Data center energy efficiency optimization method and system

Abstract

According to the data center energy efficiency optimization method and system provided by the invention, a comprehensive and accurate data basis is provided for formulating an intelligent scheduling strategy by collecting multi-modal data and carrying out fusion processing. And an intelligent scheduling strategy making and self-adaptive learning mechanism is adopted, so that continuous optimization of an energy efficiency management strategy and dynamic adjustment of system parameters can be realized. According to the method integrating multi-modal data, intelligent scheduling and adaptive learning, the energy efficiency management level of the data center can be effectively improved, the overall energy efficiency of the data center can be remarkably improved, and the operation cost is reduced.

Description

Technical Field

[0001] The present invention relates to data center environment management and control technology, and in particular to a data center energy efficiency optimization method and system. Background Art

[0002] With the rapid development of information technology, data centers, as the core infrastructure for information storage and processing, have become increasingly energy-intensive. Currently, data centers not only undertake a large number of computing and storage tasks, but are also one of the important sources of global energy consumption. Data shows that the proportion of energy consumption of data centers in the world's total energy consumption continues to rise. How to effectively improve the energy efficiency of data centers has become a key issue that needs to be solved urgently.

[0003] In the prior art, energy efficiency management methods for data centers usually rely on static resource allocation strategies. These strategies often fail to fully consider the dynamic changes in the operating environment and load of the data center during the design process, making it difficult to achieve optimal resource allocation and maximize energy efficiency in actual operation. For example, traditional static optimization methods cannot effectively cope with load fluctuations during peak and non-peak hours, and cooling systems usually operate based on preset temperature thresholds, which makes it difficult to flexibly adjust according to real-time changes in server load and ambient temperature, leading to energy waste and local overheating problems.

[0004] Although some energy efficiency optimization methods have attempted to dynamically adjust the data center by real-time monitoring of the operating status, most of these methods only focus on a single type of data (such as temperature or load data) and lack the comprehensive use of multiple data sources. In addition, existing systems often rely on preset parameters and rules, lack adaptive capabilities, and are difficult to achieve real-time adjustments when the environment and load change.

[0005] In addition, many existing optimization systems only focus on overall energy efficiency optimization, while ignoring the fine-grained optimization of each subsystem and component within the data center, resulting in low resource utilization. For example, the energy efficiency optimization of servers, storage devices, and network devices is often handled uniformly, while ignoring their respective energy consumption characteristics and optimization requirements. The problem of too coarse optimization granularity seriously affects the effect of energy efficiency optimization.

[0006] At the same time, existing technologies also face many challenges in processing and integrating data from different sources and types, making it difficult to achieve global optimal energy efficiency optimization. It is difficult to effectively integrate various types of data such as temperature sensors, humidity sensors, and power metering equipment, making it difficult for the system to comprehensively and accurately evaluate the energy efficiency status and optimization needs of the data center, thus affecting the overall optimization effect.

[0007] Therefore, there is an urgent need to develop a new method to dynamically adapt to changes in the operating environment and load of the data center, comprehensively improve the efficiency of resource allocation and optimize energy consumption management. Existing static strategies cannot cope with fluctuations in actual operation, resulting in resource waste and reduced energy efficiency, while traditional optimization methods are mostly limited to a single data source and lack the comprehensive utilization of multi-dimensional data. In addition, the lack of fine-grained optimization means that the energy efficiency improvement potential of each subsystem has not been fully explored. Therefore, the new method should have real-time monitoring and adaptive adjustment capabilities, and be able to integrate multiple data types for global optimization to comprehensively improve the energy efficiency performance of the data center. Summary of the invention

[0008] The purpose of the present invention is to provide a data center energy efficiency optimization method and system, aiming to solve the problem that the data center energy efficiency management method in the prior art is single and lacks self-adaptation capability.

[0009] To achieve the above object, the present invention adopts the following technical solution:

[0010] The present invention provides a data center energy efficiency optimization method, comprising the following steps: data collection and transmission, data fusion processing, intelligent scheduling strategy formulation and adaptive learning.

[0011] In the data collection and transmission step, multimodal data of the data center is collected, wherein the multimodal data includes temperature data, humidity data, power consumption data and server load data.

