CN107784440A - A kind of power information system resource allocation system and method - Google Patents

A kind of power information system resource allocation system and method Download PDF

Info

Publication number
CN107784440A
CN107784440A CN201710989365.5A CN201710989365A CN107784440A CN 107784440 A CN107784440 A CN 107784440A CN 201710989365 A CN201710989365 A CN 201710989365A CN 107784440 A CN107784440 A CN 107784440A
Authority
CN
China
Prior art keywords
model
resource
index
data
sequence
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.)
Pending
Application number
CN201710989365.5A
Other languages
Chinese (zh)
Inventor
葛维春
李钊
雷振江
范士新
刘树吉
乔林
任相儒
刘为
刘颖
王飞
孟威
宁亮
曹智
刘雪松
刘坤
曹国强
胡小磊
祁奕霏
王刚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
State Grid Liaoning Electric Power Co Ltd
Electric Power Research Institute of State Grid Liaoning Electric Power Co Ltd
Nanjing NARI Group Corp
Original Assignee
State Grid Corp of China SGCC
State Grid Liaoning Electric Power Co Ltd
Electric Power Research Institute of State Grid Liaoning Electric Power Co Ltd
Nanjing NARI Group Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, State Grid Liaoning Electric Power Co Ltd, Electric Power Research Institute of State Grid Liaoning Electric Power Co Ltd, Nanjing NARI Group Corp filed Critical State Grid Corp of China SGCC
Priority to CN201710989365.5A priority Critical patent/CN107784440A/en
Publication of CN107784440A publication Critical patent/CN107784440A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06313—Resource planning in a project environment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
    • G06Q10/06375—Prediction of business process outcome or impact based on a proposed change
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Educational Administration (AREA)
  • Tourism & Hospitality (AREA)
  • Theoretical Computer Science (AREA)
  • Marketing (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Physics & Mathematics (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention discloses a kind of power information system resource allocation system and method, belongs to power information system field.The system includes power information system resource device, memory capacity controller and system mode precaution device.This method is for systematic function present in current power information system scheduling of resource and dynamic resource allocation, hidden danger in terms of off-capacity, it is proposed a kind of cloud resource capacity resource concocting method based on time series, passage time series model, the situation of change that power information system resource uses in the prediction future services cycle, for in electrical network business the characteristics of each information system, it is accurate to formulate, flexible resource allocation strategy, the utilization rate of information system performance is set to lift 30%, make the unplanned interruption duration of information system reduce by 50% simultaneously, ensure effective operating of each operation flow of power network, work for grid information system operation maintenance personnel provides reference frame.

