Stock machine room energy-saving reconstruction scheme generation method and system

Abstract

The invention relates to the technical field of energy-saving transformation, in particular to a stock machine room energy-saving transformation scheme generation method and system, and the method comprises the steps: multi-source data collection and standardization, current situation intelligent diagnosis, dynamic scheme generation, benefit simulation and optimization, and tracking and closed-loop feedback implementation. Compared with the prior art which mainly depends on a general energy consumption monitoring platform or a self-research tool with a single function to carry out isolated data acquisition, the system has the defects of incomplete data dimensions and limited analysis view angle, and adopts a multi-source data acquisition scheme combining an Internet of Things sensor and a system interface; all-dimensional data such as energy consumption, environment, equipment operation and green electricity use are subjected to standardized fusion processing, a unified high-quality analysis basis is formed, and the method has the advantages of being comprehensive in data, unified in standard and reliable in quality.

Description

Technical Field

[0001] This invention relates to the field of energy-saving renovation technology, and in particular to a method and system for generating energy-saving renovation schemes for existing computer rooms. Background Technology

[0002] With the acceleration of global digitalization, energy consumption in communications, the internet, and enterprise data centers continues to rise, making energy-saving retrofitting of existing data centers a crucial step in achieving "dual carbon" goals. Traditional retrofitting processes rely heavily on expert experience. For large-scale and diverse existing data centers, there is an urgent need for scientific and standardized decision-making methods to systematically improve energy efficiency, reduce operating costs, and decrease carbon emissions.

[0003] However, existing technical solutions have significant limitations. General-purpose energy consumption monitoring platforms often only achieve basic data collection and lack in-depth analysis capabilities specific to data center scenarios; enterprise-developed tools have limited functionality, are difficult to be compatible with diverse equipment protocols, and generally do not incorporate environmental benefits such as green energy consumption and carbon emission reduction into the evaluation system; while industry standards mainly stipulate energy efficiency benchmark values ​​and fail to provide transformation paths that can be dynamically adapted to different data center types and regional characteristics. This often leads to problems such as insufficient adaptability of transformation solutions, single evaluation dimensions, and a lack of long-term effect tracking mechanisms, resulting in poor investment returns or energy efficiency rebound.

[0004] To address the aforementioned issues, this invention proposes a method and system for generating energy-saving retrofit solutions for existing data centers. By integrating IoT data acquisition, machine learning-based intelligent diagnostics, dynamic technology matching, and multi-objective optimization simulation, a fully automated decision support system is constructed, encompassing current status analysis, solution generation, and continuous optimization. This method not only accurately identifies high-energy-consuming components and generates retrofit packages suitable for different scenarios but also accurately predicts long-term economic and environmental benefits. Furthermore, a closed-loop feedback mechanism ensures the sustainability of the retrofit results, ultimately achieving a scientific, refined, and sustainable energy-saving retrofit process. Summary of the Invention

[0005] To overcome the problems mentioned in the background art, the present invention proposes a method and system for generating energy-saving renovation schemes for existing computer rooms.

[0006] The technical solution of this invention is: a method for generating an energy-saving renovation plan for existing computer rooms, comprising the following steps:

[0007] S11: Multi-source data acquisition and standardization. Through IoT sensors and system interfaces, data such as energy consumption, environmental data, equipment operation data and green electricity usage data of the computer room are collected, and the data is cleaned and normalized according to preset standards to form a standardized dataset.

[0008] S12: Current Status Intelligent Diagnosis. Based on machine learning algorithms, it analyzes standardized datasets, identifies high-energy-consuming links, and calls the rule engine to quantify and score the energy efficiency level of the data center according to industry standards.

[0009] S13: Dynamic solution generation, which matches the diagnostic results with the preset energy-saving technology library and generates a renovation solution package based on the technology applicability score, data center type and budget constraints;

[0010] S14: Benefit Simulation and Optimization. The simulation algorithm is used to predict the long-term benefits of the generated renovation plan package, and the optimization algorithm is used to iteratively optimize the technology combination and output key economic indicators.

[0011] S15: Implement tracking and closed-loop feedback. After the renovation plan is implemented, continuously collect operational data through monitoring terminals to verify the renovation effect and automatically trigger optimization alarms when energy efficiency indicators rebound.

