CN111045907B - System capacity prediction method based on traffic - Google Patents
System capacity prediction method based on traffic Download PDFInfo
- Publication number
- CN111045907B CN111045907B CN201911273253.5A CN201911273253A CN111045907B CN 111045907 B CN111045907 B CN 111045907B CN 201911273253 A CN201911273253 A CN 201911273253A CN 111045907 B CN111045907 B CN 111045907B
- Authority
- CN
- China
- Prior art keywords
- model
- data
- prediction
- system capacity
- day
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3466—Performance evaluation by tracing or monitoring
- G06F11/3476—Data logging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3447—Performance evaluation by modeling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3452—Performance evaluation by statistical analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention discloses a system capacity prediction method based on traffic, which is characterized by comprising the following steps: the method comprises the steps of collecting original data of more than one month, firstly removing seasonal factor influence, then selecting a model group based on a linear model, a moving average model and an ARIMA model, calculating a weight index of each model by using a multiple regression model, then introducing genetic algorithm machine learning to carry out iterative optimization on the model group to obtain optimal model group parameters, calculating a daily dimension prediction result and a minute dimension prediction result, and calculating the system capacity demand based on the two-dimension prediction result by using the fitting relation of service and historical data of the system capacity. The application of the prediction method of the invention comprehensively considers the influence of external factors on the prediction, predicts through the model group and introduces the genetic algorithm to automatically optimize parameters, but can flexibly deal with the change of transaction structure and traffic trend, and improves the prediction result to ensure that the average error is less than 5 percent, thereby being beneficial to ensuring the service quality of Internet application.
Description
Technical Field
The invention relates to the technical field of software of computers and network products, in particular to a prediction method for exploring the relation between service access and hardware resources occupied by an application software system and planning the capacity of required hardware resources (such as a CPU (Central processing Unit), a memory, network flow and the like).
Background
With the rapid development of IT technology, the Internet application gradually becomes an important driving force for business innovation and enterprise quality improvement and efficiency improvement. Therefore, it is important to guarantee the quality of service of internet applications. Under the rapid popularization of mobile and internet, higher requirements are provided for the access of various festivals and holidays and marketing days to internet application services; under high-frequency and high-concurrency access, the internet application needs more hardware resources to support and guarantee the normal operation of the service; however, since the languages, frameworks and services implemented by various internet applications are different, various hardware resources on which the applications depend are different, for example, the types of the applications can be classified into CPU-intensive type, memory-intensive type, flow-intensive type and the like according to occupied resources, which provides a higher challenge for planning the hardware resource capacity required by the applications by the service department.
At present, in order to better improve the utilization and use effects of hardware resources and provide a more reasonable basis for hardware resource planning, operation and maintenance product manufacturers are exploring the realization of hardware resource capacity prediction based on artificial intelligence and a machine learning framework. Although the current frames for machine learning and deep learning are Caffe (http:// cafe. bergeleyvision. org /), Trensorflow (https:// tensorflow. org), and the like, all provide a general algorithm, a model and a frame, the prediction of the capacity is strong in correlation with the access condition of the internet application service, so that the algorithm cannot be directly applied to be directly adapted, the direct application of an actual service scene is lacked, and a general method for predicting the system capacity of the service is lacked.
Disclosure of Invention
The invention aims to provide a general method for predicting the capacity of an application system for the requirements of system hardware resources based on the traffic of Internet application.
The technical solution of the present invention for achieving the above object is a traffic-based system capacity prediction method, characterized in that: the method comprises the steps of collecting original data of more than one month, firstly removing seasonal factor influence, then selecting a model group based on a linear model, a moving average model and an ARIMA model, calculating a weight index of each model by using a multiple regression model, then introducing genetic algorithm machine learning to carry out iterative optimization on the model group to obtain optimal model group parameters, calculating a daily dimension prediction result and a minute dimension prediction result, and calculating the system capacity demand based on the two-dimension prediction result by using the fitting relation of service and historical data of the system capacity.
