Computing method for power load demand data of computing power based on flywheel energy storage system

Abstract

The application relates to the technical field of electric power energy storage, and particularly discloses a computing method for power demand data of computing power load based on a flywheel energy storage system, which comprises the following steps: obtaining real-time computing power load data of a data center server through a data center monitoring device, obtaining computing power load of a processor through analysis and calculation of the computing power load data, and calculating a comprehensive computing power load value in combination with the computing power load of the processor; based on comprehensive analysis results of the comprehensive computing power load value and environmental parameters of the server, establishing a power demand model of the server, and obtaining a power demand value through analysis; based on the obtained power demand value, obtaining a demand change gradient through analysis, and obtaining power demand data of the flywheel energy storage system through comprehensive analysis; and optimizing and adjusting the power demand data of the flywheel energy storage system according to dynamic changes of the power demand value in combination with an environmental temperature compensation function.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power storage, in particular to a computing power load power demand data calculation method based on a flywheel energy storage system. BACKGROUND

[0002] In the prior art, a single indicator is used to evaluate power demand, ignoring the influence of memory load, network bandwidth load and environmental parameters on power demand. Single-dimensional analysis cannot comprehensively and accurately reflect the actual power demand of the data center, leading to unreasonable power supply planning and problems such as insufficient supply or waste. In terms of data processing, there is a lack of systematicness and scientificity, and the data is not effectively filtered, cleaned and standardized. Noise and redundancy in the data may interfere with the analysis results, leading to deviations in the evaluation of power demand. The time series characteristics of the data are not fully utilized, and the changing trend of power demand cannot be accurately grasped. The complex situation of server operation and the influence of environmental factors are not fully considered, and the actual power demand change cannot be accurately simulated, resulting in a large error between the calculated power demand value and the actual demand, which cannot provide reliable guidance for power supply. In the management of the flywheel energy storage system, there is a lack of fine-grained strategy, and the charging and discharging strategy cannot be dynamically adjusted, which cannot fully utilize the advantages of the flywheel energy storage system, resulting in low energy utilization efficiency. The influence of environmental factors on the performance of the energy storage system is not considered, and when the server load or environmental conditions change, the power supply and the operation strategy of the energy storage system cannot be adjusted in time, leading to insufficient or excessive power supply, increasing energy cost and system operation instability. Therefore, a computing power load power demand data calculation method based on a flywheel energy storage system is needed. SUMMARY

[0003] The purpose of the present application is to provide a computing power load power demand data calculation method based on a flywheel energy storage system. To solve the above-mentioned problems in the prior art, the present application realizes the following technical scheme:

[0004] In a first aspect, the present application provides a computing power load power demand data calculation method based on a flywheel energy storage system, which specifically includes the following steps:

[0005] Real-time computing power load data of the data center server is obtained through a data center monitoring device, and the computing power load of the processor is calculated. The comprehensive computing power load value is calculated by comprehensively calculating the computing power load.

[0006] Based on the comprehensive computing power load value and the comprehensive analysis result of the environmental parameters of the server, a power demand model of the server is established, and the power demand value is analyzed.

[0007] Based on the obtained power demand value, a power demand change curve is established to analyze the demand change gradient, and the power demand data of the flywheel energy storage system is comprehensively analyzed.

[0008] According to the dynamic change of the power demand value, the power demand data of the flywheel energy storage system is optimized and compensated.

[0009] In a second aspect, the power demand data calculation system based on the flywheel energy storage system is provided, and specifically includes the following modules:

[0010] The data collection module: through the monitoring tool of the data center server operating system, the CPU usage, the memory usage, the network bandwidth usage and the process running state information are obtained, the user state, the kernel state and the idle time proportion are captured from the kernel state file, the environmental parameters of the server are collected, including the environmental temperature of the server and the cooling power of the server cooling system;

