CN112232985B - Power distribution and utilization data monitoring method and device for ubiquitous power Internet of things - Google Patents
Power distribution and utilization data monitoring method and device for ubiquitous power Internet of things Download PDFInfo
- Publication number
- CN112232985B CN112232985B CN202011103619.7A CN202011103619A CN112232985B CN 112232985 B CN112232985 B CN 112232985B CN 202011103619 A CN202011103619 A CN 202011103619A CN 112232985 B CN112232985 B CN 112232985B
- Authority
- CN
- China
- Prior art keywords
- model
- optimized
- training
- lightgbm
- xgboost
- 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
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000009826 distribution Methods 0.000 title claims abstract description 25
- 238000012544 monitoring process Methods 0.000 title claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 65
- 238000012360 testing method Methods 0.000 claims abstract description 42
- 238000012795 verification Methods 0.000 claims abstract description 20
- 238000002790 cross-validation Methods 0.000 claims abstract description 18
- 238000010276 construction Methods 0.000 claims abstract description 9
- 230000005611 electricity Effects 0.000 claims description 33
- 230000008569 process Effects 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 3
- 238000012806 monitoring device Methods 0.000 claims description 2
- 230000006855 networking Effects 0.000 claims description 2
- 238000002474 experimental method Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
Images
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
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Software Systems (AREA)
- Marketing (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Public Health (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a power distribution and utilization data monitoring method, which comprises the following steps: performing time sequence characteristic construction on the training sample and the test sample, and dividing the training sample into a training set and a verification set; obtaining an optimized LightGBM model and an optimized XGboost model; respectively inputting the test samples into the two optimized models for 5-fold cross validation to obtain new characteristics n 1 And n 2 (ii) a N is to be 1 And n 2 Respectively merging the training sets to the last column of the training set to obtain a new training set; repeating the step of obtaining the optimized LightGBM model and the step of obtaining the optimized XGboost model; test sample and feature n 1 Test sample and feature n 2 Respectively inputting the current data into new optimized LightGBM and XGBoost models, performing 5-fold cross validation, and fusing the obtained results to obtain a prediction result; according to the method, the power consumption at the time to be predicted is predicted through the machine model, and support is provided for optimal scheduling of power distribution and utilization.
Description
Technical Field
The invention belongs to the technical field of communication, and particularly relates to a power distribution and utilization data monitoring method and device for a ubiquitous power Internet of things.
Background
The ubiquitous power internet of things is increasingly concerned by technicians in the world, is a connection network between people, people and objects and between objects in the power field, and extends and expands information exchange user sides. As an important application of the ubiquitous power Internet of things, a monitoring system occupies an increasingly important position in power production practice, and the problems of the traditional power monitoring system are generally concentrated on the problems of small number of connectable monitoring points, single type of monitoring data, lack of a data processing function, no timely warning function and the like.
With the arrival of the world of everything interconnection, the data transmission quantity of the ubiquitous power internet of things will be increased in a blowout mode in the future, and the requirements of large-scale, multi-dimensional and intelligentization of monitoring points of the ubiquitous power internet of things monitoring system are more and more urgent. The ubiquitous power Internet of things has the characteristics of a large number of connecting nodes, wide distribution of monitoring points and a large number of acquired data types in order to realize ubiquitous Internet of things. How to realize the warning function of the monitoring data of the mass sensor is a key problem for solving the ubiquitous power internet of things deployment in the future.
Therefore, based on the problems, the power distribution and utilization data monitoring method and device for the ubiquitous power internet of things are provided, the power consumption at the moment to be predicted is predicted through the machine model, warning information is sent according to the analysis result, and support is provided for power distribution and utilization optimization scheduling.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a power distribution and utilization data monitoring method and device for a ubiquitous power internet of things, which are used for predicting power consumption at a time to be predicted through a machine model, sending warning information according to an analysis result and providing support for optimal scheduling of power distribution and utilization.
