Green intelligent computing center energy efficiency optimization method and system
By constructing an energy consumption pattern recognition and anomaly detection model through unsupervised learning methods and combining it with multi-energy collaborative scheduling, the problem of inefficient energy scheduling in computing centers is solved. This enables adaptive detection and early warning of sudden energy consumption fluctuations and abnormal operating states, thereby improving energy efficiency and system stability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-30
- Publication Date
- 2026-04-10
AI Technical Summary
Traditional power load management methods are ill-equipped to handle the complex load characteristics of computing centers, resulting in inefficient energy dispatch, impacting grid stability and renewable energy absorption capacity, and lacking in-depth research on the grid operation characteristics after the computing center is connected.
An unsupervised learning method is used to construct an energy consumption pattern recognition and anomaly detection model. Combined with multi-energy collaborative scheduling, an intelligent decision-making model and an automated optimization method for the energy management system are designed. Through data analysis and model optimization, adaptive detection and early warning of sudden energy consumption fluctuations and abnormal operating states are achieved, and energy scheduling strategies are dynamically adjusted.
It improves the energy efficiency of computing centers under different load conditions, reduces dependence on traditional fossil fuels, enhances the intelligence and greenness of energy management, and achieves efficient utilization of renewable energy and stable system operation.
Smart Images
Figure CN121836232A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of energy management, in particular to a green intelligent computing center energy efficiency optimization method and system. BACKGROUND
[0002] Under the background of global energy structure transformation and response to climate change, green computing center as an important part of promoting digital economy and low-carbon development has become the core content of the energy policy and strategic layout of various countries. Green computing center is not only the infrastructure supporting the development of emerging technologies such as big data, artificial intelligence and cloud computing, but also the key platform to promote green and low-carbon transformation and achieve sustainable development.
[0003] However, the traditional power load management method mainly targets relatively stable load. For computing center, a new type of power consumption facility with complex load characteristics, it is difficult to achieve efficient energy scheduling and management, resulting in that simple time-based time-of-use pricing strategy cannot fully mobilize the enthusiasm of computing center to increase computing tasks during off-peak periods of power grid, and cannot effectively respond to the rapid fluctuations of computing center load, so that the traditional management method cannot accurately grasp its inherent law. In addition, the large-scale access of computing center will affect the load distribution, voltage stability and power quality of power grid, so that the access of computing center changes the load characteristics and operation mode of power grid and affects the accommodation capacity of new energy. However, due to the lack of in-depth study on the operation characteristics of power grid after the access of computing center, it hinders the efficient energy scheduling and management of computing center. SUMMARY
[0004] The present application proposes a green intelligent computing center energy efficiency optimization method and system, which has the advantages of realizing adaptive detection and early warning of sudden energy consumption fluctuations and abnormal operating state, reducing dependence on traditional fossil energy, and improving energy use efficiency of computing center under different load conditions, solving the technical problems proposed in the above background technology.
[0005] In order to achieve the above purpose, the present application adopts the following technical solution: a green intelligent computing center energy efficiency optimization system, the system includes data analysis, model construction, algorithm optimization, system integration;
[0006] The data analysis includes energy consumption mode analysis and data mining, energy consumption mode analysis and abnormal energy consumption mode mining;
[0007] The model construction includes the construction of unsupervised learning model, the research of multi-energy collaborative scheduling method, the evaluation of energy system stability and optimization benefit;
[0008] The algorithm optimization includes the construction of intelligent decision model, the research of energy management system automation optimization method, the research of energy efficiency evaluation and optimization feedback mechanism;
[0009] The system integration includes the design of the energy optimization system integration architecture, technical application cases and field verification, and research on technology promotion paths and industry influence.
[0010] Preferably, the energy consumption pattern analysis and data mining specifically refer to: analyzing the relationship between computing load fluctuations and energy demand, collecting and processing historical operating data of the green computing center, using unsupervised learning models to identify energy consumption characteristics under different operating conditions, mining potential regularities and abnormal patterns, and providing data support for subsequent energy optimization strategies;
[0011] The energy consumption pattern analysis and abnormal energy consumption pattern mining specifically refer to: using clustering algorithms and anomaly detection algorithms in unsupervised learning to analyze time series data of energy consumption, identify common energy consumption patterns and sudden fluctuations in energy demand, and study how to adaptively discover abnormal situations in system operation through unsupervised learning models, provide early warnings of potential energy efficiency problems, and adjust energy dispatch strategies in a timely manner.