[0012] As a preferred option, the current intelligent diagnostic steps specifically include:

[0013] S21: Analyze the standardized multi-source dataset of the data center based on machine learning algorithms to identify high-energy-consuming links;

[0014] S22: Call the rules engine to quantitatively score the overall energy efficiency level of the data center based on the preset industry standard library;

[0015] The analysis results from machine learning algorithms and the quantitative scores from the rule engine together constitute the current status diagnosis report.

[0016] As a preferred approach, the current intelligent diagnostic process employs machine learning algorithms to analyze standardized multi-source datasets from the data center to identify high-energy-consuming processes. Specifically:

[0017] The random forest algorithm is used as the core machine learning model to perform supervised learning and pattern recognition on historical and real-time time-series data in a standardized dataset. The time-series data includes at least sub-item energy consumption data, equipment load rate, and ambient temperature and humidity. By analyzing the correlation characteristics between data and abnormal fluctuations that deviate from the normal energy efficiency mode, the algorithm outputs the probabilistic identification results and their confidence levels for specific high-energy-consuming operation modes. The identified high-energy-consuming links include at least: hot and cold mixing caused by unreasonable airflow organization, efficiency degradation of uninterruptible power supplies (UPS) under low load rate, and mismatch between the capacity of the cooling system and the actual heat generation of IT equipment.

[0018] As a preferred method, when quantitatively scoring the overall energy efficiency level of the data center based on a pre-defined industry standard library, the comprehensive quantitative score is calculated using the following formula:

[0019] S=∑(w i ·Ni );

[0020] Where S represents the overall energy efficiency score of the data center, w i N represents the weight coefficient of the i-th evaluation index. i Let represent the standardized score of the i-th evaluation indicator, and Among them, A i B represents the actual measured value of the i-th evaluation index, including the actual value of power utilization efficiency (PUE), water resource utilization efficiency (WUE), and the actual value of equipment aging degree calculated from equipment operating parameters. i This represents the benchmark value corresponding to the i-th evaluation indicator, which is derived from the industry standard library.

[0021] As a preferred embodiment, the dynamic scheme generation steps specifically include:

[0022] S31: Match the identified high-energy-consuming links and the quantitative scores of the overall energy efficiency level of the computer room with the preset energy-saving technology library;

[0023] S32: Calculate the applicability score of each technology in the energy-saving technology library based on the preset weight formula;

[0024] S33: Based on the technology suitability score, the preset type of the data center, and the user's budget constraints, automatically generate a transformation solution package that includes multiple technology combinations.

[0025] Preferably, the pre-defined energy-saving technology library includes at least three technology categories:

[0026] Refrigeration technology includes indirect evaporative cooling systems, closed cold aisle technology, and variable frequency air conditioning technology.

[0027] Power supply and distribution technologies, including modular UPS and high-efficiency transformer technology;

[0028] IT technologies include energy-efficient server replacement and virtualization integration technologies.

[0029] Preferably, when calculating the applicability score of each technology in the energy-saving technology library based on a preset weighting formula, the technology applicability score is calculated using the following weighting formula:

[0030] Score tech =ω1*S diagnosis +ω2*S region +ω3*S cost ;

[0031] Among them, Score tech For the technology suitability score, S diagnosis S is the score representing the degree of match between the technology and the energy efficiency issues identified in the diagnostic results. regionS is the adaptability score based on the climate characteristics of the data center's location. cost The economic score represents the technology, with ω1, ω2, and ω3 being the weighting factors for each score item, which are dynamically adjusted according to the data center type: core data center, general data center, and edge data center.

[0032] As a preferred option, the benefit simulation and optimization steps specifically include:

[0033] S41: Obtain a renovation package that includes a combination of various technologies;

[0034] S42: Input the technical combination and related parameters of the solution package into the simulation algorithm model to predict long-term benefits and generate a prediction report that includes key economic and environmental benefit indicators.

[0035] S43: Input the results of the prediction report into the optimization algorithm model, iteratively optimize the initial combination of technologies, and search for a better solution set under multiple objectives;

[0036] S44: Output the final optimized renovation plan and its complete benefit evaluation indicators.