[0012] Furthermore, in the data fusion processing step, the multimodal data is fused. Specifically, the fusion processing includes: preprocessing the multimodal data; reducing the dimension of the preprocessed data using the principal component analysis method; and performing weighted average fusion on the reduced-dimensional data.

[0013] In the intelligent scheduling strategy formulation step, an intelligent scheduling strategy is generated based on the fused data. Furthermore, the step of generating an intelligent scheduling strategy includes: constructing an objective function including energy consumption minimization, performance maximization and temperature balance; optimizing the objective function using a genetic algorithm to generate a resource allocation strategy. The objective function is: f(x) = w1·E(x)+w2·P(x)+w3·T(x), where x is the resource allocation strategy, E(x) is the energy consumption function, P(x) is the performance function, T(x) is the temperature balance function, and w1, w2, and w3 are weight coefficients.

[0014] In the adaptive learning step, the intelligent scheduling strategy is executed and adaptively adjusted. Specifically, the adaptive adjustment includes: based on historical data and real-time data, using a machine learning algorithm to continuously optimize the energy efficiency management strategy; dynamically adjusting system parameters according to the optimized strategy. The machine learning algorithm is a long short-term memory network or a reinforcement learning algorithm.

[0015] Furthermore, the present invention also includes the step of real-time monitoring of the execution results, and the real-time monitoring includes: building a status monitoring dashboard to display key indicators; setting warning thresholds to trigger an alarm when an abnormal situation is detected. In addition, it also includes the step of dynamically optimizing the intelligent scheduling strategy according to the monitoring results.

[0016] The present invention also provides a data center energy efficiency optimization system, including: a data acquisition module for collecting multimodal data of a data center; a data fusion module for fusion processing of the multimodal data; a strategy generation module for generating an intelligent scheduling strategy based on the fusion processed data; and an execution control module for executing the intelligent scheduling strategy and performing adaptive adjustment. The modules communicate through standard interfaces to achieve modular design.

[0017] Compared with the prior art, the present invention has the following beneficial effects: it provides a data center energy efficiency optimization method and system, which provides a comprehensive and accurate data basis for the formulation of intelligent scheduling strategies by collecting multimodal data and performing fusion processing. By adopting intelligent scheduling strategy formulation and adaptive learning mechanism, continuous optimization of energy efficiency management strategies and dynamic adjustment of system parameters can be achieved. This method of integrating multimodal data, intelligent scheduling and adaptive learning can effectively improve the energy efficiency management level of data centers, and can significantly improve the overall energy efficiency of data centers and reduce operating costs.

[0018] In addition, the present invention can generate a resource allocation strategy that takes into account energy minimization, performance maximization and temperature balance by combining the objective function optimized by genetic algorithm. The real-time monitoring and early warning mechanism can detect and handle abnormal situations in a timely manner, further ensuring the stable operation of the data center. The use of long short-term memory network or reinforcement learning algorithm for adaptive adjustment can improve the system's ability to adapt to environmental changes. The modular system architecture can improve the scalability and maintainability of the system. The comprehensive application of these innovative technologies provides strong support for the development of green data centers. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The present invention is a flow chart of the data center energy efficiency optimization method.

[0020] Figure 2 A more detailed flow chart of the data center energy efficiency optimization method of the present invention. DETAILED DESCRIPTION

[0021] The embodiments of the present invention will be described in more detail below in conjunction with the schematic diagram, wherein the preferred embodiments of the present invention are shown, and it should be understood that those skilled in the art can modify the present invention described herein, and still achieve the beneficial effects of the present invention. Therefore, the following description should be understood as being widely known to those skilled in the art, and not as a limitation of the present invention.

[0022] The present invention is described in more detail in the following paragraphs by way of example with reference to the accompanying drawings. The advantages and features of the present invention will become more apparent from the following description and claims. It should be noted that the drawings are in very simplified form and in non-precise proportions, and are only used to facilitate and clearly assist in illustrating the purpose of the embodiments of the present invention.

[0023] Example 1

[0024] like Figure 1 As shown, this embodiment proposes a data center energy efficiency optimization method, including the following steps, specifically:

[0025] S100: Collect multimodal data from a data center.