[0012] Preferably, the clustering algorithm includes K-means, hierarchical clustering, and Gaussian mixture model. The clustering algorithm discovers the energy consumption patterns of different groups by dividing the dataset.
[0013] Preferably, the anomaly detection algorithm includes isolated forest and density peak clustering, and the anomaly detection algorithm is used to detect abnormal energy consumption patterns.
[0014] Preferably, the construction of the unsupervised learning model specifically refers to: researching and constructing an energy scheduling optimization model based on unsupervised learning, and using clustering algorithms, dimensionality reduction techniques, and anomaly detection algorithms to analyze the energy demand data of the green computing center;
[0015] The research on the multi-energy collaborative scheduling method specifically refers to: studying intelligent collaborative scheduling methods for several energy resources in green computing centers, and proposing an application framework for unsupervised learning models in multi-energy environments;
[0016] The evaluation of energy system stability and optimization benefits specifically refers to: studying the optimization effect and system stability evaluation method of unsupervised learning models in energy dispatch, and evaluating the improvement effect of the models on energy dispatch efficiency, system stability and renewable energy utilization under different load and energy configuration conditions through model verification and evaluation.
[0017] Preferably, the dimensionality reduction techniques include principal component analysis and t-SNE, which reduce the data dimensionality while retaining the main features, and are used for energy consumption pattern analysis.
[0018] Preferably, the energy resources include wind power, solar power, and conventional electricity.
[0019] Preferably, the construction of the intelligent decision-making model specifically refers to: based on unsupervised learning methods, studying intelligent decision support systems in energy management, and proposing a model that can adaptively adjust energy dispatching strategies based on real-time data;
[0020] The research on the automated optimization method of the energy management system specifically refers to: optimizing the energy allocation scheme through model optimization, and proposing an intelligent energy dispatch strategy by combining real-time load demand and renewable energy fluctuations;
[0021] The research on the energy efficiency assessment and optimization feedback mechanism specifically refers to: dynamically adjusting the energy management model through real-time monitoring and feedback mechanisms to ensure that the computing center achieves optimal energy utilization efficiency under different workloads.
[0022] Preferably, the system integration includes the design of an energy optimization system integration architecture, specifically referring to: designing a scalable system architecture to adapt to computing centers of different sizes and types, in order to ensure the universality and operability of energy optimization;
[0023] The aforementioned technology application cases and field verifications specifically refer to: through testing and verification, evaluating the stability and optimization benefits of the system under different energy loads and climatic conditions, in order to provide data support for the subsequent technology promotion and application;
[0024] The research on the technology promotion path and industry influence specifically refers to: studying the promotion and application path of unsupervised learning models in other green computing centers, data centers and large-scale energy systems, and analyzing their impact on the development of energy optimization technology in the entire industry.
[0025] A method for optimizing energy efficiency in a green intelligent computing center, applied to the aforementioned energy efficiency optimization system for a green intelligent computing center, includes the following steps:
[0026] S1. Energy Consumption Pattern Analysis and Data Mining
[0027] 1.1 Data Acquisition and Preprocessing
[0028] Historical operational data of the green computing center is collected, including load, power consumption, and energy distribution. Meteorological data, including light intensity, wind speed, and temperature, are also introduced. The data is then cleaned, denoised, and normalized to ensure the accuracy and consistency of the data.
[0029] 1.2 Analysis of Energy Consumption Patterns
[0030] Unsupervised learning algorithms, including K-means and DBSCAN, are used to perform cluster analysis on energy consumption data to uncover the energy consumption characteristics and patterns of computing centers under different time periods and load conditions. Combined with meteorological conditions, the temporal correlation between energy consumption and renewable energy supply is analyzed.
[0031] 1.3. Mining Abnormal Energy Consumption Patterns
[0032] Unsupervised anomaly detection methods, including Autoencoder and Isolation Forest, are used to identify abnormal energy consumption patterns and analyze the potential factors that cause anomalies, providing a reference for subsequent scheduling optimization.
