, a real-time stream processing engine, and an adaptive decision-maker for performing local decisions and emergency control.

[0014] The dynamic optimization platform includes a digital twin engine, a multi-objective optimizer, and a knowledge graph library for global policy optimization and model updates.

[0015] As a further aspect of the present invention: the multimodal sensing network satisfies:

[0016] The environmental sensing unit includes a temperature and humidity composite sensor (accuracy ±0.3℃ / ±2%RH, sampling rate 10Hz), a micro airflow velocity sensor (range 0-5m / s, resolution 0.01m / s), and a particle counter.

[0017] The sensor adopts a reconfigurable topology deployment method with 30% fixed nodes and 70% mobile robots to achieve three-dimensional dynamic scanning in space;

[0018] The equipment condition monitoring unit calculates equipment health based on a multi-source fusion of vibration, temperature, and current. The health model is as follows:

[0019] health index =1-0.4·S v -0.3·D t -0.3·THD i ;

[0020] In the formula, S v D is the entropy value of the vibration spectrum. t For temperature deviation, THD i denoted as the current harmonic distortion rate.

[0021] As a further aspect of the present invention: the adaptive decision-maker executes a multi-objective optimization algorithm.

[0022] The objective function is to minimize:

[0023] F(x) = [total energy consumption, absolute value of temperature deviation, absolute value of humidity deviation, and sum of equipment life loss factor];

[0024] The constraints include: temperature 20-25℃, humidity 40-60%RH, and equipment start-up and shutdown frequency ≤ 5 times / hour;

[0025] The NSGA-II algorithm is used to solve for the Pareto optimal solution set, and the weights of the objective function can be adaptively adjusted.

[0026] As a further aspect of the present invention: in the dynamic optimization platform:

[0027] The digital twin engine constructs a real-time physical-virtual mapping model of the data center, updating globally every minute, with a spatial resolution of 0.1m. 3 ;

[0028] The knowledge graph database stores the optimal operating point database for storage devices, a fault handling case library, and an energy-saving strategy effectiveness evaluation matrix.

[0029] As a further aspect of the present invention: the workflow of the management system includes:

[0030] The sensor network collects data at a 200ms interval;

[0031] When the edge controller immediately executes local control when the emergency constraint conditions are met, it optimizes the air conditioning air volume and the opening of the cooling tower valves.

[0032] The dynamic optimization platform receives edge layer data, updates the model, generates global strategies, and distributes parameters every 5 minutes.

[0033] As a further aspect of the present invention: the dynamic optimization platform and the intelligent edge controller are configured with an adaptive learning mechanism to perform the following steps:

[0034] Collect operational data (environmental parameters, equipment status, control commands);

[0035] Evaluate the effectiveness of the strategy (energy efficiency improvement rate, temperature stability, equipment life loss);

[0036] Dynamically adjust model parameters (neural network weights, objective function weights);

[0037] Update the optimization cases and fault handling records in the knowledge graph database.

[0038] As a further aspect of the present invention: the intelligent execution system includes:

[0039] Adjustable air conditioner: It adopts a combination of variable frequency compressor and EC fan to achieve stepless adjustment of cooling capacity from 30% to 100% with a response time of <10s;

[0040] Dynamic airflow wall: Equipped with 0-90° rotatable guide vanes, with airflow organization optimized based on CFD simulation;

[0041] Phase change energy storage device: adopts lauric acid-palmitic acid eutectic material (phase change temperature 23±2℃, cold storage density 180kJ / kg), and implements economic dispatch in combination with peak and off-peak electricity prices.

[0042] As a further aspect of the present invention: the adaptive adjustment of the objective function weights satisfies:

[0043]

[0044] In the formula, w i (t) represents the real-time weight of the i-th target at time t, where i∈{1,2,3,4}, corresponding to energy consumption, temperature deviation, humidity deviation, and equipment lifespan loss, respectively. i0 represents the preset baseline weight for the i-th objective, and E represents the system energy efficiency. α is the real-time energy efficiency change rate of the system, and α is the energy efficiency change sensitivity coefficient, which is a preset constant.