[0026] High-precision sensors are deployed at key locations in the data center, including temperature sensors, humidity sensors, power monitoring equipment, and server load monitoring equipment. The sensors collect multimodal data of the data center in real time, including at least one of temperature data, humidity data, power consumption data, and server load data.

[0027] Specifically, it includes:

[0028] a) Deploy high-precision temperature sensors in server racks, cooling equipment, power distribution units, air flow paths, etc.

[0029] b) Install power monitoring equipment near servers and cooling equipment to monitor power consumption in real time.

[0030] Sensors and humidity sensors monitor changes in ambient temperature and humidity in real time.

[0031] c) Install server load monitoring equipment at key locations to monitor the server's operating status and load conditions in real time.

[0032] d) Place air flow sensors in the data center to monitor air flow and optimize cooling strategies.

[0033] Preferably, the sampling frequency of the temperature sensor and humidity sensor is once every 10 seconds, with an accuracy of ±0.1°C and ±1%RH respectively. The sampling frequency of the power monitoring device is once every 1 second, with an accuracy of ±0.5%. The sampling frequency of the server load monitoring device is once every 5 seconds.

[0034] Data transmission: Use an encrypted secure network to transmit the collected data to the central data processing system to ensure the security and integrity of the data; use the highly reliable transmission protocol MQTT to ensure the real-time and accuracy of data transmission.

[0035] Data storage and management: Data transmitted to the central data processing system is stored in a distributed database to ensure high availability and scalability of the data; data is uniformly managed and scheduled to provide basic support for subsequent data fusion and analysis; the data collection and transmission process ensures comprehensive, real-time and accurate monitoring of the data center environment and load data, providing a solid foundation for intelligent scheduling and energy efficiency optimization.

[0036] S200: performing fusion processing on the multimodal data.

[0037] Multimodal data fusion is one of the core steps of this embodiment. By cleaning, synchronizing and fusing different types of data, comprehensive status indicators are generated to provide support for subsequent intelligent scheduling and optimization. The specific implementation steps are as follows:

[0038] Data cleaning:

[0039] a) De-noising: Use the Kalman filter algorithm to denoise the collected raw data, eliminate sensor noise and outliers, and ensure data accuracy.

[0040] b) Missing value filling: Linear interpolation or multiple interpolation methods are used to fill in missing values ​​in the data to ensure data integrity.

[0041] c) Data standardization: standardize temperature, humidity, power consumption and load data.

[0042] Unify data dimensions to facilitate subsequent fusion processing.

[0043] Data Synchronization:

[0044] Timestamp alignment: align the timestamps of data from different sensors to ensure that all data have a unified time reference; data segmentation: segment the data according to time windows (such as 1 minute or 5 minutes) to ensure that the data in the same time window can be synchronously fused and processed.

[0045] Data fusion algorithm:

[0046] a) Principal Component Analysis (PCA): PCA algorithm is used to perform dimensionality reduction processing on the standardized multimodal data, extract the main components, reduce the data dimension, and improve the computational efficiency.

[0047] Specific implementation method: First, construct a data matrix, use the data of each sensor as the column of the matrix, then centralize the data matrix, calculate the covariance matrix, solve the eigenvalues ​​and eigenvectors, and select the first k main components.

[0048] b) Weighted average fusion: Different weights are assigned to the data of each sensor according to its importance, and data fusion is performed through the weighted average method.

[0049] Specific implementation method: Set the weight coefficient of each sensor data, perform weighted average calculation on each data according to the weight, and obtain the comprehensive status index.

[0050] The fusion algorithm formula is as follows:

[0051]

[0052] In the formula, w i is the weight of the i-th sensor data, x i is the standardized data of the ith sensor.

[0053] Comprehensive status indicator generation:

[0054] Based on the fused data, comprehensive status indicators such as temperature index, humidity index, energy consumption index and load index are generated, providing key references for intelligent scheduling and optimization.

[0055] Specifically, in a typical data center, multimodal data such as temperature, humidity, power consumption, and server load are fused to generate comprehensive status indicators. For example:

[0056] a) Temperature index: Based on the data from multiple temperature sensors, a unified temperature index is generated to reflect the overall temperature conditions of the data center.