Description

Power information system resource allocation system and method
Technical Field
The invention relates to the field of power information systems, in particular to a power information system resource allocation system and a power information system resource allocation method.
Background
With continuous construction of power system informatization and development and progress of cloud computing virtualization technology, more and more power information systems are deployed on a virtualization platform, and along with continuous operation of online services, the system is frequently subjected to high concurrent access and mass data streams, so that the processing capacity and the computing strength borne by resources of a bearing system are required to be enhanced.
In the prior art, resource scheduling based on a task scheduling mode is the most widely applied traditional resource scheduling technology, has strong adaptability, is an important mode for realizing large-scale computation, and has wide application under a virtual cluster environment.
However, this technique is an application-level scheduling technique with application-specific and object-distributed properties, whose set of scheduling virtual machine objects needs to be a logical cluster that provides the same computing application. For large enterprises, especially for power enterprises, which have a huge system and numerous branches and use equipment of multiple manufacturers, with the development of power enterprise business, the data scale becomes larger and larger, and a set of professional resource allocation method and system suitable for the power enterprises are needed to perform professional, efficient and simple management on huge resources.
Although the existing resource allocation method can dynamically allocate resources, the existing resource allocation method is only limited to carry out adjustment and measures aiming at results under the condition that the system resources are found to be insufficient or excessive, predictive resource assessment is lacked, the resource capacity of an information system cannot be predicted and assessed in advance, resource budget cannot be guided, and the cost of human resources is greatly wasted.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a power information system resource allocation method and a power information system resource allocation system, after configuration setting, manual interference is not needed, and power system resource matching is independently and accurately analyzed and predicted; and the resources are accurately managed, and the rationality and accuracy of power grid resource management are improved.
In order to achieve the above object, an aspect of the present invention provides a power information system resource allocation system, which includes a power information system resource device, a storage capacity controller, and a system status early-warning device, wherein the power information system resource device is connected to the storage capacity controller, and the storage capacity controller is connected to the system status early-warning device.
The power information system resource device is used for allocating virtual resources of cloud resources in real time, connecting real-time resource data in an IT operation and maintenance system, and outputting allocation schemes, so that the purposes of resource balance and stable operation of all information systems are achieved.
The storage capacity controller is used for predicting the capacity of a hard disk and a database in each information system, performing terminal alarm and prompt according to a set threshold value, and performing capacity increase or capacity reduction in time according to the resource occupancy condition.
The system state early warning device is used for evaluating and analyzing the current state of the information system, analyzing and early warning the states of the system, such as normal, attention, abnormity, danger and the like according to all comprehensive evaluation indexes of the system, and guiding operation and maintenance personnel of the information system to find and process problems in time, so that the system can operate safely and effectively.
In order to solve the above-mentioned object, another aspect of the present invention provides a power information resource allocation method, which provides a time-series-based cloud resource capacity resource allocation method for the hidden dangers in the aspects of system performance and capacity insufficiency existing in the current power information system resource scheduling and resource dynamic allocation, predicts the change situation of the power information system resource usage in the future service period through a time series model, and makes an accurate and flexible resource allocation strategy according to the characteristics of each information system in the power grid service, so that the utilization rate of the information system performance is improved by 30%, the unplanned interruption time of the information system is reduced by 50%, the effective operation of each service flow of the power grid is ensured, and a reference basis is provided for the work of the power grid information system operation and maintenance personnel.
The method comprises the following steps:
step S11, data collection and preprocessing
The number of types of the components is recorded as m, data is an acquisition index of a certain type of components of all equipment at the current moment, the sample capacity is n, namely the sample contains index data of n types of the components, and the index number on the ith type of the components is pi (i =1,2.., m). In order to measure the degree of the index deviating from the standard value or the index standard range value and further measure the health degree of the component, a conversion rule is formulated:
for the index with the standard value, the acquired value is converted into the deviation rate, and the smaller the deviation rate is, the more stable the equipment state is. Recording the acquisition value as v, the standard value as u, and the deviation ratio as x, then:
for an index with a standard interval, the degree of deviation of the collected value from the interval can be converted into:
recording the acquisition value as v, the upper limit of the standard interval as uu, the lower limit of the standard interval as ul, recording the deviation degree as x, obtaining an interval according to statistical analysis, and correcting the interval according to expert experience judgment and industry standards.
For the same type of equipment, there are a plurality of parts of a certain type in number, and there are also a plurality of corresponding index values, in which case it is necessary to convert the plurality of index values into 1 composite index value. The final inputs to the model are as follows:
X i ＝(x 1i x 2i … x ni ) T ,i＝1,2,...,p i
wherein Xi is the sample value of the index i, and xij is the value of the jth index of the ith component.
The magnitude and dimension of different indicators may vary, and a normalization process is required to eliminate the effect of the magnitude and dimension.
For the forward direction index, i.e. the larger and better index, the following conversion is made:
for negative indicators, i.e. smaller and better indicators, the following conversion is made:
the normalized indexes all belong to the interval [0,1], and for the sake of understanding, the normalized data is still denoted as xij.
The conflict quantization index of the jth index and other indexes is as follows: (1-rij), wherein rij evaluates the correlation coefficient between the indexes i and j, and if the correlation coefficient between the indexes Xi and Xj is rij, then rij is:
whereinIs an average value of the index i,the average value of index j is shown.
The result of calculating r is in the interval [ -1,1], where r <0 considers Xi and Xj to be negatively correlated, r >0 considers Xi and Xj to be positively correlated, r =0 indicates that Xi and Xj have no linear relationship, and the closer r is to 0, the lower the correlation between Xi and Xj is.
The objective weight of each index is comprehensively measured by the contrast intensity and the conflict. Assuming that Cj represents the amount of information included in the j-th evaluation index, cj may be represented as:
wherein, the sigma j represents the standard deviation of the index Xi, and the standard deviation can be estimated and recorded according to the sample under the condition that the sigma j is unknownFor the estimate of σ j, the calculation formula is as follows:
the larger Cj is, the larger the information content contained in the jth evaluation index is, the greater the relative importance of the index is, and the objective weight of the jth index is calculated as follows:
and (4) obtaining an index weight vector of a certain type of components by adopting CRITIC algorithm modeling and marking as Wc.
And S12, constructing a model by adopting a time series algorithm, and analyzing the power service system.
Indexes related to the model research are all changed along with time, an ARIMA model, namely a differential autoregressive moving average model, is adopted, and the specific technical scheme is as follows:
the time series corresponding to the index X can be expressed as:
{X t :t＝0,±1,±2,...}
if X is a plateau sequence:
wherein E represents expectation.
If X is a non-stationary sequence then X needs to be paired first t Carrying out difference operation, recording delta as a difference operator, and carrying out the difference calculation process as follows:
ΔX t ＝X t -X t-1 ＝X t -BX t ＝(1-B)X t
Δ 2 X t ＝ΔX t -ΔX t-1 ＝(1-B)X t -(1-B)X t-1 ＝(1-B) 2 X t
Δ d X t ＝(1-B) d X t
the difference result is recorded as Wt:
W t ＝Δ d X t ＝(1-B) d X t
then Wt is a stationary sequence and the resulting model is called Xt-ARIMA (p, d, q) in the form of a model
Where δ is a constant and ut is a white noise sequence (a sequence where the expectation and variance are both constant), the model can be simplified as follows:
Φ(B)Δ d X t ＝δ+Θ(B)u t
where Φ (B) is a p-order autoregressive coefficient polynomial:
Θ (B) is a moving average coefficient polynomial of order q:
Θ(B)＝1-θ 1 B-θ 2 B 2 -…-θ q B q
when δ =0, the above model is a centered ARIMA (p, d, q) model.
For ARIMA (a, b, c), an autoregressive model AR (p) should be applied when a and c are both 0 in the model, and a moving average model MA (q) should be applied when b and c are both 0.
The autoregressive model AR (p) is:
the moving average model MA (q) is:
for the ARIMA (p, d, q) model, the autoregressive model AR (p) is obtained when d and q are both 0, and the moving average model MA (q) is obtained when d and p are both 0.
The time sequence is mainly applied to the prediction and evaluation of the capacity of storage equipment such as a database, a server hard disk, middleware and the like in the resource allocation of an information system. Firstly, judging the trend change rule through a scatter diagram, an autocorrelation graph and a partial correlation graph, identifying that the trend change rule belongs to a stationary sequence or a non-stationary sequence, if the trend change rule belongs to a non-stationary time sequence, carrying out stabilization processing on the non-stationary time sequence, and carrying out heteroscedastic elimination on data to enable the parameters to approach zero, thereby achieving the analysis requirement of the stationary sequence.
The specific construction method comprises the following steps:
the method comprises the steps of firstly, checking the variance, the trend and the seasonal change rule of an ADF unit root according to a scatter diagram, an autocorrelation function and a partial autocorrelation function diagram of a time sequence, and identifying the stationarity of the sequence.
And step two, carrying out stabilization treatment on the non-stationary sequence: if the data sequence is non-stationary and has a certain increasing or decreasing trend, the data needs to be processed differentially, and if the data has an variance, the data needs to be processed technically until the autocorrelation function value and the partial correlation function value of the processed data are not significantly different from zero.
Thirdly, establishing a corresponding model according to the identification rule of the time series model: if the partial correlation function of the stationary sequence is truncated and the autocorrelation function is trailing, it can be concluded that the sequence fits the AR model; if the partial correlation function of the stationary sequence is tail-biting and the autocorrelation function is tail-biting, it can be concluded that the sequence fits the MA model; if both the partial correlation function and the autocorrelation function of the stationary sequence are tail-shifted, the sequence fits the ARMA model.
And fourthly, performing parameter estimation and checking whether the statistical significance is achieved.
And fifthly, carrying out hypothesis test to diagnose whether the residual error sequence is white noise.
And sixthly, performing predictive analysis by using the model which passes the test.
The number of indexes needing to be predicted is p, and the index i historical data sequence is as follows:
X i ＝{x t,i ,x t-1,i ,...,x t-n,i },i＝1,2,...,p，