The technical problem to be solved by the invention is realized by adopting the following technical scheme:
a power distribution and utilization data monitoring method for a ubiquitous power Internet of things comprises the following steps:
respectively acquiring historical power consumption data before the power consumption moment to be predicted and relevant data of the power consumption moment to be predicted, and performing data screening to respectively obtain a training sample and a test sample;
respectively carrying out time sequence characteristic construction on the training sample and the test sample, and dividing the training sample into a training set and a verification set;
obtaining an optimized LightGBM model: using the model LightGBM to predict a verification set, and searching an optimal parameter combination through a grid search method to obtain an optimized LightGBM model;
obtaining an optimized XGboost model: predicting a verification set by using the model XGboost, and searching an optimal parameter combination by using a grid search method to obtain an optimized XGboost _1 model;
respectively inputting the test samples into two optimal models LightGBM _1 and XGboost _1 to perform 5-fold cross validation to respectively obtain two new features n 1 And n 2 ；
Two new features n 1 And n 2 Respectively merging the training sets to the last column of the training set to respectively obtain new training sets; repeating the step of obtaining the optimized LightGBM model and the step of obtaining the optimized XGboost model to obtain a new optimized LightGBM model and an XGboost model;
test sample and feature n 1 Test sample and feature n 2 Inputting the new optimized LightGBM model and XGboost model respectively, and performing 5-fold cross validation again to obtain two types of output results respectively; fusing the two types of output results according to a set proportion to obtain a prediction result of the power consumption at the moment to be predicted;
and comparing the prediction result with the early warning threshold value, and determining whether warning information is sent out or not according to the comparison result.
Further, the method for constructing the timing characteristics of the training samples comprises the following steps:
converting the time of historical power consumption data of a training sample into a timestamp, wherein the timestamp is a type of characteristic;
acquiring historical power consumption at a plurality of moments before each historical power consumption moment as a second type of characteristic;
and performing difference on the obtained second type of characteristics: and respectively carrying out subtraction on the first second-class characteristic and the rest second-class characteristics of each historical power consumption in the training sample to obtain a third-class characteristic.
Further, the method for performing time sequence feature construction on the test sample comprises the following steps:
converting the time of the moment to be predicted into a time stamp, wherein the time stamp is a type of characteristic;
acquiring historical electricity consumption at a plurality of moments before the moment to be predicted as a second type of characteristic;
and performing difference on the obtained second type of characteristics: and respectively subtracting the first second type characteristic of the electricity consumption at the moment to be predicted from the rest second type characteristics in the test sample to obtain a third type characteristic.
Further, when the optimized LightGBM model or the optimized LightGBM model is obtained by a grid search method, all set parameter combinations are input, the variable state is controlled in the search process, only one parameter is modified each time, and finally, the parameter combination which enables the mean square error to be minimum is output.
Further, when the prediction result is larger than the maximum value of the normal power distribution range, sending out the alarm information of the impending overload power utilization; and when the prediction result is smaller than the minimum value of the normal power distribution range, sending out the alarm information of the impending large-scale power failure.
A power distribution and utilization data monitoring devices for ubiquitous electric power thing networking includes:
the training sample and test sample acquisition module is used for respectively acquiring historical power consumption data before the power consumption time to be predicted and relevant data of the power consumption time to be predicted, and performing data screening to respectively obtain a training sample and a test sample;
the time sequence characteristic construction module is used for respectively carrying out time sequence characteristic construction on the training sample and the test sample and dividing the training sample into a training set and a verification set;
the optimized LightGBM model obtaining module is used for using the model LightGBM prediction verification set and searching the optimal parameter combination through a grid search method to obtain an optimized LightGBM model;
the optimized XGboost model acquisition module is used for predicting a verification set by using a model XGboost and searching an optimal parameter combination by a grid search method to obtain an optimized XGboost _1 model;
a new feature obtaining module, configured to input the test samples into two optimal models LightGBM _1 and XGBoost _1 respectively to perform 5-fold cross validation, so as to obtain two new features n 1 And n 2 ；
A LightGBM model and XGboost model secondary optimization module for optimizing two new characteristics n 1 And n 2 Respectively merging the training sets into the last column of the training set to respectively obtain new training sets; repeating the step of obtaining the optimized LightGBM model and the step of obtaining the optimized XGboost model to obtain a new optimized LightGBM model and an XGboost model;
a power consumption prediction module for predicting the power consumption of the test sample and the characteristic n 1 Test sample and feature n 2 Inputting the data into a new optimized LightGBM model and an XGboost model respectively, and performing 5-fold cross validation again to obtain two types of output results respectively; fusing the two types of output results according to a set proportion to obtain a prediction result of the power consumption at the moment to be predicted;
and the prediction result comparison module is used for comparing the prediction result with the early warning threshold value and determining whether to send out warning information or not according to the comparison result.
