Intelligent computing center infrastructure modular configuration method based on energy efficiency grading

Abstract

The invention belongs to the technical field of resource configuration, and discloses an intelligent computing center infrastructure modular configuration method based on energy efficiency grading. Comprising the steps of 1, collecting key information of an intelligent calculation center, and determining a set energy efficiency level of the intelligent calculation center; step 2, distributing the determined facility additional budget power consumption to each infrastructure module to obtain a budget power consumption target of each infrastructure module; 3, generating a plurality of configuration strategies of each infrastructure module, and selecting an optimal configuration strategy; 4, judging whether the selected optimal configuration strategy is appropriate or not; and 5, executing a corresponding response according to an analysis result. According to the method, comprehensive analysis is carried out in combination with the corrected dimension coefficient and the calculated weight coefficient, and the set basic weight is corrected, so that personalized requirements of project equipment, environment and load can be better met when the budget power consumption target is allocated to the infrastructure module, and the budget power consumption target is more accurately allocated.

Description

Technical Field

[0001] This invention relates to the field of resource allocation technology, and more specifically, to a modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification. Background Technology

[0002] Intelligent computing centers efficiently support data sharing, intelligent ecosystem construction, and industrial innovation clustering through the production, aggregation, scheduling, and release of computing power, effectively promoting the industrialization of AI, the AI-driven transformation of industries, and the intelligentization of government governance. They are not only the physical carriers of intelligent computing power but also a new type of infrastructure integrating computing power, data, and algorithms, driving AI technology from the laboratory to practical applications and accelerating the commercialization of large-scale models. With the rapid development of artificial intelligence, big data, and scientific computing, intelligent computing centers, as key digital infrastructure, are experiencing explosive growth in scale and energy consumption. Therefore, selecting appropriate configuration methods for the various infrastructure modules of intelligent computing centers is particularly important.

[0003] Current methods for configuring intelligent computing center infrastructure modules often rely on subjective preferences to select configuration strategies, failing to comprehensively consider factors such as energy efficiency, cost, and reliability. This makes it difficult to balance the project's multiple objectives and can easily lead to configuration defects where one aspect is prioritized at the expense of another. Furthermore, current budget allocations for intelligent computing center infrastructure modules are not strongly tied to the established energy efficiency level of the intelligent computing center, but are based solely on the rated power consumption of IT equipment. This can easily result in substandard energy efficiency or wasted resources.

[0004] In view of this, the present invention proposes a modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and achieve the above objectives, the present invention provides the following technical solution: A modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification, characterized in that the configuration method includes: Step 1: Collect key information about the intelligent computing center, and analyze the collected key information to determine the established energy efficiency level of the intelligent computing center; Step 2: Determine the additional budget power consumption of the facilities based on the established energy efficiency level, and allocate the additional budget power consumption of the facilities to each infrastructure module of the intelligent computing center to obtain the budget power consumption target of each infrastructure module. Step 3: Based on the budget power consumption targets of each infrastructure module, generate multiple configuration strategies, and select the optimal configuration strategy for each infrastructure module based on the adaptation score of each configuration strategy. Step 4: Simulate the intelligent computing center based on the optimal configuration strategy, obtain the energy consumption values ​​of each infrastructure module in the simulation, and analyze the energy consumption values ​​to determine whether the selected optimal configuration strategy is appropriate. Step 5: Execute the corresponding response based on the analysis results.

[0006] Furthermore, the method for determining the established energy efficiency level of an intelligent computing center is as follows: Feature extraction is performed on the key information collected from the intelligent computing center to obtain feature labels for multiple decision indicators. Based on the feature labels of the decision indicators, the quantitative values ​​of each decision indicator are obtained. The environmental impact information of the intelligent computing center is obtained. Based on the environmental impact information, an impact factor is generated to correct the quantitative values, and the quantitative correction values ​​of each decision indicator are obtained. Multiple energy efficiency levels are pre-defined, and each energy efficiency level has a set quantitative value standard range for various decision indicators. The quantitative correction value of each decision indicator is compared with its corresponding quantitative value standard range, and the number of qualified decisions for each decision indicator within each energy efficiency level is counted and recorded as the qualified number of decision indicators. The method for determining qualification is: if the quantitative correction value of a decision indicator is within the corresponding quantitative value standard range, it is considered qualified. The energy efficiency level with the most qualified decision indicators is selected as the predetermined energy efficiency level of the intelligent computing center.

[0007] Furthermore, the method for obtaining the quantitative correction values ​​of each decision indicator is as follows: The process involves acquiring environmental impact information from the intelligent computing center, extracting features from this information to obtain parameter values ​​for each environmental impact item, comparing these parameter values ​​with the standard parameter values ​​of the corresponding environmental impact items under standard conditions to obtain the absolute value of the deviation for each environmental impact item, normalizing the absolute value of the deviation to obtain the deviation coefficient for each environmental impact item, weighting and summing the deviation coefficients of all environmental impact items to obtain the impact factor, and multiplying the impact factor by one with the quantitative value of each decision indicator to obtain the quantitative correction value for each decision indicator.