[0057] b) Humidity Index: The data from multiple humidity sensors are integrated to generate a unified humidity index that reflects the humidity level in the data center.

[0058] c) Energy consumption index: Based on the data from the power monitoring equipment, an energy consumption index is generated to reflect the energy consumption of the data center.

[0059] d) Load index: The data from the server load monitoring device is integrated to generate a load index that reflects the load status of the server.

[0060] The above comprehensive status indicators can help the system make more accurate and comprehensive decisions during intelligent scheduling and energy efficiency optimization.

[0061] Through multi-modal data fusion, the system can fully and accurately grasp the environment and load conditions of the data center, providing a solid data foundation for subsequent intelligent scheduling and optimization. Combined with specific algorithms and data processing methods, the reliability and effectiveness of the fused data are ensured.

[0062] This step also includes advanced data processing and analysis: by deeply processing and analyzing the fused multimodal data, key information is extracted to provide data support for intelligent scheduling and optimization.

[0063] Data processing platform: Data storage and management: Use Hadoop distributed database to store and manage the integrated data to ensure high availability and scalability of the data; Data preprocessing: Preprocess the stored data, including data cleaning, missing value filling and data standardization to ensure data consistency and integrity.

[0064] Time Series Analysis:

[0065] a) Trend analysis: Use the autoregressive moving average model (ARIMA) time series analysis technique to identify long-term trends and seasonal changes in temperature, humidity, energy consumption, and load data.

[0066] b) Anomaly detection: Based on time series analysis, detect anomalies in the data and identify possible faults and abnormal situations.

[0067] c) Prediction model: Build a prediction model, long short-term memory network (LSTM), to predict the changes in data center environment and load in the future.

[0068] Among them, the ARIMA model formula is:

[0069] X t =c+φ1X t-1 +φ2X t-2 +…+φ p X t-p +ε t +θ1ε t-1 +θ2ε t-2 +…+θ q ε t-q

[0070] Where: X t represents the value at time t (e.g., temperature, humidity, energy consumption, or load), c is a constant term, representing the baseline value of the time series, and φ i is the autoregressive coefficient, which indicates the value of the previous moment to the current value X t The degree of influence, X t―i is the value of the time series at the previous time ti, where 1≤i≤q, ε tis the error term or random disturbance at time t, θ j is the moving average coefficient, which indicates the error term of the previous moment to the current value X t The degree of influence, ε t-1 ε t-n is the error term of the time series at the previous tn moments, where 1≤n≤q.

[0071] Machine Learning Model Training:

[0072] a) Feature Engineering: Extract features from the fused data, including key indicators such as temperature, humidity, energy consumption, and load, and construct a training data set.

[0073] b) Model selection and training: Use the LSTM prediction model and train the model based on the extracted features.

[0074] c) Model evaluation and optimization: Evaluate model performance, adjust model parameters, and optimize model effects through methods such as cross-validation and grid search.

[0075] Data Visualization:

[0076] a) Status monitoring dashboard: Using the data visualization tool Grafana, we build a data center status monitoring dashboard to display key indicators such as temperature, humidity, energy consumption, and load in real time.

[0077] b) Trend and anomaly display: Visualize the results of time series analysis to display data trends and anomalies, helping operation and maintenance personnel to quickly identify and handle problems.

[0078] c) Prediction result display: Visualize the prediction results of the machine learning model to show the data center environment and load changes in the future to assist decision-making.

[0079] In one example, advanced data processing and analysis can effectively identify and predict the changing trends of the environment and load, detect abnormal situations, and provide reliable prediction results. For example:

[0080] a) Trend analysis: Through the ARIMA model, the long-term trend of temperature and humidity in the data center is identified, and the seasonal variation pattern is discovered.

[0081] b) Anomaly detection: Using time series analysis technology, abnormal energy consumption of a server is detected, prompting the operation and maintenance personnel to check for possible faults.

[0082] c) Prediction model: Build an LSTM model to predict the load changes of the data center in the next week, helping operation and maintenance personnel to schedule and optimize resources in advance.

[0083] The above analysis results can help data center managers make more accurate and efficient decisions, optimize resource allocation, and improve overall energy efficiency.

