CN115994714B - IDC machine room lithium battery efficiency evaluation method based on big data statistics - Google Patents
IDC machine room lithium battery efficiency evaluation method based on big data statistics Download PDFInfo
- Publication number
- CN115994714B CN115994714B CN202310279444.2A CN202310279444A CN115994714B CN 115994714 B CN115994714 B CN 115994714B CN 202310279444 A CN202310279444 A CN 202310279444A CN 115994714 B CN115994714 B CN 115994714B
- Authority
- CN
- China
- Prior art keywords
- load
- lithium battery
- daily
- energy consumption
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 title claims abstract description 87
- 229910052744 lithium Inorganic materials 0.000 title claims abstract description 87
- 238000011156 evaluation Methods 0.000 title claims abstract description 74
- 238000005265 energy consumption Methods 0.000 claims abstract description 106
- 230000006641 stabilisation Effects 0.000 claims abstract description 63
- 238000011105 stabilization Methods 0.000 claims abstract description 63
- 238000012216 screening Methods 0.000 claims abstract description 11
- 238000000034 method Methods 0.000 claims description 39
- 230000000875 corresponding effect Effects 0.000 claims description 13
- 238000012512 characterization method Methods 0.000 claims description 8
- 230000002459 sustained effect Effects 0.000 claims description 8
- 230000000087 stabilizing effect Effects 0.000 claims description 5
- 230000002596 correlated effect Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 4
- 230000002354 daily effect Effects 0.000 description 107
- 238000004364 calculation method Methods 0.000 description 9
- 239000013598 vector Substances 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000005923 long-lasting effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000007599 discharging Methods 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 230000002688 persistence Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
Images
Landscapes
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The invention relates to the technical field of data processing, in particular to an IDC machine room lithium battery efficiency evaluation method based on big data statistics; since the lithium battery in the IDC machine room may have a continuous high load or a continuous low load in some time periods, the accuracy and the reliability of the calculated lithium battery performance evaluation value are low in the continuous high or low load. Therefore, in order to obtain a reliable evaluation value, a daily load representation value of the battery is obtained according to the energy consumption condition of the daily lithium battery; and screening out a load stabilization time sequence with stable lithium battery load state and long duration date, calculating the preference degree of the load stabilization time sequence, selecting an optimal evaluation period range as a clustering range, obtaining reference data of lithium battery efficiency evaluation through clustering and screening, and obtaining an accurate and reliable lithium battery efficiency evaluation result according to an SPUE mean value of the reference data.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an IDC machine room lithium battery efficiency evaluation method based on big data statistics.
Background
The IDC machine room is an internet data center (Internet Data Center) and is used for providing server storage and IT services for units in a data exchange set of a network operator or a government enterprise. With standardization of data center construction, lithium batteries are mostly used as a power supply system of the data center. The electric energy input by IDC to each server is not fully applied to the processor operation and storage module device, and in the process of evaluating the lithium battery efficiency of IDC machine room, the general measurement index includes the energy efficiency (SPUE) of the server device, the SPUE is the ratio of the total energy consumption of the server to the operation related energy consumption, and the reference of SPUE is 2, and the closer to 1, the better the efficiency level is indicated.
However, the task type operated by IDC every day is too complicated and the task amount is not fixed, and the calculated sple values of lithium batteries under different degrees of loads are different, so that the accuracy and reliability of the sple values of the lithium battery performance evaluation in IDC machine room are low, and it is difficult to obtain accurate and reliable performance evaluation values.
Disclosure of Invention
In order to solve the technical problem that the accuracy and the reliability of the SPUE value calculated by the lithium battery under different degrees of loads are low, the invention aims to provide an IDC machine room lithium battery efficiency evaluation method based on big data statistics, which adopts the following technical scheme:
collecting daily total energy consumption, daily task amount and daily task average energy consumption of a server battery in an evaluation period; obtaining daily task volume density according to daily task volume, and obtaining a daily load representation value according to the daily task volume density and daily task energy consumption variance;
obtaining a load characteristic time sequence according to the daily load characteristic value, dividing the load characteristic time sequence according to the element change rate of each position in the load characteristic time sequence to obtain at least two load stability time sequences, determining a load stability time sequence interval and a change rate information entropy value in the load stability time sequence, and obtaining the preference degree of the load stability time sequence according to the load stability time sequence interval and the change rate information entropy value;
obtaining the similarity of the energy consumption states among different dates according to the total daily task energy consumption and the daily task energy consumption variance of different dates; obtaining a local reachable range in the clustering process according to a load stabilization time sequence corresponding to the maximum preference degree and clustering data of different dates;
and screening each cluster according to a preset continuous high-load probability threshold value or a preset continuous low-load probability threshold value to obtain reference data, and evaluating battery energy consumption information of the reference data to obtain a battery efficiency evaluation result.