[0011] The computing power load analysis module: based on the collected data, the CPU computing power load, the memory computing power load and the network bandwidth load are calculated, the obtained CPU computing power load, the memory computing power load and the network bandwidth load are integrated, a time window matrix is constructed, m sampling points are set, the computing power load data at different time points are sampled, and the change characteristics of the computing power load data in the time sequence are captured; the sampled data is standardized, different orders of magnitude and distribution ranges of data are unified to the same scale through the standardization formula, the influence of different dimensions and value ranges of data is eliminated, based on the standardized sampling data, the information entropy of the sampling data is calculated through the formula, and the weight of the computing power load data is calculated to obtain the comprehensive computing power load value;

[0012] The power demand analysis module: based on the comprehensive computing power load value and the environmental parameters of the server, the power demand value is analyzed, the power demand change curve of the power demand value about time is established, through the curve, the change trend of the power demand value with time can be directly presented, based on the power demand change curve, the demand situation of the server in a specific working state is analyzed; based on the power demand change curve, the demand change gradient is calculated, reflecting the change rate of the power demand value with time;

[0013] The power demand data adjustment module: based on the power demand value, the flywheel energy storage system is adjusted, considering the influence of the environmental temperature on the discharge power of the flywheel energy storage system, the discharge power of the flywheel energy storage system is compensated, according to the dynamic change of the power demand value, the power demand data of the flywheel energy storage system is optimized and adjusted in real time, the change of the power demand is continuously monitored and analyzed, the charging and discharging strategy and the power of the flywheel energy storage system are adjusted, so that the whole system can adapt to the changing power demand of the data center in real time.

[0014] The beneficial effects of the present application are as follows:

[0015] 1. The server's environment parameters are considered by the computing power load data, the power demand model of the server is established based on the comprehensive computing power load value and the environment parameters by constructing the time window matrix, setting the sampling points, data standardization and calculating the information entropy, reflecting the power demand of the server under different operating conditions and environmental conditions, providing an effective basis for accurately calculating the power demand value; the power demand value is calculated in real time according to the changes of the server operating state and the environment parameters, the data obtained is effectively mined, and the quality and availability of the data are improved;

[0016] 2. The power demand change curve is constructed, the data in the power demand curve is clearly and intuitively analyzed, the demand change gradient is calculated based on the power demand change curve, and the flywheel energy storage system charging and discharging conditions are set according to the load fluctuation type and the power demand value; the flywheel energy storage system improves the energy utilization efficiency in the power supply of the data center, the discharge power is compensated by setting the environmental temperature compensation function, and the power demand data of the flywheel energy storage system is optimized and adjusted according to the dynamic change of the power demand value, so that the whole system can adapt to the changing power demand of the data center in real time. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 is the step flow chart of the computing power load power demand data calculation method based on the flywheel energy storage system provided by the embodiment 1 of the present application;

[0019] Figure 2 is the structural schematic diagram of the computing power load power demand data calculation system based on the flywheel energy storage system provided by the embodiment 2 of the present application. DETAILED DESCRIPTION

[0020] In order to make the personnel in the technical field better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0021] Embodiment 1

[0022] As Figure 1As shown, the power load demand data calculation method based on the flywheel energy storage system provided by the embodiment of the application specifically comprises the following steps:

[0023] Step one: obtain the real-time computing power load data of the data center server through the data center monitoring equipment, construct a comprehensive computing power load value through the computing power load data, comprehensively analyze the comprehensive computing power load value and the environmental parameters of the server, establish a power demand model of the server, and analyze to obtain a power demand value;

[0024] It should be noted that the computing power load data includes CPU computing power load, memory computing power load and network bandwidth load, and the environmental parameters include the environmental temperature of the server and the cooling power of the server cooling system;

[0025] In some embodiments, the CPU usage rate , memory usage rate , network bandwidth usage rate and process running state information are obtained through the monitoring tool of the data center server operating system;

[0026] It should be noted that the process running state information is a data set reflecting the current condition of the running process in the computer, covering the basic attributes, resource occupation and execution state of the process;

[0027] The user state, kernel state and idle time proportion are captured through the kernel state file, the CPU multi-core average utilization rate is calculated through normalization , and the composite weight formula is:

[0028]

[0029] The CPU computing power load is obtained , wherein, represents the number of context switches per second, represents the CPU preset context switch number reference value, represents the number of waiting processes in the running queue, represents the number of CPU physical cores, , and are preset weight coefficients, the value is 0.6, the value is 0.25, the value is 0.15;