In the traffic-based system capacity prediction method, the original data includes service access or transaction data, and the data field includes transaction occurrence time, traffic and marketing identification, wherein the data dimension of the traffic is in the minute level; and the data processing comprises classifying and marking the daily data according to the week and recording the date of the business marketing activity.
In the traffic-based system capacity prediction method, further, the processed data is stored in a time sequence database, and the data structure comprises type, week, date and transaction number; and performing data operation in the memory by adopting a DataFrame loader in the pandas framework of python.
In the traffic-based system capacity prediction method, the seasonal factor is calculated based on the overall mean value and the mean values of various categories of the original data, and the formula is shown as follows;
the eliminating season factor influence is that the data of the corresponding type in each day is divided by the season factor.
The traffic-based system capacity prediction method further comprises the following steps of introducing genetic algorithm optimization: taking the data of the day before the date to be predicted as a reference, setting basic parameters of population quantity, evolution times, the maximum value allowed to appear in genes, chromosome length, mating probability and mutation probability by adopting a genetic algorithm, carrying out combined intersection and mutation by using a genetic operator in each generation according to the fitness of an individual to obtain a next generation population, wherein the fitness of the individual is obtained by the RSS residual square sum of nearly 5 days of the comprehensive prediction evaluation result of a model group, and returning to remove the influence of seasonal factors and reaching the maximum iteration times.
In the method for predicting system capacity based on traffic, further, in the calculation process of the optimization process, a linear regression model in python framework skleern is adopted to analyze model parameters and calculate the square sum of RSS residuals, and the calculation formula is as follows:
the system capacity prediction method of the invention has the prominent substantive characteristics and remarkable progress: the method comprehensively considers the influence of external factors on prediction, predicts through a model group and introduces genetic algorithm to automatically optimize parameters, can flexibly cope with the change of transaction structures and traffic trends, improves the prediction result to enable the average error to be less than 5%, and is beneficial to guaranteeing the service quality of Internet application.
Drawings
Fig. 1 is a flow topology diagram of the system capacity prediction method of the present invention.
Detailed Description
The following detailed description of the embodiments of the present invention is provided in connection with the accompanying drawings for the purpose of understanding and controlling the technical solutions of the present invention, so as to define the protection scope of the present invention more clearly.
As described above, since the prediction of the hardware system capacity in the internet application is highly correlated with the service access situation, it is necessary to correlate the service prediction when the system capacity prediction is performed. The outline of the prediction method is as follows: the method comprises the steps of collecting original data of more than one month, firstly removing seasonal factor influence, then selecting a model group based on a linear model, a moving average model and an ARIMA model, calculating a weight index of each model by using a multiple regression model, then introducing genetic algorithm machine learning to carry out iterative optimization on the model group to obtain optimal model group parameters, calculating a daily dimension prediction result and a minute dimension prediction result, and calculating the system capacity demand based on the two-dimension prediction result by using the fitting relation of service and historical data of the system capacity.
From another perspective, the overall idea of the service prediction scheme is divided into two blocks, one block is day dimension prediction, and the other block is minute dimension prediction. In the aspect of daily dimension, after the original daily data are subjected to elimination marketing and periodic processing, daily transaction amount data are comprehensively predicted through model combination. In order to realize the self-adaptive adjustment of later-stage parameters and reduce manual operation, a genetic algorithm is adopted for machine learning to select a parameter model. In the minutes dimension, raw minutes data are scaled for the day and sorted by period. And finally, multiplying the result of the daily forecast data by the minute ratio to obtain the per minute forecast data.
For a clearer understanding, the specific solution of the substeps is implemented as follows:
the method comprises the following steps of firstly, processing basic day data, collecting original data of more than one month, namely service access or transaction data, wherein data fields comprise transaction occurrence date, transaction number (namely service volume) and marketing identification, and the data dimension of the service volume is in a minute level. The raw data is processed by the following steps: daily data are labeled by week classification. And recording the date of the business marketing campaign to judge the growth rate of the business volume on a specific date, and performing data correction aiming at a specific marketing special point.