[0037] As a preferred method, the simulation algorithm model is Monte Carlo simulation, and the specific process includes:

[0038] S51: Define input variables, including initial investment cost of equipment, local electricity price and its annual growth rate, expected life of equipment, annual rate of change of data center load, and carbon trading price;

[0039] S52: Define a probability distribution function for each input variable to characterize its uncertainty;

[0040] S53: Perform multiple random sampling iterations. For each iteration, extract specific values ​​for each input variable according to the probability distribution function, and calculate the output result of that iteration based on these values.

[0041] S54: The output results include the cumulative electricity savings, cumulative electricity cost savings, carbon emission reductions and corresponding carbon trading revenue over the next 5 years;

[0042] S55: Statistically analyze all iteration results, generate a prediction report, present the output results in the form of a probability distribution, and calculate its expected value as the final prediction value.

[0043] As a preferred choice, the optimization algorithm model is a genetic algorithm; its specific process includes:

[0044] S61: Initialization, encoding a combination of technologies into a chromosome, randomly generating an initial population containing multiple chromosomes, each chromosome representing a possible modification scheme;

[0045] S62: Evaluation involves decoding each chromosome in the population into a specific combination of technologies and calculating its fitness using Monte Carlo simulation. The fitness function is a comprehensive objective function, expressed as:

[0046] Fitness = α * (1 / Normalized) Cost )+β*Normalizd S avinge+γ *

[0047] Normalized CarbonReduction ;

[0048] Among them, Normalized Cost Normalizd is the initial investment cost for normalization. S avinge represents the long-term savings benefit of normalization. CarbonReduction The carbon emission reduction is the normalized amount, and α, β and γ are weighting coefficients.

[0049] S63: Selection, crossover, and mutation. Based on fitness scores, selection operations are performed to retain superior individuals, and crossover and mutation operations are performed on the selected individuals to generate new offspring populations, simulating the biological evolution process.

[0050] S64: Iteration, repeat steps S62 and S63 until the preset number of iterations is reached and the fitness function converges;

[0051] S65: Output, which decodes the multiple chromosomes with the highest fitness in the final population into an optimized modification scheme package.

[0052] A system for generating energy-saving retrofit solutions for existing computer rooms includes:

[0053] The data acquisition layer is used to connect to various IoT sensors and power monitoring systems via Modbus and MQTT protocols to collect multi-source data;

[0054] The algorithm engine layer integrates an LSTM model for energy consumption prediction, a machine learning algorithm for current status diagnosis, a Monte Carlo simulation for benefit simulation, and a multi-objective optimization algorithm for scheme optimization.

[0055] The application layer provides visual dashboards and mobile interfaces to display transformation plans, benefit comparisons, and alarm information.

[0056] The beneficial effects of this invention are:

[0057] 1. Compared with existing technologies that mainly rely on general energy consumption monitoring platforms or self-developed tools with single functions for isolated data collection, which have the disadvantages of incomplete data dimensions and limited analysis perspectives, this invention adopts a multi-source data collection scheme that combines IoT sensors and system interfaces to standardize and integrate data from all dimensions such as energy consumption, environment, equipment operation and green electricity use, forming a unified and high-quality analysis foundation with the advantages of comprehensive data, unified standards and reliable quality.

[0058] 2. Compared with existing technologies that typically focus only on the single indicator of PUE for rough evaluation, lacking collaborative analysis of multi-dimensional capabilities such as water resources and equipment efficiency, and having the disadvantages of single evaluation dimensions and one-sided diagnostic results, this invention constructs a comprehensive evaluation scheme that combines machine learning intelligent diagnosis with rule engine standard quantification. It can simultaneously and accurately identify hidden operational faults and perform multi-indicator compliance scoring, with the advantages of both diagnostic depth and quantitative breadth.

[0059] 3. Compared with existing technologies that often provide static and general technology recommendation lists, which cannot adapt to the differentiated needs of different data center types and regional climates, and have the disadvantages of poor solution targeting and weak technology adaptability, this invention adopts a solution package that automatically generates solution packages based on a dynamic weight formula technology applicability score and a multi-objective optimization algorithm. It can dynamically optimize the technology combination according to the data center level, regional characteristics and budget constraints, and has the advantages of personalized solutions and scientific decision-making.

