Wind-liquid homologous data center energy efficiency optimization method and system

Abstract

The invention relates to the technical field of data processing, in particular to an energy efficiency optimization method and system for a wind-liquid homologous data center, and the method comprises the steps: obtaining an acoustic spectrum signal during the operation of a cooling system through a sensor, and obtaining an original acoustic spectrum data set; performing frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; acoustic features are extracted according to the frequency band distribution feature set, and the system energy consumption state is determined in combination with a pre-established mapping model; judging the mode switching requirement of the cooling system; predicting an acoustic characteristic change trend according to a mode switching demand, and determining a mode switching time point; adjusting operation parameters of the cooling system according to the mode switching time point, and executing mode switching to obtain an updated system operation state; and updating the mapping model according to the updated system operation state to obtain an optimized energy consumption state prediction model. According to the invention, the problem of low precision of energy efficiency optimization of the data center is solved, and the precision of energy efficiency optimization is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and system for optimizing energy efficiency of a data center with air and liquid sources. Background Art

[0002] With the rapid development of information technology, the scale of facilities such as data centers and server farms has gradually expanded, generating a large number of computing tasks, which places higher demands on cooling systems. The cooling system plays a vital role in data centers. It not only ensures the normal operation of equipment but also directly affects the system's energy efficiency and energy conservation. However, with fluctuations in equipment load and changes in environmental conditions, traditional cooling methods cannot perceive the system's operating status in real time, making it difficult to accurately optimize energy efficiency, resulting in significant energy waste.

[0003] Traditional cooling methods rely primarily on preset cooling modes, such as air cooling and liquid cooling, which often fail to make timely adjustments based on real-time load conditions and system temperatures. Air cooling and liquid cooling each have their advantages, but in complex environments like data centers with drastic load fluctuations, the application of a single cooling mode often fails to achieve the optimal energy efficiency balance. Furthermore, most existing cooling systems fail to effectively utilize signals generated during system operation, such as acoustic signals, for real-time monitoring and adjustment. These cooling systems are often unable to accurately predict and determine the optimal timing for mode switching, resulting in unstable system energy efficiency and energy waste.

[0004] At the same time, acoustic signals, as a byproduct of system operation, contain a wealth of useful operational information. By analyzing the acoustic signals generated by changes in fan speed and liquid flow, rich characteristics of the system's operating status can be obtained. However, extracting effective features from this complex acoustic data and accurately mapping them to energy consumption status remain significant technical challenges. Existing technologies in this area are still in their early stages of research and application and are unable to effectively address the issues of intelligent cooling system regulation and energy efficiency optimization.

[0005] Therefore, how to adjust the operating status of the cooling system through precise real-time monitoring and intelligent algorithms, realize intelligent switching between air cooling and liquid cooling modes, avoid energy waste, and improve the overall energy efficiency of the system has become a technical challenge that needs to be solved urgently. To address the above problems, the present invention proposes an optimization method for air-liquid homogenous cooling system based on acoustic signal analysis. By collecting acoustic data in real time through sensors and combining intelligent algorithms to dynamically adjust the operating status of the cooling system, the cooling system can be intelligently controlled and efficiently managed. Summary of the Invention

[0006] The present invention provides a method and system for optimizing the energy efficiency of a data center with air and liquid co-sourced, which is used to achieve accurate monitoring, flexible adjustment, and continuous optimization of the cooling system. In a first aspect, the present invention provides a method for optimizing the energy efficiency of a data center with air and liquid co-sourced, comprising: Step S1: Acquiring an acoustic spectrum signal of the cooling system during operation through a sensor to obtain an original acoustic spectrum data set; performing frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; Step S2: extracting acoustic features based on the frequency band distribution feature set, determining the system energy consumption state in combination with a pre-established mapping model, and judging the need for cooling system mode switching based on the relationship between the system energy consumption state and a preset threshold; Step S3: predicting the acoustic feature change trend according to the mode switching demand and determining the mode switching time point; Step S4: adjusting the cooling system operating parameters according to the mode switching time point, executing the mode switching, and obtaining an updated system operating state; Step S5: updating the mapping model according to the updated system operation status to obtain an optimized energy consumption status prediction model.

