CN114912355A - Method and device for predicting short-term icing of power transmission line and storage medium - Google Patents
Method and device for predicting short-term icing of power transmission line and storage medium Download PDFInfo
- Publication number
- CN114912355A CN114912355A CN202210503214.5A CN202210503214A CN114912355A CN 114912355 A CN114912355 A CN 114912355A CN 202210503214 A CN202210503214 A CN 202210503214A CN 114912355 A CN114912355 A CN 114912355A
- Authority
- CN
- China
- Prior art keywords
- line
- prediction
- icing
- tension value
- sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 80
- 230000005540 biological transmission Effects 0.000 title claims abstract description 29
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 239000011248 coating agent Substances 0.000 claims abstract description 47
- 238000000576 coating method Methods 0.000 claims abstract description 47
- 238000012549 training Methods 0.000 claims description 98
- 238000012795 verification Methods 0.000 claims description 36
- 238000005070 sampling Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 12
- 230000001186 cumulative effect Effects 0.000 claims description 9
- 230000001364 causal effect Effects 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 17
- 230000007246 mechanism Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 239000012212 insulator Substances 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 239000004020 conductor Substances 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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"
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Water Supply & Treatment (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Primary Health Care (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Public Health (AREA)
- General Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)
Abstract
The application provides a method and a device for predicting short-term icing of a power transmission line and a storage medium, and belongs to the technical field of data processing. The method comprises the steps of obtaining a tension value data sequence of the line to be detected in observation time and various types of meteorological data sequences in the environment where the line to be detected is located in the observation time; inputting the tension value data sequence and various meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, output by the icing prediction model aiming at a line to be tested, within prediction duration; the ice coating prediction model is trained according to a tension value data sequence sample of the sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length; and predicting the ice coating amount of the line to be tested in the predicted time length according to the tensile value prediction sequence in the predicted time length. The method aims to accurately predict the ice coating amount of the power transmission line in a short period.
Description
Technical Field
The embodiment of the application relates to the technical field of data processing, in particular to a method and a device for predicting short-term icing of a power transmission line and a storage medium.
Background
The electric power system is the most important basic engineering facility in modern cities, is the life line of national economy, and the national economic production construction, the orderly and steady development of the society and the basic life of people are highly dependent on the reliable and stable operation of the electric power system, so that the attention to the safety of the electric power system is a necessary thing. However, in practice, there are many disasters threatening the safety of the power system, wherein the ice coating of the transmission line is one of the most common disasters affecting the power system, and the damage to the power grid caused by the ice coating of the transmission line includes: the icing quality is too high, so that the transmission line is overloaded to cause mechanical damages such as wire breakage, insulator damage or turnover, pole tower collapse and the like; the increase of the circuit sag causes electrical accidents such as flashover, wire blowout and the like, and causes great harm to the safe and stable operation of the power system.
An effective icing prediction mechanism is established as a main measure for ice resistance and ice prevention, various emergency plans can be preset to deal with the coming of ice disasters by mastering the future icing condition of a line in advance, the deicing means and the deicing time are planned, and the influence of the icing of the power transmission line on the electricity consumption of the industry and residents is reduced to the maximum extent. The existing icing prediction methods are mainly divided into two types: an icing prediction method based on a mechanism model and an icing prediction method based on a data driving model.
The method comprises the following steps that an icing forecasting mechanism model aims to establish a mathematical analysis model from a real icing physical process so as to forecast the line icing quality; however, because the use of the mechanism model needs to rely on some fixed assumptions and experiences, and parameter settings of most models are too ideal, data such as droplet diameter, liquid water content in air and the like are extremely difficult to acquire and measure in a real environment, the icing predicted value and the true value of the model are greatly different in actual use, so that the mechanism model is difficult to put into practical use.