Further, the step of obtaining the daily load characterization value includes:
in the formula (i),
represent the first
Daily load characterization values for the day,
the daily task amount is represented by the number of tasks,
indicating any one of the running tasks on the same day,
represent the first
The energy consumption of the individual tasks is calculated,
represents the average energy consumption of the daily task,
representing daily task volume density.
Further, the step of obtaining the load stabilization time series includes:
calculating a first derivative of the load characteristic time sequence to obtain a load change trend function; and taking a time point with the function value of 0 in the load change trend function as a division point, and dividing the load characteristic time sequence according to the division point to obtain at least two load stabilizing time sequences.
Further, the obtaining step of the preference degree includes:
in the formula (i),
is the first
The degree of preference of the segment load stabilization time series,
as a function of the hyperbolic tangent,
is an exponential function with a base of a natural constant,
is the first
The time series interval of the segment load stabilization,
is the first
The change rate information entropy value of the segment load stabilization time sequence.
Further, the step of obtaining the local reach includes:
and calculating the square root of the load stabilization time sequence interval corresponding to the maximum preference degree, and obtaining the local reachable range.
Further, the step of obtaining the sustained high load probability includes:
and calculating the Euclidean norm of the sum of the normalized number of dates and the normalized average energy consumption of the daily tasks in the cluster to obtain the continuous high-load probability.
Further, the step of obtaining the sustained low-load probability includes:
and calculating the Euclidean norm of the sum of the normalized number of dates in the cluster and the negatively correlated normalized average energy consumption of the daily task to obtain the continuous low-load probability.
Further, the step of obtaining the reference data includes:
and presetting a continuous high load probability threshold value and a continuous low load probability threshold value, wherein when the continuous high load probability value and the continuous low load probability value in the cluster are smaller than the corresponding continuous high load probability threshold value and the continuous low load probability threshold value, the data in the cluster are the reference data.
Further, the step of evaluating the battery performance includes:
and calculating the average value of the SPUE values of all the reference data to obtain a battery efficiency evaluation result.
The invention has the following beneficial effects:
obtaining a daily load representation value according to the daily task volume density and the daily task energy consumption variance, wherein the aim is to analyze whether a continuous high load or continuous low load state exists in the lithium battery according to the daily load representation value; dividing the load characteristic time sequence according to the element change rate of each position in the load characteristic time sequence to obtain at least two load stability time sequences, wherein the aim is to analyze the change condition of the load state of the lithium battery in an evaluation period and analyze the preference degree of the load stability time sequences through the change condition; obtaining the preference degree of the load stabilization time sequence according to the load stabilization time sequence interval and the change rate information entropy value, and finding the optimal evaluation cycle range with normal lithium battery load and long lasting date according to the value of the preference degree; obtaining the similarity of energy consumption states among different dates according to the total daily task energy consumption and the daily task energy consumption variance of different dates, wherein the purpose is to cluster the daily load states of the lithium batteries according to the similarity, and obtain the local reachable range in the clustering process according to the load stabilization time sequence corresponding to the maximum preference degree, and the purpose is to obtain the local reachable range used in the clustering algorithm, namely the searching radius, so that the clustering effect is better; the continuous high load probability and the continuous low load probability are obtained together according to the date number in each cluster obtained by clustering and the average energy consumption of daily tasks, so that cluster data which can be continuous high load and continuous low load in the evaluation period range are eliminated, and the accuracy and the reliability of efficiency evaluation are further improved. The battery efficiency evaluation result is obtained by evaluating the battery energy consumption information of the reference data, so that the energy consumption data of continuous high load and continuous low load are removed, the proper energy consumption evaluation data are obtained, the accuracy and the reliability of the lithium battery efficiency evaluation result in the evaluation period are higher based on the proper energy consumption evaluation data, and the error is reduced.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a method for evaluating lithium battery performance of an IDC machine room based on big data statistics according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of the specific implementation, structure, characteristics and effects of the method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to the invention by combining the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The invention provides a specific scheme of an IDC machine room lithium battery efficiency evaluation method based on big data statistics, which is specifically described below with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a method for evaluating performance of a lithium battery in an IDC machine room based on big data statistics according to an embodiment of the present invention is shown, and the method includes the following steps:
step S1, collecting total daily energy consumption, daily task amount and daily task average energy consumption of a server battery in an evaluation period; and obtaining daily task volume density according to the daily task volume, and obtaining a daily load representation value according to the daily task volume density and the daily task energy consumption variance.