[0030] Based on the obtained memory usage rate , the weight formula is:

[0031]

[0032] Get memory computing load ,in, It indicates the usage of memory swap partition. Indicates the amount of physical memory. It represents the cache hit rate. 、 and All are preset weight coefficients. The value is 0.7, The value is 0.2, The value is 0.1;

[0033] It should be noted that the memory swap partition usage indicates the amount of space occupied by the operating system when the physical memory of the system is insufficient and the operating system stores temporarily unused data in the physical memory in the swap partition. The cache hit rate indicates the proportion of data requested in the cache system that can be directly found in the cache.

[0034] Based on the obtained network bandwidth utilization rate By formula:

[0035]

[0036] Calculate the network bandwidth load ,in, To accept bandwidth utilization, is the sending bandwidth utilization and , It represents the network receiving bandwidth. It represents the network sending bandwidth. It indicates the nominal bandwidth of the network card. It represents the packet error rate and , It represents the preset correction factor. The value is 0.3;

[0037] Based on the obtained computing power load data, a time window matrix is ​​constructed , set m sampling points, the sampling data groups correspond to the sampling data of CPU computing load, memory computing load and network bandwidth load respectively, and the obtained sampling data are standardized through the standardization formula Get standardized sampling data ,in, is the sampling characteristic value of the computing load data in the i-th row and j-th column of the time window matrix, is the maximum value of the sampled eigenvalues ​​in the time window matrix, is the minimum value of the sampled eigenvalues ​​in the time window matrix;

[0038] Based on the obtained standardized sampling data, the sampling data information entropy is obtained by formula:

[0039]

[0040] Wherein, j is the data index of the sampling data group, i is the sampling data point index, and m represents the number of sampling points;

[0041] The weight of each computing power load data is calculated by the redundancy of the sampling data information entropy = , , respectively corresponding to the weight of CPU computing power load, memory computing power load and network bandwidth load, n is the number of sampling data points, is the redundancy of each group of sampling data and , is the number of discrete intervals, is the sampling data information entropy, which is obtained by formula:

[0042]

[0043] , wherein, is the CPU computing power load value, is the memory computing power load value, is the network bandwidth load value, represents the sampling time point, , and are the CPU computing power load, memory computing power load and network bandwidth load at time point t, , and are the CPU computing power load, memory computing power load and network bandwidth load at time point t-1, represents the sampling interval, represents the change rate compensation coefficient, which is 0.2;

[0044] Based on the obtained comprehensive computing power load value, the power demand model of the server is established based on the comprehensive environmental parameters;

[0045] It should be noted that the power demand model of the server is established by real-time monitoring of computing power load and environmental parameters, and the model can accurately quantify the energy consumption characteristics of the server cluster. For example, when the GPU utilization rate of AI training task reaches 90%, combined with the condition of room temperature 28℃, the model can dynamically adjust the power supply strategy: migrate non-real-time tasks to low temperature period for execution, use natural cooling to reduce air conditioning power consumption; At the same time, trigger the flywheel energy storage system to compensate for the instantaneous power gap, reduce the dependence on high-priced peak power of power grid;​​

[0046] Specifically, the idle power consumption of the server is obtained and the ambient temperature of the server , based on the comprehensive computing load value, the basic power consumption of the server is obtained by the formula , wherein is the linear coefficient of computing load and power consumption, and the power consumption increment corresponding to the unit computing load is obtained by fitting historical data;

[0047] Based on the obtained ambient temperature of the server , the formula is:

[0048]

[0049] The power demand model of the server is established and the power demand value is obtained , wherein represents the temperature coefficient, and the value is 0.02 / , represents the heat dissipation efficiency coefficient, the value is 0.05, represents the maximum cooling power of the cooling system, represents the ambient reference temperature;