[0033] S2. Application Research of Unsupervised Learning Model in Energy Scheduling of Green Computing Center
[0034] 2.1 Construction of Unsupervised Learning Models
[0035] Select unsupervised learning algorithms suitable for computing centers, including PCA and deep clustering, to reduce the dimensionality and extract features from energy consumption and supply data. Then, combine multidimensional data, including load, weather, and energy status, to train an unsupervised model and establish a mapping relationship between energy consumption patterns and supply status.
[0036] 2.2 Research on Multi-Energy Coordinated Scheduling Methods
[0037] Based on unsupervised learning results, a multi-energy scheduling optimization algorithm is designed to cover the coordinated use of several energy sources, including photovoltaic, wind, and grid power. The optimization algorithm aims to minimize energy costs and ensure the stable operation of the computing center, while also considering the fluctuation characteristics of different energy supplies.
[0038] 2.3 Energy System Stability and Optimization Benefit Assessment
[0039] This study investigates the impact of unsupervised learning models on the stability of energy systems under different load conditions, establishes a stability evaluation index system, analyzes the improvement effects of optimized scheduling on energy utilization efficiency and the proportion of renewable energy, and evaluates the cost-saving effect of the optimization scheme on operating costs in conjunction with cost analysis.
[0040] S3. Research on Intelligent Decision-Making and Automated Optimization of Unsupervised Learning Models in Energy Management
[0041] 3.1 Construction of Intelligent Decision-Making Model
[0042] Based on unsupervised learning methods, a real-time energy management intelligent decision-making model is developed to support the automatic adjustment of scheduling strategies. The model considers dynamic factors such as real-time load and meteorological conditions to achieve accurate decision-making. It also comprehensively considers the multi-objective requirements of energy efficiency, cost, and stability to design a multi-objective optimization decision-making model.
[0043] 3.2 Research on Automation Optimization Methods for Energy Management Systems
[0044] By utilizing unsupervised learning models, an automated optimization strategy for energy allocation is developed to dynamically respond to changes in energy demand. Real-time monitoring data is integrated to dynamically adjust energy allocation and scheduling strategies, thereby ensuring stable system operation under load fluctuations.
[0045] 3.3 Energy Efficiency Assessment and Optimization Feedback Mechanism
[0046] Construct a real-time energy efficiency evaluation index system, including indicators of energy efficiency improvement rate and load balance, develop an automated feedback module, dynamically adjust model parameters based on evaluation results, and optimize scheduling strategies;
[0047] S4. Research on the Integration and Application of Unsupervised Learning Models in Energy Optimization Systems of Green Computing Centers
[0048] 4.1 Energy Optimization System Integration Architecture Design
[0049] Design an energy optimization system architecture based on an unsupervised learning model, integrating data acquisition, model analysis, optimization decision-making, and feedback modules to ensure the scalability and operability of the architecture;
[0050] 4.2 Technical Application Cases and Field Verification
[0051] The system was tested and optimized in the actual environment of the green computing center to verify its stability and effectiveness. Practical cases of technology application were summarized to provide a reference for subsequent technology promotion.
[0052] 4.3 Research on Technology Promotion Path and Industry Influence
[0053] Develop a technology promotion plan, analyze its application potential in different computing centers and industries, analyze the technology's role in promoting the development of the green digital economy, and provide a reference for energy transition.
[0054] The present invention has the following beneficial effects:
[0055] 1. This invention constructs an energy consumption pattern recognition and anomaly detection model by using clustering algorithms and anomaly detection methods in unsupervised learning. The model can discover typical patterns and potential laws of energy consumption in green computing centers without the need for labels, and achieve adaptive detection and early warning of sudden energy consumption fluctuations and abnormal operating states, effectively supporting rapid response and scheduling optimization of energy efficiency issues.
[0056] 2. This invention utilizes dimensionality reduction, clustering, and anomaly detection techniques in unsupervised learning, combined with the multi-energy supply characteristics of green computing centers, including wind power, solar power, and grid power, to propose a multi-energy collaborative scheduling optimization model. Furthermore, through dynamic allocation analysis of different energy resources, it explores the multi-energy collaborative scheduling mechanism to achieve efficient utilization of renewable energy, reduce dependence on traditional fossil fuels, and improve the intelligence and greening level of energy management in computing centers.