In the embodiment of the invention, since the performance of the lithium battery gradually decays along with the increase of the service time, in order to reduce the error of the final calculation of the SPUE value, one year is selected as an evaluation period, namely, the efficiency evaluation value is obtained according to the machine room energy consumption data in one year, and it is required to be explained that an implementer can determine the time range of the evaluation period according to the implementation scene in the implementation process. Because the SPUE is the ratio of the total energy consumption of the server to the operation-related energy consumption, in an IDC machine room, a power supply, a voltage stabilizing module, a fan and a lighting lamp device can occupy more than twenty percent of power supply energy consumption of the whole machine room, and besides, the power supply, the voltage stabilizing module, the fan and the lighting lamp device are net energy consumption generated by the task operation of the server. Before calculating the SPUE value, acquiring total energy consumption data of the server and energy consumption data for performing task operation in a lithium battery efficiency evaluation period, wherein the total energy consumption data of the server per day can be acquired and exported by an energy consumption monitoring module at the background of server equipment of a machine room, the total energy consumption comprises various electric equipment energy consumption related to the server, and the total energy consumption data of the server per day is called total energy consumption per day for short; the energy consumption data of task operation needs to be added with a power consumption task log storage module in a server, the daily task quantity in an evaluation period and the energy consumption of each task operation are obtained through the power consumption task log storage module, the energy consumption of each task operation every day is simply called daily task energy consumption, and the sum of the daily task energy consumption is called daily task total energy.
Although the spee value can represent the percentage of net energy consumption in the total energy consumption, the different number or size of tasks operated by the IDC machine room daily can cause the different loads of the lithium battery, resulting in different recorded spee values of the lithium battery under different degrees of load. Because the best working mode of the lithium battery is continuous shallow charging and shallow discharging, the loss of the continuous shallow charging and shallow discharging to the lithium battery is minimum. The IDC machine room uses an uninterruptible power system, and if the lithium battery continuously works in a high-load state, the lithium battery is in a state of overcharge and overdischarge and the internal temperature of the battery is high. If the charging current of the lithium battery is too large or the charging time is too long, the situation that oxygen generated in the battery is delayed and cannot be consumed, the internal pressure is increased and internal ion discharge is accelerated is caused, the power consumption recorded by the power consumption task log storage module is finally increased, and meanwhile, the capacity of the battery is reduced due to frequent overcharging and overdischarging. The SPUE value calculated by the lithium battery working under the high load state can lead the energy consumption duty ratio of the task operation to be increased, namely the denominator in the SPUE calculation formula is increased, so that the SPUE value is smaller. If the lithium battery is continuously operated in a lower load state, part of the lithium battery pack is in an idle state, internal electric energy is not frequently used, the lithium battery capacity can still be lost due to insufficient discharge, and related data show that the lithium battery can lose about one percent to two percent of the battery capacity per month when the lithium battery is idle. The reduction of the capacity of the lithium battery can lead to the increase of the total energy consumption ratio of the server, namely the increase of molecules in the calculation formula of the SPUE, so that the calculated SPUE value is larger.
Therefore, in the process of evaluating the energy efficiency of the lithium battery, if the energy consumption data obtained by the lithium battery in the continuous high-load state or the continuous low-load state is added to the calculation of the SPUE value, the accuracy and the reliability of the performance evaluation data of the lithium battery are reduced. The existing evaluation mode is to participate all energy consumption data in an evaluation period in the calculation of the efficiency evaluation of the lithium battery, and the energy consumption data under the continuous high-load or continuous low-load state work is not removed. In order to obtain accurate and reliable evaluation results, data in an evaluation period need to be screened, an evaluation period range with stable and normal lithium battery load and long lasting date is screened, the screening process needs to be calculated according to daily task energy consumption data at first, and the load condition of daily server task operation energy consumption on the lithium battery is analyzed.
Obtaining the average daily task energy consumption according to the ratio of the total daily task energy consumption to the daily task energy consumption, wherein the average daily energy consumption can be expressed asThe day server performs the task operation process on the load of the lithium battery, but the average daily energy consumption cannot indicate the condition that the lithium battery is continuously or indirectly loaded within a period of time. Therefore, in order to analyze the persistence of the lithium battery load state in a period of time, a daily load characteristic value needs to be calculated, and before the daily load characteristic value is calculated, the daily task volume density needs to be obtained, and if the total daily time is 1, then
Representing total daily time and daily task amount
When the daily task amount density value is lower, the more the daily task amount is, the higher the working frequency of the lithium battery is. Obtaining a daily load representation value by calculating the reciprocal of the product of the daily task volume density and the daily task energy consumption variance; the specific formula for acquiring the daily load characterization value comprises the following steps:
in the method, in the process of the invention,
represent the first
Daily load characterization values for the day,
the daily task amount is represented by the number of tasks,
indicating any one of the running tasks on the same day,
represent the first
The energy consumption of the individual tasks is calculated,
representing the average energy consumption of the daily task.