[0045] As a further aspect of the present invention: the hardware deployment satisfies:

[0046] The edge controller uses an NVIDIA Jetson AGX Orin compute unit, and is equipped with 32GB of LPDDR5 memory and 1TB of NVMe SSD storage;

[0047] The intelligent execution system includes a variable frequency drive (VFD) installed in the air conditioner, a guide vane array in the cabinet aisle, and a top-mounted phase change energy storage unit (200kg / m²). 2 ).

[0048] Compared with the prior art, the beneficial effects of the present invention by adopting the above technical solution are as follows:

[0049] 1. This invention utilizes a three-tiered cloud-edge-device collaborative architecture to push core decision-making down to edge nodes, compressing response latency to less than 50ms. The edge controller executes emergency control in real time based on a local stream processing engine, and combined with a reconfigurable sensor network to dynamically eliminate monitoring blind spots, achieving a spatial resolution of 0.1m. 3 Compared to traditional centralized systems, hotspot detection accuracy is improved, compressor start-up and shutdown frequency is reduced, and equipment lifespan is avoided from the root cause.

[0050] 2. This invention innovatively establishes a four-dimensional objective function of temperature, humidity, energy consumption, and equipment lifespan, and adopts an adaptive weighting mechanism to dynamically balance the conflicts among the objectives. Combined with an intelligent execution system, including stepless air conditioning, dynamic airflow guide walls, and peak-valley scheduling of phase change energy storage, it achieves a PUE value of less than 1.25, simultaneously ensuring precise temperature and humidity control and extending equipment lifespan, reducing air conditioning energy consumption, and solving the energy efficiency imbalance problem caused by traditional single-dimensional optimization.

[0051] 3. This invention is based on a multi-source fusion equipment health assessment model, which can provide early warning of faults 30 minutes in advance with a low false alarm rate. The knowledge graph library continuously accumulates cross-data center optimization cases, and the AI ​​model parameters are dynamically adjusted through an adaptive learning mechanism, enabling the system to have continuous evolution capabilities, reducing the need for operation and maintenance manpower, shortening the fault response time, and forming a closed loop of full life cycle management of "perception-decision-optimization-learning". Attached Figure Description

[0052] Figure 1 This is a flowchart of the cloud-edge-device three-level collaborative architecture of the present invention;

[0053] Figure 2 This is a flowchart of the management system of the present invention;

[0054] Figure 3 This is a flowchart of the adaptive learning mechanism of the present invention. Detailed Implementation

[0055] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0056] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0057] Please see the appendix Figure 1 - Appendix Figure 3 This invention discloses an energy-saving and consumption-reducing high-efficiency data center management system, comprising a management system, which includes a three-level collaborative architecture of cloud, edge, and terminal.

[0058] End layer: Composed of multimodal sensor networks and intelligent execution systems;

[0059] Edge layer: Intelligent edge controllers that connect to the end layer via LoRa / Industrial Ethernet;

[0060] Cloud Layer: A dynamically optimized platform that connects to the edge layer via 5G / fiber optics;

[0061] Multimodal sensor networks include environmental sensing units, equipment status monitoring units, and energy metering units, which are used to collect environmental parameters, equipment status, and energy consumption data in real time.

[0062] The intelligent edge controller includes a lightweight AI inference module, a real-time stream processing engine, and an adaptive decision-maker for performing local decisions and emergency control.

[0063] The dynamic optimization platform includes a digital twin engine, a multi-objective optimizer, and a knowledge graph library for global policy optimization and model updates.

[0064] In one embodiment of the present invention: the multimodal sensing network satisfies:

[0065] The environmental sensing unit includes a temperature and humidity composite sensor (accuracy ±0.3℃ / ±2%RH, sampling rate 10Hz), a micro airflow velocity sensor (range 0-5m / s, resolution 0.01m / s), and a particle counter.

[0066] The sensor adopts a reconfigurable topology deployment method with 30% fixed nodes and 70% mobile robots to achieve three-dimensional dynamic scanning in space;

[0067] The equipment condition monitoring unit calculates equipment health based on a multi-source fusion of vibration, temperature, and current. The health model is as follows:

[0068] health index =1-0.4·S v -0.3·D t -0.3·THD i ;

[0069] In the formula, S v D is the entropy value of the vibration spectrum. t For temperature deviation, THD i denoted as the current harmonic distortion rate.