[0060] 4. Compared with existing technologies that are mostly limited to static calculations of initial investment costs and simple electricity savings, failing to cover comprehensive benefits such as green electricity consumption and carbon trading, and having the disadvantages of narrow benefit assessment and high investment risk, this invention introduces a benefit simulation and optimization scheme that integrates Monte Carlo simulation and genetic algorithm. It performs probabilistic prediction and iterative optimization of the combination of technologies for long-term electricity saving, carbon emission reduction and carbon trading benefits, and has the advantages of comprehensive assessment, controllable risk and maximized benefits.

[0061] 5. Compared with existing technologies, which generally suffer from one-time modification problems and lack a mechanism for continuous tracking and optimization of the effects after modification, resulting in management disconnect and easy rebound of energy efficiency, this invention establishes a real-time tracking and closed-loop feedback scheme based on a lightweight monitoring terminal and cloud platform. It can verify the modification effect in real time and automatically trigger optimization suggestions when energy efficiency exceeds the standard, realizing continuous optimization throughout the entire life cycle. It has the advantages of closed-loop management and sustainable effects. Attached Figure Description

[0062] Figure 1 The diagram shown is a flowchart illustrating the method for generating energy-saving retrofit solutions for existing computer rooms according to the present invention.

[0063] Figure 2The diagram shown is a structural schematic of the energy-saving renovation scheme generation system for existing computer rooms according to the present invention. Detailed Implementation

[0064] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0065] Please see Figure 1 This invention provides an embodiment: a method for generating an energy-saving renovation plan for an existing computer room, comprising the following steps:

[0066] S11: Multi-source data acquisition and standardization. Through IoT sensors and system interfaces, data such as energy consumption, environmental data, equipment operation data and green electricity usage data of the computer room are collected, and the data is cleaned and normalized according to preset standards to form a standardized dataset.

[0067] S12: Current Status Intelligent Diagnosis. Based on machine learning algorithms, it analyzes standardized datasets, identifies high-energy-consuming links, and calls the rule engine to quantify and score the energy efficiency level of the data center according to industry standards.

[0068] S13: Dynamic solution generation, which matches the diagnostic results with the preset energy-saving technology library and generates a renovation solution package based on the technology applicability score, data center type and budget constraints;

[0069] S14: Benefit Simulation and Optimization. The simulation algorithm is used to predict the long-term benefits of the generated renovation plan package, and the optimization algorithm is used to iteratively optimize the technology combination and output key economic indicators.

[0070] S15: Implement tracking and closed-loop feedback. After the renovation plan is implemented, continuously collect operational data through monitoring terminals to verify the renovation effect and automatically trigger optimization alarms when energy efficiency indicators rebound.

[0071] As described above, this invention constructs a complete process chain of "data acquisition - diagnosis - generation - simulation - feedback," realizing standardized processing of multi-source heterogeneous data, intelligent diagnosis based on machine learning and rule engines, dynamic adaptation of solution generation and Monte Carlo benefit simulation, and finally forming a management closed loop through IoT monitoring. This integrated process transforms discrete energy-saving technologies, extensive benefit assessments, and static transformation decisions into a continuously optimized dynamic systems engineering, significantly improving the scientific nature, technical adaptability, investment security, and long-term operational sustainability of energy-saving transformation solutions for existing data centers.

[0072] As a preferred option, the current intelligent diagnostic steps specifically include:

[0073] S21: Analyze the standardized multi-source dataset of the data center based on machine learning algorithms to identify high-energy-consuming links;

[0074] S22: Call the rules engine to quantitatively score the overall energy efficiency level of the data center based on the preset industry standard library;

[0075] The analysis results from machine learning algorithms and the quantitative scores from the rule engine together constitute the current status diagnosis report.

[0076] As described above, this invention achieves the dual goals of in-depth exploration and objective evaluation of data center energy efficiency issues by clearly decomposing the current status diagnosis into two collaborative stages: "machine learning anomaly identification" and "rule engine standard quantification." This design not only utilizes data-driven models to discover hidden faults and operational degradation, but also relies on industry standard libraries for compliance benchmarking and rating, significantly improving the comprehensiveness of the diagnostic process, the accuracy of the results, and the smoothness of the connection with subsequent technology recommendation stages.