[0007] As a preferred technical solution of the present invention, the method of acquiring the acoustic spectrum signal of the cooling system during operation by a sensor to obtain the original acoustic spectrum data set includes: An acoustic sensor is used to collect an audio signal generated by a change in fan speed and an acoustic wave signal generated by liquid flow; the audio signal and the acoustic wave signal are continuously sampled to generate acoustic data including multiple time points; the acoustic data is preprocessed to filter out environmental noise to obtain the original acoustic spectrum data set.

[0008] As a preferred technical solution of the present invention, the frequency domain conversion of the original acoustic spectrum data set to obtain a frequency band distribution feature set includes: The original acoustic spectrum data set is converted into a frequency domain signal by using a fast Fourier transform algorithm; the frequency domain signal is discretized into multiple frequency bands to generate a feature vector for each frequency band; and the frequency band distribution feature set including a low frequency band and a high frequency band is generated based on the feature vector.

[0009] As a preferred technical solution of the present invention, extracting acoustic features based on the frequency band distribution feature set and determining the system energy consumption state in combination with a pre-established mapping model includes: Extract low-frequency band features corresponding to fan speed changes and high-frequency band features corresponding to liquid flow from the frequency band distribution feature set; input the low-frequency band features and the high-frequency band features into the pre-established acoustic feature and energy consumption status mapping model; and determine the current system energy consumption status based on the output of the mapping model.

[0010] As a preferred technical solution of the present invention, judging the need for switching the cooling system mode according to the relationship between the system energy consumption state and a preset threshold value includes: If the system energy consumption state exceeds the preset energy consumption threshold, the frequency band distribution feature set is classified by the support vector machine algorithm; based on the classification result, it is determined whether the cooling system is in a critical state of air cooling or liquid cooling mode; and the mode switching requirement is generated based on the critical state.

[0011] As a preferred technical solution of the present invention, predicting the acoustic feature change trend according to the mode switching demand and determining the mode switching time point includes: According to the mode switching requirements, the dynamic changes of the acoustic features in the frequency band distribution feature set are analyzed; the dynamic changes of the acoustic features are modeled using a long short-term memory network algorithm; and the acoustic feature change trend in the future is predicted based on the modeling results to generate the mode switching time point.

[0012] As a preferred technical solution of the present invention, adjusting the cooling system operating parameters according to the mode switching time point, performing the mode switching, and obtaining the updated system operating state includes: According to the mode switching time point combined with load monitoring information and temperature sensor data, the optimal switching parameters of the air cooling and liquid cooling modes are determined; the operating parameters of the cooling system are adjusted through a preset control logic; the switching between the air cooling and liquid cooling modes is executed to generate the updated system operating status.

[0013] As a preferred technical solution of the present invention, the updating of the mapping model according to the updated system operation state to obtain an optimized energy consumption state prediction model includes: Acquire the acoustic spectrum signal under the updated system operation state to generate a new frequency band distribution feature set; extract new acoustic features based on the new frequency band distribution feature set; update the mapping model between the acoustic features and the energy consumption state based on the new acoustic features to generate the optimized energy consumption state prediction model.

[0014] In a second aspect, the present invention further provides an energy efficiency optimization system for a data center with air and liquid co-sources, for implementing the above method, the system comprising: A conversion unit is used to obtain an acoustic spectrum signal of the cooling system during operation through a sensor to obtain an original acoustic spectrum data set; perform frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; a judgment unit, configured to extract acoustic features based on the frequency band distribution feature set, determine the system energy consumption state in combination with a pre-established mapping model, and judge the need for switching the cooling system mode based on a relationship between the system energy consumption state and a preset threshold; a determination unit, configured to predict a change trend of acoustic characteristics according to the mode switching requirement and determine a mode switching time point; an updating unit, configured to adjust the operating parameters of the cooling system according to the mode switching time point, execute the mode switching, and obtain an updated system operating state; The optimization unit is used to update the mapping model according to the updated system operation state to obtain an optimized energy consumption state prediction model.

[0015] In a third aspect, the present invention further provides a computer-readable storage medium having instructions stored thereon, and the instructions implement the above method when executed by a processor.