[0084] S300, based on the fused data, generate an intelligent scheduling strategy. The intelligent scheduling strategy is formulated by using genetic algorithms, combining the results of multimodal data fusion and advanced data processing and analysis, to generate the optimal resource allocation strategy and achieve energy efficiency optimization of the data center. Specifically:

[0085] Scheduling algorithm selection: Genetic Algorithm (GA) is used to find the optimal solution for resource allocation by simulating the natural selection process and using selection, crossover and mutation operations.

[0086] Objective function design:

[0087] a) Minimizing energy consumption: Design an objective function to minimize the system energy consumption, including the energy consumption of the cooling system and the energy consumption of the server operation.

[0088] b) Maximizing performance: Maximizing server performance and task completion efficiency while minimizing energy consumption.

[0089] c) Temperature balance: Ensure that the temperature inside the data center is evenly distributed to avoid local overheating.

[0090] Objective function:

[0091] f(x)=w1·E(x)+w2·P(x)+w3·T(x)

[0092] Where x is the resource allocation strategy, E(x) is the energy consumption function, P(x) is the performance function, T(x) is the temperature balance function, and w1 w2 w3 are weight coefficients, which represent the importance of energy consumption, performance and temperature balance, respectively.

[0093] Resource allocation strategy generation:

[0094] a) Initialization: Initialize the parameters and population of the genetic algorithm based on the current status and historical data of the data center.

[0095] b) Iterative optimization: In each iteration, the quality of the current solution is evaluated according to the objective function, and the resource allocation strategy is continuously optimized through operations such as selection, crossover, and mutation.

[0096] c) Generate optimal strategy: After multiple iterations, generate the optimal resource allocation strategy that meets the energy consumption, performance and temperature balance requirements.

[0097] In a specific example, the following goals can be achieved through the above genetic algorithm:

[0098] a) Minimize energy consumption: By optimizing the operating parameters of the cooling system and servers, the overall energy consumption of the data center can be significantly reduced.

[0099] b) Maximizing performance: While minimizing energy consumption, maximize server performance and task completion efficiency, and improve the computing power of the data center.

[0100] c) Temperature balance: optimize temperature distribution, avoid local overheating, and improve cooling efficiency.

[0101] By achieving the above goals, the intelligent scheduling strategy formulation process can effectively combine the results of multimodal data fusion and advanced data processing and analysis, and use genetic algorithms to generate the optimal resource allocation strategy to ensure the energy efficiency optimization of the data center. This implementation solution ensures that the data center can achieve optimal efficiency under various operating conditions by optimizing energy consumption, performance and temperature balance.

[0102] S400: The data center executes the intelligent scheduling strategy and performs adaptive adjustment. Specifically:

[0103] Data collection and preprocessing: Collect energy consumption data E of each server in the data center i (t), CPU usage U cpu,i (t), memory usage U mem,i (t), and network bandwidth usage U net,i (t). Data preprocessing: Clean, normalize and fill missing values ​​of the collected data to ensure data consistency and completeness.

[0104] Feature extraction and pattern recognition: Extract key features from preprocessed data, such as CPU usage, memory usage, and network bandwidth usage, and apply cluster analysis and other methods to identify energy consumption patterns and determine energy consumption peaks and valleys.

[0105] Build an energy consumption prediction model: Select the applicable long short-term memory network LSTM machine learning model, then use historical data to train the model to predict energy efficiency performance in the future. The applicable formula of the model is:

[0106] h t =σ(W h ·[h t-1 ,x t ]+b h )

[0107]

[0108] y t =σ(W y ·h t +b y )

[0109] Where: h t is the hidden state at time t, representing the output of the memory at the current moment, x t is the input data at time t, C t is the state of the memory unit at time t, f t is the output of the forget gate, which controls the degree of retention of the memory of the previous moment at the current moment. t is the output of the input gate, which controls the influence of the current input data on the state of the memory unit. is the candidate memory cell state, W c , W h , W y is the weight matrix used to connect the input, hidden state and memory unit state, b c 、b h 、b y is the bias term.

[0110] Adaptive learning mechanism: Based on real-time and historical data, continuously update and optimize energy efficiency management strategies. Use reinforcement learning algorithms to dynamically adjust system parameters to ensure optimal energy efficiency under different operating conditions.