Represents the daily task volume density,
to calculate the formula for the daily task energy consumption variance,
representing the formula of variance to be calculated
The dimensionality removal process is performed in order to make the daily task energy consumption variance the same as the dimensionality of the daily task bulk density value, and 1 is added in order to prevent the variance result from being 0. Because the smaller the product of the daily task volume density and the daily task energy consumption variance in the formula is, the more the daily task volume is represented, and the task volume energy consumption is relatively uniform, in order to make the result show positive correlation, the inverse of the product of the daily task volume density and the daily task energy consumption variance is taken as a daily load representation value. When the daily load representation value is larger, the daily task quantity is more, and the task energy consumption is more uniform; if the daily load characteristic value is higher in the sequence, the lithium battery is in a load state of high load or low load continuously in the sequence. When the daily load representation value is at a normal level, the task amount is in a normal range, and the task energy consumption difference distance is in a normal range. When the lithium battery is in a continuous high-load or continuous low-load state in a certain section of sequence, the energy consumption data of the section of sequence is required to be removed when the efficiency evaluation of the lithium battery is analyzed, so that the error of the efficiency evaluation of the lithium battery is reduced. And further calculating lithium battery efficiency evaluation according to the daily load characterization value.
Step S2, a load characteristic time sequence is obtained according to the daily load characteristic value, at least two load stabilization time sequences are obtained by dividing the load characteristic time sequence according to the element change rate of each position in the load characteristic time sequence, the load stabilization time sequence interval and the change rate information entropy value in the load stabilization time sequence are determined, and the preference degree of the load stabilization time sequence is obtained according to the load stabilization time sequence interval and the change rate information entropy value.
After the daily load representation value is obtained, the load state characteristics of the daily lithium battery can be represented according to the daily load representation value, and the lithium battery efficiency evaluation can be carried out by screening a proper evaluation period range according to the load state characteristics. Screening for a suitable evaluation cycle range first requires analysis based on the change in daily load characterization values, and a period of time during which the continuous load is more normal and the duration date is longer in an evaluation cycle of one year may be referred to as a suitable evaluation cycle range.
After the daily load representation values of all the dates in the evaluation period are obtained, a two-dimensional coordinate system of time and the daily load representation values can be constructed, the daily load representation values of different dates can be found to be densely distributed around a curve in the coordinate system, so that the daily load representation values of different dates can be fitted through a least square method to obtain a curve, curve data fitted through the least square method has the characteristics of low frequency and smoothness, the basic change trend of the data can be reflected, the normal, stable and continuous range of the load representation values can be found more easily through a data curve, and the proper evaluation period range can be further determined. The fitted data curve is called as a load characteristic time sequence, and it should be noted that the least square method belongs to the disclosure technology, and the specific fitting process is not repeated. Dividing the load characteristic time sequence according to the element change rate of each position in the load characteristic time sequence to obtain at least two load stability time sequences, namely dividing the load characteristic time sequence for analyzing the change rate of daily load characterization values of different dates. Preferably, calculating the first derivative of the data curve function fitted by the least square method, namely calculating the first derivative of the load characteristic time series function, wherein the first derivative of the function can represent the change rate of the function, and the load change trend function is obtained; and dividing the load characteristic time sequence by taking the time point with the function value of 0 in the load change function as a dividing point, namely, taking the time point with the change rate of 0 as the dividing point to obtain at least two load stabilizing time sequences. And calculating the preference degree of each load stabilization time sequence, and screening out the load stabilization time sequence with normal continuous load and long continuous date through the maximum value of the preference degree as an evaluation period range.
Calculating the preference degree of the load stabilization time series first requires calculating the load stabilization time series interval within the load stabilization time series, i.e. the number of time points per load stabilization time series, the more time points, meaning the longer the load stabilization time series. After the load stabilization time sequence interval is obtained, the change rate information entropy value of the load stabilization time sequence is required to be calculated, the information entropy value can represent the disorder of data, the larger the information entropy value is, the more unstable the change condition of the segment of sequence is, and the specific steps for obtaining the information entropy value are as follows: firstly, determining a load change trend function value corresponding to a first derivative load change trend function of a load characteristic time sequence at a time point of each load stability time sequence, wherein the load change trend function value represents the change rate of a daily load representation value at each time point, counting the change rate type and the change rate type quantity, wherein the change rate type represents different change rate values, the change rate type quantity represents the quantity of each change rate type, for example, five change rate values are obtained in a certain load stability time sequence, and are respectively 1, 2, 3 and 4, and the change rate type quantity of the change rate type 1 is 2; the probability or the duty ratio of each type can be obtained through the ratio of the number of the change rate types to the total number of the change rates, the change rate information entropy value of the load stabilization time sequence is calculated, when the change rate information entropy value is smaller, the change of the load stabilization time sequence is smaller and more stable, and a calculation formula of the information entropy value is needed to be described, which belongs to the known technology, and the specific calculation steps are not repeated.