Secondly, identifying seasonal factors, wherein different access demands, namely access volumes, can be obtained in different weeks and different time periods when a customer accesses or uses a service system; the data is classified according to the day of the week marking marked in the first step into 5 categories of working day, day of vacation (saturday), sunday, day after vacation, day of vacation, and for eliminating the influence of the season factor later.
And thirdly, selecting initial model group parameters, forming a sample amount n2 by adopting least square method first-order polynomial fitting operation according to the total sample amount n1, forming a period number n3 by adopting first-order moving smoothing and second-order moving smoothing, and finally giving the parameters of a specific model group by a genetic algorithm initial population for subsequent model group prediction based on the sample amount.
Fourthly, eliminating the seasonal factor, wherein the seasonal factor needs to be eliminated under the overall sample size due to the existence of the seasonal factor, and an overall average value and each category average value are respectively calculated under the overall sample size n1, namely
And fifthly, predicting the model group, namely adopting a model 1: linear equation (LM): linear model sample size n2, and fitting linear model prediction by using a least square method first-order polynomial; then, model 2 is adopted: second Moving Average (MA): calculating a primary moving average and a secondary moving average by using the total sample size n1 and delaying for n3 period, and calculating a moving average predicted value by using a formula; and finally adopting a model 3: ARIMA model: and predicting by using a time series model by using the total sample size n 3.
And sixthly, comprehensively predicting the model group, namely calculating the weight index of each model in the prediction model by combining three groups of models and utilizing a multiple regression model after the linear model, the moving average model and the ARIMA model are obtained in the fifth step, and using the weight index to synthesize the prediction result of the model group, so that the prediction result is more accurate.
And seventhly, adjusting parameters of the model group by a genetic algorithm, and performing combined intersection and mutation by using a genetic operator according to the fitness of the individual in each generation through parameters such as the population number, the evolution times, the maximum value allowed to appear in the gene, the chromosome length, the mating probability, the mutation probability and the like to obtain a next generation population, wherein the fitness of the individual is obtained through the RSS residual square sum of nearly 5 days of the comprehensive prediction evaluation result of the model group, namely the total sample amount n1, the linear model sample amount n2, the moving average period n3 and the like. And returning to the fourth step for calculation until the maximum iteration number is reached. The calculation formula of the square of the RSS residual error is as follows:
。
and eighthly, calculating a prediction result according to the daily dimension by using the model group and the optimal model group parameters obtained after the adjustment in the seventh step.
And step nine, processing the basic minute data. 1) Calculating the ratio of the traffic per minute in the total traffic of the day; 2) the daily data marks three categories of working days, holidays and marketing periods; 3) the average value of the traffic volume ratio per minute of nearly ten days in each type of date is selected as the reference value of the time-to-minute ratio of each type of date.
And step ten, predicting the traffic. And multiplying the day dimension prediction result of the eighth step by the minute ratio reference value of the ninth step to obtain a predicted value of the minute dimension traffic.
The tenth step, T +0 capacity prediction. And performing model fitting trial calculation by using the historical transaction data and the system capacity historical data to obtain the relation between the service volume and the system capacity, so as to predict and plan the system capacity according to the predicted value in the tenth step.
As shown in the drawings, embodiments of the present invention are described as follows:
first, data preparation
Preparing original day data and original minute data of more than one month, wherein the data format comprises transaction occurrence time, traffic and marketing identification, and the service data dimension is in a minute level; because the business data can be influenced by marketing business, causing high burst and large flow of data, the data needs to be processed, and the processing steps are as follows:
according to the service occurrence time, intercepting the week of the occurrence day, and taking the week as a seasonal factor, wherein the week is divided into 5 categories of working day, first day of vacation (saturday), sunday, days after vacation and first day after vacation; and recording the date of the business marketing activity to judge the increase of the transaction rate under a specific date, and performing data correction aiming at a specific marketing special point.