The invention has the advantages and positive effects that:
the method respectively applies machine learning models XGBost and LightGBM to predict the power consumption of users, carries out alarm prediction according to a normal power distribution range, sends out alarm information according to an analysis result, and provides support for optimal scheduling of power distribution and utilization; by comparing and analyzing the prediction results under different data set training, the XGBost and LightGBM models are superior to the traditional GBDT model, and the prediction model and the analysis results have certain guiding significance.
Drawings
The technical solution of the present invention will be described in further detail with reference to the accompanying drawings and examples, but it should be understood that these drawings are designed for illustrative purposes only and thus are not intended to limit the scope of the present invention. Furthermore, unless otherwise indicated, the drawings are intended to be illustrative of the structural configurations described herein and are not necessarily drawn to scale.
Fig. 1 is a flowchart of a power distribution and utilization data monitoring method for a ubiquitous power internet of things, provided in an embodiment of the present invention;
fig. 2 is a graph comparing the electricity consumption of a cell and the actual electricity consumption by using the electricity distribution and utilization data monitoring method provided in the embodiment of the present invention;
Detailed Description
First, it should be noted that the specific structures, features, advantages, etc. of the present invention will be specifically described below by way of example, but all the descriptions are for illustrative purposes only and should not be construed as limiting the invention in any way. Furthermore, any single feature described or implicit in the embodiments described herein or shown or implicit in the drawings may continue to be combined or subtracted from any single feature or equivalent thereof to obtain still further embodiments of the invention that may not be directly mentioned herein.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
This embodiment specifically describes the present invention with reference to fig. 1 and 2:
in this embodiment, the provided scenarios are: now 9 months, 24 days, afternoon, 15:30, need to predict 16 pm on 24 days of 9 months: 00, the specific steps are as follows:
acquiring historical power consumption data from a fusion terminal, wherein the historical power consumption data of 30 minutes from 10 am at 1 month and 1 day to 9 months and 24 days at 15 o' clock at 1 month and 1 day before the power consumption moment to be predicted is acquired, and reasonably preprocessing abnormal missing data and a plurality of data sets (the specific measures comprise adopting an average value to carry out simulated replacement on the missing data, adopting a variance to carry out effectiveness analysis on a plurality of compared data sets and the like), and taking the abnormal missing data and the plurality of data sets as training samples (TrainData);
in addition, related data of the power consumption moment to be predicted are obtained, data screening is carried out, reasonable preprocessing is carried out on abnormal missing data and a plurality of data sets (the specific measures comprise that the missing data is subjected to analog replacement by adopting an average value, effectiveness analysis is carried out on a plurality of data sets through comparison by adopting variance and the like), so that a test sample is obtained, and the initial format of the test sample (TestData) can be defined according to the actual situation;
respectively carrying out characteristic construction on TrainData and TestData, wherein 'historical electricity consumption' in the table 1 represents the current electricity consumption at the historical moment; "timestamp" represents timestamp data converted from "time"; "historical electricity consumption _1" is the electricity consumption half an hour before the current moment; the historical electricity consumption quantity _2 is the electricity consumption quantity at the previous hour, and the historical electricity consumption quantity _10 can be sequentially obtained as the electricity consumption quantity at the previous five hours; the "historical electricity consumption _ difference 1" is a difference value obtained by subtracting the "historical electricity consumption _2" from the "historical electricity consumption _1", and so on, and the "historical electricity consumption _ difference 9" is a difference value obtained by subtracting the "historical electricity consumption _10" from the "historical electricity consumption _ 1".