[0070] In one embodiment of the present invention: the adaptive decision-maker executes a multi-objective optimization algorithm:

[0071] The objective function is to minimize:

[0072] F(x) = [total energy consumption, absolute value of temperature deviation, absolute value of humidity deviation, and sum of equipment life loss factor];

[0073] The constraints include: temperature 20-25℃, humidity 40-60%RH, and equipment start-up and shutdown frequency ≤ 5 times / hour;

[0074] The NSGA-II algorithm is used to solve for the Pareto optimal solution set, and the weights of the objective function can be adaptively adjusted.

[0075] In one embodiment of the present invention: in the dynamic optimization platform:

[0076] The digital twin engine constructs a real-time physical-virtual mapping model of the data center, updating globally every minute, with a spatial resolution of 0.1m. 3 ;

[0077] The knowledge graph database stores the optimal operating point database for storage devices, a fault handling case library, and an energy-saving strategy effectiveness evaluation matrix.

[0078] In one embodiment of the present invention, the management system workflow includes:

[0079] The sensor network collects data at a 200ms interval;

[0080] When the edge controller immediately executes local control when the emergency constraint conditions are met, it optimizes the air conditioning air volume and the opening of the cooling tower valves.

[0081] The dynamic optimization platform receives edge layer data, updates the model, generates global strategies, and distributes parameters every 5 minutes.

[0082] In one embodiment of the present invention: the dynamic optimization platform and the intelligent edge controller are configured with an adaptive learning mechanism to perform the following steps:

[0083] Collect operational data (environmental parameters, equipment status, control commands);

[0084] Evaluate the effectiveness of the strategy (energy efficiency improvement rate, temperature stability, equipment life loss);

[0085] Dynamically adjust model parameters (neural network weights, objective function weights);

[0086] Update the optimization cases and fault handling records in the knowledge graph database.

[0087] In one embodiment of the present invention: the intelligent execution system includes:

[0088] Adjustable air conditioner: It adopts a combination of variable frequency compressor and EC fan to achieve stepless adjustment of cooling capacity from 30% to 100% with a response time of <10s;

[0089] Dynamic airflow wall: Equipped with 0-90° rotatable guide vanes, with airflow organization optimized based on CFD simulation;

[0090] Phase change energy storage device: adopts lauric acid-palmitic acid eutectic material (phase change temperature 23±2℃, cold storage density 180kJ / kg), and implements economic dispatch in combination with peak and off-peak electricity prices.

[0091] In one embodiment of the present invention: the adaptive adjustment of the objective function weights satisfies:

[0092]

[0093] In the formula, w i (t) represents the real-time weight of the i-th target at time t, where i∈{1,2,3,4}, corresponding to energy consumption, temperature deviation, humidity deviation, and equipment lifespan loss, respectively. i 0 represents the preset baseline weight for the i-th objective, and E represents the system energy efficiency. α is the real-time energy efficiency change rate of the system, and α is the energy efficiency change sensitivity coefficient, which is a preset constant.

[0094] In one embodiment of the present invention: the hardware deployment satisfies:

[0095] The edge controller uses an NVIDIA Jetson AGX Orin compute unit, and is equipped with 32GB of LPDDR5 memory and 1TB of NVMe SSD storage;

[0096] The intelligent execution system includes a variable frequency drive (VFD) installed in the air conditioner, a guide vane array in the cabinet aisle, and a top-mounted phase change energy storage unit (200kg / m²). 2 ).

[0097] Example 1: Dynamic Energy Saving Control of High-Load Data Centers in Summer

[0098] A data center of an internet company (2000㎡, with 500 servers) entered its high-load summer operation phase in July 2024, with an average daily server load rate of 85%. The traditional air conditioning system suffered from frequent start-ups and shutdowns, resulting in a PUE value consistently above 1.45, and an average monthly equipment failure rate of 3 times. This embodiment applies the energy-saving and consumption-reducing data center management system of the present invention, and the specific implementation process is as follows:

[0099] 1. Edge-layer perception and execution: The multimodal sensor network collects data at 200ms cycles: The environmental perception unit uses a reconfigurable topology of 30% fixed nodes (deployed between rack rows) and 70% mobile robots (inspecting along tracks) to scan the three-dimensional space temperature and humidity (accuracy ±0.3℃ / ±2%RH), micro-airflow velocity (0-5m / s), and particle concentration in real time. The equipment status monitoring unit simultaneously collects the vibration spectrum entropy (S) of the server and air conditioning compressor. v Temperature deviation (D) t ) and current harmonic distortion (THD) i ), calculate device health index =1-0.4·S v -0.3·D t -0.3·THD i The system detected that the health status of two air conditioner compressors had dropped to 0.65 (the threshold is 0.7), triggering an alert.

