Energy consumption analysis method, device, equipment and storage medium for mini computer host

Abstract

The present invention relates to the field of energy consumption analysis of mini computers, and in particular to an energy consumption analysis method, device, equipment and storage medium for a mini computer host. The method comprises the following steps: obtaining multi-dimensional real-time monitoring data, the multi-dimensional real-time monitoring data including CPU usage, memory usage, disk I / O volume and computer multi-component temperature monitoring data; performing time window decomposition on the multi-dimensional real-time monitoring data, and performing multi-load scene recognition, thereby obtaining multiple load scene feature data; performing component temperature average calculation on the computer multi-component temperature monitoring data one by one, and performing dynamic temperature trend evolution fitting to construct temperature distribution trend curves of multiple components; performing scene type matching of the current node on multiple load scene feature data based on the temperature distribution trend curves of multiple components, and then performing component heating trend calculation under different scenes, thereby generating component heating trend diagrams of multiple load scenes. Thus, the energy consumption of the mini computer is optimized.

Description

Technical Field

[0001] The present invention relates to the field of energy consumption analysis of mini computers, and in particular to an energy consumption analysis method, device, equipment and storage medium for a mini computer host. Background Art

[0002] With the development of computer technology, miniaturization and low power consumption have always been the main demands of users and manufacturers. Mini computer hosts have been widely used in home audio and video entertainment, office automation, Internet of Things and other fields due to their small size and low power consumption. However, during long-term use, due to their compact structural design and complex internal structure, this kind of mini computer host is easily affected by many factors such as ambient temperature, load changes, system software, etc., resulting in abnormal fluctuations in power consumption, thus affecting the stability of the system and energy efficiency.

[0003] Traditional computer host power consumption detection methods usually rely on professional power test instruments, require manual intervention, and have cumbersome testing processes and low data analysis efficiency. Therefore, there is an urgent need for an intelligent computer host power consumption analysis method that can monitor the power consumption change characteristics of the host in real time, detect abnormal situations in a timely manner, and provide users with accurate energy consumption diagnosis and optimization suggestions, thereby improving the energy utilization efficiency of the mini computer host and reducing operating costs. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes an energy consumption analysis method, device, equipment and storage medium for a mini computer host to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides an energy consumption method for a mini computer host, comprising the following steps:

[0006] Step S1: acquiring multi-dimensional real-time monitoring data, the multi-dimensional real-time monitoring data including CPU usage, memory usage, disk I / O volume and computer multi-component temperature monitoring data; performing time window decomposition on the multi-dimensional real-time monitoring data, and performing multi-load scene recognition, thereby obtaining multiple load scene feature data;

[0007] Step S2: Calculate the average temperature of each component of the computer multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves of multiple components;

[0008] Step S3: matching the scene type of the current node with the characteristic data of multiple load scenes based on the temperature distribution trend curves of multiple components, and then calculating the heating trend of the components under different scenes, thereby generating component heating trend diagrams of multiple load scenes;

[0009] Step S4: Obtain historical computer operation logs; perform load scenario change analysis on the historical computer operation logs within a period, and perform load scenario prediction in a future period, thereby obtaining load scenario prediction data for a future period;

[0010] Step S5: Based on the component heating trend diagrams of multiple load scenarios, the heating situation of the load scenario forecast data for the future period is calculated, and the energy consumption evolution analysis is performed to generate the energy consumption value under the predicted load scenario;

[0011] Step S6: Based on the predicted energy consumption value under the load scenario, iterative fan power adjustment decision is made for the mini computer host, and dynamic energy consumption optimization is performed to build an intelligent energy consumption optimization model.

[0012] The present invention also provides an energy consumption analysis device for a mini computer host, comprising:

[0013] The load scenario recognition module is used to obtain multi-dimensional real-time monitoring data, including CPU usage, memory usage, disk I / O volume, and computer multi-component temperature monitoring data; decompose the multi-dimensional real-time monitoring data into time windows, and perform multi-load scenario recognition to obtain multiple load scenario feature data;

[0014] The temperature distribution trend module is used to calculate the average temperature of each component of the computer's multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves for multiple components;

[0015] The heating trend module is used to match the scene type of the current node with the characteristic data of multiple load scenes based on the temperature distribution trend curves of multiple components, and then calculate the heating trend of components under different scenes, so as to generate the heating trend graphs of components for multiple load scenes;

[0016] The load scenario prediction module is used to obtain historical computer operation logs; analyze the load scenario changes within the period of the historical computer operation logs, and predict the load scenario in the future period, so as to obtain the load scenario prediction data in the future period;

[0017] The energy consumption evolution module is used to calculate the heating situation of the load scenario forecast data for the future period based on the component heating trend diagrams of multiple load scenarios, and to perform energy consumption evolution analysis, so as to generate the energy consumption value under the predicted load scenario;

[0018] The energy consumption optimization module is used to make iterative fan power adjustment decisions for the mini computer host based on the energy consumption value under the predicted load scenario, and to perform dynamic energy consumption optimization to build an intelligent energy consumption optimization model.

[0019] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the energy consumption analysis method of the mini computer host described in any one of the above items are implemented.

[0020] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the energy consumption analysis method of the mini computer host described in any one of the above are implemented.