[0077] As a preferred approach, the current intelligent diagnostic process employs machine learning algorithms to analyze standardized multi-source datasets from the data center to identify high-energy-consuming processes. Specifically:

[0078] The random forest algorithm is used as the core machine learning model to perform supervised learning and pattern recognition on historical and real-time time-series data in a standardized dataset. The time-series data includes at least sub-item energy consumption data, equipment load rate, and ambient temperature and humidity. By analyzing the correlation characteristics between data and abnormal fluctuations that deviate from the normal energy efficiency mode, the algorithm outputs the probabilistic identification results and their confidence levels for specific high-energy-consuming operation modes. The identified high-energy-consuming links include at least: hot and cold mixing caused by unreasonable airflow organization, efficiency degradation of uninterruptible power supplies (UPS) under low load rate, and mismatch between the capacity of the cooling system and the actual heat generation of IT equipment.

[0079] As described above, this invention employs the random forest algorithm for supervised learning and pattern recognition of time-series data, enabling probabilistic and precise localization of complex high-energy-consuming factors such as unreasonable airflow organization and inefficient UPS operation. This algorithm can effectively capture nonlinear correlations and abnormal fluctuations among multiple variables, and its output confidence index provides a quantitative basis for the credibility of diagnostic results, significantly improving the intelligence level of high-energy-consuming root cause analysis, the accuracy of localization, and the targeted nature of subsequent operation and maintenance interventions.

[0080] As a preferred method, when quantitatively scoring the overall energy efficiency level of the data center based on a pre-defined industry standard library, the comprehensive quantitative score is calculated using the following formula:

[0081] S=∑(w i ·N i );

[0082] Where S represents the overall energy efficiency score of the data center, w iN represents the weight coefficient of the i-th evaluation index. i Let represent the standardized score of the i-th evaluation indicator, and Among them, A i B represents the actual measured value of the i-th evaluation index, including the actual value of power utilization efficiency (PUE), water resource utilization efficiency (WUE), and the actual value of equipment aging degree calculated from equipment operating parameters. i This represents the benchmark value corresponding to the i-th evaluation indicator, which is derived from the industry standard library.

[0083] As described above, this invention constructs a multi-dimensional energy efficiency quantitative evaluation system by introducing a standardized score calculation based on "actual value / benchmark value" and a weighted comprehensive scoring formula with adjustable weight coefficients. This formulaic method transforms abstract energy efficiency status into intuitive comprehensive scores, achieving a unified, objective, and comparable quantitative assessment of energy efficiency levels in different data centers. This significantly improves the standardization of energy efficiency evaluation, the fairness of results, and the interpretability of data center energy efficiency shortcomings indicators.

[0084] As a preferred embodiment, the dynamic scheme generation steps specifically include:

[0085] S31: Match the identified high-energy-consuming links and the quantitative scores of the overall energy efficiency level of the computer room with the preset energy-saving technology library;

[0086] S32: Calculate the applicability score of each technology in the energy-saving technology library based on the preset weight formula;

[0087] S33: Based on the technology suitability score, the preset type of the data center, and the user's budget constraints, automatically generate a transformation solution package that includes multiple technology combinations.

[0088] Preferably, the pre-defined energy-saving technology library includes at least three technology categories:

[0089] Refrigeration technology includes indirect evaporative cooling systems, closed cold aisle technology, and variable frequency air conditioning technology.

[0090] Power supply and distribution technologies, including modular UPS and high-efficiency transformer technology;

[0091] IT technologies include energy-efficient server replacement and virtualization integration technologies.

[0092] As described above, this invention achieves automated and structured mapping from problem diagnosis to solutions by systematically matching diagnostic results with an energy-saving technology library and combining them in packages based on quantitative scores, data center type, and budget constraints. This process ensures that the recommended technical solutions are highly relevant to the identified energy efficiency problems and fully considers the objective conditions of the implementation scenario, significantly improving the relevance and feasibility of solution generation and providing users with diverse choices.

[0093] Preferably, when calculating the applicability score of each technology in the energy-saving technology library based on a preset weighting formula, the technology applicability score is calculated using the following weighting formula:

[0094] Score tech =ω1*S diagnosis +ω2*S region +ω3*S cost ;

[0095] Among them, Score tech For the technology suitability score, S diagnosis S is the score representing the degree of match between the technology and the energy efficiency issues identified in the diagnostic results. region S is the adaptability score based on the climate characteristics of the data center's location. cost The economic score represents the technology, with ω1, ω2, and ω3 being the weighting factors for each score item, which are dynamically adjusted according to the data center type: core data center, general data center, and edge data center.