[0016] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses a method and system for optimizing the energy efficiency of an air-liquid homogenous data center. The method collects acoustic signals during the operation of the cooling system, performs frequency domain conversion and feature extraction, and combines a pre-established mapping model to determine the system energy consumption state. According to the relationship between the energy consumption state and the preset threshold, the mode switching demand is judged, and the trend of acoustic feature changes is predicted to determine the optimal switching time point. The present invention adjusts the operating parameters according to the switching time point to achieve intelligent switching between air cooling and liquid cooling modes, and optimizes the energy consumption prediction model according to the updated operating status. The method realizes real-time monitoring and intelligent adjustment of the cooling system through acoustic analysis, effectively improves the system energy efficiency, reduces energy consumption, and provides a new technical solution for optimizing the cooling system in scenarios such as data centers. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a method for optimizing energy efficiency of a data center with air and liquid sources in accordance with an embodiment of the present invention; Figure 2 This is a structural diagram of an energy efficiency optimization system for a data center with air and liquid co-sourced according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] like Figure 1 In this embodiment, a method for optimizing energy efficiency of a data center with air and liquid co-sources may specifically include: Step S1: Acquiring an acoustic spectrum signal of the cooling system during operation through a sensor to obtain an original acoustic spectrum data set; performing frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; The acoustic spectrum signals of the cooling system during operation are acquired through sensors to obtain an original acoustic spectrum data set, including: collecting audio signals formed by changes in fan speed and acoustic wave signals generated by liquid flow through acoustic sensors; continuously sampling the audio signals and acoustic wave signals to generate acoustic data containing multiple time points; and preprocessing the acoustic data to filter out ambient noise to obtain the original acoustic spectrum data set.

[0020] Specifically, the acoustic sensor captures the low-frequency sound wave signals generated by changes in fan speed and the high-frequency sound wave signals generated by liquid flow. The changes in these signals reflect the operating status of the cooling system. By continuously sampling these acoustic signals and recording data at multiple time points, an acoustic data set containing multi-time information is generated. In order to improve the accuracy and reliability of the signal, the collected acoustic data is pre-processed and noise filtered to remove environmental noise and other interference, ensuring a high-quality original acoustic spectrum data set. The above technical solution ensures the accuracy of data collection and provides a basis for subsequent energy consumption status judgment and mode switching.

[0021] Furthermore, the original acoustic spectrum dataset is converted into a frequency domain to obtain a frequency band distribution feature set, including: The original acoustic spectrum data set is converted into a frequency domain signal by fast Fourier transform; the frequency domain signal is discretized into multiple frequency bands to generate a feature vector for each frequency band; and the frequency band distribution feature set including a low frequency band and a high frequency band is generated according to the feature vector.

[0022] Specifically, the characteristics of the cooling system operation status are extracted by performing frequency domain conversion on the original acoustic spectrum data set, that is, the original time domain acoustic signal is converted into a frequency domain signal through fast Fourier transform, so as to obtain the energy distribution of the signal at different frequencies. The signal after frequency domain conversion is discretized into multiple frequency bands, which represent the energy characteristics within different frequency ranges. The frequency band distribution feature set is generated by calculating the eigenvector of each frequency band, wherein the above eigenvector includes the amplitude spectrum, energy distribution, frequency distribution, spectral entropy and root mean square value of the corresponding frequency band. The above amplitude spectrum represents the strength of the signal in a certain frequency band, which is usually obtained by calculating the amplitude of the signal in the frequency band. The above energy distribution represents the total energy distribution of the corresponding frequency band, which indicates the overall energy quantization value of the signal in the frequency band. The above frequency distribution indicates the frequency distribution of the acoustic signal in a specific frequency band, including at least information such as the peak frequency and the center value of the frequency. The above spectral entropy reflects the complexity and information content of the frequency band signal. The higher the spectral entropy, the higher the signal complexity of the frequency band. The root mean square value describes the energy intensity of the signal in the frequency band, which is usually calculated by the square root average value of the signal in the frequency band. The above frequency band distribution feature set contains feature information of low-frequency bands and high-frequency bands, which respectively reflect the different acoustic characteristics of fan speed changes and liquid flow. The low-frequency band mainly corresponds to the operation of the fan, while the high-frequency band corresponds to the state of liquid flow. The feature vectors of these frequency bands provide a basis for the subsequent prediction of the energy consumption status of the data center.