The original data is imported into the system through the text in the csv format, the data is processed in the importing process, and the processed data is stored in the time sequence database, so that the retrieval and processing of the data can be accelerated. The data structure in the time-series database includes type (season classification), week, date (transaction occurrence date in the format of yyyyMMdd), and transaction number. For convenience of data processing, a DataFrame in the pandas frame of python is used to load data into a program memory for data manipulation.
Second, optimal model set parameters
After the data is prepared, the optimal model group parameters are obtained according to the selected initial model so as to be used for realizing a transaction prediction model. The common prediction models comprise a fitting linear prediction model, a primary moving average prediction method, a secondary moving average prediction method, an ARIMA (differential integrated moving average autoregressive model) prediction model and other various modes.
The method comprises the steps of adjusting the weight indexes of three calculation models in a multiple regression model according to a genetic algorithm, namely, taking the data of the day before the date to be predicted as a reference, setting the population number to be 10, the evolution frequency to be 5, the maximum value allowed to appear in a gene to be [16,32,4], the chromosome length to be 4+5+2, the mating probability to be 0.6 and the variation probability to be 0.05 as basic parameters by adopting the genetic algorithm, and after an initial population is formed by encoding an actual parameter set, carrying out combined intersection and variation by means of a genetic operator according to the evolution frequency and the fitness of an individual to obtain a next generation population, wherein the fitness of the individual can be obtained by the RSS residual square sum of nearly 5 days of the comprehensive prediction evaluation result of the model group, so that the optimal solution of the weight indexes in the three calculation models is obtained. The calculation process adopts a Linear Regression model in python framework skleern to analyze the model parameters and calculate RSS (residual square sum).
Third, transaction prediction
And calculating prediction result data of the final day dimension according to given basic day dimension data through the optimal individuals (parameter sets) obtained from the optimal model group. In order to obtain traffic minute data prediction, firstly, processing the minute-based data, and calculating the proportion of the traffic per minute in the total traffic of the day; marking daily data into three categories of working days, holidays and marketing periods; and selecting the average value of business ratios per minute of nearly ten days in each class of dates as a reference value of the minute ratio of each class of dates, and finally multiplying the day dimension prediction result by the reference value of the minute ratio to obtain a predicted value of the minute dimension transaction.
Volume prediction
In order to obtain the result of predicting the system capacity, a linear relation between the capacity and the traffic needs to be calculated by calculating the fitting dependency of the historical data of each type of traffic or traffic flow data on various system resources (cpu, memory, disk, flow and other system indexes), and then the required capacity in the future is calculated according to the prediction result of the traffic in the third stage, so as to realize a traffic flow-based system capacity prediction model.
In summary, the system capacity prediction scheme and the embodiment of the present invention can be seen in detail, and the scheme has substantial features and progressions: the technical scheme comprehensively considers the influence of external factors such as seasons, marketing and the like on the transaction or the visit amount, a model group is formed by a plurality of models for prediction, and the model parameters are continuously adjusted every day to adapt to the setting of optimized parameters, so that the result accuracy is ensured; meanwhile, in order to solve the maintenance work of the later-stage model parameters and reduce the manual maintenance amount, a genetic algorithm method is introduced for machine learning, automatic parameter tuning is realized, the problems of change of a transaction structure and change of the traffic trend in the future can be flexibly solved, and the parameter self-adaption work is completed; finally, the average error of the prediction results of the daily model prediction and the minute model prediction is less than 5 percent. After the service prediction is completed, the future system capacity requirement is calculated based on the fitting relation between the service data and the system capacity data, and the purpose of predicting the future system capacity is achieved.
In addition to the above embodiments, the present invention may have other embodiments, and any technical solutions formed by equivalent substitutions or equivalent transformations are within the scope of the present invention as claimed.