After the load stabilization time series interval and the change rate information entropy value of the load stabilization time series are obtained, the load stabilization time series can be considered to have smaller change and longer duration date when the load stabilization time series interval is longer and the change rate information entropy value of the load stabilization time series is smaller according to the load stabilization time series interval and the change rate information entropy value of the load stabilization time series. The obtaining step of the preference degree is as follows: calculating the European norm of the normalized load stabilization time sequence interval and the inversely related normalized change rate information entropy value to obtain the optimization degree of the load stabilization time sequence; the specific formula for obtaining the preference degree is as follows:
in the method, in the process of the invention,
is the first
The degree of preference of the segment load stabilization time series,
as a function of the hyperbolic tangent,
is an exponential function with a base of a natural constant,
is the first
The time series interval of the segment load stabilization,
is the first
The change rate information entropy value of the segment load stabilization time series is subjected to the negative correlation normalization processing by an exponential function based on a natural constant, because a smaller information entropy value of the change rate means a smaller change of the load stabilization time series.
Since the normal load state of the lithium battery cannot be ensured on all dates in the load stabilization time series, in order to further reduce the evaluation error, the evaluation data needs to be further screened. In the subsequent step, reference data of the evaluation cycle range are screened through clustering, wherein the reference data refer to energy consumption data which can normally participate in lithium battery efficiency evaluation.
Step S3, obtaining the similarity of the energy consumption states among different dates according to the total daily task energy consumption and the daily task energy consumption variance of different dates; and obtaining a local reachable range in the clustering process according to the load stabilization time sequence corresponding to the maximum preference degree, and clustering the data on different dates.
In order to screen out the reference data further describing the evaluation period range, firstly, the daily task total energy consumption and the daily task energy consumption variance of different dates in the evaluation period range can be clustered according to the date that the daily task total energy consumption and the daily task energy consumption variance are close, and the load states of the lithium batteries are similar, so that the daily task total energy consumption and the daily task energy consumption variance of the same day can be constructed into a daily load state vector:
，
represent the first
Daily load status of dayThe vector quantity is used to determine the vector quantity,
represent the first
Total daily task energy vector component of the day,
represent the first
Daily task energy consumption variance vector component of the day. And calculating cosine similarity through evaluating daily load state vectors between any two different dates in the range period, wherein the larger the cosine similarity is, the similar is considered to be the load states of the lithium batteries on any two dates. It should be noted that cosine similarity is a disclosure technology, and specific calculation steps are not described again.
In the embodiment of the invention, after the cosine similarity of the lithium battery load state is calculated, the similar lithium battery load states can be clustered into a cluster through a DBSCAN clustering algorithm, wherein the DBSCAN clustering algorithm is essentially a clustering algorithm based on local density, is more sensitive to the data which are locally continuous and scattered in time sequence, and is more accurate in segmentation. The different load states within the evaluation period range, such as continuous high or low load, indirect high or low load, etc., are distinguished by clustering. The result of the DBSCAN clustering algorithm depends on the setting of a local reachable range, namely the search radius in the DBSCAN clustering algorithm, a load stabilization time sequence with the maximum preference degree is obtained in the step S2 as an evaluation range period and a corresponding load stabilization time sequence interval, and the square root of the load stabilization time sequence interval corresponding to the maximum preference degree is calculated to obtain the local reachable range. The purpose of calculating the square root of the load stabilization time series interval corresponding to the maximum degree of preference is to: assuming that the time series is a region in a data set, if a window is needed to completely traverse the region, then the optimal window size is the square root of the area of the region, analogized to the load stabilization time series, the optimal window is the local reach is the square root of the load stabilization time series interval, and if the local reach is not an integer, then the window is rounded down. And clustering based on the load state similarity of the lithium battery after the optimal local reachable range of the DBSCAN clustering algorithm is obtained, so as to obtain at least two clusters with different load states. It should be noted that, the DBSCAN clustering algorithm is a public technology, specific clustering steps are not repeated, and in the embodiment of the present invention, the clustered objects are in a sequence form, so that only local reachable distances required by clustering are determined, and data densities within the local reachable distances are determined in sequence without further calculation and determination.
After the clustering is completed, clusters of different lithium battery load states are obtained, and energy consumption data of the suitable lithium battery load states need to be screened out for efficiency evaluation.