[0008] Furthermore, the method for determining the facility's additional budgeted power consumption is as follows: Obtain the power consumption of each IT device in the intelligent computing center, add up the power consumption of all IT devices to get the total power consumption of the devices, obtain the PUE corresponding to the predetermined energy efficiency level of the intelligent computing center, subtract a constant from the PUE and multiply it by the total power consumption of the devices to obtain the facility additional budget power consumption.

[0009] Furthermore, the method for obtaining the budgeted power consumption targets for each infrastructure module is as follows: Obtain the typical energy consumption percentage of each infrastructure module in the industry report and set basic weights for each infrastructure module; Energy consumption correlation data for each infrastructure module is collected, and features are extracted from the energy consumption correlation data to obtain the actual parameter values ​​of each correlation feature. Each correlation feature corresponds to an industry benchmark value. The actual parameter value is divided by the industry benchmark value to obtain the correction dimension coefficient of each correlation feature. The weight coefficient of each correlation feature is calculated, and the weight coefficient and the correction dimension coefficient are combined to accumulate the correlation features to obtain the comprehensive correction coefficient of each infrastructure module. The comprehensive correction coefficient of each infrastructure module is multiplied by the corresponding basic weight to obtain the initial correction weight of each infrastructure module. The initial correction weight is normalized to obtain the final weight of each infrastructure module. The budgeted power consumption near the facility is multiplied by the final weight of each infrastructure module to obtain the budgeted power consumption target of each infrastructure module.

[0010] Furthermore, the weight coefficients of each associated feature are calculated as follows: Historical energy consumption correlation data of each infrastructure module under different predetermined energy efficiency levels are obtained. Feature extraction is performed on the historical energy consumption correlation data to obtain historical parameter values ​​of each correlation feature under each predetermined energy efficiency level. The historical parameter values ​​of each correlation feature are standardized to obtain historical quantitative parameter values ​​of each correlation feature. The associated features are combined in pairs to obtain multiple associated feature combinations. Based on the historical quantification parameter values ​​of each associated feature, the correlation coefficient of each associated feature combination is calculated to obtain the correlation coefficient of each associated feature combination under each given energy efficiency level. The mean of the correlation coefficients of each associated feature combination under all given energy efficiency levels is calculated and denoted as the mean correlation coefficient. The feature matrix of associated feature combinations is constructed using the mean correlation coefficient as the element. Calculate the sum of each row in the feature matrix to obtain the first weight value of each associated feature. Calculate the sum of each column in the feature matrix to obtain the second weight value of each associated feature. Add the first weight value and the second weight value to obtain the weight value of each associated feature. Add the weight values ​​of each associated feature to obtain the total weight value. Calculate the ratio of the weight value of each associated feature to the total weight value to obtain the weight coefficient of each associated feature.

[0011] Furthermore, the method for selecting the optimal configuration strategy for each infrastructure module is as follows: Based on the configurable options for each key step within the infrastructure module, the various configuration options for each key step are combined in the order of the steps to obtain multiple configuration strategies.

[0012] Calculate the adaptation score for each configuration strategy, sort the configuration strategies from highest to lowest according to the adaptation score to obtain a configuration strategy ranking table, and select the configuration strategy with the highest ranking in the configuration strategy ranking table for configuration.

[0013] Furthermore, the adaptation score calculation method for each configuration strategy is as follows: Obtain positive and negative influencing factor data for each configuration strategy, normalize the positive and negative influencing factor data to obtain the positive impact value of each positive influencing factor and the negative impact value of each negative influencing factor. The first impact value is obtained by weighting and summing the positive impact values, and the second impact value is obtained by weighting and summing the negative impact values. The first impact value is divided by the second impact value to obtain the adaptation score of each configuration strategy.

[0014] Furthermore, the method for determining whether the selected optimal configuration strategy is appropriate is as follows: In the simulation, the energy consumption values ​​of each infrastructure module under time series are obtained, and the standard deviation of the energy consumption values ​​is calculated. An exponential function value is constructed with the natural constant e as the base and the standard deviation as the exponent. Simultaneously, based on the energy consumption values ​​obtained under time series, a real-time energy consumption value change function is constructed, and an ideal energy consumption value change function under time series is pre-constructed. The integral difference between the ideal energy consumption value change function and the real-time energy consumption value change function under time series is calculated, and the integral difference is normalized to obtain the deviation coefficient. The deviation coefficient is multiplied by the exponential function value to obtain the judgment coefficient. When the judgment coefficient is greater than the preset judgment coefficient threshold, the selected optimal configuration strategy is judged to be inappropriate; when the judgment coefficient is less than or equal to the preset judgment coefficient threshold, the selected optimal configuration strategy is judged to be appropriate.

[0015] Furthermore, the method for executing the corresponding response based on the analysis results is as follows: When it is determined that the optimal configuration strategy is appropriate, select that configuration strategy for configuration; When it is determined that the selected optimal configuration strategy is not suitable, iterative simulation is performed on each configuration strategy according to the order of arrangement in the configuration strategy sorting table until a suitable configuration strategy is selected.