The model based on data driving is the main choice for solving the existing icing prediction problem, and the purpose of the model is to use a specific statistical method to find the law from meteorological factors influencing icing and to fit and predict the targets such as icing thickness or quality. At present, most of common methods used by people, such as Support Vector Regression (SVR) or multilayer perceptron (MLP), utilize meteorological point data of a current observation time point to predict icing, that is, it is assumed that the future icing condition is only related to the meteorological state of the current time, but the method can predict the icing quantity in a future ultra-short period, such as 15 minutes to 1 hour, but the icing quantity in a future longer time is often required to be predicted in actual use, and when the method is used for predicting the short-term icing in a future 6 hours, errors can rise rapidly, so that the existing transmission line short-term icing prediction method is difficult to popularize and apply in an actual industrial scene.
Therefore, how to predict the ice coating amount of the power transmission line in a short term is an urgent problem to be solved.
Disclosure of Invention
The embodiment of the application provides a method and a device for predicting short-term ice coating of a power transmission line and a storage medium, aiming at accurately predicting the ice coating of the power transmission line in a short term.
In a first aspect, an embodiment of the present application provides a method for predicting short-term icing of a power transmission line based on a meteorological time series, where the method includes:
acquiring a tension value data sequence of a line to be detected in observation time and various types of meteorological data sequences in the environment where the line to be detected is located in the observation time;
inputting the tension value data sequence and the various types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, which is output by the icing prediction model aiming at the line to be tested and is within a prediction duration; the icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length;
and predicting the ice coating amount of the line to be tested in the predicted time length according to the tension value prediction sequence in the predicted time length.
Optionally, the icing prediction model is obtained by training a plurality of different samples obtained by the line to be tested in a plurality of historical ice date data by taking the line to be tested as the sample line;
each sample of the line to be tested comprises a tension value data sequence sample of the line to be tested in the observation time length, a plurality of types of meteorological data sequence samples in the observation time length and a tension value sequence of the line to be tested in the prediction time length.
Optionally, obtaining a tension value data sequence of the line to be measured in the observation time length and multiple types of meteorological data sequences in the environment where the line to be measured is located in the observation time length includes:
acquiring a tension value data sequence of the line to be detected within the observation time length through a tension sensor arranged at the terminal of the line to be detected;
acquiring various types of meteorological data sequences in the environment of the line to be detected within the observation time length through sensors which are respectively corresponding to various types of meteorological data in the environment of the line to be detected;
and the tension sensor and the sensors corresponding to the various types of meteorological data synchronously acquire data according to a preset sampling interval.
Optionally, the multiple types of meteorological data sequences in the environment where the line to be tested is located include: a temperature data sequence, a humidity data sequence, a wind speed data sequence, a horizontal wind direction data sequence, and a vertical wind direction data sequence.
Optionally, the observed duration and the predicted duration are both durations in hours.
Optionally, the icing prediction model is trained by the following steps:
obtaining a training sample data set, wherein the training sample data set comprises a plurality of training samples, and each training sample in the plurality of training samples comprises: a tension value data sequence sample of a sample line in the observation duration, a plurality of types of meteorological data sequence samples in the observation duration and a label formed by the tension value sequence in the prediction duration;
constructing a preset model consisting of stacked TCN residual modules connected by residual errors, wherein the TCN residual modules are used for expanding a causal convolution network core, and the residual errors are connected in a manner of tensile value data in the observation time length;
inputting the training sample data set into the preset model for training, wherein the prediction model obtains a prediction result of a tension value sequence of each training sample in the training sample data set within the prediction duration by learning the relation between the cumulative effect of the weather within the observation duration and the tension value;
and calculating a loss value corresponding to each training sample according to the prediction result and the label thereof corresponding to each training sample, and updating the model parameters of the preset model based on the loss of the preset model to the training sample data set.
Optionally, the training process of the icing prediction model further includes:
iteratively updating the calibration times of the preset model by using the training sample data set;
verifying the preset model after each updating by using a verification sample data set for the preset model after each updating;
if the average absolute error of the updated preset model on the verification sample data set is smaller than the average absolute error of the updated preset model on the verification sample data set, saving the updated preset model as a candidate model;
and if the average absolute error of the preset model on the verification sample data set obtained by each updating is larger than the average absolute error corresponding to the candidate model in the iteration updating of the subsequent preset times, stopping the iteration updating, and taking the candidate model as the trained icing prediction model.