And S4, obtaining continuous high load probability and continuous low load probability according to the date number in each cluster obtained by clustering and average energy consumption of daily tasks, screening each cluster according to a preset continuous high load probability threshold or a preset continuous low load probability threshold to obtain reference data, and obtaining a battery efficiency evaluation result by evaluating battery energy consumption information of the reference data.
Because in step S1, the accuracy and reliability of the SPUE value are affected by the analysis of the energy consumption data of the lithium battery in the continuous high or low load, in order to obtain an accurate and reliable performance evaluation SPUE value, the energy consumption data of the continuous high or low load needs to be removed. The step of calculating the continuous high load probability of the data in the cluster is as follows: calculating the Euro norm of the sum of the normalized number of dates and the normalized average energy consumption of daily tasks in the cluster to obtain continuous high-load probability; the specific formula for obtaining the continuous high load probability comprises the following steps:
in the method, in the process of the invention,
is the first
A sustained high load probability value for each cluster,
is the first
The number of all dates within a cluster of the clusters,
is the first
The sum of the average energy consumption of all daily tasks within a cluster of the clusters,
the hyperbolic tangent function is normalized.
When the number of dates in the cluster is larger and the average energy consumption sum of the daily tasks is larger, then
The larger the lithium battery load state in this cluster can be considered as a continuous high load, and rejection is required.
The step of calculating the sustained low load probability of the data in the cluster is as follows: calculating the Euclidean norm of the sum of the normalized date number in the cluster and the negatively-correlated normalized daily task average energy consumption to obtain the continuous low-load probability; the specific formula for obtaining the sustained low load probability comprises the following steps:
in the method, in the process of the invention,
is the first
A sustained low load probability value for each cluster,
is the first
The number of all dates within a cluster of the clusters,
is the first
The sum of the average energy consumption of all daily tasks within a cluster of the clusters,
for the hyperbolic tangent function of the normalization process,
an exponential function based on a natural constant normalized for negative correlation.
When the number of dates in the cluster is larger and the average energy consumption sum of the daily tasks is smaller, then
The larger the lithium battery load state in this cluster can be considered as a continuous low load, and rejection is required.
In order to reject the clusters with the continuous high load and the continuous low load, a continuous high load probability threshold value and a continuous low load probability threshold value can be preset, and when the continuous high load probability value and the continuous low load probability value in the clusters are smaller than the corresponding continuous high load probability threshold value and the continuous low load probability threshold value, the data in the clusters are reference data. In the embodiment of the invention, the preset continuous high load probability threshold and the preset continuous low load probability threshold are both set to 0.8, and it is noted that an implementer can determine the values of the preset continuous high load probability threshold and the preset continuous low load probability threshold according to implementation scenes in the implementation process.
After the reference data are obtained, according to the average value of the ratio of the total daily energy consumption of the reference data to the total daily task energy consumption, namely, the average value of all SPUEs of the reference data is used as an evaluation result of the lithium battery efficiency of the IDC machine room, compared with the evaluation of all the energy consumption data in the evaluation period, the accuracy and the reliability of the average value of the SPUEs of the reference data are higher.
In summary, the embodiment of the invention provides a big data statistics-based lithium battery performance evaluation method for an IDC machine room, and because a lithium battery in the IDC machine room may have a continuous high load or a continuous low load in a certain period of time, the accuracy and the reliability of the calculated lithium battery performance evaluation value are low in the continuous high load or the continuous low load. Therefore, in order to obtain a reliable evaluation value, a daily load representation value of the battery is obtained according to the energy consumption condition of the daily lithium battery; and screening out a load stabilization time sequence with stable lithium battery load state and long duration date, calculating the preference degree of the load stabilization time sequence, selecting an optimal evaluation period range as a clustering range, obtaining reference data of lithium battery efficiency evaluation through clustering and screening, and obtaining an accurate and reliable lithium battery efficiency evaluation result according to an SPUE mean value of the reference data.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (9)
1. The method for evaluating the lithium battery efficiency of the IDC machine room based on big data statistics is characterized by comprising the following steps of:
collecting daily total energy consumption, daily task amount and daily task average energy consumption of a server battery in an evaluation period; obtaining daily task volume density according to daily task volume, and obtaining a daily load representation value according to the daily task volume density and daily task energy consumption variance;
obtaining a load characteristic time sequence according to the daily load characteristic value, dividing the load characteristic time sequence according to the element change rate of each position in the load characteristic time sequence to obtain at least two load stability time sequences, determining a load stability time sequence interval and a change rate information entropy value in the load stability time sequence, and obtaining the preference degree of the load stability time sequence according to the load stability time sequence interval and the change rate information entropy value;
obtaining the similarity of the energy consumption states among different dates according to the total daily task energy consumption and the daily task energy consumption variance of different dates; obtaining a local reachable range in the clustering process according to a load stabilization time sequence corresponding to the maximum preference degree and clustering data of different dates;
and screening each cluster according to a preset continuous high-load probability threshold value or a preset continuous low-load probability threshold value to obtain reference data, and evaluating battery energy consumption information of the reference data to obtain a battery efficiency evaluation result.
2. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the daily load representation value comprises the steps of:
in the formula (i),
indicate->
Daily load characterization value of day,/->
Representing daily task volume, +.>
Representing any one of the running tasks on the same day, +.>
Indicate->
Energy consumption of individual tasks->
Represents the average energy consumption of daily tasks, +.>
Representing daily task volume density.
3. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the load stabilization time sequence comprises the steps of:
calculating a first derivative of the load characteristic time sequence to obtain a load change trend function; and taking a time point with the function value of 0 in the load change trend function as a division point, and dividing the load characteristic time sequence according to the division point to obtain at least two load stabilizing time sequences.
4. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the preference degree comprises the steps of:
in the formula (i),
is->
Preference degree of segment load stabilization time series, +.>
As hyperbolic tangent function, +.>
Is an exponential function based on natural constant, < ->
Is->
Segment load stabilization time sequence interval,/->
Is->
The change rate information entropy value of the segment load stabilization time sequence.
5. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the local reach comprises the steps of:
and calculating the square root of the load stabilization time sequence interval corresponding to the maximum preference degree, and obtaining the local reachable range.
6. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the sustained high load probability comprises the steps of:
and calculating the Euclidean norm of the sum of the normalized number of dates and the normalized average energy consumption of the daily tasks in the cluster to obtain the continuous high-load probability.
7. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the sustained low load probability comprises the steps of:
and calculating the Euclidean norm of the sum of the normalized number of dates in the cluster and the negatively correlated normalized average energy consumption of the daily task to obtain the continuous low-load probability.
8. The method for evaluating lithium battery performance of an IDC machine room based on big data statistics according to claim 1, wherein the step of obtaining the reference data comprises:
and presetting a continuous high load probability threshold value and a continuous low load probability threshold value, wherein when the continuous high load probability value and the continuous low load probability value in the cluster are smaller than the corresponding continuous high load probability threshold value and the continuous low load probability threshold value, the data in the cluster are the reference data.
9. The method for evaluating the lithium battery performance of the IDC machine room based on big data statistics according to claim 1, wherein the step of evaluating the battery performance comprises the steps of:
and calculating the average value of the SPUE values of all the reference data to obtain a battery efficiency evaluation result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310279444.2A CN115994714B (en) | 2023-03-22 | 2023-03-22 | IDC machine room lithium battery efficiency evaluation method based on big data statistics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310279444.2A CN115994714B (en) | 2023-03-22 | 2023-03-22 | IDC machine room lithium battery efficiency evaluation method based on big data statistics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115994714A CN115994714A (en) | 2023-04-21 |
| CN115994714B true CN115994714B (en) | 2023-05-19 |
Family
ID=85993684
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310279444.2A Active CN115994714B (en) | 2023-03-22 | 2023-03-22 | IDC machine room lithium battery efficiency evaluation method based on big data statistics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115994714B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108898273A (en) * | 2018-05-29 | 2018-11-27 | 国网能源研究院有限公司 | A kind of user side load characteristic clustering evaluation method based on morphological analysis |
| CN109683091A (en) * | 2017-10-19 | 2019-04-26 | 沈慕仲 | The detection method and system and cell performance evaluation method of the performance parameter of battery |
| CN109980742A (en) * | 2019-04-15 | 2019-07-05 | 广东电网有限责任公司电网规划研究中心 | Meter and power battery echelon utilize the charging station optimization method based on MMC structure |
| CN110097474A (en) * | 2019-03-25 | 2019-08-06 | 国网辽宁省电力有限公司 | A kind of duty control method for realizing power distribution network peak load shifting |
| CN113077122A (en) * | 2021-03-12 | 2021-07-06 | 上海电力大学 | Space-time distribution charging load evaluation method based on electric automobile |
| CN216490104U (en) * | 2021-11-15 | 2022-05-10 | 广东宏达通信有限公司 | Power energy-saving device for server |