[0016] The technical effects and advantages of this invention's modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification are as follows: This invention combines the correction dimension coefficient and the calculated weight coefficient for comprehensive analysis to correct the basic weight set according to the typical energy consumption ratio of the industry. This makes the allocation of budget power consumption targets for infrastructure modules more closely aligned with the personalized needs of project equipment, environment, and load, and makes the allocation of budget power consumption targets more accurate. This invention selects appropriate configuration strategies by combining key step selection with comprehensive analysis of multi-dimensional influencing factors. It breaks away from the subjective preferences of traditional experience-based decision-making, avoids excessive reliance on a single indicator, and can balance multi-dimensional requirements such as energy efficiency, cost, reliability, and power consumption. It not only ensures the established energy efficiency level of the intelligent computing center, but also takes into account long-term operating costs and environmental adaptability. Attached Figure Description

[0017] Figure 1 This is a flowchart of the modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0019] This embodiment discloses a modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification, such as... Figure 1 As shown, the configuration methods mainly include: Step 1: Collect key information about the intelligent computing center, and analyze the collected key information to determine the established energy efficiency level of the intelligent computing center; Step 2: Determine the additional budget power consumption of the facilities based on the established energy efficiency level, and allocate the additional budget power consumption of the facilities to each infrastructure module of the intelligent computing center to obtain the budget power consumption target of each infrastructure module. Step 3: Based on the budget power consumption targets of each infrastructure module, generate multiple configuration strategies, and select the optimal configuration strategy for each infrastructure module based on the adaptation score of each configuration strategy. Step 4: Simulate the intelligent computing center based on the optimal configuration strategy, obtain the energy consumption values ​​of each infrastructure module in the simulation, and analyze the energy consumption values ​​to determine whether the selected optimal configuration strategy is appropriate. Step 5: Execute the corresponding response based on the analysis results.

[0020] Through the above technical solution, this application first collects key information of the intelligent computing center, including business requirements and power consumption information of IT equipment. Features are extracted from the collected key information to obtain quantitative values ​​for each decision indicator. Then, the quantitative values ​​of each decision indicator are corrected based on the environmental impact of the intelligent computing center. The predetermined energy efficiency level of the intelligent computing center is determined based on the corrected quantitative values. Next, the additional budget power consumption for the facilities is determined based on the predetermined energy efficiency level. Based on the energy consumption correlation data analysis of each infrastructure module, the additional budget power consumption is allocated to each infrastructure module. This allows for a more precise allocation of budget power consumption targets to infrastructure modules, better aligning with the personalized needs of project equipment, environment, and load. After allocating budget power consumption targets to each infrastructure module, the feasibility of each key step within the infrastructure module is further refined. The configuration selection process combines multiple configuration strategies, and the optimal strategy is chosen for each infrastructure module based on the adaptability score of each strategy. This approach breaks away from the subjective preferences of traditional experience-based decision-making, avoids over-reliance on a single indicator, and balances multiple dimensions such as energy efficiency, cost, reliability, and power consumption. It ensures the intelligent computing center's predetermined energy efficiency level while also considering long-term operating costs and environmental adaptability. The intelligent computing center is then simulated based on the optimal configuration strategy to obtain the energy consumption values ​​of each infrastructure module in the simulation. Analysis of these energy consumption values ​​determines whether the selected optimal configuration strategy is suitable for the current infrastructure module. If suitable, the configuration strategy is adopted; otherwise, a new configuration strategy is selected and iterative simulations are performed until a suitable strategy is found, thus completing the modular configuration of the intelligent computing center's infrastructure.

[0021] The method for determining the predetermined energy efficiency level of the intelligent computing center is as follows: Feature extraction is performed on the collected key information of the intelligent computing center to obtain feature labels for multiple decision indicators. Based on these feature labels, the quantitative values ​​of each decision indicator are obtained. Environmental impact information of the intelligent computing center is acquired, and impact factors are generated based on this information to correct the quantitative values, resulting in corrected quantitative values ​​for each decision indicator. Multiple energy efficiency levels are pre-defined, with standard ranges for the quantitative values ​​of each decision indicator set for each level. The corrected quantitative values ​​of each decision indicator are compared with their corresponding standard ranges, and the number of qualified indicators within each energy efficiency level is counted, recorded as the qualified quantity of the decision indicator. The qualification method is as follows: if the corrected quantitative value of a decision indicator falls within its corresponding standard range, it is considered qualified. The energy efficiency level with the highest number of qualified decision indicators is selected as the predetermined energy efficiency level of the intelligent computing center. The method for obtaining the quantitative correction values ​​of each decision indicator is as follows: obtain the environmental impact information of the intelligent computing center, extract features from the environmental impact information to obtain the parameter values ​​of each environmental impact item, compare the parameter values ​​of each environmental impact item with the standard parameter values ​​of the corresponding environmental impact item under the standard environment to obtain the absolute value of the deviation of each environmental impact item, normalize the absolute value of the deviation to obtain the deviation coefficient of each environmental impact item, weight and sum the deviation coefficients of all environmental impact items to obtain the impact factor, add one to the impact factor and multiply it by the quantitative value of each decision indicator to obtain the quantitative correction value of each decision indicator.