[0023] Step S2: extracting acoustic features based on the frequency band distribution feature set, and determining the system energy consumption state in combination with a pre-established mapping model; judging the need for cooling system mode switching based on the relationship between the system energy consumption state and a preset threshold; The method of extracting acoustic features based on the frequency band distribution feature set and determining the system energy consumption state in combination with a pre-established mapping model includes: Extract low-frequency band features corresponding to fan speed changes and high-frequency band features corresponding to liquid flow from the frequency band distribution feature set; input the low-frequency band features and the high-frequency band features into the pre-established acoustic feature and energy consumption status mapping model; and determine the current energy consumption status based on the output of the mapping model.

[0024] Specifically, by extracting acoustic features from the frequency band distribution feature set and combining it with a pre-established mapping model, the energy consumption status of the cooling system is determined, thereby solving the problem that traditional cooling systems cannot monitor and optimize energy efficiency in real time. The above method collects the acoustic signals of the cooling system during operation, performs frequency domain conversion, and obtains a frequency band distribution feature set, where the low frequency band corresponds to the change in fan speed, and the high frequency band corresponds to the acoustic signal generated by liquid flow. After feature extraction, these acoustic signals respectively obtain low-frequency and high-frequency feature vectors reflecting the fan operation and liquid flow status. The low-frequency band features caused by the change in fan speed can reflect the air cooling mode of the cooling system, while the high-frequency band features generated by the liquid flow reflect the system status under the liquid cooling mode; by inputting the extracted low-frequency band features and high-frequency band features into the pre-established acoustic feature and energy consumption status mapping model, the model outputs the energy consumption status of the current system based on these input information, i.e. Power consumption. The mapping model establishes a mapping relationship between acoustic features and energy consumption status based on the learning of historical data. It can accurately judge the current energy consumption level of the system according to the input acoustic features. For example, in the application scenario of the data center, when the fan speed increases, the low-frequency band features will increase accordingly, indicating that the system is running in air cooling mode; if the liquid flow rate increases, the high-frequency band features will increase, indicating that the system is running in liquid cooling mode. By monitoring these changes in real time and combining with the pre-established mapping model, the cooling mode can be dynamically adjusted to optimize energy efficiency. The mapping model is trained using the historical frequency band features and corresponding energy consumption status of the cooling system as training data. The above technical solution solves the problem of insufficient energy efficiency of the cooling system under different loads and environments, ensuring that the cooling system always operates in the most appropriate mode, reducing energy waste and improving the overall efficiency of the system.

[0025] Furthermore, judging the need for switching the cooling system mode according to the relationship between the system energy consumption state and a preset threshold value includes: If the system energy consumption state exceeds a preset energy consumption threshold, the frequency band distribution feature set is classified by inputting the frequency band distribution feature set and the frequency band feature vector into a vector machine model; based on the classification result, it is determined whether the cooling system is in a critical state of air cooling or liquid cooling mode; and the mode switching requirement is generated based on the critical state.

[0026] Specifically, by comparing the energy consumption state of the system with a preset threshold, it is determined whether the cooling system needs to switch modes, thereby optimizing the cooling efficiency. The above energy consumption state is power consumption. In the implementation of the above technical solution, it is determined whether the current energy consumption state exceeds the set energy consumption threshold; if the system energy consumption state exceeds the threshold, it will be further determined whether the cooling system is in a critical state of air cooling or liquid cooling mode. The critical state usually refers to the system energy consumption approaching or exceeding the optimal operating state, which may lead to decreased efficiency or excessive energy consumption; by inputting the acoustic feature vector corresponding to the current data center cooling system into the above vector machine model, the energy consumption state classification result is obtained, and it can be determined whether there is a need for mode switching. If the classification indicates that the system is in a critical state, a mode switching demand is generated. Signals are sent to indicate the need to switch from one mode to another, such as from air cooling to liquid cooling, thereby improving the overall energy efficiency of the cooling system. The technical solution achieves intelligent control of the cooling system through the close integration of algorithm optimization and data analysis, reduces unnecessary energy waste and ensures the stable operation of the system. For example, when the load on the data center increases, the air cooling mode may not be able to meet the heat load demand. At this time, the system predicts this through the changing trend of acoustic characteristics and generates a switching demand for liquid cooling mode, thereby avoiding overheating, reducing energy consumption, and maintaining effective cooling of the system. The above technical solution solves the problems of low energy efficiency of the cooling system and untimely mode switching by dynamically judging and optimizing the timing of mode switching, thereby improving overall energy efficiency and reducing unnecessary energy waste.