[0057] 3. This invention combines unsupervised learning time series analysis with load forecasting methods based on meteorological factors, integrates computing load, meteorological conditions and energy consumption data to predict future load change trends, innovatively combines the forecast results with energy efficiency assessment, and improves the energy efficiency of computing centers under different load conditions by dynamically adjusting energy scheduling strategies, and realizes a real-time assessment and optimization feedback mechanism for system energy efficiency. Attached Figure Description
[0058] Fig. 1 This is a technical roadmap for the energy efficiency optimization system of the green intelligent computing center of the present invention;
[0059] Fig. 2 This is a diagram illustrating the composition of the energy consumption pattern analysis and abnormal energy consumption pattern mining of this invention.
[0060] Fig. 3 This is a technical roadmap for the energy efficiency optimization method of the green intelligent computing center of the present invention. Detailed Implementation
[0061] The technical solution of the present invention will now be clearly and completely described in conjunction with preferred embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0062] like Figs. 1 to 3 A green intelligent computing center energy efficiency optimization system, comprising data analysis, model building, algorithm optimization, and system integration;
[0063] Data analysis includes energy consumption pattern analysis and data mining, as well as energy consumption pattern analysis and the mining of abnormal energy consumption patterns;
[0064] Model building includes the construction of unsupervised learning models, research on multi-energy coordinated scheduling methods, and evaluation of energy system stability and optimization benefits;
[0065] Algorithm optimization includes the construction of intelligent decision-making models, research on automated optimization methods for energy management systems, and research on energy efficiency assessment and optimization feedback mechanisms.
[0066] System integration includes the design of energy optimization system integration architecture, technology application cases and field verification, and research on technology promotion paths and industry impact.
[0067] Energy consumption pattern analysis and data mining specifically refer to: analyzing the relationship between computing load fluctuations and energy demand; collecting and processing historical operating data of green computing centers; using unsupervised learning models to identify energy consumption characteristics under different operating conditions; mining potential regularities and abnormal patterns; and providing data support for subsequent energy optimization strategies.
[0068] Energy consumption pattern analysis and abnormal energy consumption pattern mining specifically refer to: using clustering algorithms and anomaly detection algorithms in unsupervised learning to analyze time series data of energy consumption, identify common energy consumption patterns and sudden fluctuations in energy demand, and study how to adaptively discover abnormal situations in system operation through unsupervised learning models, provide early warnings of potential energy efficiency problems, and adjust energy dispatch strategies in a timely manner.
[0069] Clustering algorithms include K-means, hierarchical clustering, and Gaussian mixture models. Clustering algorithms discover energy consumption patterns of different groups by dividing the dataset.
[0070] Anomaly detection algorithms include Isolation Forest and Density Peak Clustering. These algorithms are used to detect abnormal energy consumption patterns. The Isolation Forest algorithm can efficiently detect abnormal fluctuations in energy consumption time series.
[0071] The construction of unsupervised learning models specifically refers to: researching and constructing energy scheduling optimization models based on unsupervised learning, and using clustering algorithms, dimensionality reduction techniques, and anomaly detection algorithms to analyze energy demand data of green computing centers; unsupervised learning models can automatically mine hidden patterns and regularities in data without external label data, which helps to deeply understand the energy consumption patterns and load fluctuation characteristics of computing centers. Through unsupervised learning models, the energy demand of computing centers can be predicted more accurately, providing a scientific basis for energy scheduling of computing centers, improving energy utilization efficiency, reducing operating costs, and also helping to optimize the operation strategy of computing centers, reduce energy waste, and promote green and low-carbon development;
[0072] The research on multi-energy collaborative scheduling methods specifically refers to: studying intelligent collaborative scheduling methods for several energy resources in green computing centers, proposing an application framework for unsupervised learning models in multi-energy environments, exploring how to maximize the utilization of renewable energy through intelligent scheduling, reduce dependence on traditional fossil energy, and improve the stability and reliability of the system.
[0073] The evaluation of energy system stability and optimization benefits specifically refers to: studying the optimization effect of unsupervised learning models in energy dispatch and the evaluation method of system stability, and evaluating the role of models in improving energy dispatch efficiency, system stability and renewable energy utilization under different load and energy configuration conditions through model verification and evaluation.