[0022] The above scheme provides a specific method for determining the predetermined energy efficiency level of an intelligent computing center. First, the predetermined energy efficiency level of the intelligent computing center is determined based on key information, such as the business requirements of the intelligent computing center and the power consumption information of the IT equipment. Then, features are extracted from the key information to obtain feature labels for multiple decision indicators. Based on the feature labels of the decision indicators, the quantitative values ​​of each decision indicator are obtained. The decision indicators can be computing power density, power supply efficiency, renewable resource ratio, cooling COP, etc., and the feature labels are used to indicate the key attributes of each decision indicator.For example, the feature label for computing power density is the corresponding rack load power; the feature label for power supply efficiency is the corresponding power supply efficiency; the feature label for renewable resource proportion is the proportion of green energy; and the feature label for cooling COP is the energy efficiency ratio of the cooling module. Finally, these feature labels are uniformly quantified to obtain their respective quantified values ​​for convenient subsequent calculation and analysis. However, various decision indicators are affected by the actual environment, which can lead to inaccurate quantified values. Therefore, it is necessary to obtain environmental impact information for the intelligent computing center, including information such as temperature, humidity, and wind speed. Feature extraction is then performed on this environmental impact information to obtain various environmental impact items. The parameter values ​​correspond to environmental impact items such as annual average temperature, annual average wind speed, and annual average relative humidity. The parameter values ​​for each environmental impact item are compared with the standard parameter values ​​for the corresponding environmental impact item under a standard environment to obtain the absolute value of the deviation for each environmental impact item. The standard environment and the standard parameter values ​​for each environmental impact item under the standard environment can be determined based on the experience and professional knowledge of personnel in this field. The larger the absolute value of the deviation, the greater the difference from the standard environmental impact item, and therefore the greater the impact. The absolute values ​​of the deviations are normalized to obtain the deviation coefficients for each environmental impact item. The deviation coefficients of all environmental impact items are then weighted and summed. The impact factor is obtained; then, the impact factor is incremented by one and multiplied by the quantified value of each decision indicator to obtain the quantified correction value of each decision indicator. This allows for adjustments to the quantified values ​​of each decision indicator based on environmental impact, making the data more closely reflect the actual situation of the intelligent computing center. Multiple energy efficiency levels are pre-defined, with standard ranges for the quantified values ​​of each decision indicator set under each energy efficiency level. The energy consumption levels are set with reference to the "Data Center Energy Efficiency Limits and Energy Efficiency Levels". Under each energy efficiency level, standard ranges for the quantified values ​​of each decision indicator are pre-set based on empirical data. The quantified correction value of each decision indicator is then multiplied by its corresponding quantified value. The standard range is compared, and the number of each decision indicator that meets the requirements within each energy efficiency level is counted and recorded as the number of qualified decision indicators. The method for determining qualification is: if the quantitative correction value of the decision indicator is within the corresponding quantitative value standard range, it is considered qualified. The energy efficiency level with the most qualified decision indicators is selected as the predetermined energy efficiency level of the intelligent computing center. The more qualified decision indicators there are, the more suitable the energy efficiency level is for the current intelligent computing center, and the energy efficiency level is selected as the predetermined energy efficiency level of the intelligent computing center. In this way, the misjudgment of energy efficiency level caused by regional environmental differences can be avoided, and the determination of the energy efficiency level of the intelligent computing center is more in line with the actual energy efficiency performance.