[0100] 2. Edge-layer Real-time Decision Making: The intelligent edge controller receives edge-layer data via the LoRa network. After analysis by the real-time stream processing engine, it identifies that the temperature in a local area (rack A) reaches 26.5℃ (exceeding the upper limit of 25℃), and the compressor start-stop frequency has reached 4 times / hour (approaching the constraint value of 5 times / hour). Based on the lightweight AI inference module, it calls the NSGA-II algorithm to solve the multi-objective optimization problem (objective function: minimize total energy consumption, temperature deviation, humidity deviation, and equipment life loss), and generates an emergency control strategy: the dynamic air wall guide vanes are rotated to 60° (optimizing airflow organization and reducing hot spots in area A), the adjustable air conditioning cooling output is reduced from 70% to 55% (stepless adjustment response time <10s), and the cooling tower valve opening is increased by 15% to enhance heat dissipation.

[0101] 3. Cloud-based global optimization: The dynamic optimization platform receives edge layer data every 5 minutes and updates the data center physical-virtual mapping model (spatial resolution 0.1m) through a digital twin engine. 3Based on similar data center summer operation cases in the knowledge graph database, the weights of the objective function are adjusted (based on the energy efficiency change rate dE / dt, the energy consumption weight is increased from 0.4 to 0.5, and the equipment life loss weight is decreased from 0.2 to 0.1), and a global strategy is generated simultaneously: when the electricity price is low at night, the phase change energy storage device (lauric acid-palmitic acid eutectic material) starts the cold storage mode to store cold energy for use during the high load period of the next day, reducing the running time of the air conditioner.

[0102] Implementation results: After one month of system operation, the data center PUE value dropped to 1.23 (below the standard of 1.25), air conditioning energy consumption decreased by 42%, the compressor start-stop frequency stabilized at 3 times / hour, and the equipment failure rate dropped to once per month, verifying the system's multi-objective collaborative optimization capability under high load scenarios.

[0103] Example 2: Peak-valley energy storage scheduling for low-load data centers in winter

[0104] A financial institution's disaster recovery data center (1500㎡, deploying 300 servers) entered its low-load winter period in December 2024 (average daily load rate 40%). Traditional systems, due to static threshold control, resulted in redundant air conditioning operation at night (60% of energy consumption during peak electricity hours), and a localized area (rack B) experienced low temperatures (18℃) due to uneven airflow, affecting equipment stability. This embodiment applies the system of the present invention, and the specific implementation process is as follows:

[0105] 1. End-layer perception and execution: Data collected by the multimodal sensor network shows that the ambient temperature is 19.5℃ (close to the lower limit of 20℃), the humidity is 38%RH (close to the lower limit of 40%), and the equipment health is above 0.8 (no faults). The energy metering unit monitors that the electricity price is low from 23:00 to 7:00 the next day (0.3 yuan / kWh), and high from 10:00 to 15:00 during the day (1.2 yuan / kWh).

[0106] 2. Edge-layer real-time decision-making: The intelligent edge controller analysis found that the current temperature was slightly below the threshold but did not trigger emergency constraints (the number of device start-stop times was only 1 per hour). Therefore, the basic strategy of the adaptive decision-maker was executed: the dynamic wind wall guide vanes were rotated to 30° (to reduce the direct blowing of cold air in Zone B), the adjustable air conditioning cooling output was reduced to 30% (the minimum adjustment value), and the demand during the low-temperature period at night was predicted through the real-time stream processing engine.

