CN112835949A - Load prediction method - Google Patents
Load prediction method Download PDFInfo
- Publication number
- CN112835949A CN112835949A CN202010944952.4A CN202010944952A CN112835949A CN 112835949 A CN112835949 A CN 112835949A CN 202010944952 A CN202010944952 A CN 202010944952A CN 112835949 A CN112835949 A CN 112835949A
- Authority
- CN
- China
- Prior art keywords
- data
- load
- prediction
- model
- effect
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000000007 visual effect Effects 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000000694 effects Effects 0.000 claims description 28
- 238000004458 analytical method Methods 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 238000011144 upstream manufacturing Methods 0.000 claims description 12
- 238000013499 data model Methods 0.000 claims description 10
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000012417 linear regression Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000012423 maintenance Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000004140 cleaning Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000011002 quantification Methods 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 3
- 230000002596 correlated effect Effects 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims 2
- 238000013277 forecasting method Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000013461 design Methods 0.000 description 6
- 238000011161 development Methods 0.000 description 3
- 230000005611 electricity Effects 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000012550 audit Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/248—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- General Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Mathematical Physics (AREA)
- Fuzzy Systems (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Software Systems (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a load forecasting method, which comprises the following steps: the method comprises the following steps: collecting data; step two: preprocessing data; step three: analyzing load characteristics; step four: self-adaptive load prediction; step five: displaying diversified results; step six: and (4) visual estimation after prediction. The invention continuously improves the load prediction accuracy of the whole network and each province, improves the safety check accuracy of the dispatching operation department in the spot market environment, and ensures the safe, stable and economic operation of the power system.
Description
Technical Field
The invention relates to the technical field of scheduling and distribution network load prediction, in particular to a load prediction method.
Background
With the continuous enlargement of the power grid scale and the increasing of the influence of economy, weather, industry, festivals and holidays, the load analysis and prediction difficulty is continuously increased. With the deep advance of electric power reform, the enthusiasm and interaction capacity of each market subject are greatly stimulated. Demand side response, market bidding strategies and distributed clean energy can obviously change the time-space characteristics of the load, influence the result of power grid security check, and foresee that power system reform will bring unprecedented new challenges to load prediction work. In recent years, the load prediction accuracy of the incoming call network shows a gradual decline trend, and a new technology and a new algorithm are urgently needed to be applied to improve the scientific and fine level of load prediction analysis. Therefore, intensive research on the aspects of influence factor mining, prediction technology improvement, hierarchical cooperation mechanism and the like is urgently needed to promote comprehensive enhancement and improvement of load prediction work.
With the increasingly complex national economic situation and the increasingly close relationship between economic development and energy consumption in recent years, various data of the modern economic society are not simple human society data any more, but are closely related to power load prediction and rich in key information required by the load prediction. The information can be obtained by observation, mining and summarization, and the defects of the traditional prediction method are overcome.
Disclosure of Invention
The invention aims to provide a load forecasting method, which continuously improves the load forecasting accuracy of the whole network and each province, improves the safety check accuracy of a dispatching operation department in a spot market environment, and ensures the safe, stable and economic operation of a power system so as to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme:
a method of load prediction comprising the steps of:
the method comprises the following steps: data acquisition
The data acquisition module is actively connected with an upstream system to acquire data, and stores the result on a large data platform HDFS for subsequent data processing;
step two: data pre-processing
Null values and abnormal data may exist in the acquired data, and in order to ensure the prediction accuracy, a set of proper cleaning rules needs to be established to provide an accurate data source for load analysis and prediction;
step three: load characteristic analysis
The preprocessed unified data model is analyzed, the advantages of a big data platform are fully utilized, the load characteristics are analyzed based on multiple dimensions of climate, solar terms, holidays and maintenance plans, whether the data characteristics are correlated with a prediction target or not is checked, effective data characteristics are reserved in the modeling data, irrelevant data characteristics are eliminated, the accuracy of the model can be improved, and the selection of a prediction algorithm is determined according to the data characteristics;
step four: adaptive load prediction
After the load characteristic is analyzed, a model wide table is generated, various machine learning algorithms such as linear regression, time sequence, neural network and the like are adopted for modeling, the effect of each model is verified, the measurement goodness of fit R2 and variance are adopted for quantification, the effect of the model is accurately judged, and visual display is carried out by adopting modes such as a chart, a curve graph and the like;
step five: diversified result display
Displaying the model prediction result number according to a hierarchy, supporting search query of regions and buses, and displaying regional load data of the regions when the regions are selected; when a bus is selected, displaying the bus load data, displaying the prediction data in a list and graph mode, and supporting data export;
step six: visual post-prediction assessment
In the operation process, the prediction accuracy is continuously monitored, the prediction effect is quantified by adopting the measurement goodness of fit R2 and the variance, and the visualization of the estimation after prediction is realized by using a chart for display.