The characteristic constructed TrainData is shown in Table 1:
TABLE 1
The characteristic-constructed TestData is shown in table 2 (note: 16:00 in the table is future time of power consumption to be predicted, 15:30 is current time, and here, historical power consumption _1 in the table is power consumption of the current time 15):
TABLE 2
Next, we divide the training sample TrainData into a training set Train and a verification set Val; considering that the amount of data used in training a model is large, in order to allow the model to perform sufficient learning and prevent the model from being over-fitted, we follow 2: trainData is divided by a ratio of 1. The method comprises the following specific steps:
the training set Train is the data of the first 2/3 of the training sample TrainData (TrainData line 1 to line 8520), and the verification set Val is the data of the last 1/3 of the training sample TrainData (TrainData line 8521 to line 12780).
A grid search method is adopted, all set parameter combinations are input, the variable state is controlled in the search process, only one parameter is modified each time, and finally the parameter combination which enables the error function (mean square error rmse) to be minimum is output. And finally, after traversing all parameter combinations, saving the parameter combination with the minimum error as the optimal model LightGBM _1 or XGboost _1.
Inputting test data into an XGboost _1 model and a LightGBM _1 model for 5-fold cross validation respectively to obtain an output result characteristic n 1 And feature n 2 The five-fold cross validation process is as follows:
the k-fold is to divide the data set into k groups, extract one of k data as a verification set from the training set every time, and take the rest data as a test set. The test results were averaged over the k sets of data. If the training set is large, k is small, the training cost is reduced, and if the training set is small, k is large, and the training data are increased. If k =10, 90% of the data is trained. Through multiple times of analysis and verification, the best analysis effect and the highest efficiency on the electricity utilization data are found in the verification of 5 folds. On the basis, the test set and the verification set are mixed into an intersection process, the verification set and the test set mutually form a complementary set, and the cycle is alternated. We use 5-fold cross-validation for model tuning, i.e. if all data is used for training, overfitting will result, and 5-fold cross-validation can mitigate overfitting.
The 5-fold cross validation is to divide the data set into 5 parts according to a certain proportion (equal proportion or non-equal proportion), take one part of the data set as test data, and take the other 4 parts of the data set as training data. And then repeating the experiment, wherein the 5-fold cross validation completes one complete experiment only after 5 times of the experiment, namely the cross validation actually repeats the experiment 5 times, each experiment selects one different data part from the 5 parts as test data, the remaining 4 data parts are used as training data, and finally, the obtained 5 experiment results are divided equally.
Will be characteristic n 1 Merging the training data into the last column of TrainData to obtain TrainData _1, using the TrainData _1 as a new training set, and repeating the model optimizing step to obtain an optimal model LightGBM _2; will be characteristic n 2 Merging the training data into the last column of TrainData to obtain TrainData _2, using the TrainData _2 as a new training set, repeating the model optimizing step to obtain an optimal model XGboost _2;
test sample TestData and characteristic n 1 Input into LightGBM _2 model, test sample TestData and feature n 2 And inputting the XGboost _2 model, respectively performing 5-fold cross validation to obtain an output result _1 and an output result _2, and finally weighting the output result _1 and the output result _2 to obtain the output predicted power consumption.
For example, in this embodiment, the predicted cell power consumption and the actual power consumption change trend are substantially the same by using the method of this embodiment, and the accuracy is high. And when the power consumption is higher than the overload power consumption threshold value of the cell, the automatic reporting system provides support for the optimal scheduling of power distribution and utilization, so that the operation reliability of the power grid is improved.
The present invention has been described in detail with reference to the above examples, but the above description is only for the purpose of describing the preferred embodiments of the present invention, and should not be construed as limiting the scope of the present invention. All equivalent changes and modifications made within the scope of the present invention shall fall within the scope of the present invention.