[0027] Step S3: predicting the acoustic feature change trend according to the mode switching demand and determining the mode switching time point; specifically comprising: According to the mode switching requirements, the dynamic changes of the acoustic features in the frequency band distribution feature set are analyzed; the dynamic changes of the acoustic features are modeled using a long short-term memory network model; and the acoustic feature change trend in the future is predicted based on the modeling results to generate the mode switching time point.

[0028] Specifically, the above technical solution is to determine the mode switching time point of the cooling system by predicting the changing trend of acoustic characteristics, thereby realizing intelligent cooling system optimization. The implementation of this step relies on the long short-term memory network model, which predicts the changing trend of acoustic characteristics in the future by modeling the dynamic changes of acoustic characteristics, and generates the mode switching time point based on this trend; in the implementation process, according to the mode switching requirements, the system analyzes the frequency band distribution characteristics of the collected acoustic data. The characteristic vectors formed by these characteristics after Fourier transform represent the dynamic changes of low-frequency and high-frequency bands. The above-mentioned low-frequency band usually corresponds to fan speed, and the above-mentioned high-frequency band usually corresponds to liquid flow; by analyzing the changes in these acoustic characteristics, energy consumption fluctuations or changing trends of system status can be identified; for example, in air cooling mode, an increase in fan speed will lead to enhanced low-frequency characteristics, while in liquid cooling mode, the characteristics of liquid flow will be enhanced. Changes can indicate whether the system is about to reach a critical energy efficiency state. A long short-term memory (LSTM) network model is used. This model is suitable for processing time series data and can capture long-term dependencies between features over time. Through its unique network structure, the LSTM model can learn and memorize the temporal patterns of acoustic features, predict future trends, and thus predict when a mode switch will be needed. Specifically, the LSTM model is trained on historical acoustic feature data, learning how acoustic features change under different loads and temperatures, predicting future acoustic signal changes, and determining whether a mode switch is necessary. For example, in a data center environment, when the system load increases, the increased fan speed gradually strengthens the acoustic features in the low-frequency band. The LSTM model can predict this trend and determine when to switch to liquid cooling mode to avoid excessive energy consumption and system overheating. When making predictions, the LSTM model comprehensively considers factors such as fan speed, liquid flow rate, and temperature to analyze system state changes over a period of time. In this way, the system can predict changes in cooling demand in advance and determine the optimal switching time to achieve energy-saving optimization.

[0029] Step S4: adjusting the cooling system operating parameters according to the mode switching time point, executing the mode switching, and obtaining an updated system operating state, including: According to the mode switching time point combined with load monitoring information and temperature sensor data, the optimal switching parameters of the air cooling and liquid cooling modes are determined; the operating parameters of the cooling system are adjusted through a preset control logic; the switching between the air cooling and liquid cooling modes is executed to generate the updated system operating status.

[0030] Specifically, the technical solution of step S4 adjusts the cooling system operating parameters at the mode switching time point by combining load monitoring information and temperature sensor data, and performs mode switching to generate an updated system operating status. During the implementation process, the load monitoring information and temperature sensor data are used as input data sources to provide real-time feedback on the operating status of the cooling system. The load monitoring information provides real-time change data of the system load, reflecting indicators such as the CPU utilization and memory usage of the server in the data center, thereby indirectly inferring changes in cooling demand; and the temperature sensor data provides temperature changes in the cooling system and data center environment, which directly affects the judgment of the cooling effect; when the load monitoring information and temperature sensor data fluctuate abnormally, the system needs to dynamically adjust the cooling mode to ensure optimal energy efficiency and temperature control effects.

[0031] After the mode switching time point is determined, the preset control logic begins execution. The control logic determines whether it is necessary to switch from air cooling mode to liquid cooling mode, or vice versa, based on the set parameter range, such as the upper temperature limit and the load threshold. The execution of the control logic is not only based on static rules, but also adjusted according to the analysis results of the previous step and the real-time changes in the data. For example, when the server load continues to increase and the ambient temperature exceeds the effective operating range of the air cooling mode, the control logic will automatically calculate the optimal time to switch to liquid cooling mode, thereby starting the liquid cooling system to prevent the system from overheating.