[0050] It should be noted that the idle power consumption of the server represents the idle power consumption of the server at the ambient reference temperature , and the coupling relationship between the environmental parameters and the power demand directly affects the stability of the equipment, predicts the corrosion risk of electrical components, automatically reduces the GPU frequency to control the temperature rise, and migrates high-priority tasks to stable area servers. By analyzing the load-failure correlation in historical data, the model generates maintenance warnings in advance, considers multiple key factors such as the idle power consumption of the server, the comprehensive computing load value and the ambient temperature to calculate the basic power consumption and the power demand value. Compared with the method of simply estimating by considering only a single factor, the power consumption of the server during actual operation is more comprehensively reflected, the calculation result is closer to the true value, and more accurate data support is provided for power supply and energy consumption management; In the formula for calculating the power demand value, the temperature coefficient and the heat dissipation efficiency coefficient are explicitly introduced, the influence of the ambient temperature on the power demand is quantitatively evaluated, the corresponding change of the server power demand when the ambient temperature changes is accurately measured, and the calculation deviation of the power demand caused by the temperature is avoided;

[0051] ​The technical scheme of the present application is: by taking the environmental parameters of the server into consideration through the computing power load data, by constructing a time window matrix, setting sampling points, performing data standardization and calculating information entropy, establishing a power demand model of the server based on the comprehensive computing power load value and the environmental parameters, reflecting the power demand of the server under different operating states and environmental conditions, and providing an effective basis for accurately calculating the power demand value; the power demand value is calculated in real time according to the changes of the server operating state and the environmental parameters, the obtained data is effectively mined, and the quality and availability of the data are improved.

[0052] Embodiment 2

[0053] As Figure 1 shown, the computing power load power demand data calculation method based on the flywheel energy storage system provided by the embodiment of the present application specifically includes the following steps:

[0054] Step two: based on the obtained power demand value, a power demand change curve is established to analyze the demand change gradient, and the power demand data of the flywheel energy storage system is obtained through comprehensive analysis; the power demand data of the flywheel energy storage system is optimized and adjusted according to the dynamic changes of the power demand value;

[0055] It should be noted that the power demand data includes the discharge power and the charge power of the flywheel energy storage system.

[0056] In some embodiments, the power demand value of the server is obtained, and a power demand change curve of the power demand value with respect to time is established with time as the X-axis and the power demand value as the Y-axis;

[0057] It should be noted that the fluctuation of the server power demand with time is observed intuitively through the curve, the change type and the increase and decrease of the power demand value are clearly understood, and the abnormal value or mutation point of the power demand is easily identified; when the curve shows obvious peaks or troughs, which do not conform to the normal trend, it can be quickly detected that there may be server failure, business anomaly and other problems, which is convenient for timely investigation and processing; for example, if the curve shows periodic peaks and troughs, it means that the business has periodic busy and idle periods.

[0058] Based on the obtained power demand change curve, the peak value generated by the instantaneous power surge of the CPU full load in the power demand curve is analyzed , the high-load average power demand value of the continuous high-load section during batch task processing is analyzed , and the power demand value of the trough section during the night low-load period is analyzed .

[0059] It should be noted that by analyzing the instantaneous power consumption surge peak value when the CPU is full, the power consumption of the server hardware under extreme load is understood, thereby evaluating the performance boundary of the hardware, which helps to determine whether the server can withstand sudden high-load tasks, provides a basis for hardware upgrade or expansion, analyzes the high-load average power demand value of the continuous high-load section when processing batch tasks, understands the demand characteristics of batch tasks for power resources, optimizes the task scheduling strategy, improves the overall performance and stability of the system, and analyzes the power demand value of the low-load section at night, discovers whether there is power waste or resource idling in the low-load state of the system, and provides a direction for further performance optimization.