Furthermore, the data acquisition module in the first step has a plurality of functions of data vacancy warning and data quality auditing, and generates warning information when the data has quality problems and displays the warning information on an interface.
Furthermore, in the second step, the source data are distributed in different tables, and the data need to be integrated to generate a uniform data model, which contains load values and other data characteristics and is used for model establishment and load prediction.
Further, if the data in the third step conforms to the linear relationship, a linear regression algorithm can be considered; if the data has periodicity, various load analysis reports can be generated by considering a time series algorithm.
Further, in the fourth step, based on the development of the adaptive optimal model evaluation technology, the model with the highest score is automatically selected for prediction according to R2 and the variance comprehensive score.
Furthermore, the graph in the fifth step can simultaneously display historical predicted data, historical real data and current predicted data, and visually see the load change trend.
Furthermore, a threshold value of the prediction effect is set in the sixth step, when the threshold value is exceeded, an alarm is automatically given, an alarm result is displayed on an interface, once the model effect is reduced, the model effect can be found in time, and modeling optimization can be carried out again.
Compared with the prior art, the invention has the beneficial effects that: the invention carries out unified management and distributed storage of mass data based on a unified load data model of a big data platform, realizes abnormal data cleaning and data standardization processing through the technologies of memory calculation, stream processing and the like, intelligently identifies different types of loads by adopting big data analysis technologies of deep learning, correlation analysis and the like, analyzes the spatial distribution characteristics of various loads and the influence mode of meteorological information on load prediction, establishes a high-precision load prediction model and a practical prediction method considering numerical weather information, and realizes multidimensional refined load prediction.
Drawings
FIG. 1 is a flow chart of a method of load prediction according to the present invention;
FIG. 2 is a data acquisition module layout of the present invention;
FIG. 3 is a block diagram of the data preprocessing module of the present invention;
FIG. 4 is a design of a load characteristic analysis module of the present invention;
FIG. 5 is a chart of annual load variation of the present invention
FIG. 6 is a load period analysis chart of the present invention;
FIG. 7 is a diagram illustrating an example of the correlation between the current load and the previous period according to the present invention;
FIG. 8 is an exemplary graph of the temperature versus load dependence of the present invention;
FIG. 9 is an exemplary graph of wind speed versus load dependence of the present invention;
FIG. 10 is an exemplary graph of humidity versus load dependence of the present invention;
FIG. 11 is a holiday load analysis chart of the present invention;
FIG. 12 is a block diagram of the adaptive load prediction module of the present invention;
FIG. 13 is a diagram illustrating an exemplary effect of a prediction model according to the present invention;
FIG. 14 is a design diagram of a diversified result display module according to the present invention;
FIG. 15 is a visual post-prediction evaluation plan of the present invention;
FIG. 16 is a graph comparing the effect of the historical model with the effect of the current model according to the present invention;
FIG. 17 is a flow chart of model training of the present invention;
FIG. 18 is a flow chart of model prediction according to the present invention;
FIG. 19 is a post-model evaluation flow diagram of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
A method of load prediction, the flow chart being as in fig. 1, comprising the steps of:
the method comprises the following steps: data acquisition
The data acquisition module is actively connected with an upstream system to acquire data, and stores results on a large data platform HDFS for subsequent data processing. The module has the functions of data vacancy warning, data quality audit and the like, generates warning information when the data has quality problems, and displays the warning information on an interface; and starting an acquisition program through ETL scheduling, wherein the data acquisition module is actively connected with an upstream system to acquire data, and stores the result on a large data platform HDFS (distributed file system) for subsequent data processing. After the data generation is completed, the upstream system writes a log with standard data completion in a designated log table, and the acquisition process judges whether the data are in order through the data log provided by the upstream. And if the data is not complete, generating data which is not provided with a log on time, and reminding an upstream system through the log. And the program enters a cyclic waiting process, and the data is automatically acquired after the data is in order. And auditing the data quality, if the data has quality problem, generating alarm information, and displaying on an interface, wherein the design flow is shown in figure 2.
The data acquisition module only needs an upstream system to be matched in a small amount, and the specific development content is as follows:
upstream system orchestration content: after the data is provided, a log message is generated.
Developing contents by a data acquisition module:
(1) and accessing the message log table of the upstream database to check whether the data are aligned.
(2) And after the data are collected, collecting the data to a load prediction system.
(3) And auditing the data, and generating alarm information to inform the upstream if the data is abnormal.