[0074] Dimensionality reduction techniques include principal component analysis (PCA) and t-SNE. Dimensionality reduction techniques are used to analyze energy consumption patterns by reducing the dimensionality of data while retaining the main features.
[0075] Energy resources include wind power, solar power, and traditional electricity.
[0076] The construction of intelligent decision-making models specifically refers to: based on unsupervised learning methods, studying intelligent decision support systems in energy management, proposing models that can adaptively adjust energy dispatch strategies based on real-time data, analyzing how to use unsupervised learning models to automatically identify energy demand patterns and formulate optimization schemes to achieve automation and intelligence in energy management.
[0077] The research on automated optimization methods for energy management systems specifically refers to: studying the application of unsupervised learning models in the automated optimization of energy management systems, focusing on how to optimize energy allocation schemes through models, avoid energy waste, improve energy efficiency, and propose intelligent energy dispatch strategies by combining real-time load demand and renewable energy fluctuations to reduce manual intervention;
[0078] The research on energy efficiency assessment and optimization feedback mechanisms specifically refers to: exploring how unsupervised learning models can achieve real-time energy efficiency assessment in green computing centers, automatically provide feedback and adjust energy optimization strategies, and studying how to dynamically adjust energy management models through real-time monitoring and feedback mechanisms to ensure that computing centers achieve optimal energy efficiency under different workloads.
[0079] System integration includes the design of an energy optimization system integration architecture, specifically referring to: designing a scalable system architecture to adapt to computing centers of different sizes and types, ensuring the universality and operability of energy optimization, researching an energy optimization system integration architecture based on an unsupervised learning model, and proposing how to effectively integrate this technology into the existing energy management system of the Qinghai Provincial Green Computing Center;
[0080] The technical application cases and field verification specifically refer to: evaluating the stability and optimization benefits of the system under different energy loads and climate conditions through testing and verification, so as to provide data support for the subsequent technology promotion and application; and studying the application effect of unsupervised learning models in real environments through actual cooperation with green computing centers.
[0081] The research on technology promotion paths and industry impact specifically refers to: studying the promotion and application paths of unsupervised learning models in other green computing centers, data centers, and large-scale energy systems, analyzing their impact on the development of energy optimization technologies across the industry, and proposing how to utilize this technology to promote the development of green digital economy nationwide and globally, and help achieve global energy transition goals.
[0082] A method for optimizing energy efficiency in a green intelligent computing center, applied to an energy efficiency optimization system for a green intelligent computing center, includes the following steps:
[0083] S1. Energy Consumption Pattern Analysis and Data Mining
[0084] By using clustering algorithms and anomaly detection methods in unsupervised learning, an energy consumption pattern recognition and anomaly detection model is constructed. The model can uncover typical patterns and potential laws of energy consumption in green computing centers without labels, and achieve adaptive detection and early warning of sudden energy consumption fluctuations and abnormal operating states, effectively supporting rapid response and scheduling optimization of energy efficiency issues.
[0085] 1.1 Data Acquisition and Preprocessing
[0086] Historical operational data of the green computing center is collected, including load, power consumption, and energy distribution. Meteorological data, including light intensity, wind speed, and temperature, are also introduced. The data is then cleaned, denoised, and normalized to ensure the accuracy and consistency of the data.
[0087] 1.2 Analysis of Energy Consumption Patterns
[0088] Unsupervised learning algorithms, including K-means and DBSCAN, are used to perform cluster analysis on energy consumption data to uncover the energy consumption characteristics and patterns of computing centers under different time periods and load conditions. Combined with meteorological conditions, the temporal correlation between energy consumption and renewable energy supply is analyzed.
[0089] 1.3. Mining Abnormal Energy Consumption Patterns
[0090] Unsupervised anomaly detection methods, including Autoencoder and Isolation Forest, are used to identify abnormal energy consumption patterns and analyze the potential factors that cause anomalies, providing a reference for subsequent scheduling optimization.
[0091] S2. Application Research of Unsupervised Learning Model in Energy Scheduling of Green Computing Center
[0092] By utilizing dimensionality reduction, clustering, and anomaly detection techniques in unsupervised learning, and combining the multi-energy supply characteristics of the Qinghai Green Computing Center (wind, solar, and grid power), a multi-energy collaborative scheduling optimization model is proposed. Through dynamic allocation analysis of different energy resources, the multi-energy collaborative scheduling mechanism is explored to achieve efficient utilization of renewable energy, reduce dependence on traditional fossil fuels, and improve the intelligence and greening level of energy management in the computing center.