Claims (4)
1. A power distribution and utilization data monitoring method for a ubiquitous power Internet of things is characterized by comprising the following steps:
respectively acquiring historical power consumption data before the power consumption moment to be predicted and relevant data of the power consumption moment to be predicted, and performing data screening to respectively obtain a training sample and a test sample;
respectively carrying out time sequence characteristic construction on the training sample and the test sample, and dividing the training sample into a training set and a verification set;
obtaining an optimized LightGBM model: using a model LightGBM prediction verification set, and searching an optimal parameter combination through a grid search method to obtain an optimized LightGBM model;
obtaining an optimized XGboost model: predicting a verification set by using the model XGboost, and searching an optimal parameter combination by using a grid search method to obtain an optimized XGboost _1 model;
respectively inputting the test samples into two optimal models LightGBM _1 and XGboost _1 to perform 5-fold cross validation to respectively obtain two new features n 1 And n 2 ；
Two new features n 1 And n 2 Respectively merging the training sets to the last column of the training set to respectively obtain new training sets; repeating the step of obtaining the optimized LightGBM model and the step of obtaining the optimized XGboost model to obtain a new optimized LightGBM model and an XGboost model;
test sample and feature n 1 Test sample and feature n 2 Inputting the new optimized LightGBM model and XGboost model respectively, and performing 5-fold cross validation again to obtain two types of output results respectively; fusing the two types of output results according to a set proportion to obtain a prediction result of the power consumption at the moment to be predicted;
comparing the prediction result with an early warning threshold value, and determining whether warning information is sent out or not according to the comparison result;
the method for constructing the time sequence characteristics of the training samples comprises the following steps:
converting the time of the historical electricity consumption data of the training sample into a timestamp, wherein the timestamp is a class of characteristics;
acquiring historical electricity consumption at a plurality of moments before each historical electricity consumption moment as a second type of characteristic;
and performing difference on the obtained second type of characteristics: respectively subtracting the first second-class characteristic of each historical power consumption in the training sample from the rest second-class characteristics to obtain a third-class characteristic;
the method for constructing the time sequence characteristics of the test sample comprises the following steps:
converting the time of the moment to be predicted into a timestamp which is a type of characteristic;
acquiring historical electricity consumption at a plurality of moments before the moment to be predicted as a second type of characteristic;
and performing difference on the obtained second type of characteristics: and respectively carrying out difference on the first second type characteristic and the rest second type characteristics of the electricity consumption at the moment to be predicted in the test sample to obtain a third type characteristic.
2. The power distribution and utilization data monitoring method for the ubiquitous power internet of things according to claim 1, wherein: when the optimized LightGBM model or the optimized LightGBM model is obtained through a grid search method, all set parameter combinations are input, the variable state is controlled in the search process, only one parameter is modified each time, and finally the parameter combination enabling the mean square error to be minimum is output.
3. The power distribution and utilization data monitoring method for the ubiquitous power internet of things according to claim 1, wherein: when the prediction result is larger than the maximum value of the normal power distribution range, sending out the information of about overload power utilization; and when the prediction result is smaller than the minimum value of the normal power distribution range, sending out the alarm information of the impending large-scale power failure.