[0021] The beneficial effects of the present invention are as follows: through time window decomposition and multi-load scenario identification, the characteristic data under different load scenarios can be accurately identified, the working status of the system can be understood, the CPU usage rate, memory usage and other data can be monitored, resource allocation can be optimized, and system performance can be improved. The average temperature of each component can be calculated and dynamic temperature trend evolution fitting can be performed to build the temperature distribution trend curve of the component, the component status can be monitored, and the component temperature trend can be monitored to help discover potential faults in advance and avoid hardware damage. According to the component temperature trend and the load scenario characteristic data, the scenario type of the current node can be matched, the component heating situation under different load scenarios can be understood, and the component heating trend graph can be calculated. Help optimize heat dissipation design, improve system stability, analyze load scenario changes and predict future load scenarios through historical computer operation logs, predict the future working status of the system, predict load scenarios to help optimize system configuration, improve performance and efficiency, generate energy consumption values ​​under predicted load scenarios through heating situation calculations and energy consumption evolution analysis, help evaluate the energy efficiency of the system, analyze energy consumption values ​​to guide energy-saving measures, reduce energy consumption, and reduce carbon emissions. Iterate fan power adjustment decisions for mini computer hosts based on predicted energy consumption values, achieve dynamic energy optimization, improve energy efficiency, and build an intelligent energy optimization model to balance performance requirements and energy consumption to provide the best working status. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic flow chart of the steps of an energy consumption method for a mini computer host according to the present invention;

[0023] Figure 2 Detailed implementation flow chart of step S1;

[0024] Figure 3 Detailed implementation flow chart of step S2;

[0025] Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0026] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0027] The present application example provides an energy consumption analysis method, device, equipment and storage medium for a mini computer host. The execution subjects of the energy consumption analysis method, device, equipment and storage medium for the mini computer host include but are not limited to: mechanical equipment, data processing platform, cloud server node, network upload equipment, etc. equipped with the system can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to: at least one of an audio and image management system, an information management system, and a cloud data management system.

[0028] See also Figures 1 to 4 The present invention provides an energy consumption analysis method for a mini computer host, and the energy consumption analysis method for the mini computer host comprises the following steps:

[0029] Step S1: acquiring multi-dimensional real-time monitoring data, the multi-dimensional real-time monitoring data including CPU usage, memory usage, disk I / O volume and computer multi-component temperature monitoring data; performing time window decomposition on the multi-dimensional real-time monitoring data, and performing multi-load scene recognition, thereby obtaining multiple load scene feature data;

[0030] Step S2: Calculate the average temperature of each component of the computer multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves of multiple components;

[0031] Step S3: matching the scene type of the current node with the characteristic data of multiple load scenes based on the temperature distribution trend curves of multiple components, and then calculating the heating trend of the components under different scenes, thereby generating component heating trend diagrams of multiple load scenes;

[0032] Step S4: Obtain historical computer operation logs; perform load scenario change analysis on the historical computer operation logs within a period, and perform load scenario prediction in a future period, thereby obtaining load scenario prediction data for a future period;

[0033] Step S5: Based on the component heating trend diagrams of multiple load scenarios, the heating situation of the load scenario forecast data for the future period is calculated, and the energy consumption evolution analysis is performed to generate the energy consumption value under the predicted load scenario;

[0034] Step S6: Based on the predicted energy consumption value under the load scenario, iterative fan power adjustment decision is made for the mini computer host, and dynamic energy consumption optimization is performed to build an intelligent energy consumption optimization model.

[0035] The present invention accurately identifies characteristic data under different load scenarios through time window decomposition and multi-load scenario identification, understands the working status of the system, monitors data such as CPU usage and memory occupancy, helps optimize resource allocation, improves system performance, calculates the average temperature of each component one by one and performs dynamic temperature trend evolution fitting, builds a temperature distribution trend curve of the component, monitors the component status, monitors the component temperature trend to help discover potential faults in advance and avoid hardware damage, matches the scenario type of the current node according to the component temperature trend and load scenario characteristic data, understands the component heating situation under different load scenarios, calculates the component heating trend chart to help optimize the heat dissipation design and improve the system stability, performs load scenario transformation analysis and future load scenario prediction through historical computer operation logs, predicts the future working status of the system, predicts the load scenario to help optimize the system configuration and improve performance and efficiency, generates energy consumption values ​​under predicted load scenarios through heating situation calculation and energy consumption evolution analysis, helps evaluate the energy efficiency of the system, analyzes the energy consumption value to guide energy-saving measures, reduces energy consumption and reduces carbon emissions, makes iterative fan power adjustment decisions for the mini computer host based on the predicted energy consumption value, realizes dynamic energy consumption optimization, improves energy efficiency, builds an intelligent energy consumption optimization model to balance performance requirements and energy consumption, and provides the best working status.

[0036] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of an energy consumption analysis method of a mini computer host of the present invention. In this example, the steps of the energy consumption analysis method of the mini computer host include:

[0037] Step S1: acquiring multi-dimensional real-time monitoring data, the multi-dimensional real-time monitoring data including CPU usage, memory usage, disk I / O volume and computer multi-component temperature monitoring data; performing time window decomposition on the multi-dimensional real-time monitoring data, and performing multi-load scene recognition, thereby obtaining multiple load scene feature data;

[0038] In this embodiment, the source of the data is determined, such as a system monitoring tool (such as Prometheus, Zabbix) or a log system, to ensure that the CPU usage, memory usage, disk I / O volume and temperature data can be obtained in real time. A programming language (such as Python) is used to write a script to regularly extract the following multi-dimensional real-time monitoring data from the data source: CPU usage: obtain the real-time CPU usage percentage through the system API, memory usage: obtain the current memory usage, including used memory and total memory, disk I / O volume: obtain the read and write speed per second through system commands, component temperature monitoring data: use a hardware monitoring tool (such as lm-sensors) to obtain the temperature of each component (such as CPU, GPU, hard disk), determine the appropriate time window size, such as every 5 minutes, 10 minutes or 30 minutes, and select it according to the change frequency of the monitoring data and the analysis requirements, use Pandas's resample function to decompose the real-time monitoring data into time windows, aggregate the data according to the defined time windows (such as calculating the average, maximum and minimum values ​​within the window), and store the decomposed time window data in In a new data frame for subsequent analysis, ensure that the timestamps and aggregate indicators corresponding to each time window are clearly visible, select key features for load scenario identification from the data after the time window is decomposed, such as CPU usage, memory usage, disk I / O and temperature, and select appropriate load scenario identification methods. Commonly used methods include: Threshold method: set a threshold and judge the scenario type (such as high load, low load) according to the feature value. Cluster analysis: use algorithms such as K-means or hierarchical clustering to cluster data and identify different load scenarios. Classification algorithm: use supervised learning methods (such as decision trees and random forests) to train and identify existing scenario labels, apply the selected method to analyze multi-dimensional monitoring data, identify multiple load scenario feature data (such as normal operation, overload, idle, etc.), and store the identified load scenario feature data in the data frame, including the time range, scenario type and related indicator values ​​of each scenario.