Claims (3)
1. A traffic-based system capacity prediction method is characterized in that: acquiring raw data of more than one month, and performing basic day data processing: firstly, eliminating the influence of seasonal factors, intercepting the week of a day of occurrence according to the occurrence time of a service, taking the week as a seasonal factor, and dividing the week into five categories including a working day, a first day of vacation, a weekday, a plurality of days after vacation and a first day after vacation, wherein the seasonal factors are obtained by calculation based on the overall mean value and the mean values of all categories of original data, and the formula is as follows:
dividing the data of the corresponding type every day by a season factor; then selecting a model group based on a linear model, a moving average model and an ARIMA model, calculating the weight index of each model by using a multiple regression model, and introducingThe method comprises the steps of performing iterative optimization on a model group by genetic algorithm machine learning to obtain the optimal model group parameters, calculating a daily dimension prediction result and a minute dimension prediction result, calculating the system capacity demand based on the prediction results of two dimensions by utilizing the fitting relation of historical data of service and system capacity, wherein the introduced genetic algorithm optimization process comprises the following steps: taking the data of the day before the date to be predicted as a reference, setting basic parameters of population quantity, evolution times, the maximum value allowed to appear in genes, chromosome length, mating probability and mutation probability by adopting a genetic algorithm, carrying out combined intersection and mutation by using a genetic operator in each generation according to the fitness of an individual to obtain a next generation population, wherein the fitness of the individual is obtained by the RSS residual square sum of nearly 5 days of the comprehensive prediction evaluation result of a model group, and returning to remove the influence of seasonal factors and reaching the maximum iteration times; and the optimized calculation process adopts a Linear regression model of Linear regression in python framework skleern to analyze the model parameters and calculate the square sum of RSS residual errors, and the calculation formula is as follows:
2. The traffic-based system capacity prediction method of claim 1, wherein: the original data comprises access or transaction data of the business, and the data field comprises transaction occurrence time, business volume and marketing identification, wherein the data dimension of the business volume is in a minute level; and the basic day data processing comprises classifying and marking the data of each day according to the week and recording the date of the business marketing activity.
3. The traffic-based system capacity prediction method of claim 2, wherein: the processed data is stored in a time sequence database, and the data structure comprises the type, the week, the date and the number of transaction strokes; and performing data operation in the memory by adopting a DataFrame loader in the pandas framework of python.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911273253.5A CN111045907B (en) | 2019-12-12 | 2019-12-12 | System capacity prediction method based on traffic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911273253.5A CN111045907B (en) | 2019-12-12 | 2019-12-12 | System capacity prediction method based on traffic |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111045907A CN111045907A (en) | 2020-04-21 |
| CN111045907B true CN111045907B (en) | 2020-10-09 |
Family
ID=70235996
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911273253.5A Active CN111045907B (en) | 2019-12-12 | 2019-12-12 | System capacity prediction method based on traffic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111045907B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112035067A (en) * | 2020-09-04 | 2020-12-04 | 平安科技（深圳）有限公司 | Event-driven cloud storage capacity prediction method, system and device |
| 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 |
| CN112215444B (en) * | 2020-12-09 | 2021-04-02 | 平安科技（深圳）有限公司 | Prediction method, device and equipment based on distribution curve and storage medium |
| CN113077107A (en) * | 2021-04-19 | 2021-07-06 | 国网浙江省电力有限公司物资分公司 | Electric power material configuration demand prediction system |