[0032] During this process, the optimal switching parameters between air-cooling and liquid-cooling modes are determined through comprehensive calculations, taking into account both real-time temperature data from temperature sensors and load monitoring information to optimize the balance between the air-cooling and liquid-cooling systems. For example, when the temperature in air-cooling mode approaches a set threshold, the system adjusts the operating parameters of the liquid-cooling mode, such as the liquid flow rate and pump speed, to ensure that excessive energy consumption is not caused during the switch. At the same time, the liquid-cooling mode may also be optimized based on changes in load information, allowing the liquid-cooling mode to effectively leverage its advantages under high-load conditions and reduce energy waste.

[0033] By executing the control logic for these mode switches, the system can ultimately automatically switch between air cooling and liquid cooling modes at precise timing, achieving optimal cooling effects and avoiding energy inefficiencies caused by excessively high or low temperatures. This optimization approach not only improves the overall energy efficiency of the data center cooling system through precise mode switching timing and parameter adjustment, but also effectively reduces energy consumption while maintaining system stability. For example, during a data center's peak load period, if load monitoring information indicates a sharp increase in server processing power demand, and the temperature sensor indicates that the ambient temperature has reached a preset critical value, the system will promptly determine and automatically adjust the switching time of the cooling mode. Through the optimization of the control logic, the air cooling system will enter liquid cooling mode in advance when the temperature approaches the critical point, and accurately adjust the liquid flow rate and pump speed of the liquid cooling system to avoid overheating and ensure the stability and efficiency of the cooling effect.

[0034] The above technical solution solves the problem that traditional cooling modes in data center cooling systems cannot be flexibly adjusted and respond to load changes in a timely manner. By comprehensively using load monitoring information and temperature sensor data, the system can ensure the optimal cooling effect under different load and temperature conditions without increasing additional energy consumption, thereby improving the system's energy efficiency and reducing energy waste.

[0035] Step S5: updating the mapping model according to the updated system operation status to obtain an optimized energy consumption state prediction model, including: Acquire the acoustic spectrum signal under the updated system operation state to generate a new frequency band distribution feature set; extract new acoustic features based on the new frequency band distribution feature set; update the mapping model between the acoustic features and the energy consumption state based on the new acoustic features to generate the optimized energy consumption state prediction model.

[0036] Specifically, the above-mentioned step S5 involves updating the mapping model of acoustic characteristics and energy consumption status based on the updated system operating status to obtain an optimized energy consumption status prediction model; this process uses a series of data collection, processing and algorithm optimization operations to further improve the system energy efficiency and solve the technical problem that traditional cooling systems cannot be flexibly adjusted. During the operation of the cooling system, the acoustic spectrum signals collected by the sensor in real time reflect the dynamic changes in the system operating status. The frequency band distribution feature set generated based on these signals includes acoustic features in different frequency ranges, among which the low frequency band mainly corresponds to the sound wave signal generated by the change in fan speed, while the high frequency band reflects the noise generated by the flow of liquid. These features provide an important basis for the judgment of system energy efficiency and mode switching.

[0037] When the system operating status changes, new acoustic spectrum signals are collected and processed to generate a new frequency band distribution feature set. This feature set not only includes the acoustic characteristics of the fan and liquid flow, but also converts these time domain signals into frequency domain signals through fast Fourier transform, thereby providing richer data input for subsequent energy efficiency status prediction. The new frequency band distribution feature set provides features that more accurately reflect the operating status of the cooling system, and can better match the actual energy consumption performance of the system under different load and environmental conditions.

[0038] Using the new frequency distribution feature set, new acoustic signatures were extracted, representing the energy consumption characteristics of the cooling system under its current operating mode. These acoustic signatures were combined with the previously established energy consumption state mapping model and input into the model for calculation and prediction, resulting in the current system energy consumption state. To adapt to the dynamic changes in the cooling system, the mapping model needs to be updated with new data to ensure that it can more accurately predict the system's energy efficiency level.