for the sequence Xi, predicting the index xt +1,I at the t +1 moment according to ARIMA, and finally obtaining the index estimation vector of the t +1 moment as
WhereinIs the predicted value of the ith index.
The model suitable for the method is judged according to the formula, if the partial correlation function is truncated, the model is suitable for an AR model (autoregressive model), if the autocorrelation truncation and the partial correlation tailing are ended, the model is suitable for an MA model (moving average model), and if the autocorrelation coefficient and the partial correlation coefficient are both truncated, the model is suitable for an ARMA model (differential autoregressive moving average model).
Step S13, carrying out resource allocation of the power information system in a hierarchical manner
Through the analysis of a service system, resource scheduling related to the system is analyzed and classified, and the resource scheduling technology in a virtualization environment is divided into physical layer resource scheduling, virtual layer resource scheduling and application layer resource scheduling from bottom to top according to the difference of the characteristics of various scheduling technologies, such as resource abstraction levels, targets and the like.
The target object of the physical layer resource scheduling is the load of a physical machine, the resource load of the physical machine is changed by adjusting the distribution of virtual machines, and the method is classified into load balancing or energy saving; the load balancing scheduling enables the load of the physical machine cluster to tend to be balanced by adjusting the distribution of the virtual machines; and the energy-saving scheduling enables the load of the physical machine cluster to tend to be saturated by adjusting the distribution of the virtual machines, releases unnecessary resources and achieves the effect of energy saving.
The resource scheduling of the virtual layer mainly aims at the load of the virtual machine, and the load condition of the virtual machine is changed by adjusting the resources of the virtual machine, so that the application performance in the virtual machine is changed.
The resource scheduling of the application layer is mainly aimed at specific applications, and the traditional task scheduling is based on the resource scheduling of the application layer, and the overall service capacity of the whole application is changed by adjusting the task amount applied to different nodes.
In order to achieve the aim, the invention also provides an effective capacity evaluation method, which is used for analyzing and evaluating the capacity of the cloud resources based on a cloud resource application state evaluation model, and a capacity evaluation model construction method which is used for determining system application indexes and resource index baselines according to historical data training results and in combination with business system index management requirements; predicting the data of the next service period according to a large amount of historical data and by combining a service time window, and mastering the change condition of the data of the next service period; through the comparison of the predicted data of the next service period of the service and the baseline data, whether the temporary data of the next service period exceeds the historical baseline data or not is found, and if the temporary data of the next service period exceeds the historical baseline data, resources are required to be adjusted to meet the requirements of the service; and after the adjustment time is mastered, analyzing the adjustment quantity, defining a service unit by combining a system service architecture, finding a relation between the server unit and system application data and resource data, and obtaining the specific adjustment quantity by combining a service period prediction result.
The method comprises the following specific steps:
step S21, establishing a baseline
(1) Screening online user number data of a section of history period of a service system;
(2) Drawing a scatter diagram of the number of online users, and observing the distribution condition of the number of online users;
(3) And determining a system application index and a resource index baseline according to a historical data training result and in combination with a service system index management requirement, wherein after the baseline is determined, the later-stage system operation is operated within the range of the baseline, and if the baseline requirement is not met, the system is considered to be unstable.
Step S22, service cycle data prediction
The time sequence algorithm is used for finding the rule of the time index changing along with the time, and the index value of the future time can be predicted so as to solve the practical problem, and the capacity evaluation and resource adjustment process comprises the following steps:
(1) Establishment of an index baseline
Capacity assessment requires establishing a baseline to determine whether resource adjustment is required, the established baseline including: an online user quantity baseline and a resource utilization rate baseline. Firstly, screening online user number data of a service system in a historical period; secondly, drawing a scatter diagram of the number of the online users, and observing the distribution condition of the number of the online users; secondly, calculating the standard deviation of the index value, and establishing the base line can be constructed by adding or subtracting the standard deviation to the index predicted value.
(2) Index prediction
According to a scatter diagram, an autocorrelation function and a partial autocorrelation function diagram of the time sequence, the variance, the trend and the seasonal change rule of the time sequence are tested by using an ADF unit root, and the stationarity of the sequence is identified; carrying out stabilization treatment on the non-stationary sequence; and establishing a corresponding model according to the identification rule of the time series model, carrying out parameter estimation, checking whether the model has statistical significance, and carrying out prediction analysis by using the model which passes the check.
(3) Resource adjustment judgment
The adjustment of system resources is performed when one service period ends and the next service period starts, the prediction of the next service period is already realized through an ARIMA model, and whether the resource adjustment is performed or not is determined according to the prediction result.
The judgment is based on the following:
when either the predicted number of online users or the resource utilization exceeds a limit (the maximum number of online users supported or the resource utilization), adjustment (increase in resources) is required.
When the predicted resource utilization is below the standard resource utilization minimum, an adjustment (resource release) is required.
The service system has the following conditions for the use of resources:
insufficient resources: by using the load test and pressure test method in the performance test method, the maximum number of users supported by the system is tested within the range of the average response time of the system and the utilization rate of system resources in the operating performance parameters, and is represented by U.
According to the judgment basis, when R '> R (b) or U' > U, resource adjustment is needed, and the adjustment strategy is as follows: the resources are increased.
Excessive resources: when the average response time of the system is in a specified range, as long as the resource utilization rate is less than the standard minimum value, the resource release needs to be considered, and the method is irrelevant to the number of online users.
According to the judgment basis, when R' < R (a), resource adjustment is needed, and the adjustment strategy is as follows: and releasing the resources.
The resources satisfy: and when the number of the online users and the resource utilization rate are within the standard value range, the adjustment is not carried out.
According to the judgment basis, when S <5 and R' is belonged to [ a, b ], no resource adjustment is needed.
(4) Resource adjustment
After the situation of the next service period is predicted and judged, a corresponding adjustment strategy needs to be provided according to the judgment result, and a quantitative strategy can be provided by combining the system architecture.
The business system of the invention is a decoupled system with a horizontal expansion capability framework, in the framework, a concept of a service unit is defined in a state evaluation model, a server in each service unit consists of a plurality of servers with the same configuration information, and the system has a load balancing capability, namely, all loads are shared by each service unit. Quantitative evaluations will be given below for resource adjustments.
The dynamic adjustment of the system resources is based on a large amount of historical operation and maintenance data, a model for diagnosing and predicting the state of the system resources is constructed by utilizing a big data technology and a visualization technology, the blank of asset management of operation and maintenance data of a power grid information system, architecture evaluation of the information system and prediction of the operation state is filled, decision analysis is made for operation management of the information system, the operation and maintenance management level is improved, the increase of the system resources is estimated by utilizing the diagnosis and prediction model, whether the future data capacity meets the application requirement or not is judged, and the resource allocation is adjusted in time.
Step 23, capacity evaluation
The adjustment of the information system resources is performed at the end of one service period and at the beginning of the next service period, the prediction of the next service period has been performed above by means of the ARIMA model, and it is now determined whether to perform resource adjustment based on the result of the prediction.
Through the dynamic analysis and allocation tool of system resources, system operation and maintenance personnel can monitor and analyze space occupation conditions of a database, a server hard disk, middleware and the like in real time, make correct and timely judgment according to the increase and time assessment, and improve the system operation and maintenance level.
For interactive applications, the experience of system performance to the user is directly embodied by the "system response time", and for systems the performance is expressed by the "number of concurrent users" or "throughput". Throughput in interactive applications, which reflects the server's bearing pressure, is used to explain the system-level load capacity.
And according to the system service period, the resources are dynamically adjusted, and the resource utilization rate is improved. And the model evaluation adopts a simulation experiment to adjust resources, the effective rate evaluation of the model adopts an average value of the accuracy, and the stability evaluation adopts an effective mean square error. And (3) recording that N times of experiments are carried out, wherein the resource adjustment times of a certain experiment are Ni, the effective rate is Ri, and then:
wherein xj represents the j-th resource adjustment is effective or not
The average effective rate of N experiments is:
the closer to 100%, the higher the model prediction accuracy.
The mean square error of the accuracy is:
the closer the MSE is to 0, the more stable the model is.
In the test, the average response time of the system is less than 5 seconds, the predicted resource utilization rate is 7.4%, and resource adjustment is not needed.
Step S24, capacity adjustment
After predicting the condition of the next service period and making a judgment, a corresponding adjustment strategy needs to be provided according to the judgment result, and a quantitative strategy can be provided by combining the system architecture.
In the state evaluation model, the concept of service units is defined, a server in each service unit consists of a plurality of servers with the same configuration information, and the system has load balancing capability, namely, each service unit can share all loads. Quantitative evaluations will be given below for resource adjustments.
Defining global parameters:
n0: indicating the current number of service units
Nt: representing the number of service units at time t in the future
M: indicating the adjusted number of service units
Increasing resource condition analysis
The following parameters are defined:
umax: representing the maximum number of supported online users or the maximum concurrency of the system
Then, the service capability of each unit we denote by P: p = U/N
Ut: and expressing the predicted value of the online user number or the predicted value of the system concurrency in the next service period.
Then, the number of cells that needs to be increased at least is:
when M is a decimal, we consider that we all adjust to the service unit, and therefore should adjust to an integer, rounding up.
Second, analysis of released resource condition
The following parameters are defined:
rmax: representing resource utilization ceiling