| CN114493367A (en) * | 2022-03-08 | 2022-05-13 | 国网辽宁省电力有限公司经济技术研究院 | Power supply reliability assessment method considering differentiated user load probability characteristics |
| CN115015789A (en) * | 2022-04-02 | 2022-09-06 | 中国人民解放军国防科技大学 | Lithium ion battery health state estimation method and device based on incremental capacitance curve |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| NL2018056B1 (en) * | 2016-12-23 | 2018-07-02 | Univ Delft Tech | Hybrid battery and electrolyser |
| WO2021108745A1 (en) * | 2019-11-27 | 2021-06-03 | Cruickshank Iii Robert F | An optimized load shaping system, method & apparatus for optimizing production and consumption of energy |
-
2023
- 2023-03-22 CN CN202310279444.2A patent/CN115994714B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109683091A (en) * | 2017-10-19 | 2019-04-26 | 沈慕仲 | The detection method and system and cell performance evaluation method of the performance parameter of battery |
| CN108898273A (en) * | 2018-05-29 | 2018-11-27 | 国网能源研究院有限公司 | A kind of user side load characteristic clustering evaluation method based on morphological analysis |
| CN110097474A (en) * | 2019-03-25 | 2019-08-06 | 国网辽宁省电力有限公司 | A kind of duty control method for realizing power distribution network peak load shifting |
| CN109980742A (en) * | 2019-04-15 | 2019-07-05 | 广东电网有限责任公司电网规划研究中心 | Meter and power battery echelon utilize the charging station optimization method based on MMC structure |
| CN113077122A (en) * | 2021-03-12 | 2021-07-06 | 上海电力大学 | Space-time distribution charging load evaluation method based on electric automobile |
| CN216490104U (en) * | 2021-11-15 | 2022-05-10 | 广东宏达通信有限公司 | Power energy-saving device for server |
| CN114493367A (en) * | 2022-03-08 | 2022-05-13 | 国网辽宁省电力有限公司经济技术研究院 | Power supply reliability assessment method considering differentiated user load probability characteristics |
| CN115015789A (en) * | 2022-04-02 | 2022-09-06 | 中国人民解放军国防科技大学 | Lithium ion battery health state estimation method and device based on incremental capacitance curve |
Non-Patent Citations (2)
| Title |
|---|
| 通信蓄电池健康度综合评价方法研究;缪巍巍;吴海洋;施健;吕顺利;;计算机与数字工程(第04期);251-256 * |
| 铅酸蓄电池生命周期评价;张晗;王红梅;马聪丽;徐艳;;中国环境管理(第03期);45-54 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115994714A (en) | 2023-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114114039B (en) | Method and device for evaluating consistency of single battery cells of battery system | |
| CN117493921B (en) | Artificial intelligence energy-saving management method and system based on big data | |
| CN114090646A (en) | Abnormal electricity utilization identification method and system | |
| CN115423013A (en) | Power system operation mode classification method, device, equipment and medium | |
| CN115825747A (en) | Method, device, equipment, storage medium and program product for detecting lithium deposition of battery | |
| CN114331238B (en) | Intelligent model algorithm optimization method, system, storage medium and computer equipment | |
| CN115994714B (en) | IDC machine room lithium battery efficiency evaluation method based on big data statistics | |
| CN116883059B (en) | Distribution terminal management method and system | |
| CN113945852A (en) | Storage battery pack inconsistency evaluation method | |
| CN116743180B (en) | Intelligent storage method for energy storage power supply data | |
| CN111797899B (en) | Low-voltage transformer area kmeans clustering method and system | |
| CN107274025B (en) | System and method for realizing intelligent identification and management of power consumption mode | |
| CN116226468B (en) | Service data storage management method based on gridding terminal | |
| US11847619B2 (en) | System-state monitoring method and device and storage medium | |
| CN117192391A (en) | Energy storage battery abnormality identification method, system, device and energy storage station | |
| CN116990697A (en) | Method for detecting abnormal single body in lithium battery pack based on probability distribution | |
| CN115618229A (en) | Line loss allocation algorithm based on big data | |
| CN117706377B (en) | Battery inconsistency identification method and device based on self-adaptive clustering | |
| Tianjiao et al. | Clustering algorithm based battery energy storage performance analysis method | |
| CN110297145B (en) | Voltage sag detection method based on multi-user electric energy data deep analysis | |
| CN117972463A (en) | Improved load adjustable capacity clustering method | |
| CN113673551B (en) | Power metering bad data identification method and system | |
| CN115563193B (en) | Big data analysis processing method for digital information | |
| CN111007418B (en) | Lithium battery residual life prediction method | |
| Orhan et al. | Gravitational fuzzy clustering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A method for evaluating the effectiveness of lithium batteries in IDC data centers based on big data statistics Granted publication date: 20230519 Pledgee: China Postal Savings Bank Co.,Ltd. Nanjing Branch Pledgor: Jiangsu Jinyu Information Technology Co.,Ltd. Registration number: Y2024980002775 |
|
| PE01 | Entry into force of the registration of the contract for pledge of patent right |