[0039] Through the above process, the mapping model continuously adjusts its parameters according to the latest acoustic characteristics, making the predicted energy consumption status more accurate and able to respond promptly to changes in system operation. This optimization process can effectively solve the problems of inflexible and untimely energy efficiency optimization of cooling systems in existing technologies. The optimized energy consumption status prediction model can achieve more intelligent and accurate switching between air cooling and liquid cooling modes. When the load increases, it can predict and switch to liquid cooling mode in advance, thereby avoiding excessive energy consumption and improving cooling efficiency and system stability. For example, in an actual application scenario, when the fan speed in the cooling system changes due to increased load, the low-frequency band features with concentrated frequency band distribution characteristics will show an enhanced trend. The updated mapping model can accurately judge the system energy efficiency status at this time and predict whether it is necessary to switch to liquid cooling mode to adapt to higher cooling requirements. If the mapping model detects that the energy efficiency threshold is approaching, the system can switch modes at the appropriate time, thereby achieving energy saving and optimization effects.

[0040] The present invention also provides an energy efficiency optimization system for a data center with air and liquid homogenization, which is used to implement the above method, such as Figure 2 As shown, the system includes: A conversion unit is used to obtain an acoustic spectrum signal of the cooling system during operation through a sensor to obtain an original acoustic spectrum data set; perform frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; a judgment unit, configured to extract acoustic features based on the frequency band distribution feature set, determine the system energy consumption state in combination with a pre-established mapping model, and judge the need for switching the cooling system mode based on a relationship between the system energy consumption state and a preset threshold; a determination unit, configured to predict a change trend of acoustic characteristics according to the mode switching requirement and determine a mode switching time point; an updating unit, configured to adjust the operating parameters of the cooling system according to the mode switching time point, execute the mode switching, and obtain an updated system operating state; The optimization unit is used to update the mapping model according to the updated system operation state to obtain an optimized energy consumption state prediction model.

[0041] The present invention provides a computer-readable storage medium having instructions stored thereon, and the above-mentioned method is implemented when the instructions are executed by a processor.

[0042] In summary, the acoustic spectrum signals collected by the sensors in this invention provide accurate feedback on the real-time status of the cooling system. By converting these signals into the frequency domain, a frequency distribution feature set is generated, which can clearly distinguish the different frequency signals generated by changes in fan speed and liquid flow. These signals not only represent the operating status of the cooling system but also reflect energy efficiency fluctuations caused by changes in system load. This step provides an accurate, real-time data foundation for subsequent optimization, addressing the inherent inability of traditional cooling systems to adapt flexibly to changing loads. Combining acoustic features extracted from the frequency distribution feature set with a pre-established energy consumption state mapping model effectively assesses the system's current energy consumption. By mapping these features, the system can determine in real time whether to switch operating modes, avoiding excessive energy waste. For example, when air cooling fails to meet cooling requirements, the system automatically switches to liquid cooling, improving cooling efficiency and reducing energy consumption. Furthermore, by introducing a predictive algorithm for mode switching needs and using a long-short-term memory network to model acoustic feature trends, the system can predict changes in the cooling system's operating state in advance and optimize the timing of mode switching. This ensures precise switching between cooling modes, avoids over-reliance on fixed rules and manual intervention, and enhances the system's adaptive regulation capabilities. By continuously optimizing the energy consumption state prediction model, the system continuously updates the mapping model based on real-time operating conditions, making energy efficiency predictions more accurate and adaptable. This optimization cycle ensures that the cooling system always operates in the most appropriate mode, avoiding unnecessary energy waste and improving the system's long-term stability and efficiency through a self-updating mapping model.

[0043] The above only lists some preferred embodiments of the present invention, but the present invention is not limited thereto, and many improvements and modifications can be made. As long as the improvements and modifications are made on the basis of the basic principles of the present invention, they should be considered to fall within the scope of protection of the present invention.

Claims

1. A method for optimizing energy efficiency of a data center with air and liquid sources, characterized in that: include: Step S1: Acquire the acoustic spectrum signal of the cooling system during operation through a sensor to obtain an original acoustic spectrum data set; Performing frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; Step S2: extracting acoustic features based on the frequency band distribution feature set, determining the system energy consumption state in combination with a pre-established mapping model, and judging the need for cooling system mode switching based on the relationship between the system energy consumption state and a preset threshold; Step S3: predicting the acoustic feature change trend according to the mode switching demand and determining the mode switching time point; Step S4: adjusting the cooling system operating parameters according to the mode switching time point, executing the mode switching, and obtaining an updated system operating state; Step S5: updating the mapping model according to the updated system operation status to obtain an optimized energy consumption status prediction model.