Rmin: indicating a lower bound on resource utilization
M: indicating adjusted number of service units
Rt: indicating the predicted value of the system resource utilization rate of the next service period
Mmax: the maximum number of units that need to be released is:
mmin: the minimum number of units that need to be released is:
rounding is also used if Mmax, mmin is fractional, since it is the release of resources at this time, rounding down.
Drawings
Fig. 1 is a structural framework diagram of a resource allocation system of an electrical information system according to the present invention.
Fig. 2 is a flowchart of allocating power information resources according to the present invention.
FIG. 3 is a flow chart of the construction of the difference autoregressive moving average model according to the present invention
FIG. 4 is a diagram of a capacity estimation model construction method provided by the present invention
FIG. 5 is a scatter diagram of online user number distribution according to the present invention
FIG. 6 is a flow chart of capacity estimation and resource adjustment according to the present invention
Detailed Description
As shown in fig. 1, the present invention provides a power information system resource allocation system, which includes a power information system resource device, a storage capacity controller, and a system status early-warning device, wherein the power information system resource device is connected to the storage capacity controller, and the storage capacity controller is connected to the system status early-warning device.
The power information system resource device is used for allocating virtual resources of cloud resources in real time, connecting real-time resource data in an IT operation and maintenance system, and outputting allocation schemes, so that the purposes of resource balance and stable operation of all information systems are achieved.
The storage capacity controller is used for predicting the capacity of a hard disk and a database in each information system, carrying out terminal alarm and prompt according to a set threshold value, and carrying out capacity increase or reduction in time according to the resource occupancy condition.
The system state early warning device is used for evaluating and analyzing the current state of the information system, analyzing and early warning the states of the system, such as normal, attention, abnormity, danger and the like according to all comprehensive evaluation indexes of the system, and guiding operation and maintenance personnel of the information system to find and process problems in time, so that the system can operate safely and effectively.
As shown in fig. 2, the present invention discloses a power information resource allocation method, which includes:
step S11, data collection and pretreatment
The number of types of the components is recorded as m, data is an acquisition index of a certain type of components of all equipment at the current moment, the sample capacity is n, namely the sample contains index data of n types of the components, and the index number on the ith type of the components is pi (i =1,2.., m). In order to measure the degree of deviation of the index from the standard value or the index standard range value and further measure the health degree of the component, the following conversion rule is formulated:
for indices having a standard value. The acquired value can be converted into a deviation rate, and the smaller the deviation rate is, the more stable the equipment state is.
Recording the acquisition value as v, the standard value as u and the deviation ratio as x, then:
for the index with the standard interval. The degree to which the collected values deviate from the interval can be converted into:
recording the acquisition value as v, the upper limit of the standard interval as uu, the lower limit of the standard interval as ul, and the deviation degree as x, then:
TABLE-1 index conversion rules
The standard value and the standard interval are determined according to expert experience judgment and industry standards, and the interval with the highest value occurrence frequency or concentrated value can be found out as the standard value and the standard interval according to the statistical index data distribution of a large amount of historical data. In practical operation, we can combine the two to use, first obtain an interval according to statistical analysis, and then correct the interval according to expert's experience judgment and industry standard.
For the same type of equipment, there may be a plurality of parts of a certain type in number, and there may be a plurality of corresponding index values, in which case it is necessary to convert the plurality of index values into 1 integrated index value. The conversion rules are as follows:
TABLE-2 case where there are a plurality of parts of a certain type of the same equipment
The model after the above processing is finally input as follows:
X i ＝(x 1i x 2i … x ni ) T ,i＝1,2,...,p i
wherein Xi is the sample value of the index i, and xij is the value of the jth index of the ith component.
The magnitude and dimension of different indicators may vary, and a normalization process is required to eliminate the effect of the magnitude and dimension.
For the forward direction index, i.e. the larger and better index, the following conversion is made:
for negative indicators, i.e. the smaller the better the indicator, the following conversion is made:
the normalized indexes all belong to the interval [0,1], and for the sake of understanding, the normalized data is still denoted as xij.
The conflict quantization index of the jth index and other indexes is as follows: (1-rij), wherein rij evaluates the correlation coefficient between the indices i and j, and if the correlation coefficient between the indices Xi and Xj is rij, then rij is:
whereinIs the average value of the index i,the average value of index j is shown.
The result of r is calculated in the interval [ -1,1], where r <0 considers Xi and Xj to be negatively correlated, r >0 considers Xi and Xj to be positively correlated, r =0 indicates that Xi and Xj have no linear relationship, and the closer r is to 0, the lower the correlation between Xi and Xj is considered.
The objective weight of each index is comprehensively measured by the contrast intensity and the conflict. Assuming that Cj represents the amount of information included in the j-th evaluation index, cj may be represented as:
wherein, the sigma j represents the standard deviation of the index Xi, and the standard deviation can be estimated and recorded according to the sample under the condition that the sigma j is unknownFor the estimate of σ j, the calculation formula is as follows:
the larger Cj is, the larger the information content contained in the jth evaluation index is, the greater the relative importance of the index is, and the objective weight of the jth index is calculated as follows:
and (4) obtaining an index weight vector of a certain type of component by adopting CRITIC algorithm modeling and recording the index weight vector as Wc.
And S12, constructing a model by adopting a time series algorithm, and analyzing the power service system.
All indexes related to the model change along with time, and taking an ARIMA model, namely a differential autoregressive moving average model as an example, the specific method is as follows:
the time series corresponding to the index X can be expressed as:
{X t :t＝0,±1,±2,...}
if X is a plateau sequence:
wherein E represents expectation.
If X is a non-stationary sequence then X needs to be paired first t Carrying out difference operation, recording delta as a difference operator, and carrying out the difference calculation process as follows:
ΔX t ＝X t -X t-1 ＝X t -BX t ＝(1-B)X t
Δ 2 X t ＝ΔX t -ΔX t-1 ＝(1-B)X t -(1-B)X t-1 ＝(1-B) 2 X t
Δ d X t ＝(1-B) d X t
the difference result is recorded as Wt:
W t ＝Δ d X t ＝(1-B) d X t
then Wt is a stationary sequence and the resulting model is called Xt-ARIMA (p, d, q) in the form of a model
Where δ is a constant and ut is a white noise sequence (a sequence where the expectation and variance are both constant), the model can be simplified as follows:
Φ(B)Δ d X t ＝δ+Θ(B)u t
where Φ (B) is a p-order autoregressive coefficient polynomial:
Θ (B) is a moving average coefficient polynomial of order q:
Θ(B)＝1-θ 1 B-θ 2 B 2 -…-θ q B q
when δ =0, the above model is a centered ARIMA (p, d, q) model.
For ARIMA (a, b, c), when a and c are both 0 in the model, an autoregressive model should be applied
AR (p), when b and c are both 0, the moving average model MA (q) should be applied.
The autoregressive model AR (p) is:
the moving average model MA (q) is:
for the ARIMA (p, d, q) model, an autoregressive model AR (p) is obtained when d and q are both 0, and a moving average model MA (q) is obtained when d and p are both 0.
The time sequence is mainly applied to the prediction and evaluation of the capacity of storage equipment such as a database, a server hard disk, middleware and the like in the resource allocation of an information system. Firstly, judging the trend change rule through a scatter diagram, an autocorrelation graph and a partial correlation graph, identifying that the trend change rule belongs to a stationary sequence or a non-stationary sequence, if the trend change rule belongs to a non-stationary time sequence, carrying out stabilization processing on the non-stationary time sequence, and carrying out heteroscedastic elimination on data to enable the parameters to approach zero, thereby achieving the analysis requirement of the stationary sequence.
As shown in fig. 3, the specific construction method is as follows:
the method comprises the steps of firstly, checking the variance, the trend and the seasonal change rule of an ADF unit root according to a scatter diagram, an autocorrelation function and a partial autocorrelation function diagram of a time sequence, and identifying the stationarity of the sequence.
And step two, carrying out stabilization treatment on the non-stationary sequence. If the data sequence is non-stationary and has a certain increasing or decreasing trend, the data needs to be processed differentially, and if the data has an variance, the data needs to be processed technically until the autocorrelation function value and the partial correlation function value of the processed data are not significantly different from zero.
And thirdly, establishing a corresponding model according to the identification rule of the time series model. If the partial correlation function of the stationary sequence is truncated and the autocorrelation function is trailing, it can be concluded that the sequence fits the AR model; if the partial correlation function of the stationary sequence is tail-biting and the autocorrelation function is tail-biting, it can be concluded that the sequence fits the MA model; if both the partial correlation function and the autocorrelation function of the stationary sequence are tail-shifted, the sequence fits the ARMA model.
And fourthly, performing parameter estimation and checking whether the statistical significance is achieved.
And fifthly, performing hypothesis test to diagnose whether the residual error sequence is white noise.
And sixthly, performing predictive analysis by using the model which passes the test.
The number of indexes needing to be predicted is p, and the index i historical data sequence is as follows:
X i ＝{x t,i ,x t-1,i ,...,x t-n,i },i＝1,2,...,p，
for the sequence Xi, predicting the index xt +1,I at the t +1 moment according to ARIMA, and finally obtaining the index estimation vector of the t +1 moment as
WhereinIs the predicted value of the ith index.
The appropriate model is determined according to the above formula, and is adapted to an AR model (autoregressive model) if the partial correlation function is truncated, to an MA model (moving average model) if the autocorrelation is truncated and the partial correlation is trailing, and to an ARMA model (differential autoregressive moving average model) if both the autocorrelation and the partial correlation coefficients are truncated.
Step S13, carrying out resource allocation of the power information system in a hierarchical manner
Through the analysis of a service system, resource scheduling related to the system is analyzed and classified, and the resource scheduling technology in a virtualization environment is divided into physical layer resource scheduling, virtual layer resource scheduling and application layer resource scheduling from bottom to top according to the difference of the characteristics of various scheduling technologies, such as resource abstraction levels, targets and the like.
The target object of the physical layer resource scheduling is the load of a physical machine, the resource load of the physical machine is changed by adjusting the distribution of virtual machines, and the load balancing or energy saving is achieved; the load balancing scheduling enables the load of the physical machine cluster to tend to be balanced by adjusting the distribution of the virtual machines; and the energy-saving scheduling enables the load of the physical machine cluster to tend to be saturated by adjusting the distribution of the virtual machines, releases unnecessary resources and achieves the effect of energy saving.