Optionally, the method further comprises:
acquiring original data, wherein the original data comprises tension value data of different lines in the same region in a plurality of ice periods and a plurality of types of meteorological data, or the tension value data of the same line in the plurality of ice periods and the plurality of types of meteorological data;
preprocessing the acquired original data;
determining a plurality of icing events from the preprocessed original data, and dividing the icing events into the icing events corresponding to a training sample data set and the icing events corresponding to a verification sample data set;
selecting a plurality of training samples from the icing events corresponding to the training sample data set through a set sliding window, and selecting a plurality of verification samples from the icing events corresponding to the verification sample data set, wherein the length of the sliding window is the sum of the observation duration and the prediction duration.
In a second aspect, an embodiment of the present application provides a device for predicting short-term icing of a power transmission line based on a meteorological time series, where the device includes:
the acquisition module is used for acquiring a tension value data sequence of a line to be detected in observation time and various types of meteorological data sequences in the environment where the line to be detected is located in the observation time;
the tension value prediction module is used for inputting the tension value data sequence and the various types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, which is output by the icing prediction model aiming at the line to be tested and is within a prediction duration; the icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length;
and the ice coating amount prediction module is used for predicting the ice coating amount of the line to be measured in the prediction time length according to the tension value prediction sequence in the prediction time length.
In a third aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for predicting short-term icing on a power transmission line based on a meteorological time series according to the first aspect of the embodiment.
Has the advantages that:
acquiring a tension value data sequence of a line to be detected in observation time and various types of meteorological data sequences in the environment where the line to be detected is located in the observation time; inputting the tension value data sequence and the various types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, which is output by the icing prediction model aiming at the line to be tested and is within a prediction duration; the icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length; and predicting the ice coating amount of the line to be tested in the predicted time length according to the tension value prediction sequence in the predicted time length.
When the method is used for predicting the ice coating amount in the prediction time length, firstly, the tension value of the line to be tested in the prediction time length is predicted, a more accurate prediction sequence of the tension value in the prediction time length can be obtained by adopting an ice coating prediction model, and then the ice coating amount of the line to be tested in the prediction time length can be determined according to the prediction sequence of the tension value in the prediction time length with better accuracy, so that the ice coating amount of the power transmission line in a short time length can be accurately predicted.
In the training process of the ice coating prediction model, based on various types of meteorological data sequence samples and tension value data sequence samples within a period of time, namely within an observation duration, the deep level relation between the cumulative effect of the weather and the tension of the line can be excavated, the ice coating amount of the line in a longer time in the future can be predicted, and the obtained prediction result is more accurate.
Compared with an icing prediction mechanism model, meteorological data and tension data in the method are easier to collect, and compared with the method that ice covering prediction is carried out only according to the meteorological data observed currently by Support Vector Regression (SVR) or multilayer perceptron (MLP), the method can only carry out prediction in a super short-term range in the future, the method can carry out prediction on the ice covering amount in a longer time in the future by digging the influence of the cumulative effect of the meteorological data in an observation time period on the line tension, and the prediction result is more accurate; the method is based on the data which is easier to obtain, and can more accurately predict the ice coating amount in a longer time in the future, so the method can be widely applied to actual industrial scenes.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the description of the embodiments of the present application will be briefly described below, and it is obvious that the drawings in the description below are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings may be obtained according to these drawings without inventive labor.
Fig. 1 is a flowchart illustrating steps of a method for predicting short-term icing of a power transmission line based on a meteorological time series according to an embodiment of the present application;
FIG. 2 is a flowchart illustrating the steps for training an icing prediction model according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a preset model according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of a TCN residual block according to an embodiment of the present application;
FIG. 5 is a schematic diagram of a test result of an icing prediction model provided in an embodiment of the present application in a real scene;
fig. 6 is a functional block diagram of a short-term icing prediction device for a power transmission line based on a meteorological time series according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. 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 application.