(4) The system has the functions of judging whether the data are in order and waiting.
The scheme has the advantages that: the upstream system is much less modified. The data transfer links are few, and the speed is high.
Step two: data pre-processing
Null values and abnormal data may exist in the acquired data, and in order to ensure the prediction accuracy, a set of proper cleaning rules needs to be established, so that an accurate data source is provided for load analysis and prediction. In addition, the source data are distributed in different tables, and the data need to be integrated to generate a uniform data model which comprises a load value and other data characteristics (such as weather, maintenance and the like) and is used for model establishment and load prediction; fig. 3 shows a design diagram of a data preprocessing module, wherein data anomaly detection refers to a situation that statistical data is compared with a previous time ring and is compared with yesterday at the same time, and a threshold value needs to be established. And when the data exceeds the threshold value during detection, the data is considered to be abnormal. When data is abnormal, the data needs to be cleaned, and methods such as mean filling, filling with previous data (FFILL), filling with next data (BFILL) and the like can be adopted. The data integration means that load data, climate data, holiday conditions and a work table account are integrated into a unified data model for load forecasting modeling.
Step three: load characteristic analysis
And analyzing the preprocessed unified data model, fully utilizing the advantages of a large data platform, and analyzing the load characteristics based on multiple dimensions such as climate, solar terms, holidays, maintenance plans and the like. Whether the data characteristics are relevant to the prediction target or not is checked, effective data characteristics are reserved in the modeling data, irrelevant data characteristics are eliminated, and the accuracy of the model can be improved. According to the data characteristics, the selection of the prediction algorithm is determined, such as: if the data conforms to the linear relationship, a linear regression algorithm can be considered; if the data is periodic, a time series algorithm may be considered. Generating various load analysis reports, such as: the year-round trend of load, the correlation condition of new energy power generation and load, the trend of industrial power utilization, the trend of resident power utilization and the like; as shown in fig. 4, the preprocessed unified data model is analyzed, the advantages of the large data platform are fully utilized, and the load characteristics are analyzed based on multiple dimensions such as climate, solar terms, holidays, maintenance plans and the like. Whether each characteristic dimension of the data is relevant to the load value of the next day or not is checked, relevant data dimensions are reserved in the unified data model, the data dimensions without relevance are eliminated, and the accuracy of the model can be improved. Meanwhile, various load analysis reports are generated through data statistics, such as: the year-round trend of load, the correlation condition of new energy power generation and load, the trend of industrial power utilization, the trend of residential power utilization and the like.
A. Load periodicity analysis
Analyzing the annual load condition (as shown in fig. 5), and researching the annual load cycle condition, growth condition and the like to obtain the conclusion that: the electricity consumption peaks in 7 and 8 months every year, the electricity consumption valleys in 1 and 2 months every year, and the electricity consumption shows an increasing trend every year.
The daily load periodicity is analyzed (as in fig. 6) to draw conclusions such as: load peaks were 18 o 'clock each day and load valleys were 4 o' clock each morning.
B. Load dependency analysis
As shown in fig. 7, the correlation between the current collection point load data and the historical collection point load data is analyzed to draw conclusions such as: the current load exhibits a first order linear correlation with the last acquisition point.
As shown in fig. 8, the temperature dependence of the load is analyzed to draw conclusions such as: the temperature and the load show a second order linear dependence.
As shown in fig. 9, the correlation between wind speed and load is analyzed to draw conclusions such as: wind speed has no significant dependence on load.
As shown in fig. 10, the correlation between humidity and load is analyzed to draw conclusions such as: humidity has no significant dependence on load.
C. Holiday load analysis
The holiday and the working day load condition are compared, as shown in fig. 11, the holiday type (long holiday, small long holiday, weekend) is subdivided, the holiday boundary condition is analyzed, and when the spring festival returns to the countryside, the returning army may start to return to the countryside before the holiday, so that the power utilization load is influenced.
Step four: adaptive load prediction
After the load characteristic analysis, a model wide table is generated, various machine learning algorithms such as linear regression, time series and neural network are adopted for modeling respectively, the effect of each model is verified, the measurement goodness of fit R2 and the variance are adopted for quantification, the effect of the model is accurately judged, and visual display is carried out by adopting modes such as a chart and a curve graph. Based on the self-adaptive optimal model evaluation technology, automatically selecting a model with the highest score for prediction according to R2 and the variance comprehensive score; as shown in fig. 12, the load data and the climate data are cleaned, integrated with the data of the work station account and the data of the holiday and the festival, stored in the big data cluster HDFS system, and processed by using HIVE and SPARK.