The resource scheduling of the virtual layer mainly aims at the load of the virtual machine, and the load condition of the virtual machine is changed by adjusting the resources of the virtual machine, so that the application performance in the virtual machine is changed.
The resource scheduling of the application layer is mainly aimed at specific applications, and the traditional task scheduling is based on the resource scheduling of the application layer, and the overall service capacity of the whole application is changed by adjusting the task amount applied to different nodes.
In order to achieve the above object, the present invention further provides an effective capacity evaluation method, as shown in fig. 4, based on a cloud resource application state evaluation model, cloud resource capacity is analyzed and evaluated, a capacity evaluation model construction method determines a system application index and a resource index baseline according to a historical data training result and in combination with a business system index management requirement, after the baseline is determined, the system operation in the later stage will operate within the range of the baseline, and if the baseline requirement is not met, the system is considered to be unstable.
And predicting the data of the next service period according to a large amount of historical data and by combining a service time window, and mastering the change condition of the data of the next service period.
The two steps realize the prediction of the data of the next service period, and establish a baseline at the same time, and by comparing the predicted data of the next service period with the baseline data, whether the temporary data of the next service period exceeds the historical baseline data can be found, if so, the resource is required to be adjusted to meet the requirement of the service. After the adjustment time point is mastered, the adjustment quantity needs to be analyzed, a service unit is defined by combining a system service architecture, the relation between the server unit and system application data and resource data is found, and the specific adjustment quantity is obtained by combining a service period prediction result.
The method comprises the following steps of establishing a base line, predicting data of the next service period, comparing the data with the base line and evaluating the capacity, and comprises the following specific steps:
step S21, establishing a baseline
Firstly, screening online user number data of a service system in a historical period; and secondly, drawing a scatter diagram of the number of the online users, and observing the distribution condition of the number of the online users. As shown in fig. 5, the horizontal axis represents time (day), the vertical axis represents an index value, and the yellow curve represents a predicted value curve of the index; again, the standard deviation of the indicator value is calculated and the establishment of the baseline can be constructed by adding or subtracting the standard deviation to the predicted indicator value, as shown in fig. 5.
And determining a system application index and a resource index baseline according to a historical data training result and in combination with a service system index management requirement, wherein after the baseline is determined, the later-stage system operation is operated within the range of the baseline, and if the baseline requirement is not met, the system is considered to be unstable.
S22, service period data prediction
The time-series algorithm is used to find the rule of the time index changing along with the time, and the index value of the future time can be predicted to solve the practical problem, and the flow of capacity evaluation and resource adjustment is shown in fig. 6, and is as follows:
(1) Establishment of an index baseline
Capacity assessment requires establishing a baseline to determine whether resource adjustment is required, the established baseline including: an online user quantity baseline and a resource utilization rate baseline. Firstly, screening online user number data of a service system in a historical period; secondly, drawing a scatter diagram of the number of the online users, and observing the distribution condition of the number of the online users. As shown in fig. 5, the horizontal axis represents time (day), the vertical axis represents an index value, and the yellow curve represents a predicted value curve of the index; secondly, calculating the standard deviation of the index value, and establishing the base line can be constructed by adding or subtracting the standard deviation to the index predicted value.
(2) Index prediction
And (4) checking the variance, the trend and the seasonal change rule of the dispersion, the autocorrelation function and the partial autocorrelation function graph of the time sequence by using an ADF unit root, and identifying the stationarity of the sequence. And carrying out smoothing treatment on the non-stationary sequence. And establishing a corresponding model according to the identification rule of the time series model, carrying out parameter estimation, checking whether the model has statistical significance, and carrying out prediction analysis by using the model which passes the check.
(3) Resource adjustment judgment
The adjustment of system resources is performed when one service period ends and the next service period starts, the prediction of the next service period is already realized through an ARIMA model, and whether the resource adjustment is performed or not is determined according to the prediction result.
Whether a system is over-resource or idle can be judged according to the two indexes, and the judgment is as follows:
when either the predicted number of online users or the resource utilization exceeds a limit (the maximum supported number of online users or the resource utilization), adjustment (resource increase) is required.
When the predicted resource utilization is below the standard resource utilization minimum, an adjustment (resource release) is required.
The service system has the following conditions for the use of resources:
insufficient resources: by using the load test and pressure test method in the performance test method, the maximum number of users supported by the system is tested within the range of the average response time of the system and the utilization rate of system resources in the operating performance parameters, and is represented by U.
According to the judgment basis, when R '> R (b) or U' > U, resource adjustment is needed, and the adjustment strategy is as follows: the resources are increased.
Excessive resources: when the average response time of the system is in a specified range, as long as the resource utilization rate is less than the standard minimum value, the resource release needs to be considered, and the method is irrelevant to the number of online users.
According to the judgment basis, when R' < R (a), resource adjustment is needed, and the adjustment strategy is as follows: and releasing the resources.
The resources satisfy: and when the number of the online users and the resource utilization rate are within the standard value range, the adjustment is not carried out.
According to the judgment basis, when S <5 and R' is belonged to [ a, b ], no resource adjustment is needed.
(4) Resource adjustment policy
After predicting the condition of the next service period and making a judgment, a corresponding adjustment strategy needs to be provided according to the judgment result, and a quantitative strategy can be provided by combining the system architecture.
The business system of the invention is a decoupled system with a horizontal expansion capability framework, in the framework, a concept of a service unit is defined in a state evaluation model, a server in each service unit consists of a plurality of servers with the same configuration information, and the system has a load balancing capability, namely, all loads are shared by each service unit. Quantitative evaluations will be given below for resource adjustments.
The dynamic adjustment of the system resources is based on a large amount of historical operation and maintenance data, a model for diagnosing and predicting the state of the system resources is constructed by utilizing a big data technology and a visualization technology, the blank of asset management of operation and maintenance data of a power grid information system, architecture evaluation of the information system and prediction of the operation state is filled, decision analysis is made for operation management of the information system, the operation and maintenance management level is improved, the increase of the system resources is estimated by utilizing the diagnosis and prediction model, whether the future data capacity meets the application requirement or not is judged, and the resource allocation is adjusted in time.
Step 23, capacity evaluation
As shown in fig. 4, the adjustment of the information system resources is performed at the end of one service cycle and at the beginning of the next service cycle, and the prediction of the next service cycle has been performed by the ARIMA model above, and it is now determined whether to perform resource adjustment based on the result of the prediction.
Through the dynamic analysis and allocation tool of system resources, system operation and maintenance personnel can monitor and analyze the space occupation conditions of a database, a server hard disk, middleware and the like in real time, make correct and timely judgment according to the increase and time evaluation, and improve the system operation and maintenance level.
For interactive applications, the experience of system performance to the user is directly reflected by the "system response time", for systems performance is expressed by the "number of concurrent users" or "throughput". Throughput in interactive applications, which reflects the server's bearing pressure, is used to explain the system-level load capacity.
And according to the system service period, the resources are dynamically adjusted, and the resource utilization rate is improved. And the model evaluation adopts a simulation experiment to adjust resources, the effective rate evaluation of the model adopts an average value of the accuracy, and the stability evaluation adopts an effective mean square error. And (3) recording that N times of experiments are carried out, wherein the resource adjustment times of a certain experiment are Ni, the effective rate is Ri, and then:
wherein xj represents the j-th resource adjustment is valid or not
The average effective rate of N experiments is:
the closer to 100%, the higher the model prediction accuracy.
The mean square error of the accuracy is:
the closer the MSE is to 0, the more stable the model is.
In the test, the average response time of the system is less than 5 seconds, the predicted resource utilization rate is 7.4%, and resource adjustment is not needed.
Step S24, capacity adjustment strategy
After predicting the condition of the next service period and making a judgment, a corresponding adjustment strategy needs to be provided according to the judgment result, and a quantitative strategy can be provided by combining the system architecture.
In the state evaluation model, the concept of service units is defined, a server in each service unit consists of a plurality of servers with the same configuration information, and the system has load balancing capability, namely, each service unit can share all loads. Quantitative evaluations will be given below for resource adjustments.
Defining global parameters:
n0: indicating the current number of service units
Nt: representing the number of service units at time t in the future
M: indicating the adjusted number of service units
1. Augmenting resource condition analysis
The following parameters are defined:
umax: representing the maximum number of supported online users or the maximum concurrency of the system
Then, the service capability of each unit we denote by P: p = U/N
Ut: and expressing the predicted value of the online user number or the predicted value of the system concurrency in the next service period.
Then, the number of cells that needs to be increased at least is:
when M is a decimal, we consider that we all adjust to the service unit, and therefore should adjust to an integer, rounding up.
2. Released resource condition analysis
The following parameters are defined:
rmax: representing resource utilization cap
Rmin: indicating a lower bound on resource utilization
M: indicating adjusted number of service units
Rt: indicating the predicted value of the system resource utilization rate of the next service period
Mmax: the maximum number of units that need to be released is:
mmin: the minimum number of units that need to be released is:
rounding is also used if Mmax, mmin is fractional, since it is the release of resources at this time, rounding down.
The specific embodiments are given above, but the present invention is not limited to the described embodiments. The basic idea of the present invention lies in the above basic scheme, and it is obvious to those skilled in the art that no creative effort is needed to design various modified models, formulas and parameters according to the teaching of the present invention. Variations, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention, and still fall within the scope of the invention.