[0093] 2.1 Construction of Unsupervised Learning Models
[0094] Select unsupervised learning algorithms suitable for computing centers, including PCA and deep clustering, to reduce the dimensionality and extract features from energy consumption and supply data. Then, combine multidimensional data, including load, weather, and energy status, to train an unsupervised model and establish a mapping relationship between energy consumption patterns and supply status.
[0095] 2.2 Research on Multi-Energy Coordinated Scheduling Methods
[0096] Based on unsupervised learning results, a multi-energy scheduling optimization algorithm is designed to cover the coordinated use of several energy sources, including photovoltaic, wind, and grid power. The optimization algorithm aims to minimize energy costs and ensure the stable operation of the computing center, while also considering the fluctuation characteristics of different energy supplies.
[0097] 2.3 Energy System Stability and Optimization Benefit Assessment
[0098] This study investigates the impact of unsupervised learning models on the stability of energy systems under different load conditions, establishes a stability evaluation index system, analyzes the improvement effects of optimized scheduling on energy utilization efficiency and the proportion of renewable energy, and evaluates the cost-saving effect of the optimization scheme on operating costs in conjunction with cost analysis.
[0099] S3. Research on Intelligent Decision-Making and Automated Optimization of Unsupervised Learning Models in Energy Management
[0100] Combining unsupervised learning time series analysis with load forecasting methods based on meteorological factors, this paper integrates computing load, meteorological conditions, and energy consumption data to predict future load change trends. It innovatively combines the forecast results with energy efficiency assessment, and improves the energy efficiency of computing centers under different load conditions by dynamically adjusting energy dispatch strategies. It also realizes a real-time assessment and optimization feedback mechanism for system energy efficiency.
[0101] 3.1 Construction of Intelligent Decision-Making Model
[0102] Based on unsupervised learning methods, a real-time energy management intelligent decision-making model is developed to support the automatic adjustment of scheduling strategies. The model considers dynamic factors such as real-time load and meteorological conditions to achieve accurate decision-making. It also comprehensively considers the multi-objective requirements of energy efficiency, cost, and stability to design a multi-objective optimization decision-making model.
[0103] 3.2 Research on Automation Optimization Methods for Energy Management Systems
[0104] By utilizing unsupervised learning models, an automated optimization strategy for energy allocation is developed to dynamically respond to changes in energy demand. Real-time monitoring data is integrated to dynamically adjust energy allocation and scheduling strategies, thereby ensuring stable system operation under load fluctuations.
[0105] 3.3 Energy Efficiency Assessment and Optimization Feedback Mechanism
[0106] Construct a real-time energy efficiency evaluation index system, including indicators of energy efficiency improvement rate and load balance, develop an automated feedback module, dynamically adjust model parameters based on evaluation results, and optimize scheduling strategies;
[0107] S4. Research on the Integration and Application of Unsupervised Learning Models in Energy Optimization Systems of Green Computing Centers
[0108] 4.1 Energy Optimization System Integration Architecture Design
[0109] Design an energy optimization system architecture based on an unsupervised learning model, integrating data acquisition, model analysis, optimization decision-making, and feedback modules to ensure the scalability and operability of the architecture;
[0110] 4.2 Technical Application Cases and Field Verification
[0111] The system was tested and optimized in the actual environment of the green computing center to verify its stability and effectiveness. Practical cases of technology application were summarized to provide a reference for subsequent technology promotion.
[0112] 4.3 Research on Technology Promotion Path and Industry Influence
[0113] Develop a technology promotion plan, analyze its application potential in different computing centers and industries, analyze the technology's role in promoting the development of the green digital economy, and provide a reference for energy transition.
[0114] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention. The invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A green intelligent computing center energy efficiency optimization system, characterized in that: The system includes data analysis, model building, algorithm optimization, and system integration; The data analysis includes energy consumption pattern analysis and data mining, energy consumption pattern analysis and abnormal energy consumption pattern mining; The model construction includes the construction of unsupervised learning models, research on multi-energy coordinated scheduling methods, and evaluation of energy system stability and optimization benefits; The algorithm optimization includes the construction of intelligent decision-making models, research on automated optimization methods for energy management systems, and research on energy efficiency assessment and optimization feedback mechanisms. The system integration includes the design of the energy optimization system integration architecture, technical application cases and field verification, and research on technology promotion paths and industry influence.