[0096] As described above, this invention calculates the technology suitability score by designing a weighted formula that includes matching degree, regional adaptability, and economy, thus upgrading technology recommendation from "simple enumeration" to "multi-factor weighted selection". This formula accurately reflects the core requirements of different data center types by dynamically adjusting the weight factors (such as core data centers emphasizing reliability and edge data centers emphasizing return on investment), significantly improving the refinement of technology selection, the scientific nature of decision-making, and the dynamic response capability to differentiated needs.

[0097] As a preferred option, the benefit simulation and optimization steps specifically include:

[0098] S41: Obtain a renovation package that includes a combination of various technologies;

[0099] S42: Input the technical combination and related parameters of the solution package into the simulation algorithm model to predict long-term benefits and generate a prediction report that includes key economic and environmental benefit indicators.

[0100] S43: Input the results of the prediction report into the optimization algorithm model, iteratively optimize the initial combination of technologies, and search for a better solution set under multiple objectives;

[0101] S44: Output the final optimized renovation plan and its complete benefit evaluation indicators.

[0102] As described above, this invention introduces an independent "simulation-optimization" closed loop after the scheme is generated, uses Monte Carlo simulation to predict long-term benefits and uses optimization algorithms to iteratively search for better solutions, thereby realizing the quantification of financial risks and multi-objective Pareto optimization of the preliminary scheme. This step transforms the transformation decision from a judgment based on static experience to a precise calculation based on dynamic probability simulation, significantly improving the foresight, risk controllability, and potential for maximizing the overall return of the final scheme.

[0103] As a preferred method, the simulation algorithm model is Monte Carlo simulation, and the specific process includes:

[0104] S51: Define input variables, including initial investment cost of equipment, local electricity price and its annual growth rate, expected life of equipment, annual rate of change of data center load, and carbon trading price;

[0105] S52: Define a probability distribution function for each input variable to characterize its uncertainty;

[0106] S53: Perform multiple random sampling iterations. For each iteration, extract specific values ​​for each input variable according to the probability distribution function, and calculate the output result of that iteration based on these values.

[0107] S54: The output results include the cumulative electricity savings, cumulative electricity cost savings, carbon emission reductions and corresponding carbon trading revenue over the next 5 years;

[0108] S55: Statistically analyze all iteration results, generate a prediction report, present the output results in the form of a probability distribution, and calculate its expected value as the final prediction value.

[0109] As described above, this invention, by employing Monte Carlo simulation to define probability distributions for key input variables and conducting extensive random sampling iterations, achieves probabilistic predictions of key indicators such as investment returns and energy-saving benefits, rather than single-value estimations. This probabilistic model effectively quantifies the risks brought about by uncertainties such as future market electricity prices and load changes, outputting results in the form of expected values ​​and confidence intervals, significantly improving the robustness of benefit predictions, the ability to characterize the complexity of the real world, and the comprehensiveness of investment risk assessment.

[0110] As a preferred choice, the optimization algorithm model is a genetic algorithm; its specific process includes:

[0111] S61: Initialization, encoding a combination of technologies into a chromosome, randomly generating an initial population containing multiple chromosomes, each chromosome representing a possible modification scheme;

[0112] S62: Evaluation involves decoding each chromosome in the population into a specific combination of technologies and calculating its fitness using Monte Carlo simulation. The fitness function is a comprehensive objective function, expressed as:

[0113] Fitness = α * (1 / Normalized) Cost )+β*Normalizd S avinge+γ *

[0114] Normalized CarbonReduction ;

[0115] Among them, Normalized Cost Normalizd is the initial investment cost for normalization. S avinge represents the long-term savings benefit of normalization. CarbonReduction The carbon emission reduction is the normalized amount, and α, β and γ are weighting coefficients.

[0116] S63: Selection, crossover, and mutation. Based on fitness scores, selection operations are performed to retain superior individuals, and crossover and mutation operations are performed on the selected individuals to generate new offspring populations, simulating the biological evolution process.

[0117] S64: Iteration, repeat steps S62 and S63 until the preset number of iterations is reached and the fitness function converges;

[0118] S65: Output, which decodes the multiple chromosomes with the highest fitness in the final population into an optimized modification scheme package.