[0111] System integration and performance monitoring: The integrated management platform realizes unified management and real-time monitoring of all monitoring and control devices. Real-time monitoring of the system's operating status ensures that the system can achieve optimal energy efficiency under different operating conditions.

[0112] In another specific example, the following goals can be achieved through the adaptive learning mechanism:

[0113] a) Energy efficiency optimization: By collecting and analyzing the server's energy consumption data, CPU usage, memory usage, and network bandwidth usage, the adaptive learning mechanism can identify high energy consumption patterns and optimize based on historical and real-time data. The LSTM model is used to predict the data center resource demand in the future to ensure dynamic adjustment of resource allocation strategies when the load changes.

[0114] b) Performance monitoring and anomaly detection: Real-time monitoring of the system's operating status, using the LSTM model and Q-learning algorithm to detect anomalies such as server overload, memory leak, etc. Through the reinforcement learning algorithm, the system parameters are dynamically adjusted to ensure optimal energy efficiency under different operating conditions.

[0115] c) Adaptive adjustment strategy: The adaptive learning mechanism can continuously adjust the energy efficiency management strategy according to the changes in real-time data and historical data to improve the adaptability and flexibility of the system.

[0116] The learning algorithm optimizes resource allocation strategies to ensure that data centers can achieve optimal energy efficiency under various load and environmental conditions.

[0117] Through the above steps, the adaptive learning mechanism can effectively combine the results of multimodal data fusion and advanced data processing and analysis, and use machine learning and reinforcement learning algorithms to continuously optimize the energy efficiency management strategy of the data center to ensure that the system can achieve optimal energy efficiency under different operating conditions.

[0118] This step also includes: automated resource scheduling and control, through intelligent algorithms and real-time monitoring, to achieve dynamic allocation and optimization of data center resources. The specific implementation steps are as follows:

[0119] a) Resource monitoring and data collection: Deploy monitoring equipment at key nodes in the data center to collect key data such as CPU usage, memory usage, network bandwidth usage, etc. in real time. The data is transmitted to the central control system through a secure network for processing and storage.

[0120] b) Resource demand forecasting: Based on historical data and real-time data, resource demand forecasting is performed using time series analysis and machine learning models (such as LSTM).

[0121] c) Optimization algorithm and scheduling strategy: Use optimization algorithms (such as dynamic resource allocation algorithm, DRCA)

[0122] Dynamically adjust resource allocation strategies based on resource demand forecast results.

[0123] The dynamic resource allocation algorithm formula is: R opt =argmin R CEOF(R;θ) where R opt is the optimal resource allocation, Θ is the parameter set, and CEOF is the comprehensive energy efficiency optimization function.

[0124] The formula of comprehensive energy efficiency optimization function is: CEOF=λ1·ERM+λ2·AVPI+λ3·IRI+λ4·RFI, where λ1, λ2, λ3, and λ4 are weight coefficients, ERM is the energy efficiency-resource matching function, AVPI is the audio and video processing intensity index, IRI is the interactive response index, and RFI is the resource flexibility index.

[0125] d) Execution and feedback adjustment: Execute the optimized resource allocation strategy in real time, and automatically adjust the resource allocation of each device through the control system. According to the real-time monitoring data, evaluate the adjustment effect, perform feedback optimization, and ensure the accuracy and efficiency of resource allocation.

[0126] In another embodiment, the following objectives can be achieved through the above-mentioned automated resource scheduling and control steps:

[0127] a) Maximizing resource utilization: By dynamically adjusting resource allocation strategies, efficient resource utilization can be ensured under different time periods and load conditions.

[0128] b) Minimize energy consumption: Optimize the operating parameters of the cooling system and servers to significantly reduce the overall energy consumption of the data center.

[0129] c) Performance optimization: Maximize server performance and task completion efficiency while minimizing energy consumption, and improve the computing power of the data center.

[0130] d) Temperature balance: optimize temperature distribution, avoid local overheating, and improve cooling efficiency.