And selecting a proper prediction algorithm from the diversified prediction algorithm library according to the data characteristics, and verifying the effect of each prediction model respectively. And quantizing the prediction accuracy by adopting an R square and a variance, and realizing self-adaptive selection of an optimal model for prediction. Examples of the predicted effect are shown in fig. 13, in which (a) a linear regression model effect graph, (b) a Lasso regression model effect graph, and (c) a time series model effect graph.
Step five: diversified result display
And displaying the model prediction results according to levels, and supporting search and query of regions and buses. When a zone is selected, regional load data of the zone is displayed; when a bus is selected, the bus load data is presented. Predictive data may be shown in tabular and graphical form and support data derivation. The curve graph can simultaneously display historical predicted data, historical real data and current predicted data, and visually see the load change trend; the load prediction data and the load analysis report can be displayed in various modes such as a list, a chart, a large screen and the like, and export is supported. The load forecast data is transmitted to the downstream system, which can be used as data display, data base for planning mode, etc., and the module design is shown in fig. 14.
The load characteristic analysis report supports customized large-screen display, a plurality of analysis reports can be displayed simultaneously, and each report can be displayed in various forms such as a histogram, a curve graph, a pie chart and the like.
Step six: visual post-prediction assessment
In the operation process, the prediction accuracy is continuously monitored, the prediction effect is quantified by adopting the measurement goodness of fit R2 and the variance, and the visualization of the estimation after prediction is realized by using a chart for display. And setting a threshold value of the prediction effect, automatically alarming when the threshold value is exceeded, and displaying an alarm result on an interface. Once the model effect is reduced, the model effect can be found in time, and modeling optimization can be carried out again. Closed loop feedback: once the prediction accuracy is reduced, the prediction model can be found in time and adaptively optimized. The module design flow is shown in fig. 15.
As shown in fig. 16, by comparing the quantization indexes of the history model and the current model: such as variance, R2, and evaluating whether the model effect is reduced or not, and continuously monitoring and timely optimizing the model.
FIG. 17 is a flowchart illustrating the model training process of the present invention.
FIG. 18 is a flow chart of model prediction according to the present invention.
FIG. 19 is a flow chart of the post-model evaluation of the present invention.
By combining a big data analysis technology, massive historical data accumulated by a power grid company can be fully utilized, industrial relations, social activities, economic laws and the like can be explored and mined in a larger data range, factors which influence future load demands and are complex and difficult to model by a traditional method are obtained, influence relevance of various relevant factors on load increase and decrease is established, electric power and electric quantity prediction based on credible data is realized, and accuracy and reliability of electric power and electric quantity demand prediction are further improved.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be able to cover the technical solutions and the inventive concepts of the present invention within the technical scope of the present invention.
Claims (7)
1. A method of load prediction, comprising the steps of:
the method comprises the following steps: data acquisition
The data acquisition module is actively connected with an upstream system to acquire data, and stores the result on a large data platform HDFS for subsequent data processing;
step two: data pre-processing
Null values and abnormal data may exist in the acquired data, and in order to ensure the prediction accuracy, a set of proper cleaning rules needs to be established to provide an accurate data source for load analysis and prediction;
step three: load characteristic analysis
The preprocessed unified data model is analyzed, the advantages of a big data platform are fully utilized, the load characteristics are analyzed based on multiple dimensions of climate, solar terms, holidays and maintenance plans, whether the data characteristics are correlated with a prediction target or not is checked, effective data characteristics are reserved in the modeling data, irrelevant data characteristics are eliminated, the accuracy of the model can be improved, and the selection of a prediction algorithm is determined according to the data characteristics;
step four: adaptive load prediction
After the load characteristic is analyzed, a model wide table is generated, various machine learning algorithms such as linear regression, time sequence, neural network and the like are adopted for modeling, the effect of each model is verified, the measurement goodness of fit R2 and variance are adopted for quantification, the effect of the model is accurately judged, and visual display is carried out by adopting modes such as a chart, a curve graph and the like;
step five: diversified result display
Displaying the model prediction result number according to a hierarchy, supporting search query of regions and buses, and displaying regional load data of the regions when the regions are selected; when a bus is selected, displaying the bus load data, displaying the prediction data in a list and graph mode, and supporting data export;
step six: visual post-prediction assessment
In the operation process, the prediction accuracy is continuously monitored, the prediction effect is quantified by adopting the measurement goodness of fit R2 and the variance, and the visualization of the estimation after prediction is realized by using a chart for display.
2. The method for load forecasting according to claim 1, wherein the data collection module in step one has a plurality of functions of data vacancy warning and data quality auditing, and when there is a quality problem in the data, the data collection module generates warning information and displays the warning information on the interface.