[0023] The method for determining the facility additional budget power consumption is as follows: obtain the power consumption of each IT device in the intelligent computing center, add up the power consumption of all IT devices to obtain the total power consumption of the devices, obtain the PUE corresponding to the predetermined energy efficiency level of the intelligent computing center, subtract a constant from the PUE and multiply it by the total power consumption of the devices to obtain the facility additional budget power consumption. The method for obtaining the budgeted power consumption targets for each infrastructure module is as follows: Obtain the typical energy consumption ratio of each infrastructure module from industry reports and set basic weights for each infrastructure module; collect relevant energy consumption correlation data for each infrastructure module, extract features from the energy consumption correlation data to obtain the actual parameter values ​​of each correlation feature (each correlation feature corresponds to an industry benchmark value), divide the actual parameter values ​​by the industry benchmark value to obtain the corrected dimension coefficient of each correlation feature; calculate the weight coefficient of each correlation feature, and sum the weight coefficient and the corrected dimension coefficient to obtain the comprehensive correction coefficient of each infrastructure module; multiply the comprehensive correction coefficient of each infrastructure module by the corresponding basic weight to obtain the initial corrected weight of each infrastructure module; normalize the initial corrected weight to obtain the final weight of each infrastructure module; multiply the budgeted power consumption near the facility by the final weight of each infrastructure module to obtain the budgeted power consumption target of each infrastructure module; the method for calculating the weight coefficient of each correlation feature is as follows: obtain the historical energy consumption of each infrastructure module under different predetermined energy efficiency levels. Historical energy consumption correlation data is used to extract features from historical energy consumption correlation data to obtain historical parameter values ​​of each correlation feature under each given energy efficiency level. These historical parameter values ​​are then standardized to obtain historical quantified parameter values. Correlation features are paired to obtain multiple correlation feature combinations. Based on the historical quantified parameter values ​​of each correlation feature, the correlation coefficient of each correlation feature combination under each given energy efficiency level is calculated. The mean of the correlation coefficients of all correlation feature combinations under all given energy efficiency levels is calculated and denoted as the mean correlation coefficient. A feature matrix of correlation feature combinations is constructed using the mean correlation coefficient as elements. The sum of each row in the feature matrix is ​​calculated to obtain the first weight value of each correlation feature, and the sum of each column in the feature matrix is ​​calculated to obtain the second weight value of each correlation feature. The first and second weight values ​​are added together to obtain the weight value of each correlation feature. Finally, the weight values ​​of all correlation features are added together to obtain the total weight value. The ratio of the weight value of each correlation feature to the total weight value is calculated to obtain the weight coefficient of each correlation feature.

[0024] The above scheme provides a specific method for allocating budgeted power consumption targets to each infrastructure module. First, the power consumption of each IT device in the intelligent computing center is obtained. The power consumption of all IT devices is summed to obtain the total power consumption. Then, the PUE corresponding to the predetermined energy efficiency level of the intelligent computing center is obtained. The PUE is subtracted by a constant and multiplied by the total power consumption to obtain the facility-added budgeted power consumption, which is represented as the total budgeted power consumption of all infrastructure modules. Next, the total budgeted power consumption is allocated to each infrastructure module. Specifically, the typical energy consumption ratio of each infrastructure module is obtained from industry reports in the corresponding field, and a basic weight is set for each infrastructure module. Finally, relevant energy consumption correlation data for each infrastructure module is collected. Based on data including battery loss rate and HVDC conversion efficiency of power supply infrastructure modules, and cooling COP and natural cooling time percentage of cooling infrastructure modules, features are extracted from energy consumption-related data to obtain the actual parameter values ​​of each corresponding related feature. The related features are the energy consumption-related features extracted from the corresponding data. For example, the related features for power supply infrastructure modules are battery loss rate and HVDC conversion efficiency, while the related features for cooling infrastructure modules are cooling COP and natural cooling time percentage. Each related feature corresponds to an industry benchmark value. Dividing the actual parameter value by the industry benchmark value yields the corrected dimension coefficient for each related feature. The weight coefficient for each related feature is calculated, and the weight coefficient and corrected dimension coefficient are combined to adjust the values ​​of each related feature. The associated features are accumulated to obtain the comprehensive correction coefficient for each infrastructure module. The comprehensive correction coefficient of each infrastructure module is multiplied by its corresponding basic weight to obtain the initial correction weight for each infrastructure module. The initial correction weights are normalized to obtain the final weight for each infrastructure module. The budgeted power consumption near the facility is multiplied by the final weight for each infrastructure module to obtain the budgeted power consumption target for each infrastructure module. Since there may be signal interference between different associated features, in order to accurately obtain the weight coefficients of each associated feature, historical energy consumption correlation data of each infrastructure module under different given energy efficiency levels are used. Feature extraction is performed on the historical energy consumption correlation data to obtain the weight coefficients for each infrastructure module under each given energy efficiency level. Historical parameter values ​​of associated features are obtained; these historical parameter values ​​are standardized to obtain historical quantified parameter values ​​for each associated feature; associated features are paired to obtain multiple associated feature combinations; based on the historical quantified parameter values ​​of each associated feature, the correlation coefficients of each associated feature combination are calculated to obtain the correlation coefficients of each associated feature combination under each given energy efficiency level; the mean of the correlation coefficients of each associated feature combination under all given energy efficiency levels is calculated and denoted as the mean correlation coefficient; this method of collecting the correlation coefficients of each associated feature under all energy efficiency levels and calculating the mean value better reflects the mutual influence between different associated features; a feature matrix of associated feature combinations is constructed using the mean correlation coefficient as elements.The sum of each row in the feature matrix is ​​calculated to obtain the first weight value of each associated feature. The sum of each column in the feature matrix is ​​calculated to obtain the second weight value of each associated feature. The first weight value and the second weight value are added together to obtain the weight value of each associated feature. The first weight value indicates the influence of the associated feature, and the second weight value indicates the influence of the associated feature. Combining the two provides a better representation of the mutual influence between associated features. The total weight value is obtained by adding the weight values ​​of each associated feature. The ratio of the weight value of each associated feature to the total weight value is calculated to obtain the weight coefficient of each associated feature. By combining this method with the corrected dimensionality coefficient and the calculated weight coefficient for comprehensive analysis, the basic weights set according to the typical energy consumption ratio of the industry can be corrected. This allows for a more accurate allocation of budget power consumption targets for infrastructure modules, better aligning with the personalized needs of project equipment, environment, and load.