[0060] Calculate the demand change gradient based on the power demand change curve , wherein, represents the difference value of the power demand value between adjacent sampling time points, represents the sampling time interval, and identifies sudden load fluctuation and gradual load fluctuation;

[0061] It should be noted that the sampling time is set by the professional technical personnel of the present application according to historical experience, and can be adjusted according to actual conditions;

[0062] If the absolute value of the demand change gradient is greater than or equal to the preset demand change gradient threshold , it indicates that the load fluctuation is sudden load fluctuation;

[0063] If the absolute value of the demand change gradient is less than or equal to the preset demand change gradient threshold , it indicates that the load fluctuation is gradual load fluctuation;

[0064] If the absolute value of the demand change gradient is greater than the preset demand change gradient threshold and less than the preset demand change gradient threshold , it indicates that the load fluctuation is mixed load fluctuation;

[0065] It should be noted that the preset demand change gradient threshold is set by the professional technical personnel in the field of the present application according to historical experience, and can be adjusted according to actual conditions;

[0066] Based on the identified load fluctuation type and the obtained power demand value, the power demand data of the flywheel energy storage system is analyzed;

[0067] Specifically, if the demand change gradient is greater than or equal to the preset demand change gradient threshold and the power demand value is greater than or equal to the preset power demand threshold , the flywheel energy storage system triggers the discharge condition, and the discharge power is calculated by the formula:

[0068]

[0069] Discharge power of flywheel energy storage system , wherein, represents the rated maximum power of the flywheel, represents the current energy storage state of the flywheel energy storage system, and ;

[0070] If the power demand value is less than or equal to the preset power demand threshold and the grid is redundant, the flywheel energy storage system triggers a charging condition, and the charging power is calculated by the formula:

[0071]

[0072] Charging power of flywheel energy storage system , wherein, represents the charging efficiency of the flywheel energy storage system;

[0073] If the power demand value is greater than the preset power demand threshold and less than the preset power demand threshold , the flywheel energy storage system does not discharge to the server, and charges based on the rated charging power of the flywheel energy storage system until the energy storage state is 100%;

[0074] It should be noted that by calculating the demand change gradient of the power demand change curve and comparing it with the preset demand change gradient threshold, the load fluctuation can be accurately divided into sudden, gradual and mixed types, and accurate identification of the fluctuation type helps the operation and maintenance personnel to deeply understand the change characteristics of the server load, and different coping strategies are adopted for different types of fluctuations;

[0075] Based on the obtained discharge power of the flywheel energy storage system , the discharge power of the flywheel energy storage system is compensated by the ambient temperature, and the ambient temperature compensation function is calculated by the formula:

[0076]

[0077] The discharge power of the flywheel energy storage system is compensated, wherein, represents the ambient temperature of the server, represents the preset temperature compensation coefficient, the value is 0.6, represents the preset compensation reference temperature, and ;

[0078] Based on the obtained ambient temperature compensation function, the formula is:

[0079]

[0080] Compensated optimized compensation discharge power wherein, represents the server no-load power consumption, represents the power demand value of the server, represents the preset power demand threshold value, represents the flywheel rated maximum power, represents the current energy storage state of the flywheel energy storage system and , represents the ambient temperature compensation function;

[0081] It should be noted that the ambient temperature will affect the performance of the flywheel energy storage system. By introducing the ambient temperature compensation function, the discharge power is adjusted in real time according to the actual ambient temperature, so that the flywheel energy storage system can maintain a relatively stable discharge power output under different temperature conditions, enhance the adaptability of the system to the environment, reduce the performance fluctuation of the system caused by temperature change, and ensure that the flywheel energy storage system can reliably provide stable power support for the server under different temperature environments. Considering the compensation of the discharge power according to the ambient temperature, the flywheel energy storage system can reasonably discharge under different temperatures, avoid abnormal discharge power caused by too high or too low temperature, thereby improving the overall energy storage and discharge efficiency of the flywheel energy storage system, fully utilizing the energy storage capacity of the flywheel energy storage system, realizing the optimal allocation of resources, and combining the server no-load power consumption and the preset power demand threshold value to calculate the compensation discharge power, so that the discharge power of the flywheel energy storage system is better matched with the actual power demand of the server.

[0082] The technical scheme of the embodiment of the application is: by constructing a power demand change curve, the data in the power demand curve is clearly and intuitively analyzed, the demand change gradient is calculated based on the power demand change curve, and the sudden load fluctuation, gradual load fluctuation and mixed load fluctuation are identified based on the demand change gradient. According to the load fluctuation type and the power demand value, the flywheel energy storage system charging and discharging conditions are set; the flywheel energy storage system improves the energy utilization efficiency in the power supply of the data center, considers the influence of the ambient temperature, and compensates the discharge power by setting the ambient temperature compensation function. According to the dynamic change of the power demand value, the power demand data of the flywheel energy storage system is optimized and adjusted, so that the whole system can adapt to the changing power demand of the data center in real time. The accurate analysis of the power demand and the reasonable management of the flywheel energy storage system improve the stability of the data center power supply system.