3. The method for load forecasting according to claim 1, wherein in step two, the source data are distributed in different tables, and data are required to be integrated to generate a unified data model, which includes load values and other data characteristics, for model building and load forecasting.
4. The method of claim 1, wherein in step three, if the data satisfy the linear relationship, a linear regression algorithm is considered; if the data has periodicity, various load analysis reports can be generated by considering a time series algorithm.
5. The method of claim 1, wherein in step four, based on the adaptive optimal model evaluation technique, the model with the highest score is automatically selected for prediction according to R2 and the variance comprehensive score.
6. The method for load prediction according to claim 1, wherein in the fifth step, the graph can simultaneously display historical prediction data, historical real data and current prediction data, and visually see the load change trend.
7. The method for load forecasting according to claim 1, wherein a threshold value for forecasting the effect is set in the sixth step, when the threshold value is exceeded, an alarm is automatically given, the alarm result is displayed on an interface, and once the model effect is reduced, the model effect can be found timely, and modeling optimization can be carried out again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010944952.4A CN112835949A (en) | 2020-09-10 | 2020-09-10 | Load prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010944952.4A CN112835949A (en) | 2020-09-10 | 2020-09-10 | Load prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112835949A true CN112835949A (en) | 2021-05-25 |
Family
ID=75923269
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010944952.4A Pending CN112835949A (en) | 2020-09-10 | 2020-09-10 | Load prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112835949A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113792828A (en) * | 2021-11-18 | 2021-12-14 | 成都数联云算科技有限公司 | Power grid load prediction method, system, equipment and medium based on deep learning |
| CN114936683A (en) * | 2022-05-11 | 2022-08-23 | 国网宁夏电力有限公司银川供电公司 | Power grid bus load analysis and prediction assessment management method, device and system |
-
2020
- 2020-09-10 CN CN202010944952.4A patent/CN112835949A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113792828A (en) * | 2021-11-18 | 2021-12-14 | 成都数联云算科技有限公司 | Power grid load prediction method, system, equipment and medium based on deep learning |
| CN114936683A (en) * | 2022-05-11 | 2022-08-23 | 国网宁夏电力有限公司银川供电公司 | Power grid bus load analysis and prediction assessment management method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7409303B2 (en) | Identifying energy drivers in an energy management system | |
| CN116646933A (en) | Big data-based power load scheduling method and system | |
| CN110807550B (en) | Distribution transformer overload recognition and early warning method based on neural network and terminal equipment | |
| CN114676883A (en) | Power grid operation management method, device and equipment based on big data and storage medium | |
| Serrano-Guerrero et al. | Statistical methodology to assess changes in the electrical consumption profile of buildings | |
| CN112835949A (en) | Load prediction method | |
| CN118094262B (en) | Load abnormal value detection method based on spatial clustering | |
| CN114254806A (en) | Power distribution network heavy overload early warning method and device, computer equipment and storage medium | |
| CN116976707A (en) | User electricity consumption data anomaly analysis method and system based on electricity consumption data acquisition | |
| CN115063058A (en) | Comprehensive energy situation perception system based on model driving and data driving | |
| CN117955245B (en) | Method and device for determining running state of power grid, storage medium and electronic equipment | |
| CN117853275B (en) | Method and device for electricity utilization prediction | |
| CN108346009A (en) | A kind of power generation configuration method and device based on user model self study | |
| CN113642933A (en) | Power distribution station low-voltage diagnosis method and device | |
| CN111144628A (en) | Distributed energy supply type cooling, heating and power load prediction model system and method | |
| Deng et al. | Medium-term rolling load forecasting based on seasonal decomposition and long short-term memory neural network | |
| CN115689801A (en) | Electricity consumption information acquisition system based on carbon measurement | |
| Yanhui et al. | Forecasting hourly electricity demand with nonparametric functional data analysis | |
| Hu et al. | Big data management and application research in power load forecasting and power transmission and transformation equipment evaluation | |
| Li et al. | Research on Key Technology of State Evaluation and Replacement Acquisition System of Intelligent Electric Energy Meter Based on Smart Grid | |
| Rama et al. | A predictive analysis of residential electrical load demand in Mauritius | |
| CN118487276B (en) | Power grid safety dynamic management and control method and system for power guarantee object | |
| Zhao et al. | Flight Incidents Prediction Based on Model of X-12 and ARIMA | |
| Wang et al. | Overload analysis of distribution transformers based on relevance analyse and machine learning | |
| Zhao et al. | Regional photovoltaic installed capacity forecasting based on granger causality test and grey support vector machine |
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 |