[0131] In this example, the data center energy efficiency optimization method further includes the following steps:

[0132] Real-time monitoring of execution results: System integration and performance monitoring realize unified management and real-time monitoring of all monitoring and control devices through the integrated management platform to ensure efficient operation of the system and timely warning. The specific implementation steps are as follows:

[0133] Integrated management platform: All monitoring and control equipment (including sensors, servers, cooling systems and power distribution units) are integrated into a management platform through standardized interfaces. All monitoring data is aggregated through the integrated management platform for centralized processing and storage. The management platform is designed to be modular and scalable, and can adapt to the expansion of the data center and equipment updates.

[0134] Real-time data visualization: Use the Grafana data visualization tool to build a status monitoring dashboard to display key indicators such as temperature, humidity, energy consumption, and load in real time. Analyze and display data in real time to identify and display the changing trends of the environment and load. Display the results of the prediction model and provide a prediction of the data center environment and load changes in the future.

[0135] System status monitoring and early warning: Through the integrated management platform, the system's operating status is monitored in real time to ensure that all equipment is operating normally. Thresholds for key indicators are set, and when abnormal conditions (such as overtemperature, power overload) are detected, alarms are automatically triggered and logs are recorded. Fault diagnosis and troubleshooting functions are provided to help operation and maintenance personnel discover and solve problems in a timely manner to ensure stable system operation.

[0136] In a specific example, the following goals can be achieved through system integration and performance monitoring in this step:

[0137] a) Unified management: The integrated management platform realizes unified management of all monitoring and control devices, reduces management complexity and improves management efficiency. Operation and maintenance personnel can view all key indicators of the data center in real time through one interface, including temperature, humidity, energy consumption and server load.

[0138] b) Real-time monitoring: By monitoring the system's operating status in real time, maintenance personnel can promptly detect and handle abnormal situations. For example, when the temperature of a server exceeds the preset threshold, the system will automatically trigger an alarm, and maintenance personnel can take immediate measures to prevent equipment from overheating and causing failures.

[0140] c) Data visualization: The status monitoring dashboard built with data visualization tools can intuitively display the operating status and historical trends of the data center, helping operation and maintenance personnel to better understand and analyze the performance of the system.

[0141] d) Early warning and troubleshooting: The system has set up a variety of early warning mechanisms. When key indicators reach or exceed the threshold, the system will automatically record logs and trigger alarms. Operation and maintenance personnel can quickly locate problems based on early warning information and troubleshoot through fault diagnosis functions to ensure stable operation of the system.

[0142] Through the above steps, system integration and performance monitoring can effectively improve the management efficiency and operation stability of the data center, ensuring optimal energy efficiency under various operating conditions. This implementation solution ensures efficient operation and timely troubleshooting of the data center through unified management, real-time monitoring and early warning mechanisms, significantly improving the overall operational efficiency of the data center.

[0143] In order to more clearly show the detailed steps of the present invention, as shown in Figure 2 , which is a more detailed flow chart of the data center energy efficiency optimization method of the present invention.

[0144] In summary, the core method of the present invention is: by collecting multimodal data of the data center, fusing the multimodal data, and generating an intelligent scheduling strategy based on the fused data, and finally executing the intelligent scheduling strategy and making adaptive adjustments. The system significantly improves the resource utilization efficiency and overall energy efficiency of the data center by calling the constantly changing optimal comprehensive strategy, reduces operating costs, and promotes the development of green data centers.

[0145] Example 2

[0146] The present invention also discloses a data center energy efficiency optimization system, comprising:

[0147] Data acquisition module, used to collect multimodal data from the data center;

[0148] A data fusion module, used for fusing the multimodal data;

[0149] A strategy generation module is used to generate intelligent scheduling strategies based on the fused data;

[0150] An execution control module, used to execute the intelligent scheduling strategy and perform adaptive adjustments;

[0151] Among them, each module communicates through a standard interface to achieve modular design.

[0152] Through modular design, a scalability strategy can be implemented: add or reduce modules according to demand. For example, as the scale of the data center expands, more data acquisition and processing modules can be added.