[0119] As described above, this invention employs a genetic algorithm to encode technology combinations into a chromosome population. Based on a multi-objective fitness function that includes cost, benefit, and environmental considerations, it performs selection, crossover, and mutation iterations to automatically search for technology combinations that approximate the Pareto optimal frontier in a vast solution space. This evolutionary optimization mechanism can effectively escape local optima, significantly improving the global search capability of the optimization process, the ability to seek the best balance under multiple constraints (such as budget and PUE objectives), and the engineering superiority of the final recommended solution.

[0120] like Figure 2 As shown, this embodiment also provides a system for generating energy-saving renovation plans for existing computer rooms, including:

[0121] The data acquisition layer is used to connect to various IoT sensors and power monitoring systems via Modbus and MQTT protocols to collect multi-source data;

[0122] The algorithm engine layer integrates an LSTM model for energy consumption prediction, a machine learning algorithm for current status diagnosis, a Monte Carlo simulation for benefit simulation, and a multi-objective optimization algorithm for scheme optimization.

[0123] The application layer provides visual dashboards and mobile interfaces to display transformation plans, benefit comparisons, and alarm information.

[0124] As described above, this invention materializes the above method into a deployable and runnable software platform by constructing a system architecture that includes a data acquisition layer, an algorithm engine layer, and an application layer. The system achieves seamless access to multi-source data through protocol compatibility, provides core decision intelligence by integrating multiple advanced algorithm models, and realizes a closed-loop management system through a visual dashboard and mobile terminal interface. This significantly improves the automation level, system integration integrity, and user convenience and experience of the entire energy-saving renovation scheme generation and management process.

[0125] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for generating an energy-saving renovation plan for an existing computer room, characterized in that: Includes the following steps: S11: Multi-source data acquisition and standardization. Through IoT sensors and system interfaces, data such as energy consumption, environmental data, equipment operation data and green electricity usage data of the computer room are collected, and the data is cleaned and normalized according to preset standards to form a standardized dataset. S12: Current Status Intelligent Diagnosis. Based on machine learning algorithms, it analyzes standardized datasets, identifies high-energy-consuming links, and calls the rule engine to quantify and score the energy efficiency level of the data center according to industry standards. S13: Dynamic solution generation, which matches the diagnostic results with the preset energy-saving technology library and generates a renovation solution package based on the technology applicability score, data center type and budget constraints; S14: Benefit Simulation and Optimization. The simulation algorithm is used to predict the long-term benefits of the generated renovation plan package, and the optimization algorithm is used to iteratively optimize the technology combination and output key economic indicators. S15: Implement tracking and closed-loop feedback. After the renovation plan is implemented, continuously collect operational data through monitoring terminals to verify the renovation effect and automatically trigger optimization alarms when energy efficiency indicators rebound.

2. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 1, characterized in that: The specific steps of intelligent diagnosis of the current situation include: S21: Analyze the standardized multi-source dataset of the data center based on machine learning algorithms to identify high-energy-consuming links; S22: Call the rules engine to quantitatively score the overall energy efficiency level of the data center based on the preset industry standard library; The analysis results from machine learning algorithms and the quantitative scores from the rule engine together constitute the current status diagnosis report.

3. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 2, characterized in that: The intelligent diagnostic process employs machine learning algorithms to analyze standardized multi-source datasets from the data center to identify high-energy-consuming processes. Specifically: The random forest algorithm is used as the core machine learning model to perform supervised learning and pattern recognition on historical and real-time time-series data in a standardized dataset. The time-series data includes at least sub-item energy consumption data, equipment load rate, and ambient temperature and humidity. By analyzing the correlation characteristics between data and abnormal fluctuations that deviate from the normal energy efficiency mode, the algorithm outputs the probabilistic identification results and their confidence levels for specific high-energy-consuming operation modes. The identified high-energy-consuming links include at least: hot and cold mixing caused by unreasonable airflow organization, efficiency degradation of uninterruptible power supplies (UPS) under low load rate, and mismatch between the capacity of the cooling system and the actual heat generation of IT equipment.

4. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 3, characterized in that: When quantitatively scoring the overall energy efficiency level of a data center based on a pre-defined industry standard library, the comprehensive quantitative score is calculated using the following formula: S=∑(w i ·N i ); Where S represents the overall energy efficiency score of the data center, w i N represents the weight coefficient of the i-th evaluation index. i Let represent the standardized score of the i-th evaluation indicator, and Among them, A i B represents the actual measured value of the i-th evaluation index, including the actual value of power utilization efficiency (PUE), water resource utilization efficiency (WUE), and the actual value of equipment aging degree calculated from equipment operating parameters. i This represents the benchmark value corresponding to the i-th evaluation indicator, which is derived from the industry standard library.

5. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 4, characterized in that: The dynamic scheme generation steps specifically include: S31: Match the identified high-energy-consuming links and the quantitative scores of the overall energy efficiency level of the computer room with the preset energy-saving technology library; S32: Calculate the applicability score of each technology in the energy-saving technology library based on the preset weight formula; S33: Based on the technology suitability score, the preset type of the data center, and the user's budget constraints, automatically generate a transformation solution package that includes multiple technology combinations.

6. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 5, characterized in that: When calculating the applicability score of each technology in the energy-saving technology library based on a preset weighting formula, the technology applicability score is calculated using the following weighting formula: Score tech =ω1*S diagnosis +ω2*S region +ω3*S cost ; Among them, Score tech For the technology suitability score, S diagnosis S is the score representing the degree of match between the technology and the energy efficiency issues identified in the diagnostic results. region S is the adaptability score based on the climate characteristics of the data center's location. cost The economic score represents the technology, with ω1, ω2, and ω3 being the weighting factors for each score item, which are dynamically adjusted according to the data center type: core data center, general data center, and edge data center.

7. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 6, characterized in that: The specific steps of benefit simulation and optimization include: S41: Obtain a renovation package that includes a combination of various technologies; S42: Input the technical combination and related parameters of the solution package into the simulation algorithm model to predict long-term benefits and generate a prediction report that includes key economic and environmental benefit indicators. S43: Input the results of the prediction report into the optimization algorithm model, iteratively optimize the initial combination of technologies, and search for a better solution set under multiple objectives; S44: Output the final optimized renovation plan and its complete benefit evaluation indicators.

8. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 7, characterized in that: The simulation algorithm model is a Monte Carlo simulation, and the specific process includes: S51: Define input variables, including initial investment cost of equipment, local electricity price and its annual growth rate, expected life of equipment, annual rate of change of data center load, and carbon trading price; S52: Define a probability distribution function for each input variable to characterize its uncertainty; S53: Perform multiple random sampling iterations. For each iteration, extract specific values ​​for each input variable according to the probability distribution function, and calculate the output result of that iteration based on these values. S54: The output results include the cumulative electricity savings, cumulative electricity cost savings, carbon emission reductions and corresponding carbon trading revenue over the next 5 years; S55: Statistically analyze all iteration results, generate a prediction report, present the output results in the form of a probability distribution, and calculate its expected value as the final prediction value.

9. The method for generating an energy-saving renovation scheme for an existing computer room according to claim 8, characterized in that: The optimization algorithm model is a genetic algorithm; its specific process includes: S61: Initialization, encoding a combination of technologies into a chromosome, randomly generating an initial population containing multiple chromosomes, each chromosome representing a possible modification scheme; S62: Evaluation, which decodes each chromosome in the population into a specific combination of technologies and uses Monte Carlo simulation to calculate its fitness, with the fitness function being a comprehensive objective function; S63: Selection, crossover, and mutation. Based on fitness scores, selection operations are performed to retain superior individuals, and crossover and mutation operations are performed on the selected individuals to generate new offspring populations, simulating the biological evolution process. S64: Iteration, repeat steps S62 and S63 until the preset number of iterations is reached and the fitness function converges; S65: Output, which decodes the multiple chromosomes with the highest fitness in the final population into an optimized modification scheme package.

10. A system for generating energy-saving renovation plans for existing computer rooms, used to implement the method for generating energy-saving renovation plans for existing computer rooms as described in any one of claims 1-9, characterized in that: include: The data acquisition layer is used to connect to various IoT sensors and power monitoring systems via Modbus and MQTT protocols to collect multi-source data; The algorithm engine layer integrates an LSTM model for energy consumption prediction, a machine learning algorithm for current status diagnosis, a Monte Carlo simulation for benefit simulation, and a multi-objective optimization algorithm for scheme optimization. The application layer provides visual dashboards and mobile interfaces to display transformation plans, benefit comparisons, and alarm information.

Images