2. The green intelligent computing center energy efficiency optimization system according to claim 1, characterized in that: The energy consumption pattern analysis and data mining specifically refer to: analyzing the relationship between computing load fluctuations and energy demand; collecting and processing historical operating data of the green computing center; using unsupervised learning models to identify energy consumption characteristics under different operating conditions; mining potential regularities and abnormal patterns; and providing data support for subsequent energy optimization strategies. The energy consumption pattern analysis and abnormal energy consumption pattern mining specifically refer to: using clustering algorithms and anomaly detection algorithms in unsupervised learning to analyze time series data of energy consumption, identify common energy consumption patterns and sudden fluctuations in energy demand, and study how to adaptively discover abnormal situations in system operation through unsupervised learning models, provide early warnings of potential energy efficiency problems, and adjust energy dispatch strategies in a timely manner.
3. The green intelligent computing center energy efficiency optimization system according to claim 2, characterized in that: The clustering algorithms include K-means, hierarchical clustering, and Gaussian mixture model. These clustering algorithms discover energy consumption patterns in different groups by dividing the dataset.
4. The green intelligent computing center energy efficiency optimization system according to claim 2, characterized in that: The anomaly detection algorithm includes isolated forest and density peak clustering, and is used to detect abnormal energy consumption patterns.
5. The green intelligent computing center energy efficiency optimization system according to claim 1, characterized in that: The construction of the unsupervised learning model specifically refers to: researching and constructing an energy dispatch optimization model based on unsupervised learning, and using clustering algorithms, dimensionality reduction techniques, and anomaly detection algorithms to analyze the energy demand data of the green computing center; The research on the multi-energy collaborative scheduling method specifically refers to: studying intelligent collaborative scheduling methods for several energy resources in green computing centers, and proposing an application framework for unsupervised learning models in multi-energy environments; The evaluation of energy system stability and optimization benefits specifically refers to: studying the optimization effect and system stability evaluation method of unsupervised learning models in energy dispatch, and evaluating the improvement effect of the models on energy dispatch efficiency, system stability and renewable energy utilization under different load and energy configuration conditions through model verification and evaluation.
6. The green intelligent computing center energy efficiency optimization system according to claim 5, characterized in that: The dimensionality reduction techniques include principal component analysis and t-SNE. These techniques reduce the dimensionality of data while retaining key features, and are used for energy consumption pattern analysis.
7. The green intelligent computing center energy efficiency optimization system according to claim 5, characterized in that: The energy resources mentioned include wind power, solar power, and traditional electricity.
8. The green intelligent computing center energy efficiency optimization system according to claim 1, characterized in that: The construction of the intelligent decision-making model specifically refers to: based on unsupervised learning methods, studying intelligent decision support systems in energy management, and proposing a model that can adaptively adjust energy dispatch strategies based on real-time data; The research on the automated optimization method of the energy management system specifically refers to: optimizing the energy allocation scheme through model optimization, and proposing an intelligent energy dispatch strategy by combining real-time load demand and renewable energy fluctuations; The research on the energy efficiency assessment and optimization feedback mechanism specifically refers to: dynamically adjusting the energy management model through real-time monitoring and feedback mechanisms to ensure that the computing center achieves optimal energy utilization efficiency under different workloads.
9. The green intelligent computing center energy efficiency optimization system according to claim 1, characterized in that: The system integration mentioned above includes the design of an energy optimization system integration architecture, specifically referring to the design of a scalable system architecture to adapt to computing centers of different sizes and types, in order to ensure the universality and operability of energy optimization. The aforementioned technology application cases and field verifications specifically refer to: through testing and verification, evaluating the stability and optimization benefits of the system under different energy loads and climatic conditions, in order to provide data support for the subsequent technology promotion and application; The research on the technology promotion path and industry influence specifically refers to: studying the promotion and application path of unsupervised learning models in other green computing centers, data centers and large-scale energy systems, and analyzing their impact on the development of energy optimization technology in the entire industry.