[0025] The method for selecting the optimal configuration strategy for each infrastructure module is as follows: Based on the configurable options for each key step within the infrastructure module, the configuration options for each key step are combined in sequence to obtain multiple configuration strategies; the adaptation score of each configuration strategy is calculated, and the configuration strategies are sorted from largest to smallest according to their adaptation scores to obtain a configuration strategy ranking table; the configuration strategy ranked highest in the ranking table is selected for configuration; the adaptation score of each configuration strategy is calculated as follows: the positive and negative influencing factor data of each configuration strategy are obtained, and the positive and negative influencing factor data are normalized to obtain the positive influence value of each positive influence factor and the negative influence value of each negative influence factor; the positive influence values ​​are weighted and accumulated to obtain the first influence value, and the negative influence values ​​are weighted and accumulated to obtain the second influence value; the first influence value is divided by the second influence value to obtain the adaptation score of each configuration strategy.

[0026] The above technical solution provides a specific method for selecting the optimal configuration strategy for each infrastructure module. First, based on the configurable options for each key step within the infrastructure module, the configuration options for each key step are combined sequentially to obtain multiple configuration strategies. For example, the cooling infrastructure module is divided into three key steps: cooling method, cooling mode, and control strategy. The cooling method includes three configuration options: cold plate liquid cooling, air-cooled precision air conditioning, and immersion liquid cooling. The cooling mode includes three configuration options: natural cooling, mechanical cooling, and hybrid cooling. The control strategy includes three configuration options: dynamic control, timed control, and manual control. These are arranged and combined according to the step sequence to obtain multiple configuration strategies. Then, the positive and negative influencing factor data for each configuration strategy are obtained. The positive and negative influencing factor data are normalized to obtain the positive influence value of each positive influencing factor and the negative influence value of each negative influencing factor. The weighted sum of the positive influence values ​​yields... The first influence value is obtained by weighting and accumulating all negative influence values ​​to get the second influence value. The first influence value is then divided by the second influence value to obtain the adaptation score of each configuration strategy. Positive factors indicate that the higher the value, the better, such as cooling energy efficiency and scalability. Negative factors indicate that the lower the value, the better, such as budget cost, failure rate, and noise figure. Thus, the larger the first influence value and the smaller the second influence value, the better the configuration strategy. Therefore, the higher the adaptation score of the configuration strategy, the better the configuration strategy. So, the configuration strategies are sorted from largest to smallest according to their adaptation scores to obtain a configuration strategy ranking table. The configuration strategy ranked first in the ranking table is selected for configuration. By adopting a combination of key steps and multi-dimensional influencing factors, a suitable configuration strategy is selected. This breaks away from the subjective preferences of traditional experience-based decision-making and avoids over-reliance on a single indicator. It can balance the multi-dimensional requirements of energy efficiency, cost, reliability, and power consumption, ensuring the established energy efficiency level of the intelligent computing center while taking into account long-term operating costs and environmental adaptability.

[0027] The method for determining whether the selected optimal configuration strategy is appropriate is as follows: In the simulation, the energy consumption values ​​of each infrastructure module under the time series are obtained, the standard deviation of the energy consumption values ​​is calculated, and an exponential function value is constructed with the natural constant e as the base and the standard deviation as the exponent. Simultaneously, based on the energy consumption values ​​obtained under the time series, a real-time energy consumption value change function is constructed, and an ideal energy consumption value change function under the time series is pre-constructed. The integral of the difference between the ideal energy consumption value change function and the real-time energy consumption value change function under the time series is calculated, and the integral is normalized to obtain the deviation coefficient. The deviation coefficient is multiplied by the exponential function value to obtain the judgment coefficient. When the judgment coefficient is greater than a preset judgment coefficient threshold, the selected optimal configuration strategy is deemed inappropriate; when the judgment coefficient is less than or equal to the preset judgment coefficient threshold, the selected optimal configuration strategy is deemed appropriate. The method for executing the corresponding response based on the analysis results is as follows: When the selected optimal configuration strategy is deemed appropriate, it is selected for configuration; when the selected optimal configuration strategy is deemed inappropriate, iterative simulations are performed on each configuration strategy according to the order in the configuration strategy ranking table until a suitable configuration strategy is selected.