[0039] Step S2: Calculate the average temperature of each component of the computer multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves of multiple components;

[0040] In this embodiment, ensure that the existing multi-component temperature monitoring data includes timestamps and temperature values ​​of each component. These data come from hardware monitoring tools (such as lm-sensors, HWMonitor, etc.). Arrange the temperature monitoring data into an appropriate data frame format (such as Pandas DataFrame), ensuring that each column represents the temperature of a component and each row corresponds to a timestamp. Example code:

[0041] import pandas as pd

[0042] # Assume df is a DataFrame containing temperature data

[0043] df['timestamp'] = pd.to_datetime(df['timestamp'])

[0044] df.set_index('timestamp', inplace=True)

[0045] average_temp = df.resample('5T').mean() # Calculate the average value every 5 minutes

[0046] Select a suitable fitting method, for example: Linear regression: suitable for simple linear relationships. Polynomial regression: suitable for nonlinear relationships, use numpy.polyfit. Exponential smoothing: such as using ExponentialSmoothing in the statsmodels library. Perform trend fitting on the average temperature data of each component. Generate a fitting model based on the selected fitting method and output the fitting results.

[0047] Sample code:

[0048] import numpy as np

[0049] import matplotlib.pyplot as plt

[0050] # Get time index and temperature data

[0051] time_indices = np.arange(len(average_temp)) # time index

[0052] for column in average_temp.columns:

[0053] coeffs = np.polyfit(time_indices, average_temp[column], 2) # quadratic polynomial fitting

[0054] fitted_curve = np.polyval(coeffs, time_indices)

[0055] # Plot the original data and the fitted curve

[0056] plt.plot(average_temp.index, average_temp[column], label=f'original data({column})')

[0057] plt.plot(average_temp.index, fitted_curve, label=f'fitted curve ({column})')

[0058] plt.legend()

[0059] plt.show()

[0060] Use the temperature distribution trend curve generated by the fitting model to ensure that the trend curve of each component can reflect its temperature change law. Plot the temperature distribution trend curves of each component in the same chart for easy comparison and analysis. Use Matplotlib or Seaborn for visualization.

[0061] Sample code:

[0062] plt.figure(figsize=(12, 6))

[0063] for column in average_temp.columns:

[0064] plt.plot(average_temp.index, average_temp[column], label=f'{column}trend')

[0065] plt.title('Temperature distribution trend of multiple components')

[0066] plt.xlabel('time')

[0067] plt.ylabel('Temperature (°C)')

[0068] plt.legend()

[0069] plt.show()

[0070] Step S3: matching the scene type of the current node with the characteristic data of multiple load scenes based on the temperature distribution trend curves of multiple components, and then calculating the heating trend of the components under different scenes, thereby generating component heating trend diagrams of multiple load scenes;

[0071] In this embodiment, the temperature distribution trend curve data of multiple components are collected to ensure that the data contains timestamps, temperature values ​​and corresponding load scene feature data. The timestamp of the current node is determined, which is usually the latest monitoring time, so as to perform scene matching. A scene matching method is selected, such as using time window matching or nearest neighbor matching: Time window matching: set a time window (such as ±5 minutes) and search for load scene feature data closest to the current node timestamp within this range. Nearest neighbor matching: use a distance metric (such as Euclidean distance) to find the most similar scene in the feature space. According to the selected matching method, the timestamp of the current node is matched with the load scene feature data to identify the current scene type. The matched current scene type is recorded in a data frame, including timestamps, scene types and related feature values. The current scene type is integrated with the temperature distribution trend curve data of multiple components to ensure that the temperature change of each component is associated with the scene type. Select a suitable heating trend calculation method, such as: Linear regression: by fitting the relationship between temperature and time, Calculate the slope and intercept of the heating trend. Moving average: smooth the temperature data to identify long-term trends. Thermal model: use the heat conduction model to consider the power input and heat dissipation characteristics of the component and calculate the heating change. For each component, apply the selected calculation method to generate the component heating trend data under different load scenarios, and ensure that the temperature change, heating rate and stable temperature under each scenario are recorded. Store the calculated heating trend data in a data frame, including the heating characteristics of each component under different load scenarios. Select appropriate visualization tools (such as Matplotlib, Seaborn or Plotly) to draw the heating trend graph. Design the heating trend graph to ensure that the graph can clearly show the temperature changes of each component under different load scenarios, including: X-axis: time, Y-axis: temperature, different colors or line types represent different load scenarios, use the selected visualization tool to draw the component heating trend graph, and ensure that the graph is beautiful and easy to understand. Save the generated heating trend graph as an image file and record it in the document for subsequent analysis and reporting.

[0072] Step S4: Obtain historical computer operation logs; perform load scenario change analysis on the historical computer operation logs within a period, and perform load scenario prediction in a future period, thereby obtaining load scenario prediction data for a future period;

[0073] In this embodiment, the sources of historical operation logs are confirmed, including operating system logs, application logs, and performance monitoring tools (such as Windows Event Viewer and Linux Syslog). The historical operation logs are extracted from the specified location using programming tools (such as Python scripts), which are text files, CSV files, or databases. The extracted data formats are ensured to be consistent, including timestamps, CPU usage, memory usage, disk I / O, and load conditions. The extracted log data is cleaned to remove missing values ​​and outliers to ensure data integrity and accuracy. The Pandas library is used for data cleaning to determine the analysis cycle, such as by day, week, or month. According to the system usage pattern and load change frequency, a suitable time window is selected to extract key features from the cleaned operation logs, such as CPU usage, memory usage, disk I / O, etc., to generate time series data, and threshold method, cluster analysis, or classification algorithm (such as K-means clustering, decision tree) is used for Classify load changes, identify different load scenarios (such as high load, low load, idle, etc.), record the identified load scenario changes in the data frame, ensure that each scenario contains the time range, scenario type and related indicators, select a suitable load scenario prediction model, such as a time series prediction model (such as ARIMA, SARIMA), regression model or machine learning model (such as LSTM), use historical load scenario data to train the selected model, ensure that the model can learn the pattern and law of load changes, apply the trained model to the current load scenario data, and predict the load scenario in the future period. Select an appropriate prediction duration (such as 1 hour, 1 day, 1 week) according to the prediction requirements, and record the predicted future load scenario data in the data frame, ensuring that the prediction scenario type and related indicators at each time point are included.