Claims (11)

1. A power information system resource allocation system is characterized by comprising a power information system resource device, a storage capacity controller and a system state early warning device, wherein the power information system resource device is connected with the storage capacity controller;
the power information system resource device allocates the cloud resource virtualization resources in real time, connects real-time resource data in the IT operation and maintenance system, and outputs an allocation scheme, so that the resources of all information systems are balanced and stably run;
the storage capacity controller predicts the capacity of a hard disk and a database in each information system, performs terminal alarm and prompt according to a set threshold value, and performs capacity increase or reduction in time according to the resource occupancy condition;
the system state early warning device evaluates and analyzes the current state of the information system, analyzes and warns the normal, attention, abnormal and dangerous states of the system according to all comprehensive evaluation indexes of the system, guides operation and maintenance personnel of the information system to find and process problems in time, and enables the system to operate safely and effectively.
2. A power information resource allocation method is characterized in that the method adopts the power information system resource allocation system to predict the change situation of the power information system resource usage in the future service period through a time sequence model, and a resource allocation strategy is formulated according to the characteristics of each information system in the power grid service, and the specific method comprises the following steps: step one, data collection and pretreatment; secondly, a model is constructed by adopting a time series algorithm, and an electric power service system is analyzed; and step three, carrying out resource allocation of the power information system in a hierarchical mode.
3. The method according to claim 2, wherein the step of data collection and pre-processing comprises the following steps:
recording the number of types of components as m, data as acquisition indexes of certain types of components of all equipment at the current moment, wherein the sample capacity is n, namely the sample comprises index data of n types of components, and the index number on the ith type of components is pi (i =1,2., m); in order to measure the degree of the index deviating from the standard value or the index standard range value and further measure the health degree of the component, a conversion rule is formulated:
for the index with the standard value, the acquired value is converted into the deviation rate, and the smaller the deviation rate is, the more stable the equipment state is. Recording the acquisition value as v, the standard value as u, and the deviation ratio as x, then:
for the same type of equipment, a plurality of parts exist in quantity, and a plurality of corresponding index values also exist, so that a plurality of index values need to be converted into 1 comprehensive index value; the final inputs to the model are as follows:
X i ＝(x 1i x 2i … x ni ) T ,i＝1,2,...,p i
wherein Xi is the sample value of the index i, and xij is the value of the jth index of the ith component. The magnitude order and the dimension of different indexes can be different, and standardization treatment is needed to eliminate the influence of the magnitude order and the dimension; for the forward direction index, i.e. the larger and better index, the following conversion is made:
for negative indicators, i.e. the smaller the better the indicator, the following conversion is made:
the indexes after standardization all belong to an interval [0,1], and for the sake of understanding, the data after standardization is still denoted as xij; the objective weight of each index is comprehensively measured by the contrast strength and the conflict; assuming that Cj represents the amount of information included in the j-th evaluation index, cj may be represented as:
wherein, the sigma j represents the standard deviation of the index Xi, and the standard deviation can be estimated and recorded according to the sample under the condition that the sigma j is unknownFor the estimate of σ j, the calculation formula is as follows:
4. the method as claimed in claim 2, wherein in the second step, a model is constructed by using a time series algorithm, the power service system is analyzed, all indexes related to the model research change with time, and an ARIMA model, i.e., a differential autoregressive moving average model, is used.
5. The method for allocating power information resources according to claim 4, wherein the difference autoregressive moving average model in the second step has the following specific technical scheme:
the time series corresponding to the index X can be expressed as:
{X t :t＝0,±1,±2,...}
if X is a plateau sequence:
wherein E represents desired;
if X is a non-stationary sequence then X needs to be paired first t Carrying out difference operation, recording delta as a difference operator,
the difference calculation process is as follows:
ΔX t ＝X t -X t-1 ＝X t -BX t ＝(1-B)X t
Δ 2 X t ＝ΔX t -ΔX t-1 ＝(1-B)X t -(1-B)X t-1 ＝(1-B) 2 X t
Δ d X t ＝(1-B) d X t
the difference result is recorded as Wt:
W t ＝Δ d X t ＝(1-B) d X t
then Wt is a stationary sequence and the resulting model is called Xt-ARIMA (p, d, q) in the form:
where δ is a constant and ut is a white noise sequence (a sequence where the expectation and variance are both constant), the model can be simplified to the following form:
Φ(B)Δ d X t ＝δ+Θ(B)u t
where Φ (B) is a p-order autoregressive coefficient polynomial:
Θ (B) is a moving average coefficient polynomial of order q:
Θ(B)＝1-θ 1 B-θ 2 B 2 -…-θ q B q
when δ =0, the above model is a centralized ARIMA (p, d, q) model;
for ARIMA (a, b, c), when a and c are both 0 in the model, an autoregressive model AR (p) should be applied, and when b and c are both 0, a moving average model MA (q) should be applied;
the autoregressive model AR (p) is:
the moving average model MA (q) is:
for the ARIMA (p, d, q) model, an autoregressive model AR (p) is obtained when d and q are both 0, and a moving average model MA (q) is obtained when d and p are both 0.
6. The method for allocating electric power information resources according to claim 4, wherein the method for allocating time series in information system resources according to the second step comprises the steps of firstly judging trend change rules through a scatter diagram, autocorrelation and partial correlation graphs, identifying whether the trend change rules belong to a stationary sequence or a non-stationary sequence, and if the trend change rules belong to the non-stationary time series, performing stabilization processing on the non-stationary time series, and eliminating variance on data to enable the parameters to approach zero, thereby achieving the analysis requirements of the stationary sequence, and the specific construction method comprises the following steps:
firstly, checking the variance, the trend and the seasonal change rule of a time sequence by using an ADF unit root according to a scatter diagram, an autocorrelation function and a partial autocorrelation function diagram of the time sequence, and identifying the stationarity of the sequence;
and step two, carrying out stabilization treatment on the non-stationary sequence: if the data sequence is non-stationary and has a certain increasing or decreasing trend, the data needs to be subjected to differential processing, if the data has an variance, the data needs to be subjected to technical processing until the autocorrelation function value and the partial correlation function value of the processed data are not remarkably different from zero;
thirdly, establishing a corresponding model according to the identification rule of the time series model: if the partial correlation function of the stationary sequence is truncated and the autocorrelation function is trailing, it can be concluded that the sequence fits the AR model; if the partial correlation function of the stationary sequence is tailing and the autocorrelation function is truncation, the sequence can be judged to be suitable for the MA model; if the partial correlation function and the autocorrelation function of the stationary sequence are both trailing, the sequence is suitable for the ARMA model;
fourthly, performing parameter estimation and checking whether the statistical significance is achieved;
fifthly, performing hypothesis test to diagnose whether the residual sequence is white noise;
sixthly, performing predictive analysis by using the model which passes the inspection;
the number of indexes needing to be predicted is p, and the index i historical data sequence is as follows:
X i ＝{x t,i ,x t-1,i ,...,x t-n,i },i＝1,2,...,p，
for the sequence Xi, predicting the index xt +1,I at the t +1 moment according to ARIMA, and finally obtaining the index estimation vector of the t +1 moment as
WhereinIs the predicted value of the ith index;
and judging a suitable model according to the formula, if the partial correlation function is truncated, the model is suitable for an AR model, namely an autoregressive model, if the autocorrelation is truncated and the partial correlation is trailing, the model is suitable for an MA model, namely a moving average model, and if the autocorrelation and the partial correlation coefficient are both truncated, the model is suitable for an ARMA model, namely a differential autoregressive moving average model.
7. A cloud resource capacity evaluation method is characterized in that cloud resource capacity analysis and evaluation are performed based on a cloud resource application state evaluation model, a capacity evaluation model construction method is used for determining system application indexes and resource index baselines according to historical data training results and in combination with business system index management requirements; predicting the data of the next service period according to a large amount of historical data and by combining a service time window, and mastering the change condition of the data of the next service period; comparing the predicted data of the next service period with the baseline data to find whether the temporary data of the next service period exceeds the historical baseline data or not, and if so, adjusting resources to meet the service requirement; and after the adjustment time is mastered, analyzing the adjustment quantity, defining a service unit by combining a system service architecture, finding the relation between the server unit and system application data and resource data, and obtaining the specific adjustment quantity by combining a service period prediction result.
8. The method for evaluating the capacity of the cloud resource according to claim 7, wherein the method comprises the following specific steps: step one, establishing a base line; step two, service period data prediction; step three, capacity evaluation; step four, capacity adjustment.
9. The method for evaluating cloud resource capacity according to claim 8, wherein the method specifically comprises the following steps: establishing a base line by adopting the following steps: 1. screening online user number data of a service system in a historical period; 2. drawing a scatter diagram of the number of online users, and observing the distribution condition of the number of online users; 3. and determining a system application index and a resource index baseline according to a historical data training result and in combination with a service system index management requirement, wherein after the baseline is determined, the later-stage system operation is operated within the range of the baseline, and if the baseline requirement is not met, the system is considered to be unstable.
10. The method for evaluating the capacity of the cloud resource according to claim 8, wherein the method comprises the following specific steps: step two, the service period data prediction adopts the following steps:
step one, establishing an index baseline: firstly, screening online user number data of a service system in a historical period; secondly, drawing a scatter diagram of the number of online users, and observing the distribution condition of the number of online users; secondly, calculating the standard deviation of the index value, and constructing the base line by adding or subtracting the standard deviation from the predicted value of the index;
step two, index prediction: according to a scatter diagram, an autocorrelation function and a partial autocorrelation function diagram of the time sequence, the variance, the trend and the seasonal change rule of the time sequence are tested by an ADF unit root, and the stationarity of the sequence is identified; carrying out stabilization treatment on the non-stationary sequence; and establishing a corresponding model according to the identification rule of the time series model, carrying out parameter estimation, checking whether the model has statistical significance, and carrying out prediction analysis by using the model which passes the check.
Step three, resource adjustment and judgment:
when any one of the predicted online user number or the resource utilization rate exceeds the limit, the adjustment is needed; when the predicted resource utilization rate is lower than the minimum value of the standard resource utilization rate, adjustment is needed; there are three situations for the service system to use the resources:
insufficient resources: using a load test method and a pressure test method in the performance test method, testing the maximum number of users supported by the system in the range of the average response time of the system and the utilization rate of system resources in the operating performance parameter, and using U to represent the maximum number of users; according to the judgment basis, when R '> R (b) or U' > U, resource adjustment is needed, and the adjustment strategy is as follows: increasing resources;
excessive resources: when the average response time of the system is in a specified range, as long as the resource utilization rate is less than the standard minimum value, the resource is considered to be released; according to the judgment basis, when R' < R (a), resource adjustment is needed, and the adjustment strategy is as follows: releasing resources;
the resources satisfy: when the number of online users and the resource utilization rate are both in the standard value range, no adjustment is carried out; according to the judgment basis, when S is less than 5 and R' belongs to [ a, b ], resource adjustment is not needed;
step four, resource adjustment: after the condition of the next service period is predicted and judged, a corresponding adjustment strategy is provided according to the judgment result, and a quantitative strategy is provided by combining the system architecture.
11. The cloud resource capacity evaluation method of claim 8, wherein in the third capacity evaluation step, the model evaluation adopts a simulation experiment to adjust the resources, the model effective rate evaluation adopts an average value of accuracy rates, and the stability evaluation adopts an effective mean square error; the specific method comprises the following steps: and (3) recording that N times of experiments are carried out, wherein the resource adjustment times of a certain experiment are Ni, the effective rate is Ri, and then:
wherein xj represents the j-th resource adjustment is effective or not
The average effective rate of N experiments is:
the closer to 100%, the higher the model prediction accuracy rate is;
the mean square error of the accuracy is:
the closer the MSE is to 0, the more stable the model is.
CN201710989365.5A 2017-10-23 2017-10-23 A kind of power information system resource allocation system and method Pending CN107784440A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710989365.5A CN107784440A (en) 2017-10-23 2017-10-23 A kind of power information system resource allocation system and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710989365.5A CN107784440A (en) 2017-10-23 2017-10-23 A kind of power information system resource allocation system and method