4. A power distribution and utilization data monitoring devices for ubiquitous electric power thing networking, its characterized in that: the method comprises the following steps:
the training sample and test sample acquisition module is used for respectively acquiring historical electricity consumption data before the electricity consumption time to be predicted and relevant data of the electricity consumption time to be predicted, and performing data screening to respectively obtain a training sample and a test sample;
the time sequence characteristic construction module is used for respectively carrying out time sequence characteristic construction on the training sample and the test sample and dividing the training sample into a training set and a verification set;
the optimized LightGBM model obtaining module is used for using the model LightGBM prediction verification set and searching the optimal parameter combination through a grid search method to obtain an optimized LightGBM model;
the optimized XGboost model acquisition module is used for predicting a verification set by using a model XGboost and searching an optimal parameter combination by a grid search method to obtain an optimized XGboost _1 model;
a new feature obtaining module, configured to input the test samples into two optimal models LightGBM _1 and XGBoost _1 respectively to perform 5-fold cross validation, so as to obtain two new features n 1 And n 2 ；
A LightGBM model and XGboost model secondary optimization module for optimizing two new characteristics n 1 And n 2 Respectively merging the training sets to the last column of the training set to respectively obtain new training sets; repeating the step of obtaining the optimized LightGBM model and the step of obtaining the optimized XGboost model to obtain a new optimized LightGBM model and an XGboost model;
a power consumption prediction module for predicting the time to be predicted 1 Test sample and feature n 2 Respectively input into a new optimized LightGBM model,The XGboost model carries out 5-fold cross validation again to respectively obtain two types of output results; fusing the two types of output results according to a set proportion to obtain a prediction result of the power consumption at the moment to be predicted;
the prediction result comparison module is used for comparing the prediction result with the early warning threshold value and determining whether to send out warning information or not according to the comparison result;
the method for constructing the time sequence characteristics of the training samples comprises the following steps:
converting the time of the historical electricity consumption data of the training sample into a timestamp, wherein the timestamp is a class of characteristics;
acquiring historical electricity consumption at a plurality of moments before each historical electricity consumption moment as a second type of characteristic;
and performing difference on the obtained second type of characteristics: respectively subtracting the first second-class characteristic of each historical power consumption in the training sample from the rest second-class characteristics to obtain a third-class characteristic;
the method for constructing the time sequence characteristics of the test sample comprises the following steps:
converting the time of the moment to be predicted into a timestamp which is a type of characteristic;
acquiring historical electricity consumption at a plurality of moments before the moment to be predicted as a second type of characteristic;
and performing difference on the obtained second type of characteristics: and respectively carrying out difference on the first second type characteristic and the rest second type characteristics of the electricity consumption at the moment to be predicted in the test sample to obtain a third type characteristic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011103619.7A CN112232985B (en) | 2020-10-15 | 2020-10-15 | Power distribution and utilization data monitoring method and device for ubiquitous power Internet of things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011103619.7A CN112232985B (en) | 2020-10-15 | 2020-10-15 | Power distribution and utilization data monitoring method and device for ubiquitous power Internet of things |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112232985A CN112232985A (en) | 2021-01-15 |
| CN112232985B true CN112232985B (en) | 2023-02-28 |
Family
ID=74117302
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011103619.7A Active CN112232985B (en) | 2020-10-15 | 2020-10-15 | Power distribution and utilization data monitoring method and device for ubiquitous power Internet of things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112232985B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113159398B (en) * | 2021-04-01 | 2023-10-24 | 国网内蒙古东部电力有限公司 | Power consumption prediction method and device and electronic equipment |
| CN113590471B (en) * | 2021-07-05 | 2024-05-24 | 陕西银河时代清洁能源有限公司 | Communication terminal equipment simulation system and application method thereof |