[0074] Step S5: Based on the component heating trend diagrams of multiple load scenarios, the heating situation of the load scenario forecast data for the future period is calculated, and the energy consumption evolution analysis is performed to generate the energy consumption value under the predicted load scenario;

[0075] In this embodiment, component heating trend chart data of multiple load scenarios are collected, including temperature change trends under current and predicted load scenarios, and the load scenario prediction data for future time periods are integrated with the component heating trend chart to ensure that each load scenario has corresponding temperature and timestamp information. For each load scenario and corresponding component, key indicators of temperature change are calculated, such as peak temperature, temperature change rate, and continuous high temperature period. The calculated heating situation characteristics are recorded in a data frame to ensure that the heating characteristics of each component and its corresponding load scenario are clearly visible, and the energy consumption values ​​obtained by analysis are recorded in a data frame to ensure that the predicted energy consumption and corresponding time information of each load scenario are included.

[0076] Step S6: Based on the predicted energy consumption value under the load scenario, iterative fan power adjustment decision is made for the mini computer host, and dynamic energy consumption optimization is performed to build an intelligent energy consumption optimization model.

[0077] In this embodiment, energy consumption values ​​under predicted load scenarios are collected, including temperature and energy consumption data corresponding to different scenarios, and the range (for example, 10%-100%) and rules of fan power adjustment are determined, for example: in high temperature scenarios, fan power is increased to reduce temperature, and in low load scenarios, fan power is reduced to save energy. Methods such as genetic algorithms, simulated annealing or greedy algorithms are used to make multiple iterative decisions to generate different fan power adjustment schemes. The corresponding energy consumption value is calculated for each iteration, and the scheme with the lowest energy consumption is selected as a candidate. The generated fan power adjustment scheme is recorded in a data frame, including the corresponding power setting, scenario type and predicted energy consumption value. A dynamic monitoring mechanism is designed to monitor system temperature, load, energy consumption and other indicators in real time. A feedback loop is set to ensure that the fan power setting is adjusted according to real-time data. A suitable optimization algorithm is selected, for example: reinforcement learning: using Q-learning or deep reinforcement learning methods to automatically adjust the fan power according to environmental feedback, PID control :Use the proportional-integral-derivative control algorithm to continuously adjust the fan power to keep the system temperature within the set range. Integrate iterative decisions and real-time monitoring data to build an intelligent energy consumption optimization model to ensure that the model can flexibly adjust the fan power under different load scenarios. Use historical data to train the model so that it can predict the optimal fan power setting under different scenarios. Record the parameters and decision-making process of the optimization model, and evaluate the optimization effect, such as the percentage of energy consumption reduction and temperature stability. Verify the intelligent energy consumption optimization model in the actual environment, monitor temperature changes and energy consumption to evaluate the effectiveness of the model, collect feedback data, and analyze the performance of the model under different load scenarios. According to the verification results, adjust the model parameters and fan power adjustment strategy to improve the accuracy and response speed of the model. Implement periodic review and update to ensure that the model adapts to new usage patterns and environmental changes. Record the results of verification and optimization in documents and generate reports to ensure information transparency and facilitate subsequent analysis and improvement.

[0078] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:

[0079] Step S11: Perform real-time multi-point monitoring sampling on the mini computer host to obtain multi-dimensional real-time monitoring data, wherein the multi-dimensional real-time monitoring data includes CPU usage, memory usage, disk I / O volume, and computer multi-component temperature monitoring data;

[0080] Step S12: Decomposing the multi-dimensional real-time monitoring data into time windows, thereby obtaining a multi-time window monitoring data sequence;

[0081] Step S13: performing personalized identification of the operating status of the multi-time window monitoring data sequence to obtain the operating status data of each time window;

[0082] Step S14: performing time series load calculation on the multi-time window monitoring data sequence to generate time series load data;

[0083] Step S15: performing dynamic operation scenario cluster analysis on the operation status data of each time window, thereby obtaining a plurality of operation scenario classification data;

[0084] Step S16: performing multiple load scenario identification on the time series load data based on the multiple operation scenario classification data, thereby obtaining multiple load scenario feature data;

[0085] The specific steps of performing time series load calculation on the multi-time window monitoring data sequence to generate time series load data are as follows:

[0086] Extract the CPU usage, memory usage, and disk I / O volume of each time window in the time window monitoring data sequence;

[0087] The CPU usage is normalized to obtain a standardized CPU usage;

[0088] Calculate the free storage location of the memory usage to generate the free storage location of the computer;

[0089] Calculate the time series peak value change of disk I / O volume and extract the disk I / O peak value change data;

[0090] Define load index weights for normalized CPU usage, computer idle storage locations, and disk I / O peak change data;

[0091] A time series load calculation is performed based on the load index weight to generate time series load data.