[0083] Example 3

[0084] As Figure 2As shown, the power load and power demand data calculation system based on the flywheel energy storage system provided by the embodiment of the application specifically comprises the following modules:

[0085] The data collection module: through the monitoring tool of the data center server operating system, the CPU usage, memory usage, network bandwidth usage and process running state information are obtained, the user state, kernel state and idle time proportion are captured from the kernel state file, and the environmental parameters of the server are collected, including the environmental temperature of the server and the cooling power of the server cooling system;

[0086] The power load analysis module: based on the collected data, the CPU power load, memory power load and network bandwidth load are calculated, the obtained CPU power load, memory power load and network bandwidth load are integrated, a time window matrix is constructed, m sampling points are set, the power load data at different time points are sampled, and the change characteristics of the power load data in the time sequence are captured; the sampled data is standardized, different magnitudes and distribution ranges of data are unified to the same scale through the standardization formula, the influence of different dimensions and value ranges of data is eliminated, based on the standardized sampling data, the information entropy of the sampling data is calculated through the formula, and the weight of the power load data is calculated to obtain the comprehensive power load value;

[0087] The power demand analysis module: based on the comprehensive power load value and the environmental parameters of the server, the power demand value is analyzed, the power demand change curve of the power demand value about time is established, through the curve, the change trend of the power demand value with time can be directly presented, based on the power demand change curve, the demand situation of the server in a specific working state is analyzed; based on the power demand change curve, the demand change gradient is calculated, reflecting the change rate of the power demand value with time;

[0088] The power demand data adjustment module: based on the power demand value, the flywheel energy storage system is adjusted, considering the influence of environmental temperature on the discharge power of the flywheel energy storage system, the discharge power of the flywheel energy storage system is compensated, according to the dynamic change of the power demand value, the power demand data of the flywheel energy storage system is optimized and adjusted in real time, the change of the power demand is continuously monitored and analyzed, the charging and discharging strategy and power of the flywheel energy storage system are adjusted, so that the whole system can adapt to the changing power demand of the data center in real time.

[0089] The above describes one embodiment of the present application in detail, but the content is only the preferred embodiment of the present application and cannot be considered to limit the implementation range of the present application; the above formulas are all de-dimensioned to calculate the numerical values, the formulas are obtained by collecting a large amount of data to simulate a formula of the most recent real situation, and the preset parameters in the formulas are set by the person skilled in the art according to the actual situation and historical experience, and can be adjusted according to the actual situation; any equivalent changes and improvements made according to the application range of the present application should still belong to the patent coverage range of the present application.

Claims

1. A method for calculating power demand data of computing power based on a flywheel energy storage system, characterized in that, Comprise the following steps: The computing power load of the processor is calculated through real-time computing power load data, and a comprehensive computing power load value is obtained through comprehensive calculation of the computing power load; The acquisition process of the computing power load of the processor is: The CPU multi-core average utilization rate, memory usage rate and network bandwidth usage rate are calculated through normalization, and a composite weight formula is used: CPU computing power load memory computing power load and network bandwidth load wherein, CPU usage rate, is the number of context switches per second, is the CPU preset context switch number reference value, is the number of processes waiting in the run queue, is the number of CPU physical cores, memory usage rate, , and are preset weight coefficients, is the memory swap partition usage amount, is the physical memory amount, is the cache hit rate, , and are preset weight coefficients, is the network receiving bandwidth, is the network sending bandwidth, is the network card nominal bandwidth, is the data packet error rate and , is the preset correction coefficient; The acquisition process of the comprehensive computing power load value is: Calculate the weight of each computing power load data by the redundancy of sampling data information entropy = , , respectively corresponding to the weight of CPU computing power load, memory computing power load and network bandwidth load, wherein j is the data index of the sampling data group, i is the sampling data point index, n is the number of sampling data points, is the redundancy of each group of sampling data and , is the number of discrete intervals, is the sampling data information entropy, which is calculated by the formula: obtaining a comprehensive computing power load value wherein, is a CPU computing power load value, is a memory computing power load value, is a network bandwidth load value, represents a sampling time point, , and are respectively a CPU computing power load, a memory computing power load and a network bandwidth load at a time point t, , and are respectively a CPU computing power load, a memory computing power load and a network bandwidth load at a time point t-1, represents a sampling interval, represents a change rate compensation coefficient; An electric power demand model of the server is established through comprehensive analysis of the comprehensive computing power load value and the environmental parameters of the server, and an electric power demand value is obtained through analysis; Based on the electric power demand value, an electric power demand change curve is established, a demand change gradient and electric power demand data of the flywheel energy storage system are obtained through analysis; The electric power demand data of the flywheel energy storage system is optimized and compensated according to the dynamic change of the electric power demand value and an environmental temperature compensation function.

2. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 1, characterized in that, The acquisition process of the sampling data information entropy is: Based on the obtained computing power load data, a time window matrix is constructed , m sampling points are set, the obtained sampling data is standardized, and the standardized sampling data is obtained through a standardization formula , wherein, is the sampling eigenvalue of the computing power load data in the ith row and jth column of the time window matrix, is the maximum value of the sampling eigenvalue in the time window matrix, is the minimum value of the sampling eigenvalue in the time window matrix;​ Based on the obtained standardized sampling data, a formula is used: Obtaining sample data information entropy where j is a sample data group data index, i is a sample data point index, and m represents the number of sample points.

3. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 1, characterized in that, The acquisition process of the electric power demand value is: Obtaining server idle power consumption and the ambient temperature of the server , based on the comprehensive computing power load value through the formula to obtain the server basic power consumption , wherein, is the linear coefficient of computing power load and power consumption; Based on the obtained ambient temperature of the server by the formula: establishing a power demand model of the server and obtaining a power demand value wherein, represents a temperature coefficient, represents a heat dissipation efficiency coefficient, represents a maximum cooling power of the cooling system, represents an ambient reference temperature.

4. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 1, characterized in that, The acquisition process of the demand change gradient is: Calculating demand change gradient based on power demand change curve wherein, represents the difference of power demand values between adjacent sampling time points, represents the sampling time interval, identifying the sudden load fluctuation and the gradual load fluctuation.

5. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 1, characterized in that, The acquisition process of the electric power demand data is: Based on the identified load fluctuation type and the obtained electric power demand value, the discharge power and the charging power of the flywheel energy storage system are obtained through analysis; When the demand change gradient is greater than or equal to a preset demand change gradient threshold value and the power demand value is greater than or equal to a preset power demand threshold value , a discharging condition is triggered, and the discharging power of the flywheel energy storage system is calculated through a discharging power formula.

6. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 5, characterized in that, The acquisition process of the charging power is: When the power demand value is less than or equal to a preset power demand threshold and the power grid is redundant, a charging condition is triggered, and the charging power of the flywheel energy storage system is calculated by a charging power formula.

7. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 1, characterized in that, The acquisition process of the environmental temperature compensation function is: Based on the obtained discharging power of the flywheel energy storage system, power compensation is performed on the discharging power of the flywheel energy storage system by an ambient temperature, and an ambient temperature compensation function is calculated by a formula The discharging power of the flywheel energy storage system is compensated by the ambient temperature, wherein, represents the ambient temperature of the server, is a preset temperature compensation coefficient, is 0.6, is a preset compensation reference temperature, and .

8. The flywheel-based energy storage system-based computing power load electricity demand data calculation method according to claim 1, characterized in that, The optimization compensation process is: Based on the obtained ambient temperature compensation function, the compensated optimized compensation discharge power is obtained by formula wherein, represents the server no-load consumption power, represents the power demand value of the server, represents the preset power demand threshold value, represents the rated maximum power of the flywheel, represents the current energy storage state of the flywheel energy storage system, and , represents the ambient temperature compensation function.

Images