| CN116384683A (en) * | 2023-04-04 | 2023-07-04 | 清华大学 | Model and data combined driven industrial load demand response characteristic describing method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909933A (en) * | 2017-01-18 | 2017-06-30 | 南京邮电大学 | A kind of stealing classification Forecasting Methodology of three stages various visual angles Fusion Features |
| CN108171379A (en) * | 2017-12-28 | 2018-06-15 | 无锡英臻科技有限公司 | A kind of electro-load forecast method |
| CN108564204A (en) * | 2018-03-23 | 2018-09-21 | 西安理工大学 | Least square method supporting vector machine power predicating method based on maximal correlation entropy criterion |
| CN109948668A (en) * | 2019-03-01 | 2019-06-28 | 成都新希望金融信息有限公司 | A kind of multi-model fusion method |
| CN110245801A (en) * | 2019-06-19 | 2019-09-17 | 中国电力科学研究院有限公司 | A kind of Methods of electric load forecasting and system based on combination mining model |
| CN110472778A (en) * | 2019-07-29 | 2019-11-19 | 上海电力大学 | A kind of short-term load forecasting method based on Blending integrated study |
| WO2020010717A1 (en) * | 2018-07-13 | 2020-01-16 | 南京理工大学 | Short-term traffic flow prediction method based on spatio-temporal correlation |
| CN111415027A (en) * | 2019-01-08 | 2020-07-14 | 顺丰科技有限公司 | Method and device for constructing component prediction model |
-
2020
- 2020-10-15 CN CN202011103619.7A patent/CN112232985B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909933A (en) * | 2017-01-18 | 2017-06-30 | 南京邮电大学 | A kind of stealing classification Forecasting Methodology of three stages various visual angles Fusion Features |
| CN108171379A (en) * | 2017-12-28 | 2018-06-15 | 无锡英臻科技有限公司 | A kind of electro-load forecast method |
| CN108564204A (en) * | 2018-03-23 | 2018-09-21 | 西安理工大学 | Least square method supporting vector machine power predicating method based on maximal correlation entropy criterion |
| WO2020010717A1 (en) * | 2018-07-13 | 2020-01-16 | 南京理工大学 | Short-term traffic flow prediction method based on spatio-temporal correlation |
| CN111415027A (en) * | 2019-01-08 | 2020-07-14 | 顺丰科技有限公司 | Method and device for constructing component prediction model |
| CN109948668A (en) * | 2019-03-01 | 2019-06-28 | 成都新希望金融信息有限公司 | A kind of multi-model fusion method |
| CN110245801A (en) * | 2019-06-19 | 2019-09-17 | 中国电力科学研究院有限公司 | A kind of Methods of electric load forecasting and system based on combination mining model |
| CN110472778A (en) * | 2019-07-29 | 2019-11-19 | 上海电力大学 | A kind of short-term load forecasting method based on Blending integrated study |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112232985A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112232985B (en) | Power distribution and utilization data monitoring method and device for ubiquitous power Internet of things | |
| Lin et al. | A general framework for quantitative modeling of dependability in cyber-physical systems: A proposal for doctoral research | |
| CN109409561B (en) | Construction method of multi-time scale time sequence collaborative prediction model | |
| WO2013189110A1 (en) | Power communication fault early warning analysis method and system | |
| CN108710990B (en) | Line transformer subscriber multilevel line loss analysis method and system based on synchronous data | |
| CN104933631A (en) | Power distribution network operation online analysis and evaluation system | |
| CN109800995A (en) | A kind of grid equipment fault recognition method and system | |
| CN106099945A (en) | A kind of big data modeling of GA for reactive power optimization and abnormal scheme detection method | |
| CN109783553A (en) | A kind of power distribution network mass data increased quality system | |
| CN105471647A (en) | Power communication network fault positioning method | |
| Cui et al. | Edge learning for surveillance video uploading sharing in public transport systems | |
| CN115842342A (en) | Topology identification method and device for distributed power distribution network | |
| CN113133038A (en) | Power Internet of things link backup method, device, equipment and storage medium | |
| CN115526265A (en) | Non-invasive load decomposition method based on progressive learning structure | |
| CN113935390A (en) | Data processing method, system, device and storage medium | |
| Soldan et al. | Short-term forecast of EV charging stations occupancy probability using big data streaming analysis | |
| CN115758151A (en) | Combined diagnosis model establishing method and photovoltaic module fault diagnosis method | |
| Sifat et al. | Design, development, and optimization of a conceptual framework of digital twin electric grid using systems engineering approach | |
| CN103258255A (en) | Knowledge discovery method applicable to power grid management system | |
| CN103336200A (en) | Device and method for predicting power distribution cabinet electric health index | |
| CN116596574A (en) | Power grid user portrait construction method and system | |
| CN104809584A (en) | Substation maintenance patrol method and system | |
| CN113965467B (en) | Power communication system reliability assessment method and system based on neural network | |
| CN117154723B (en) | Platform short-term load prediction method and system based on multi-source data and model fusion | |
| CN117764214B (en) | Multi-dimensional target dynamic early warning method and system for energy storage power station considering double-carbon targets |
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 |