Publications (1)

Publication Number Publication Date
CN107784440A true CN107784440A (en) 2018-03-09

Family

ID=61435189

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710989365.5A Pending CN107784440A (en) 2017-10-23 2017-10-23 A kind of power information system resource allocation system and method

Country Status (1)

Country Link
CN (1) CN107784440A (en)

Cited By (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108512919A (en) * 2018-03-25 2018-09-07 东莞市华睿电子科技有限公司 A kind of cloud storage space allocation method and server
CN108632365A (en) * 2018-04-13 2018-10-09 腾讯科技（深圳）有限公司 Service Source method of adjustment, relevant apparatus and equipment
CN108876157A (en) * 2018-06-25 2018-11-23 英特尔产品（成都）有限公司 Deploy the method and its management system of resources of production
CN109766234A (en) * 2018-12-11 2019-05-17 国网甘肃省电力公司信息通信公司 Disk storage capacity prediction technique based on time series models
CN110232521A (en) * 2019-06-12 2019-09-13 北京陆融通达科技有限责任公司 A kind of operation system stability assessment method and assessment device
CN110458601A (en) * 2019-07-09 2019-11-15 平安科技（深圳）有限公司 Processing method, device, computer equipment and the storage medium of resource data
CN111582530A (en) * 2019-02-18 2020-08-25 北京京东尚科信息技术有限公司 Method and device for predicting consumption of cloud product resources
CN111709779A (en) * 2020-06-01 2020-09-25 广东电网有限责任公司 Trading variety optimization system for electric power spot market
CN112085257A (en) * 2020-08-13 2020-12-15 国网上海市电力公司 Comprehensive energy service potential prediction model based on time sequence analysis
CN112269811A (en) * 2020-10-13 2021-01-26 北京同创永益科技发展有限公司 IT capacity prediction method and system based on traffic
CN112350878A (en) * 2019-08-07 2021-02-09 阿里巴巴集团控股有限公司 Pressure test system
CN112381359A (en) * 2020-10-27 2021-02-19 惠州蓄能发电有限公司 Multi-critic reinforcement learning power economy scheduling method based on data mining
CN112685173A (en) * 2020-12-22 2021-04-20 中通天鸿（北京）通信科技股份有限公司 Intelligent route distribution system based on rich media
CN112988506A (en) * 2021-02-19 2021-06-18 山东英信计算机技术有限公司 Big data server node performance monitoring method and system
CN113095566A (en) * 2021-04-08 2021-07-09 国网山东省电力公司电力科学研究院 Multi-source coordination energy supply method and system based on energy consumption behavior hierarchical analysis
CN113095669A (en) * 2021-04-08 2021-07-09 国网山东省电力公司电力科学研究院 Comprehensive energy scheduling method and system based on energy consumption coupling coordination
CN113515382A (en) * 2021-07-22 2021-10-19 中移（杭州）信息技术有限公司 Cloud resource allocation method and device, electronic equipment and program product
CN113709062A (en) * 2020-06-19 2021-11-26 天翼智慧家庭科技有限公司 Resource adjustment method and system based on business dynamic portrait
CN114390086A (en) * 2021-06-30 2022-04-22 国网新疆电力有限公司信息通信公司 Method suitable for power communication operation analysis
CN114648178A (en) * 2022-05-12 2022-06-21 武汉格蓝若智能技术有限公司 Operation and maintenance strategy optimization method of electric energy metering device based on DDPG algorithm
CN114896142A (en) * 2022-04-25 2022-08-12 矩阵时光数字科技有限公司 Performance index optimization method based on JMeter
CN115147008A (en) * 2022-08-02 2022-10-04 中国神华能源股份有限公司 Power plant unit storage resource real-time assessment method and system based on data lake technology
WO2023124469A1 (en) * 2021-12-28 2023-07-06 中兴通讯股份有限公司 Device energy saving method and system, electronic device, and storage medium
CN117170881A (en) * 2023-11-02 2023-12-05 建信金融科技有限责任公司 Resource regulation method and device, storage medium and processor

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102799512A (en) * 2012-07-04 2012-11-28 南京邮电大学 Virtual machine monitoring method in vector-autoregression-based cloud computing
CN103873498A (en) * 2012-12-11 2014-06-18 中国电信股份有限公司 Cloud platform resource self-adaptive early warning method and system
CN104836819A (en) * 2014-02-10 2015-08-12 阿里巴巴集团控股有限公司 Dynamic load balancing method and system, and monitoring and dispatching device
CN106227647A (en) * 2016-07-20 2016-12-14 云南电网有限责任公司信息中心 A kind of data analysing method for server hardware state evaluation
CN106293949A (en) * 2016-08-19 2017-01-04 浪潮电子信息产业股份有限公司 Resource dispatching strategy based on baseline analysis under a kind of computing environment
CN106447171A (en) * 2016-08-31 2017-02-22 清华大学 Power demand side scheduling resource potential modeling method and system