[0092] In this embodiment, monitoring tools and software for the mini computer host are configured, for example, performance monitoring tools (such as Prometheus, Grafana) are used for data collection. Multiple indicators are monitored simultaneously, including: CPU usage: Use system calls or libraries (such as psutil) to obtain CPU usage in real time. Memory usage: Get current memory usage. Disk I / O volume: Monitor the number and speed of read and write operations. Temperature monitoring: Use temperature sensors or software tools to read temperature data of various components (such as CPU, GPU). The collected multi-dimensional real-time monitoring data is stored in a database (such as InfluxDB) or a real-time data stream for subsequent analysis. Determine the length of the time window (such as 1 minute, 5 minutes) and the overlap method (such as overlap or no overlap) to meet data analysis requirements. Use a data processing library (such as Pandas) in a programming language (such as Python) to decompose the real-time monitoring data and generate multiple time window monitoring data sequences. Store the data sequence of each time window as a separate data frame, select a suitable personalized recognition model, and use a machine learning classification model (such as KNN, decision tree) or a rule-based method. Extract data features for each time window, such as average CPU usage, maximum memory usage, average disk I / O, etc. Input the features into the selected model to identify the operating status of each time window and output status data (such as "normal", "overloaded", "idle"). Select an appropriate load calculation method, usually using weighted average or simple average to calculate the load of each time window. Select a suitable clustering algorithm, such as K-means, hierarchical clustering, or DBSCAN. Use the operating status data of each time window as input to the clustering algorithm to perform clustering analysis of dynamic operating scenarios and obtain multiple operating scenario classifications. Extract feature data for each clustered operating scenario, such as average load, peak load, duration, etc. Select an appropriate recognition model, such as a multi-class classification model (such as support vector machine or random forest), to classify the load scenario.

[0093] Input time series load data and combine it with clustering results to identify multiple load scenario characteristics.

[0094] The specific implementation steps of performing time series load calculation on the multi-time window monitoring data sequence to generate time series load data are as follows:

[0095] Extract the CPU usage, memory usage, and disk I / O volume of each time window from the previously generated multi-window monitoring data sequence, store the extracted data in a new data frame, ensure that each time window has a corresponding timestamp and three monitoring indicators, select a standardization method, usually using Z-score standardization or Min-Max standardization, and standardize the CPU usage of each time window. The Z-score standardization formula is Z= , X is the original CPU usage, μ is the mean of CPU usage, σ is the standard deviation, the standardized CPU usage is stored in the data frame, the difference between the total memory capacity and the current occupied memory is calculated from the memory usage data to determine the free storage location, the formula is: free storage location = total memory − memory usage, the disk I / O peak change is defined as the difference between the current time window and the previous time window, by traversing the time window monitoring data sequence, the disk I / O peak change of each time window is calculated, the formula is: peak change = current time window I / O volume − previous time window I / O volume, according to actual monitoring needs and experience, define the load index weights of standardized CPU usage, computer free storage location and disk I / O peak change, standardized CPU usage weight = 0.5; computer free storage location weight = 0.3; disk I / O peak change weight = 0.2; use weights to weight the standardized CPU usage, computer free storage location and disk I / O peak change of each time window, time series load = (standardized CPU usage × )+(Computer free storage space × )+(disk I / O peak change× ),in, , , For the defined weight value, traverse each time window data and calculate the time series load data according to the above formula.

[0096] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:

[0097] Step S21: performing wavelet transform decomposition on the computer multi-component temperature monitoring data to generate a multi-component temperature frequency domain;

[0098] Step S22: performing component-by-component temperature average calculation on the temperature frequency domain of multiple components to generate an average temperature value of each component;

[0099] Step S23: performing temperature change analysis on the average temperature value of each component to extract the temperature change characteristics of each component;

[0100] Step S24: performing temperature distribution analysis on the temperature frequency domain of multiple components based on the temperature variation characteristics of each component to obtain component temperature distribution variation data;

[0101] Step S25: Perform dynamic temperature trend evolution fitting on the component temperature distribution change data to construct temperature distribution trend curves of multiple components.

[0102] In this embodiment, the temperature monitoring data of multiple components are collected to ensure that the data format is unified and contains timestamps and temperature values ​​of each component. An appropriate wavelet basis function is selected, such as Daubechies, wavelet or Meyer wavelet. The best option is determined according to the characteristics of the temperature data. A wavelet transform tool (such as the PyWavelets library in Python) is used to perform wavelet change decomposition on the temperature data of each component. The specific steps include: performing wavelet decomposition on each component to obtain its frequency domain representation, averaging the temperature frequency domain data of each component to generate an average temperature value of each component, using simple average or weighted average to perform change analysis on the average temperature value of each component, calculating characteristic data such as temperature change rate, maximum temperature, and minimum temperature, selecting an appropriate temperature distribution analysis method, such as a histogram, density estimation, or heat map, calculating temperature distribution change data based on the temperature change characteristics of each component, analyzing the temperature distribution under different time periods or different load conditions, using the numpy and matplotlib libraries in Python to draw a temperature distribution graph, selecting a suitable trend fitting model, such as linear regression, polynomial regression, or exponential smoothing, etc., using a fitting tool (such as scikit-learn in Python) or numpy.polyfit) to fit the temperature change data of each component to obtain a trend curve, visualize the fitting results, draw the temperature distribution trend curve of each component, and intuitively display the temperature change trend.

[0103] In this embodiment, refer to Figure 4 , is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:

[0104] Step S31: extracting multiple temperature change points based on the temperature distribution trend curves of multiple components, and calculating timestamp nodes;

[0105] Step S32: matching the scene type of the current node to the multiple operation scene classification data based on the timestamp node to obtain the load scene type time point;

[0106] Step S33: performing scene-temperature deep association mining on load scene type time points based on multiple temperature change points to extract load scene-temperature association data;

[0107] Step S34: Calculate component heating trends in different scenarios for multiple load scenario feature data according to the load scenario-temperature association data, thereby generating component heating trend diagrams for multiple load scenarios.

[0108] In this embodiment, the temperature distribution trend curves of multiple components are collected to ensure that the data contains timestamps and corresponding temperature values. A threshold detection method (such as the first-order derivative method) is used to identify temperature change points, which are nodes where the temperature rises, falls, or changes greatly. For each identified temperature change point, its corresponding timestamp is recorded to ensure that the timing information of the data is complete. The NumPy or Pandas library in Python is used for processing to collect multiple operating scenario classification data, including timestamps and scenario types of different scenarios, such as normal load, overload, idle, etc. For each timestamp node, a time window matching method (such as ±5 minutes or ±1 hour) is used to find the corresponding operating scenario type. The merge_asof function of Pandas is used for approximate matching, and the extracted temperature change points and load scenario type time points are integrated to form a A dataset containing timestamps, temperature values, and scene types is collected. Data mining techniques (such as association rule learning or deep learning) are used to analyze the integrated data to identify the deep association between scenes and temperatures. For example, the Apriori algorithm is used to find frequent item sets, extract scene-temperature association data, and record the temperature distribution characteristics and their change patterns in each scene. Based on the load scene-temperature association data, the heating trend of the component is calculated for each load scene type. This is achieved by performing linear regression or polynomial regression on the relationship between temperature and time. Visualization tools (such as Matplotlib, Seaborn, or Plotly) are used to draw component heating trend graphs for multiple load scenarios. Ensure that the graph includes the load scene type, timestamp, and corresponding temperature change. The generated trend graph is saved as an image file or embedded in a report to facilitate subsequent analysis and decision support.