10. A method for optimizing the energy efficiency of a green intelligent computing center, applied to the energy efficiency optimization system of a green intelligent computing center as described in any one of claims 1-9, characterized in that, The following steps are included: S1. Energy Consumption Pattern Analysis and Data Mining 1.1 Data Acquisition and Preprocessing Historical operational data of the green computing center is collected, including load, power consumption, and energy distribution. Meteorological data, including light intensity, wind speed, and temperature, are also introduced. The data is then cleaned, denoised, and normalized to ensure the accuracy and consistency of the data. 1.2 Analysis of Energy Consumption Patterns Unsupervised learning algorithms, including K-means and DBSCAN, are used to perform cluster analysis on energy consumption data to uncover the energy consumption characteristics and patterns of computing centers under different time periods and load conditions. Combined with meteorological conditions, the temporal correlation between energy consumption and renewable energy supply is analyzed. 1.
3. Mining Abnormal Energy Consumption Patterns Unsupervised anomaly detection methods, including Autoencoder and Isolation Forest, are used to identify abnormal energy consumption patterns and analyze the potential factors that cause anomalies, providing a reference for subsequent scheduling optimization. S2. Application Research of Unsupervised Learning Model in Energy Scheduling of Green Computing Center 2.1 Construction of Unsupervised Learning Models Select unsupervised learning algorithms suitable for computing centers, including PCA and deep clustering, to reduce the dimensionality and extract features from energy consumption and supply data. Then, combine multidimensional data, including load, weather, and energy status, to train an unsupervised model and establish a mapping relationship between energy consumption patterns and supply status. 2.2 Research on Multi-Energy Coordinated Scheduling Methods Based on unsupervised learning results, a multi-energy scheduling optimization algorithm is designed to cover the coordinated use of several energy sources, including photovoltaic, wind, and grid power. The optimization algorithm aims to minimize energy costs and ensure the stable operation of the computing center, while also considering the fluctuation characteristics of different energy supplies. 2.3 Energy System Stability and Optimization Benefit Assessment This study investigates the impact of unsupervised learning models on the stability of energy systems under different load conditions, establishes a stability evaluation index system, analyzes the improvement effects of optimized scheduling on energy utilization efficiency and the proportion of renewable energy, and evaluates the cost-saving effect of the optimization scheme on operating costs in conjunction with cost analysis. S3. Research on Intelligent Decision-Making and Automated Optimization of Unsupervised Learning Models in Energy Management 3.1 Construction of Intelligent Decision-Making Model Based on unsupervised learning methods, a real-time energy management intelligent decision-making model is developed to support the automatic adjustment of scheduling strategies. The model considers dynamic factors such as real-time load and meteorological conditions to achieve accurate decision-making. It also comprehensively considers the multi-objective requirements of energy efficiency, cost, and stability to design a multi-objective optimization decision-making model. 3.2 Research on Automation Optimization Methods for Energy Management Systems By utilizing unsupervised learning models, an automated optimization strategy for energy allocation is developed to dynamically respond to changes in energy demand. Real-time monitoring data is integrated to dynamically adjust energy allocation and scheduling strategies, thereby ensuring stable system operation under load fluctuations. 3.3 Energy Efficiency Assessment and Optimization Feedback Mechanism Construct a real-time energy efficiency evaluation index system, including indicators of energy efficiency improvement rate and load balance, develop an automated feedback module, dynamically adjust model parameters based on evaluation results, and optimize scheduling strategies; S4. Research on the Integration and Application of Unsupervised Learning Models in Energy Optimization Systems of Green Computing Centers 4.1 Energy Optimization System Integration Architecture Design Design an energy optimization system architecture based on an unsupervised learning model, integrating data acquisition, model analysis, optimization decision-making, and feedback modules to ensure the scalability and operability of the architecture; 4.2 Technical Application Cases and Field Verification The system was tested and optimized in the actual environment of the green computing center to verify its stability and effectiveness. Practical cases of technology application were summarized to provide a reference for subsequent technology promotion. 4.3 Research on Technology Promotion Path and Industry Influence Develop a technology promotion plan, analyze its application potential in different computing centers and industries, analyze the technology's role in promoting the development of the green digital economy, and provide a reference for energy transition.