2. The method according to claim 1, wherein The acoustic spectrum signal of the cooling system during operation is obtained by the sensor to obtain an original acoustic spectrum data set, including: An acoustic sensor is used to collect an audio signal generated by a change in fan speed and an acoustic wave signal generated by liquid flow; the audio signal and the acoustic wave signal are continuously sampled to generate acoustic data including multiple time points; the acoustic data is preprocessed to filter out environmental noise to obtain the original acoustic spectrum data set.

3. The method according to claim 1, wherein The frequency domain conversion of the original acoustic spectrum data set to obtain a frequency band distribution feature set includes: The original acoustic spectrum data set is converted into a frequency domain signal by using a fast Fourier transform algorithm; the frequency domain signal is discretized into multiple frequency bands to generate a feature vector for each frequency band; and the frequency band distribution feature set including a low frequency band and a high frequency band is generated based on the feature vector.

4. The method according to claim 1, wherein The extracting of acoustic features according to the frequency band distribution feature set and determining the system energy consumption state in combination with a pre-established mapping model includes: Extract low-frequency band features corresponding to fan speed changes and high-frequency band features corresponding to liquid flow from the frequency band distribution feature set; input the low-frequency band features and the high-frequency band features into the pre-established acoustic feature and energy consumption status mapping model; and determine the current system energy consumption status based on the output of the mapping model.

5. The method according to claim 1, wherein The determining of the cooling system mode switching requirement based on the relationship between the system energy consumption state and a preset threshold value includes: If the system energy consumption state exceeds the preset energy consumption threshold, the frequency band distribution feature set is classified by the support vector machine algorithm; based on the classification result, it is determined whether the cooling system is in a critical state of air cooling or liquid cooling mode; and the mode switching requirement is generated based on the critical state.

6. The method according to claim 1, wherein The predicting of the acoustic feature change trend according to the mode switching demand and determining the mode switching time point includes: According to the mode switching requirements, the dynamic changes of the acoustic features in the frequency band distribution feature set are analyzed; the dynamic changes of the acoustic features are modeled using a long short-term memory network algorithm; and the acoustic feature change trend in the future is predicted based on the modeling results to generate the mode switching time point.

7. The method according to claim 1, wherein The adjusting the cooling system operating parameters according to the mode switching time point, performing the mode switching, and obtaining an updated system operating state includes: According to the mode switching time point combined with load monitoring information and temperature sensor data, the optimal switching parameters of the air cooling and liquid cooling modes are determined; the operating parameters of the cooling system are adjusted through a preset control logic; the switching between the air cooling and liquid cooling modes is executed to generate the updated system operating status.

8. The method according to claim 1, wherein The updating of the mapping model according to the updated system operation state to obtain an optimized energy consumption state prediction model includes: Acquire the acoustic spectrum signal under the updated system operation state to generate a new frequency band distribution feature set; extract new acoustic features based on the new frequency band distribution feature set; update the mapping model between the acoustic features and the energy consumption state based on the new acoustic features to generate the optimized energy consumption state prediction model.

9. A wind-liquid homogenous data center energy efficiency optimization system, used to implement the method according to any one of claims 1 to 8, characterized in that: The system comprises: A conversion unit is used to obtain an acoustic spectrum signal of the cooling system during operation through a sensor to obtain an original acoustic spectrum data set; perform frequency domain conversion on the original acoustic spectrum data set to obtain a frequency band distribution feature set; a judgment unit, configured to extract acoustic features based on the frequency band distribution feature set, determine the system energy consumption state in combination with a pre-established mapping model, and judge the need for switching the cooling system mode based on a relationship between the system energy consumption state and a preset threshold; a determination unit, configured to predict a change trend of acoustic characteristics according to the mode switching requirement and determine a mode switching time point; an updating unit, configured to adjust the operating parameters of the cooling system according to the mode switching time point, execute the mode switching, and obtain an updated system operating state; The optimization unit is used to update the mapping model according to the updated system operation state to obtain an optimized energy consumption state prediction model.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images