[0109] In this embodiment, step S4 includes the following steps:

[0110] Step S41: Obtain historical computer operation logs;

[0111] Step S42: Analyze the load scenario changes within a period on the historical computer operation log to identify the historical period load scenario data;

[0112] Step S43: performing load change cycle mining on historical periodic load scenario data to generate load scenario change rules;

[0113] Step S44: simulating the evolution of load scenarios for multi-dimensional real-time monitoring data based on the load scenario change law, thereby obtaining scenario simulation data at the current monitoring time point;

[0114] Step S45: Perform load scenario prediction for a future period on the scenario simulation data at the current monitoring time point, thereby obtaining load scenario prediction data for the future period.

[0115] In this embodiment, the source of the computer operation log is confirmed. Common sources include operating system logs, application logs, and performance monitoring tools. A data extraction tool (such as a script or log analysis software) is used to extract historical operation logs from a specified location. The logs are text files, CSV files, or databases. The extracted data format is ensured to be consistent, including fields such as timestamps, CPU usage, memory usage, and load information. The historical operation logs are cleaned to remove missing values ​​and outliers to ensure data quality. Pandas and other libraries are used for processing. The analysis cycle is defined, such as dividing the data by day, week, or month. The appropriate time window is selected according to specific needs, and the threshold is used. Use cluster analysis (such as K-means) or pattern recognition technology to classify load changes, identify different load scenarios (such as high load, low load, idle, etc.), store the identified historical period load scenario data in the data frame, ensure that each scenario contains the time range, scenario type and related load indicators, integrate the historical period load scenario data into a complete data set, ensure that the timestamp and load situation of each scenario are clearly visible, select a suitable cycle mining algorithm, such as time series analysis (such as ARIMA model), frequency analysis (such as Fourier transform) or machine learning method (such as LSTM neural network), and apply the selected algorithm to the load scenario. Analyze the load scenario data, extract the law and periodic characteristics of load changes, record the generated load scenario change law in the data frame, including the law description and periodic parameters, collect the current multi-dimensional real-time monitoring data, ensure that it includes indicators such as CPU usage, memory usage, disk I / O and network traffic, select a suitable simulation model, such as a regression model based on historical data or a Markov model based on state transition, simulate the evolution of the load scenario, use the selected model, simulate the evolution of the real-time monitoring data according to the load scenario change law, generate the scenario simulation data at the current monitoring time point, and record the simulated scenario data in the data frame In the prediction model, make sure to include simulation time, scenario type and related load indicators, select a suitable load scenario prediction model, such as a time series prediction model (such as SARIMA), regression model or deep learning model (such as LSTM), use historical load scenario data to train the selected model, and ensure that the model can learn the pattern and law of load changes. Future scenario prediction implementation: input the scenario simulation data at the current monitoring time point into the trained model, perform load scenario prediction in the future period, generate future load scenario prediction data, store the predicted future load scenario data in a data frame, and generate a report to record the changing trend and expected impact of future load scenarios.

[0116] In this embodiment, step S5 includes the following steps:

[0117] Step S51: Calculating the heating situation of the load scenario prediction data for the future period based on the component heating trend diagrams of multiple load scenarios, thereby obtaining the heating situation characteristics of the predicted load scenario;

[0118] Step S52: Calculating the energy consumption of the time series load data based on the temperature monitoring data of the multiple components, thereby obtaining the energy consumption values ​​of the multiple components;

[0119] Step S53: performing energy consumption evolution analysis on the energy consumption values ​​of multiple components based on the heating situation characteristics of the predicted load scenario, thereby generating energy consumption values ​​under the predicted load scenario.

[0120] In this embodiment, component heating trend chart data of multiple load scenarios are collected to ensure that the temperature changes and timestamps under different scenarios are included. Determine the calculation method of the heating situation characteristics, such as peak temperature, temperature change rate, continuous high temperature period, and calculate the corresponding heating situation characteristics according to its heating trend chart for each component under the predicted load scenario. Use the NumPy and Pandas libraries in Python for data processing and calculation. Collect the current multi-component temperature monitoring data to ensure that the timestamp and the temperature value of each component are included. Obtain the time series load data corresponding to the temperature monitoring data, including the load conditions in each time period (such as the usage rate of CPU, memory, disk, etc.). Select a suitable energy consumption calculation model, for example: calculation based on the power formula (energy consumption = power × time) using an empirical formula (such as the energy consumption characteristics of each component under different loads). Calculate the energy consumption value of each component in each time period based on the load condition and the monitored temperature data. Calculate based on the power characteristics of each component and use Pandas for data processing. Store the calculated energy consumption value in the data frame to ensure that the energy consumption, timestamp and corresponding load condition of each component are included. Select appropriate energy consumption evolution analysis methods, such as regression analysis, time series analysis, and machine learning models (such as LSTM) for energy consumption forecasting. Combine the heating status characteristics of the predicted load scenario to analyze its impact on the energy consumption values ​​of each component and identify the reasons for the energy consumption changes. Use the selected analysis method to predict future energy consumption values ​​based on the existing energy consumption values ​​and heating status characteristics. Associate historical energy consumption data with heating characteristics and use machine learning algorithms for forecasting. Record the predicted energy consumption values ​​in a data frame, making sure to include the predicted energy consumption value of each component and the corresponding load scenario.