[0107] 3. Global Cloud Optimization: The dynamic optimization platform simulates low-load winter scenarios based on a digital twin model (updated every 1 minute). Combining this with the "Peak-Valley Energy Storage Scheduling" case study from the knowledge graph, a global strategy is generated: During nighttime off-peak hours, the phase change energy storage device (phase change temperature 23±2℃) is activated, utilizing excess cooling capacity from air conditioning to cool the energy storage material to 21℃ (stored cooling capacity 180kJ / kg). During daytime peak hours, two redundant air conditioners are shut down, and the phase change energy storage device releases cooling capacity to maintain the ambient temperature (20-25℃). The objective function weights are adjusted (based on the energy efficiency change rate). The weight of equipment lifespan loss was increased from 0.2 to 0.3 to reduce the wear and tear on the compressor caused by frequent adjustments.

[0108] Implementation results: After one month of system operation, the energy consumption of the computer room air conditioning at night decreased to 25%, the overall PUE value was 1.21, the temperature in some areas was stable at 21-23℃, there were no low temperature failures of the equipment, and the phase change energy storage device saved about 800 yuan in electricity costs per day (calculated based on an average daily operation of 10 hours), which verified the system's economic scheduling capability under low load scenarios.

[0109] Example 3: Equipment Fault Early Warning and Adaptive Learning Optimization

[0110] In March 2025, a core data center of a certain operator (3000㎡, deploying 800 servers) experienced abnormal vibration of its air conditioning compressor. Traditional systems, due to single-parameter monitoring (temperature only), experienced a delay of over 2 hours in fault detection. This embodiment utilizes the system of the present invention, and the specific implementation process is as follows:

[0111] 1. Edge-layer perception and early warning: A multimodal sensor network continuously collected the vibration spectrum entropy S of an air conditioner compressor for 30 minutes. v From 0.12 to 0.25 (threshold 0.2), temperature deviation D t From 0.05 to 0.15 (threshold 0.1), current harmonic distortion (THD) increased. i The overall calculated equipment health level decreased from 0.85 to 0.62 (threshold 0.7) when the value increased from 0.08 to 0.12 (threshold 0.1), triggering a Level 1 fault warning (30 minutes in advance).

[0112] 2. Edge-layer emergency control: After receiving the warning, the intelligent edge controller immediately initiates local decision-making: by calling the NSGA-II algorithm through the adaptive decision-maker, an emergency control strategy is generated—switching the cooling output of the adjustable air conditioner corresponding to the faulty compressor from 80% to the standby compressor (90% cooling capacity), rotating the dynamic air wall guide vanes to 90° (guiding the cold air around the faulty area), and restricting the start and stop of the faulty compressor (reducing the frequency to 2 times / hour) to avoid further damage;

[0113] 3. Cloud-based learning and updates: After receiving edge-layer fault data, the dynamic optimization platform simulates the fault propagation path (spatial resolution 0.1m) using a digital twin engine. 3 The fault was confirmed to affect only a localized area (rack C). Historical cases of "compressor vibration-temperature-current coupling faults" were retrieved from the knowledge graph database for comparison, revealing the characteristics of this fault (S...). v ↑、D t ↑、THD i (↑) Highly similar to a data center case from 2024 (92% match), generating maintenance suggestions: Check compressor bearing wear and power supply line harmonic interference. Simultaneously, trigger an adaptive learning mechanism: evaluate the effectiveness of the strategy (temperature deviation in the fault area is controlled within ±0.5℃, without triggering a cascading failure), and dynamically adjust the equipment health model parameters (S...). v The weight was increased from 0.4 to 0.45 to enhance the sensitivity to vibration anomalies, and the fault handling record was updated to the "Fault Handling Case Library" in the knowledge graph.

[0114] Implementation results: Maintenance personnel completed troubleshooting (replacing the bearing and installing a harmonic filter) within 40 minutes based on the system's suggestions, and the fault did not spread to other equipment; in the following month, the system issued two early warnings for similar faults through the optimized health model (false alarm rate <5%), reducing the need for maintenance personnel by 30% and shortening the fault response time to within 1 hour, verifying the system's fault early warning and continuous evolution capabilities.

[0115] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Any variations and modifications can be made by those skilled in the art without departing from the spirit and scope of the invention. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention, without departing from the scope of the invention, fall within the protection scope defined by the claims of the present invention.