[0153] Furthermore, the modules of this embodiment adopt distributed computing and storage technology to support the horizontal expansion of the system, ensuring high performance and high availability when processing large amounts of data. In addition, this embodiment supports dynamic adjustment of system parameters and configurations to adapt to different operating conditions and business requirements by flexibly configuring the management system. Furthermore, the configuration management interface of this embodiment is visually set, and operation and maintenance personnel can easily adjust the configuration and optimize the system.

[0154] The modular design of this embodiment can achieve the following goals:

[0155] The modular design makes each function of the system independent and easy to maintain and upgrade. For example, when a new monitoring function needs to be added, only a new monitoring module needs to be added without affecting the normal operation of other modules. The scalable system architecture supports the flexible expansion of the data center scale. For example, as the business volume increases, more data processing and storage nodes can be seamlessly added to ensure high performance and high availability of the system. The flexible configuration management system supports dynamic adjustment of system parameters to adapt to different operating conditions and business needs.

[0156] In summary, the present invention provides a method and system for optimizing energy efficiency in a data center. By collecting multimodal data (such as temperature, humidity, power consumption, and server load) and performing fusion processing, a comprehensive and accurate data basis is provided for the formulation of intelligent scheduling strategies. By adopting intelligent scheduling strategy formulation and adaptive learning mechanisms, continuous optimization of energy efficiency management strategies and dynamic adjustment of system parameters can be achieved. Resource allocation strategies are dynamically optimized based on energy consumption, performance, and environmental factors. This method of integrating multimodal data, intelligent scheduling, and adaptive learning continuously adjusts and optimizes energy efficiency strategies based on real-time data and historical data to ensure optimal energy efficiency under various operating conditions. The system has real-time monitoring and adjustment capabilities, and can respond quickly to changes in load and environment. It can effectively improve the energy efficiency management level of the data center, and can significantly improve the overall energy efficiency of the data center and reduce operating costs.

[0157] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A data center energy efficiency optimization method, characterized in that: The following steps are involved: Collect multimodal data from data centers; Performing fusion processing on the multimodal data; Generate intelligent scheduling strategies based on the fused data; and The data center executes the intelligent scheduling strategy and performs adaptive adjustments.

2. The data center energy efficiency optimization method according to claim 1, characterized in that: The multimodal data includes at least one of temperature data, humidity data, power consumption data, and server load data.

3. The data center energy efficiency optimization method according to claim 1, characterized in that: The fusion process includes: Preprocess multimodal data; Use principal component analysis to reduce the dimensionality of the preprocessed data; and Perform weighted average fusion on the reduced-dimensional data.

4. The data center energy efficiency optimization method according to claim 1, characterized in that: The steps of generating an intelligent scheduling strategy include: constructing an objective function including energy consumption minimization, performance maximization and temperature balance; optimizing the objective function by using a genetic algorithm to generate a resource allocation strategy.

5. The data center energy efficiency optimization method according to claim 4, characterized in that: The objective function is: f(x)=w1·E(x)+w2·P(x)+w3·T(x), wherein x is the resource allocation strategy, E(x) is the energy consumption function, P(x) is the performance function, T(x) is the temperature balance function, and w1, w2, and w3 are weight coefficients.

6. The data center energy efficiency optimization method according to claim 1, characterized in that: The adaptive adjustment includes: continuously optimizing the energy efficiency management strategy using a machine learning algorithm based on historical data and real-time data; and dynamically adjusting system parameters according to the optimized strategy.

7. The data center energy efficiency optimization method according to claim 6, characterized in that: The machine learning algorithm is a long short-term memory network or a reinforcement learning algorithm.

8. The data center energy efficiency optimization method according to claim 1, characterized in that: It also includes real-time monitoring of execution results, which includes: building a status monitoring dashboard to display key indicators; setting early warning thresholds to trigger alarms when abnormal conditions are detected.

9. The data center energy efficiency optimization method according to claim 8, characterized in that: It also includes dynamic optimization of intelligent scheduling strategies based on monitoring results.

10. A data center energy efficiency optimization system, characterized in that: include: Data acquisition module, used to collect multimodal data from the data center; A data fusion module, used for fusing the multimodal data; A strategy generation module is used to generate intelligent scheduling strategies based on the fused data; An execution control module, used to execute the intelligent scheduling strategy and perform adaptive adjustments; Among them, each module communicates through a standard interface.

Images