[0121] In this embodiment, step S6 includes the following steps:

[0122] Step S61: performing iterative fan power adjustment decisions on the mini computer host based on the energy consumption value under the predicted load scenario to generate multiple fan power adjustment parameter solutions;

[0123] Step S62: performing energy consumption simulation calculation based on multiple fan power adjustment parameter schemes, thereby obtaining an energy consumption simulation value for each scheme;

[0124] Step S63: extracting the fan power adjustment parameter scheme with the lowest energy consumption based on the energy consumption simulation value of each scheme;

[0125] Step S64: Dynamically optimize the energy consumption of the mini computer host based on the fan power adjustment parameter solution with the lowest energy consumption, and build an intelligent energy consumption optimization model.

[0126] In this embodiment, energy consumption values ​​under predicted load scenarios are collected to ensure that the data includes temperature and energy consumption information under different load conditions, and the strategy for fan power adjustment is determined, for example: the adjustment range of fan power is defined (such as 10% to 100%), and fan power adjustment rules are formulated under different temperature ranges, such as increasing fan power during high temperature periods and reducing power during low temperature periods. According to energy consumption values ​​and temperature information, multiple fan power adjustment parameter schemes are generated through iterative decision-making methods (such as simulated annealing and genetic algorithms), and the generated fan power adjustment parameter schemes are stored in a data frame to ensure that each scheme includes power settings, corresponding temperatures and energy consumption estimates, and a suitable energy consumption simulation model is selected, for example: simulation based on physical models (such as heat conduction models), using a regression model established using historical energy consumption data, performing energy consumption simulation calculations on each fan power adjustment parameter scheme, and evaluating energy consumption performance under different power settings. Using Python's SimPy or its Use other simulation tools to perform calculations, record the energy consumption simulation value of each solution in a data frame, make sure to include the solution description and the corresponding energy consumption value, compare all fan power adjustment parameter solutions, extract the solution with the lowest energy consumption simulation value, use Pandas' idxmin function to quickly find the lowest value, record the fan power adjustment parameter solution with the lowest energy consumption in detail, including power settings, expected energy consumption, and corresponding temperature and load scenarios, determine the framework of the intelligent energy consumption optimization model, such as selecting rule-based models, machine learning models (such as reinforcement learning), or scheduling algorithms, integrate the fan power adjustment parameter solution with the lowest energy consumption into the intelligent energy consumption optimization model, enable the model to adjust the fan power in real time to achieve dynamic optimization, set up a feedback mechanism so that the model can be adjusted according to real-time temperature and energy consumption data, optimize the fan power setting, record the constructed model parameters, and conduct preliminary tests to ensure that the model can effectively reduce energy consumption and maintain system stability.

[0127] The present invention also provides an energy consumption analysis device for a mini computer host, comprising:

[0128] The load scenario recognition module is used to obtain multi-dimensional real-time monitoring data, including CPU usage, memory usage, disk I / O volume, and computer multi-component temperature monitoring data; decompose the multi-dimensional real-time monitoring data into time windows, and perform multi-load scenario recognition to obtain multiple load scenario feature data;

[0129] The temperature distribution trend module is used to calculate the average temperature of each component of the computer's multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves for multiple components;

[0130] The heating trend module is used to match the scene type of the current node with the characteristic data of multiple load scenes based on the temperature distribution trend curves of multiple components, and then calculate the heating trend of components under different scenes, so as to generate the heating trend graphs of components for multiple load scenes;

[0131] The load scenario prediction module is used to obtain historical computer operation logs; analyze the load scenario changes within the period of the historical computer operation logs, and predict the load scenario in the future period, so as to obtain the load scenario prediction data in the future period;

[0132] The energy consumption evolution module is used to calculate the heating situation of the load scenario forecast data for the future period based on the component heating trend diagrams of multiple load scenarios, and to perform energy consumption evolution analysis, so as to generate the energy consumption value under the predicted load scenario;

[0133] The energy consumption optimization module is used to make iterative fan power adjustment decisions for the mini computer host based on the energy consumption value under the predicted load scenario, and to perform dynamic energy consumption optimization to build an intelligent energy consumption optimization model.

[0134] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the energy consumption analysis method of the mini computer host described in any one of the above items are implemented.

[0135] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the energy consumption analysis method of the mini computer host described in any one of the above are implemented.

[0136] Those skilled in the art clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0137] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it is stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or partly or all or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media for storing program codes such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks or optical disks.