Claims

1. A high-efficiency data center management system that saves energy and reduces consumption, comprising a management system, characterized in that: The management system includes a three-tiered collaborative architecture: cloud, edge, and endpoint. End layer: Composed of multimodal sensor networks and intelligent execution systems; Edge layer: Intelligent edge controllers that connect to the end layer via LoRa / Industrial Ethernet; Cloud Layer: A dynamically optimized platform that connects to the edge layer via 5G / fiber optics; The multimodal sensing network includes an environmental sensing unit, an equipment status monitoring unit, and an energy consumption metering unit, which are used to collect environmental parameters, equipment status, and energy consumption data in real time. The intelligent edge controller includes a lightweight AI inference module, a real-time stream processing engine, and an adaptive decision-maker for performing local decisions and emergency control. The dynamic optimization platform includes a digital twin engine, a multi-objective optimizer, and a knowledge graph library for global policy optimization and model updates.

2. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: The multimodal sensing network satisfies: The environmental sensing unit includes a temperature and humidity composite sensor, a micro airflow velocity sensor, and a particle counter; The sensor adopts a reconfigurable topology deployment method with 30% fixed nodes and 70% mobile robots to achieve three-dimensional dynamic scanning in space; The equipment condition monitoring unit calculates equipment health based on a multi-source fusion of vibration, temperature, and current. The health model is as follows: health index =1-0.4·S v -0.3·D t -0.3·THD i ; In the formula, S v D is the entropy value of the vibration spectrum. t For temperature deviation, THD i denoted as the current harmonic distortion rate.

3. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: The adaptive decision-maker executes a multi-objective optimization algorithm: The objective function is to minimize: F(x) = [total energy consumption, absolute value of temperature deviation, absolute value of humidity deviation, and sum of equipment life loss factor]; The constraints include: temperature 20-25℃, humidity 40-60%RH, and equipment start-up and shutdown frequency ≤ 5 times / hour; The NSGA-II algorithm is used to solve for the Pareto optimal solution set, and the weights of the objective function can be adaptively adjusted.

4. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: In the aforementioned dynamic optimization platform: The digital twin engine constructs a real-time physical-virtual mapping model of the data center, updating globally every minute, with a spatial resolution of 0.1m. 3 ; The knowledge graph database stores the optimal operating point database for storage devices, a fault handling case library, and an energy-saving strategy effectiveness evaluation matrix.

5. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: The workflow of the management system includes: The sensor network collects data at a 200ms interval; When the edge controller immediately executes local control when the emergency constraint conditions are met, it optimizes the air conditioning air volume and the opening of the cooling tower valves. The dynamic optimization platform receives edge layer data, updates the model, generates global strategies, and distributes parameters every 5 minutes.

6. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: The dynamic optimization platform and the intelligent edge controller are configured with an adaptive learning mechanism to perform the following steps: Collect operational data; Evaluate the effectiveness of the strategy; Dynamically adjust model parameters; Update the optimization cases and fault handling records in the knowledge graph database.

7. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: The intelligent execution system includes: Adjustable air conditioner: It adopts a combination of variable frequency compressor and EC fan to achieve stepless adjustment of cooling capacity from 30% to 100% with a response time of <10s; Dynamic airflow wall: Equipped with 0-90° rotatable guide vanes, with airflow organization optimized based on CFD simulation; Phase change energy storage device: adopts lauric acid-palmitic acid eutectic material, and implements economic dispatch in combination with peak and off-peak electricity prices.

8. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 3, characterized in that: The adaptive adjustment of the objective function weights satisfies: In the formula, w i (t) represents the real-time weight of the i-th target at time t, where i∈{1,2,3,4}, corresponding to energy consumption, temperature deviation, humidity deviation, and equipment lifespan loss, respectively. i 0 represents the preset baseline weight for the i-th objective, and E represents the system energy efficiency. α is the real-time energy efficiency change rate of the system, and α is the energy efficiency change sensitivity coefficient, which is a preset constant.

9. The energy-saving and consumption-reducing high-efficiency computer room management system according to claim 1, characterized in that: The hardware deployment satisfies: The edge controller uses an NVIDIA Jetson AGX Orin compute unit, and is equipped with 32GB of LPDDR5 memory and 1TB of NVMe SSD storage; The intelligent execution system includes a variable frequency drive installed in the air conditioner, a guide vane array in the cabinet aisle, and a top-mounted phase change energy storage unit.

Images