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102799512A (en) * 2012-07-04 2012-11-28 南京邮电大学 Virtual machine monitoring method in vector-autoregression-based cloud computing
CN103873498A (en) * 2012-12-11 2014-06-18 中国电信股份有限公司 Cloud platform resource self-adaptive early warning method and system
CN104836819A (en) * 2014-02-10 2015-08-12 阿里巴巴集团控股有限公司 Dynamic load balancing method and system, and monitoring and dispatching device
CN106227647A (en) * 2016-07-20 2016-12-14 云南电网有限责任公司信息中心 A kind of data analysing method for server hardware state evaluation
CN106293949A (en) * 2016-08-19 2017-01-04 浪潮电子信息产业股份有限公司 Resource dispatching strategy based on baseline analysis under a kind of computing environment
CN106447171A (en) * 2016-08-31 2017-02-22 清华大学 Power demand side scheduling resource potential modeling method and system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
向华伟 等: "基于***架构的云计算虚拟资源动态调配研究", 《电子技术与软件工程》 *
张丽梅 等: "《海洋经济评估模型与方法》", 31 October 2014, 西北农林科技大学出版社 *

Cited By (36)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108512919A (en) * 2018-03-25 2018-09-07 东莞市华睿电子科技有限公司 A kind of cloud storage space allocation method and server
CN108632365A (en) * 2018-04-13 2018-10-09 腾讯科技（深圳）有限公司 Service Source method of adjustment, relevant apparatus and equipment
CN108632365B (en) * 2018-04-13 2020-11-27 腾讯科技（深圳）有限公司 Service resource adjusting method, related device and equipment
CN108876157A (en) * 2018-06-25 2018-11-23 英特尔产品（成都）有限公司 Deploy the method and its management system of resources of production
CN108876157B (en) * 2018-06-25 2019-09-06 英特尔产品（成都）有限公司 Deploy the method and its management system of resources of production
CN109766234A (en) * 2018-12-11 2019-05-17 国网甘肃省电力公司信息通信公司 Disk storage capacity prediction technique based on time series models
CN111582530A (en) * 2019-02-18 2020-08-25 北京京东尚科信息技术有限公司 Method and device for predicting consumption of cloud product resources
CN110232521B (en) * 2019-06-12 2021-07-20 北京陆融通达科技有限责任公司 Service system stability evaluation method and evaluation device
CN110232521A (en) * 2019-06-12 2019-09-13 北京陆融通达科技有限责任公司 A kind of operation system stability assessment method and assessment device
CN110458601A (en) * 2019-07-09 2019-11-15 平安科技（深圳）有限公司 Processing method, device, computer equipment and the storage medium of resource data
CN110458601B (en) * 2019-07-09 2023-11-24 平安科技（深圳）有限公司 Method and device for processing resource data, computer equipment and storage medium
CN112350878A (en) * 2019-08-07 2021-02-09 阿里巴巴集团控股有限公司 Pressure test system
CN111709779A (en) * 2020-06-01 2020-09-25 广东电网有限责任公司 Trading variety optimization system for electric power spot market
CN113709062A (en) * 2020-06-19 2021-11-26 天翼智慧家庭科技有限公司 Resource adjustment method and system based on business dynamic portrait
CN113709062B (en) * 2020-06-19 2024-04-16 天翼视联科技有限公司 Resource adjustment method and system based on business dynamic portrayal
CN112085257A (en) * 2020-08-13 2020-12-15 国网上海市电力公司 Comprehensive energy service potential prediction model based on time sequence analysis
CN112085257B (en) * 2020-08-13 2024-04-12 国网上海市电力公司 Comprehensive energy business potential prediction model based on time sequence analysis
CN112269811A (en) * 2020-10-13 2021-01-26 北京同创永益科技发展有限公司 IT capacity prediction method and system based on traffic
CN112381359A (en) * 2020-10-27 2021-02-19 惠州蓄能发电有限公司 Multi-critic reinforcement learning power economy scheduling method based on data mining
CN112685173A (en) * 2020-12-22 2021-04-20 中通天鸿（北京）通信科技股份有限公司 Intelligent route distribution system based on rich media
CN112988506A (en) * 2021-02-19 2021-06-18 山东英信计算机技术有限公司 Big data server node performance monitoring method and system
CN112988506B (en) * 2021-02-19 2022-05-17 山东英信计算机技术有限公司 Big data server node performance monitoring method and system
CN113095566A (en) * 2021-04-08 2021-07-09 国网山东省电力公司电力科学研究院 Multi-source coordination energy supply method and system based on energy consumption behavior hierarchical analysis
CN113095669A (en) * 2021-04-08 2021-07-09 国网山东省电力公司电力科学研究院 Comprehensive energy scheduling method and system based on energy consumption coupling coordination
CN113095566B (en) * 2021-04-08 2023-02-07 国网山东省电力公司电力科学研究院 Multi-source coordination energy supply method and system based on energy consumption behavior hierarchical analysis
CN114390086A (en) * 2021-06-30 2022-04-22 国网新疆电力有限公司信息通信公司 Method suitable for power communication operation analysis
CN114390086B (en) * 2021-06-30 2023-11-17 国网新疆电力有限公司信息通信公司 Method suitable for power communication operation analysis
CN113515382B (en) * 2021-07-22 2024-04-09 中移（杭州）信息技术有限公司 Cloud resource allocation method and device, electronic equipment and storage medium
CN113515382A (en) * 2021-07-22 2021-10-19 中移（杭州）信息技术有限公司 Cloud resource allocation method and device, electronic equipment and program product
WO2023124469A1 (en) * 2021-12-28 2023-07-06 中兴通讯股份有限公司 Device energy saving method and system, electronic device, and storage medium
CN114896142A (en) * 2022-04-25 2022-08-12 矩阵时光数字科技有限公司 Performance index optimization method based on JMeter
CN114896142B (en) * 2022-04-25 2024-05-07 矩阵时光数字科技有限公司 JMeter-based performance index optimization method
CN114648178A (en) * 2022-05-12 2022-06-21 武汉格蓝若智能技术有限公司 Operation and maintenance strategy optimization method of electric energy metering device based on DDPG algorithm
CN115147008A (en) * 2022-08-02 2022-10-04 中国神华能源股份有限公司 Power plant unit storage resource real-time assessment method and system based on data lake technology
CN117170881B (en) * 2023-11-02 2024-02-02 建信金融科技有限责任公司 Resource regulation method and device, storage medium and processor
CN117170881A (en) * 2023-11-02 2023-12-05 建信金融科技有限责任公司 Resource regulation method and device, storage medium and processor

Similar Documents

Publication Publication Date Title
CN107784440A (en) A kind of power information system resource allocation system and method
CN105117602B (en) A kind of metering device running status method for early warning
CN108241901B (en) Transformer early warning evaluation method and device based on prediction data
CN107547269B (en) Method for constructing intelligent substation communication flow threshold model based on FARIMA
CN109525435A (en) A kind of electrical network business operation condition of server method for early warning
CN111881574B (en) Wind turbine generator set key component reliability modeling method based on distribution function optimization
CN112668943A (en) Distribution line health state assessment method and system
CN102354346A (en) Solar wing spreading reliability estimation method based on multilevel gray comprehensive model
CN113408869A (en) Power distribution network construction target risk assessment method
CN112116198A (en) Data-driven process industrial state perception network key node screening method
KR20200103193A (en) Methods and apparatuses for forecasting power demand using deep structure
CN114720764A (en) Harmonic analysis method and system based on real-time monitoring data of electric meter
CN116702612A (en) CVT health state prediction method and system based on multidimensional index fusion
CN104022899A (en) Three-dimensional assessment method for network management system and system
CN113742993B (en) Method, device, equipment and storage medium for predicting life loss of dry-type transformer
CN114595883A (en) Oil-immersed transformer residual life personalized dynamic prediction method based on meta-learning
CN111932081B (en) Method and system for evaluating running state of power information system
CN111985845B (en) Node priority optimization method of heterogeneous Spark cluster
Atukeren The relationship between the F-test and the Schwarz criterion: implications for Granger-causality tests
CN109165854B (en) Empty pipe operation efficiency grade evaluation method and device
CN110750848A (en) Method for estimating remaining life of software-hardware degradation system by considering software operation
CN112215394B (en) Converter transformer vibration signal prediction method, device, equipment and storage medium
CN114880067A (en) Container scheduling strategy based on container load prediction
CN108898890B (en) Empty pipe operation efficiency grade evaluation method and device
CN112529418A (en) Power distribution network investment benefit evaluation method based on matter element extension evaluation model

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20180309

RJ01 Rejection of invention patent application after publication