| CN113344282B (en) * | 2021-06-23 | 2023-01-17 | 中国光大银行股份有限公司 | Method, system and computer readable medium for capacity data processing and allocation |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8306794B2 (en) * | 2008-06-26 | 2012-11-06 | International Business Machines Corporation | Techniques for thermal modeling of data centers to improve energy efficiency |
| CN101739614A (en) * | 2009-12-08 | 2010-06-16 | 江苏省邮电规划设计院有限责任公司 | Hierarchy-combined prediction method for communication service |
| WO2011128922A1 (en) * | 2010-04-15 | 2011-10-20 | Neptuny S.R.L. | Automated upgrading method for capacity of it system resources |
| CN104899451A (en) * | 2015-06-12 | 2015-09-09 | 哈尔滨工业大学 | Time sequence prediction model selecting method based on statistical signal process |
| CN106875027B (en) * | 2016-06-06 | 2020-11-06 | 创新先进技术有限公司 | Resource request value prediction method and device, and transaction amount prediction method |
| CN107229995A (en) * | 2017-05-24 | 2017-10-03 | 腾讯科技（深圳）有限公司 | Realize method, device and computer-readable recording medium that game service amount is estimated |
| CN108022004A (en) * | 2017-11-16 | 2018-05-11 | 广东电网有限责任公司信息中心 | A kind of adaptive weighting training method of multi-model weighted array Forecasting Power System Load |
| CN109698865A (en) * | 2018-12-26 | 2019-04-30 | 苏州博纳讯动软件有限公司 | A kind of cloud application caching method based on access prediction |
| CN109816158A (en) * | 2019-01-04 | 2019-05-28 | 平安科技（深圳）有限公司 | Combined method, device, equipment and the readable storage medium storing program for executing of prediction model |
-
2019
- 2019-12-12 CN CN201911273253.5A patent/CN111045907B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN111045907A (en) | 2020-04-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111045907B (en) | System capacity prediction method based on traffic | |
| CN108846528B (en) | Long-term load prediction method for large-scale industrial power consumer | |
| CN104424598A (en) | Cash demand quantity predicating device and method | |
| US8694427B2 (en) | Time-efficient and deterministic adaptive score calibration techniques for maintaining a predefined score distribution | |
| CN105243449A (en) | Method and device for correcting prediction result of electricity selling amount | |
| CN108388974A (en) | Top-tier customer Optimum Identification Method and device based on random forest and decision tree | |
| CN111476677A (en) | Big data-based electricity consumption type electricity sales quantity analysis and prediction method and system | |
| CN108154311A (en) | Top-tier customer recognition methods and device based on random forest and decision tree | |
| CN111415192A (en) | Water quality prediction method for user based on big data | |
| CN107688901B (en) | Data adjusting method and device | |
| CN111242368A (en) | Logistics distribution scheduling algorithm based on scene deduction | |
| CN116912016A (en) | Bill auditing method and device | |
| CN112990721B (en) | Power user value analysis method and system based on payment behaviors | |
| CN114154696A (en) | Method, system, computer device and storage medium for predicting fund flow | |
| US8185227B2 (en) | Method and system for determining manufacturing throughput target | |
| CN112348601A (en) | Service development route planning method | |
| Niu et al. | Optimization of advertising budget allocation over time based on LS-SVMR and DE | |
| CN105321001A (en) | Power selling data processing method and apparatus | |
| Vázsonyi | Overview of scenario tree generation methods, applied in financial and economic decision making | |
| CN117557078B (en) | Intelligent production plan management method and system based on time nodes | |
| CN115249166B (en) | Method, device, computer equipment and storage medium for predicting clear electricity price | |
| CN112101611B (en) | Real estate customer buyback time prediction method, server and storage medium | |
| Bataev | Innovative Forms of Interaction between Financial Institutions and Clients: Automated Banking Offices | |
| CN111461377B (en) | Prediction method/system, storage medium and apparatus for total energy demand | |
| CN114239909A (en) | Multi-subject collaborative manufacturing task distribution system and method under financial service constraint |
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 | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 215213 South Floor 4, Fuhua Building, No. 199 Chongwen Road, Suzhou Industrial Park, Jiangsu Province Patentee after: Jiangsu Boyun Technology Co.,Ltd. Address before: 215213 South Floor 4, Fuhua Building, No. 199 Chongwen Road, Suzhou Industrial Park, Jiangsu Province Patentee before: SUZHOU BONA XUNDONG SOFTWARE Co.,Ltd. |