[0028] The above scheme provides a specific method for determining whether the selected optimal configuration strategy is appropriate. Since the configuration strategy used is obtained under ideal conditions, in order to determine whether the selected optimal configuration strategy is suitable for the current intelligent computing center, this embodiment simulates the intelligent computing center according to the obtained optimal configuration strategy and obtains the changes in energy consumption values ​​of each infrastructure module in the simulation. Here, the energy consumption value is the total energy consumption value of each infrastructure module; the energy consumption value of each infrastructure module under the time series is obtained, the standard deviation of the energy consumption value is calculated, and an exponential function value is constructed with the natural constant e as the base and the standard deviation as the exponent. The difference represents the fluctuation of energy consumption values; the smaller the value, the smaller the fluctuation, and the more stable the system. Constructing an exponential function can amplify the impact of this indicator. Based on real-time energy consumption values, a real-time energy consumption value change function is constructed, and an ideal energy consumption value change function under time series is pre-constructed. The ideal energy consumption value change function can be determined based on the energy consumption limit corresponding to the established energy efficiency level of the intelligent computing center, the rated power consumption of infrastructure modules, and typical energy consumption data under industry benchmark operating conditions. It is used to represent the energy consumption of infrastructure modules under ideal conditions and is an existing technology. The ideal energy consumption value change function and the real-time energy consumption value change function are calculated in time series. The difference integral listed below is normalized to obtain the deviation coefficient. The smaller the difference, the smaller the gap between the energy consumption and the ideal state, and the better the configuration strategy. The deviation coefficient is multiplied by the exponential function value to obtain the judgment coefficient. Those skilled in the art preset a judgment coefficient threshold based on experience and professional knowledge, and compare the judgment coefficient with the judgment coefficient threshold. When the judgment coefficient is greater than the preset judgment coefficient threshold, the selected optimal configuration strategy is judged to be unsuitable and needs to be changed. When the judgment coefficient is less than or equal to the preset judgment coefficient threshold, the selected optimal configuration strategy is judged to be suitable. Finally, the corresponding response is executed according to the analysis results. Specifically: when the selected optimal configuration strategy is judged to be suitable, the configuration strategy is selected and configured. When the selected optimal configuration strategy is judged to be unsuitable, the various configuration strategies are iteratively simulated according to the order of the configuration strategy sorting table until a suitable configuration strategy is selected. This embodiment analyzes the changes in energy consumption value during the simulation of the configuration strategy, which can simultaneously capture the degree of energy consumption fluctuation and the degree of deviation from the ideal target, thereby more accurately judging whether the configuration strategy is suitable for the current infrastructure module, and making the judgment result more in line with the dynamic characteristics of actual operation.

[0029] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0030] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0031] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification, characterized in that, Configuration methods include: Step 1: Collect key information about the intelligent computing center, and analyze the collected key information to determine the established energy efficiency level of the intelligent computing center; Step 2: Determine the additional budget power consumption of the facilities based on the established energy efficiency level, and allocate the additional budget power consumption of the facilities to each infrastructure module of the intelligent computing center to obtain the budget power consumption target of each infrastructure module. Step 3: Based on the budget power consumption targets of each infrastructure module, generate multiple configuration strategies, and select the optimal configuration strategy for each infrastructure module based on the adaptation score of each configuration strategy. Step 4: Simulate the intelligent computing center based on the optimal configuration strategy, obtain the energy consumption values ​​of each infrastructure module in the simulation, and analyze the energy consumption values ​​to determine whether the selected optimal configuration strategy is appropriate. Step 5: Execute the corresponding response based on the analysis results.

2. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 1, characterized in that, The method for determining the predetermined energy efficiency level of a smart computing center is as follows: Feature extraction is performed on the key information collected from the intelligent computing center to obtain feature labels for multiple decision indicators. Based on the feature labels of the decision indicators, the quantitative values ​​of each decision indicator are obtained. The environmental impact information of the intelligent computing center is obtained. Based on the environmental impact information, an impact factor is generated to correct the quantitative values, and the quantitative correction values ​​of each decision indicator are obtained. Multiple energy efficiency levels are pre-defined, and each energy efficiency level has a set quantitative value standard range for various decision indicators. The quantitative correction value of each decision indicator is compared with its corresponding quantitative value standard range, and the number of qualified decisions for each decision indicator within each energy efficiency level is counted and recorded as the qualified number of decision indicators. The method for determining qualification is: if the quantitative correction value of a decision indicator is within the corresponding quantitative value standard range, it is considered qualified. The energy efficiency level with the most qualified decision indicators is selected as the predetermined energy efficiency level of the intelligent computing center.

3. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 2, characterized in that, The method for obtaining the quantitative correction values ​​of each decision indicator is as follows: The process involves acquiring environmental impact information from the intelligent computing center, extracting features from this information to obtain parameter values ​​for each environmental impact item, comparing these parameter values ​​with the standard parameter values ​​of the corresponding environmental impact items under standard conditions to obtain the absolute value of the deviation for each environmental impact item, normalizing the absolute value of the deviation to obtain the deviation coefficient for each environmental impact item, weighting and summing the deviation coefficients of all environmental impact items to obtain the impact factor, and multiplying the impact factor by one with the quantitative value of each decision indicator to obtain the quantitative correction value for each decision indicator.

4. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 1, characterized in that, The method for determining the facility's additional budget power consumption is as follows: Obtain the power consumption of each IT device in the intelligent computing center, add up the power consumption of all IT devices to get the total power consumption of the devices, obtain the PUE corresponding to the predetermined energy efficiency level of the intelligent computing center, subtract a constant from the PUE and multiply it by the total power consumption of the devices to obtain the facility additional budget power consumption.

5. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 4, characterized in that, The method for obtaining the budgeted power consumption targets for each infrastructure module is as follows: Obtain the typical energy consumption percentage of each infrastructure module in the industry report and set basic weights for each infrastructure module; Energy consumption correlation data of each infrastructure module is collected, and features are extracted from the energy consumption correlation data to obtain the actual parameter values ​​of each correlation feature. Each correlation feature corresponds to an industry benchmark value. The actual parameter value is divided by the industry benchmark value to obtain the correction dimension coefficient of each correlation feature. The weight coefficient of each correlation feature is calculated, and the weight coefficient and the correction dimension coefficient are combined to accumulate the correlation features to obtain the comprehensive correction coefficient of each infrastructure module. The comprehensive correction coefficient of each infrastructure module is multiplied by the corresponding basic weight to obtain the primary correction weight of each infrastructure module. The initial corrected weights are normalized to obtain the final weights of each infrastructure module. The budgeted power consumption near the facility is multiplied by the final weight of each infrastructure module to obtain the budgeted power consumption target for each infrastructure module.

6. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 5, characterized in that, The weight coefficients of each associated feature are calculated as follows: Historical energy consumption correlation data of each infrastructure module under different predetermined energy efficiency levels are obtained. Feature extraction is performed on the historical energy consumption correlation data to obtain historical parameter values ​​of each correlation feature under each predetermined energy efficiency level. The historical parameter values ​​of each correlation feature are standardized to obtain historical quantitative parameter values ​​of each correlation feature. The associated features are combined in pairs to obtain multiple associated feature combinations. Based on the historical quantification parameter values ​​of each associated feature, the correlation coefficient of each associated feature combination is calculated to obtain the correlation coefficient of each associated feature combination under each given energy efficiency level. The mean of the correlation coefficients of each associated feature combination under all given energy efficiency levels is calculated and denoted as the mean correlation coefficient. The feature matrix of associated feature combinations is constructed using the mean correlation coefficient as the element. Calculate the sum of each row in the feature matrix to obtain the first weight value of each associated feature. Calculate the sum of each column in the feature matrix to obtain the second weight value of each associated feature. Add the first weight value and the second weight value to obtain the weight value of each associated feature. Add the weight values ​​of each associated feature to obtain the total weight value. Calculate the ratio of the weight value of each associated feature to the total weight value to obtain the weight coefficient of each associated feature.

7. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 1, characterized in that, The method for selecting the optimal configuration strategy for each infrastructure module is as follows: Based on the configurable options for each key step within the infrastructure module, the various configuration options for each key step are combined in the order of the steps to obtain multiple configuration strategies. Calculate the adaptation score for each configuration strategy, sort the configuration strategies from highest to lowest according to the adaptation score to obtain a configuration strategy ranking table, and select the configuration strategy with the highest ranking in the configuration strategy ranking table for configuration.

8. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 7, characterized in that, The method for calculating the adaptation score of each configuration strategy is as follows: Obtain positive and negative influencing factor data for each configuration strategy, normalize the positive and negative influencing factor data to obtain the positive impact value of each positive influencing factor and the negative impact value of each negative influencing factor. The first impact value is obtained by weighting and summing the positive impact values, and the second impact value is obtained by weighting and summing the negative impact values. The first impact value is divided by the second impact value to obtain the adaptation score of each configuration strategy.

9. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 1, characterized in that, The method to determine whether the selected optimal configuration strategy is appropriate is as follows: In the simulation, the energy consumption values ​​of each infrastructure module under the time series are obtained, the standard deviation of the energy consumption values ​​is calculated, and an exponential function value is constructed with the natural constant e as the base and the standard deviation as the exponent. At the same time, based on the energy consumption values ​​obtained under the time series, a real-time energy consumption value change function is constructed, an ideal energy consumption value change function under the time series is pre-constructed, and the integral of the difference between the ideal energy consumption value change function and the real-time energy consumption value change function under the time series is calculated. The integral of the difference is normalized to obtain the deviation coefficient. The deviation coefficient is multiplied by the exponential function value to obtain the judgment coefficient. When the judgment coefficient is greater than the preset judgment coefficient threshold, the selected optimal configuration strategy is judged to be inappropriate; when the judgment coefficient is less than or equal to the preset judgment coefficient threshold, the selected optimal configuration strategy is judged to be appropriate.

10. The modular configuration method for intelligent computing center infrastructure based on energy efficiency level classification according to claim 9, characterized in that, The method for executing the corresponding response based on the analysis results is as follows: When it is determined that the optimal configuration strategy is appropriate, select that configuration strategy for configuration; When it is determined that the selected optimal configuration strategy is not suitable, iterative simulation is performed on each configuration strategy according to the order of arrangement in the configuration strategy sorting table until a suitable configuration strategy is selected.

Images