[0138] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0139] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for analyzing energy consumption of a mini computer host, characterized in that: The following steps are involved: Step S1: acquiring multi-dimensional real-time monitoring data, the multi-dimensional real-time monitoring data including CPU usage, memory usage, disk I / O volume and computer multi-component temperature monitoring data; performing time window decomposition on the multi-dimensional real-time monitoring data, and performing multi-load scene recognition, thereby obtaining multiple load scene feature data; The specific steps of step S1 are: Step S11: Perform real-time multi-point monitoring sampling on the mini computer host to obtain multi-dimensional real-time monitoring data, wherein the multi-dimensional real-time monitoring data includes CPU usage, memory usage, disk I / O volume, and computer multi-component temperature monitoring data; Step S12: Decomposing the multi-dimensional real-time monitoring data into time windows, thereby obtaining a multi-time window monitoring data sequence; Step S13: performing personalized identification of the operating status of the multi-time window monitoring data sequence to obtain the operating status data of each time window; Step S14: performing time series load calculation on the multi-time window monitoring data sequence to generate time series load data; Step S15: performing dynamic operation scenario cluster analysis on the operation status data of each time window, thereby obtaining a plurality of operation scenario classification data; Step S16: performing multiple load scenario identification on the time series load data based on the multiple operation scenario classification data, thereby obtaining multiple load scenario feature data; The specific steps of performing time series load calculation on the multi-time window monitoring data sequence to generate time series load data are as follows: Extract the CPU usage, memory usage, and disk I / O volume of each time window in the time window monitoring data sequence; The CPU usage is normalized to obtain a standardized CPU usage; Calculate the free storage location of the memory usage to generate the free storage location of the computer; Calculate the time series peak value change of disk I / O volume and extract the disk I / O peak value change data; Define load index weights for normalized CPU usage, computer idle storage locations, and disk I / O peak change data; Performing time series load calculation based on the load index weight to generate time series load data; Step S2: Calculate the average temperature of each component of the computer multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves of multiple components; Step S3: Based on the temperature distribution trend curves of multiple components, the scene type of the current node is matched with the characteristic data of multiple load scenes, and then the heating trend of the components under different scenes is calculated, so as to generate the heating trend diagrams of the components of multiple load scenes; the specific steps of step S3 are: Step S31: extracting multiple temperature change points based on the temperature distribution trend curves of multiple components, and calculating timestamp nodes; Step S32: matching the scene type of the current node to the multiple operation scene classification data based on the timestamp node to obtain the load scene type time point; Step S33: performing scene-temperature deep association mining on load scene type time points based on multiple temperature change points to extract load scene-temperature association data; Step S34: Calculating component heating trends in different scenarios for multiple load scenario feature data according to the load scenario-temperature association data, thereby generating component heating trend graphs for multiple load scenarios; Step S4: Obtain historical computer operation logs; perform load scenario change analysis on the historical computer operation logs within a period, and perform load scenario prediction in a future period, thereby obtaining load scenario prediction data for a future period; Step S5: Based on the component heating trend diagrams of multiple load scenarios, the heating situation of the load scenario forecast data for the future period is calculated, and the energy consumption evolution analysis is performed to generate the energy consumption value under the predicted load scenario; the specific steps of step S5 are: Step S51: Calculating the heating situation of the load scenario prediction data for the future period based on the component heating trend diagrams of multiple load scenarios, thereby obtaining the heating situation characteristics of the predicted load scenario; Step S52: Calculating the energy consumption of the time series load data based on the temperature monitoring data of the multiple components, thereby obtaining the energy consumption values ​​of the multiple components; Step S53: performing energy consumption evolution analysis on the energy consumption values ​​of multiple components based on the heating situation characteristics of the predicted load scenario, thereby generating energy consumption values ​​under the predicted load scenario; Step S6: Based on the predicted energy consumption value under the load scenario, iterative fan power adjustment decision is made for the mini computer host, and dynamic energy consumption optimization is performed to build an intelligent energy consumption optimization model.

2. The energy consumption analysis method of the mini computer host according to claim 1, characterized in that: The specific steps of step S2 are: Step S21: performing wavelet transform decomposition on the computer multi-component temperature monitoring data to generate a multi-component temperature frequency domain; Step S22: performing component-by-component temperature average calculation on the temperature frequency domain of multiple components to generate an average temperature value of each component; Step S23: performing temperature change analysis on the average temperature value of each component to extract the temperature change characteristics of each component; Step S24: performing temperature distribution analysis on the temperature frequency domain of multiple components based on the temperature variation characteristics of each component to obtain component temperature distribution variation data; Step S25: Perform dynamic temperature trend evolution fitting on the component temperature distribution change data to construct temperature distribution trend curves of multiple components.

3. The energy consumption analysis method of the mini computer host according to claim 1, characterized in that: The specific steps of step S4 are: Step S41: Obtain historical computer operation logs; Step S42: Analyze the load scenario changes within a period on the historical computer operation log to identify the historical period load scenario data; Step S43: performing load change cycle mining on historical periodic load scenario data to generate load scenario change rules; Step S44: simulating the evolution of load scenarios for multi-dimensional real-time monitoring data based on the load scenario change law, thereby obtaining scenario simulation data at the current monitoring time point; Step S45: performing load scenario prediction for a future period on the scenario simulation data at the current monitoring time point, thereby obtaining load scenario prediction data for the future period.

4. The energy consumption analysis method of the mini computer host according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: performing iterative fan power adjustment decisions on the mini computer host based on the energy consumption value under the predicted load scenario to generate multiple fan power adjustment parameter solutions; Step S62: performing energy consumption simulation calculation based on multiple fan power adjustment parameter schemes, thereby obtaining an energy consumption simulation value for each scheme; Step S63: extracting the fan power adjustment parameter scheme with the lowest energy consumption based on the energy consumption simulation value of each scheme; Step S64: Dynamically optimize the energy consumption of the mini computer host based on the fan power adjustment parameter solution with the lowest energy consumption, and build an intelligent energy consumption optimization model.

5. An energy consumption analysis device for a mini computer host, characterized in that: The method for performing energy consumption analysis of a mini computer host as claimed in claim 1 comprises: The load scenario recognition module is used to obtain multi-dimensional real-time monitoring data, including CPU usage, memory usage, disk I / O volume, and computer multi-component temperature monitoring data; decompose the multi-dimensional real-time monitoring data into time windows, and perform multi-load scenario recognition to obtain multiple load scenario feature data; The temperature distribution trend module is used to calculate the average temperature of each component of the computer's multi-component temperature monitoring data, perform dynamic temperature trend evolution fitting, and construct temperature distribution trend curves for multiple components; The heating trend module is used to match the scene type of the current node with the characteristic data of multiple load scenes based on the temperature distribution trend curves of multiple components, and then calculate the heating trend of components under different scenes, so as to generate the heating trend graphs of components for multiple load scenes; The load scenario prediction module is used to obtain historical computer operation logs; analyze the load scenario changes within the period of the historical computer operation logs, and predict the load scenario in the future period, so as to obtain the load scenario prediction data in the future period; The energy consumption evolution module is used to calculate the heating situation of the load scenario forecast data for the future period based on the component heating trend diagrams of multiple load scenarios, and to perform energy consumption evolution analysis, so as to generate the energy consumption value under the predicted load scenario; The energy consumption optimization module is used to make iterative fan power adjustment decisions for the mini computer host based on the energy consumption value under the predicted load scenario, and to perform dynamic energy consumption optimization to build an intelligent energy consumption optimization model.

6. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the energy consumption analysis method of the mini computer host according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the energy consumption analysis method of the mini computer host according to any one of claims 1 to 4 are implemented.

Images