In the existing icing prediction method, data required by an icing prediction mechanism model is extremely difficult to acquire and measure in a real environment, so that in actual use, the obtained icing prediction value has a large error with a real value, and the method is difficult to be widely applied to actual industry; on the other hand, since ice coating is predicted based on only meteorological point data at the current observation time point based on a data-driven model such as Support Vector Regression (SVR) or multi-layer perceptron (MLP), and the influence of accumulation of meteorological data on ice coating prediction is ignored, the ice coating amount can be predicted only for a very short period of time, and if the ice coating amount in a longer time in the future is predicted, an error of a prediction result rapidly increases. Therefore, in order to accurately predict the ice coating amount of the power transmission line in a short term, the application provides a power transmission line short-term ice coating prediction method based on a meteorological time series.
Referring to fig. 1, a flowchart illustrating steps of a method for predicting short-term icing of a power transmission line based on a meteorological time series in an embodiment of the present invention is shown, where the method specifically includes the following steps:
s101: the method comprises the steps of obtaining a tension value data sequence of a line to be detected in observation time and multiple types of meteorological data sequences in the environment where the line to be detected is located in the observation time.
When the ice coating of the power transmission line in a short time is predicted, the method uses the line tension as an important physical quantity for measuring the ice coating of the line, firstly predicts the line tension, and then further determines the ice coating of the line according to the predicted line tension value to obtain the ice coating prediction result of the line to be measured in the predicted time.
Specifically, the observation duration refers to a time period for collecting various data of the line to be measured, the observation duration in the method is a time length in hours, for example, the observation duration may be 24 hours, and in other embodiments, the observation duration may be selected according to actual requirements.
In a feasible implementation manner, when a tension value data sequence of the line to be measured within the observation time is obtained, a tension sensor can be arranged at the tower terminal of the line to be measured.
The various types of meteorological data sequences in the environment of the line under test may include: the system comprises a temperature data sequence, a humidity data sequence, a wind speed data sequence, a horizontal wind direction data sequence and a vertical wind direction data sequence, and can be used for carrying out data acquisition by arranging sensors corresponding to various types of meteorological data in the environment of the line to be detected or at the pole tower terminal of the line to be detected, such as a temperature sensor, a humidity sensor, a wind speed sensor and a wind direction sensor.
When multiple data of a line to be measured in the observation time are acquired, the tension sensor and the sensors corresponding to the multiple types of meteorological data synchronously acquire the data according to a preset sampling interval, so that a tension value data sequence is aligned with the multiple meteorological data sequences in time; the sampling interval may be, for example, 10 minutes, or other sampling intervals may be used as desired.
After the obtained tension value data sequence and various meteorological data sequences are obtained, data cleaning can be carried out on the data sequences, and the specific adopted cleaning method can be determined according to actual conditions.
S102: and inputting the tension value data sequence and the various types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, which is output by the icing prediction model aiming at the line to be tested and is within the prediction duration, of the icing prediction model.
The icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length.
In order to further improve the accuracy of an icing prediction result of a line to be tested in a prediction time length, the icing prediction model is obtained by taking the line to be tested as the sample line and training the line to be tested on the basis of a plurality of different samples obtained by the line to be tested in a plurality of historical ice date data; each sample of the line to be tested comprises a tension value data sequence sample of the line to be tested in the observation time length, a plurality of types of meteorological data sequence samples in the observation time length and a tension value sequence of the line to be tested in the prediction time length.
By learning the relation between the meteorological data and the tension value of the line to be measured in the historical ice period, the ice coating prediction model can be more targeted for the ice coating prediction of the line to be measured, and the obtained prediction result is more accurate.
In other embodiments, different routes in the same region with the same climate can be used as a plurality of sample routes, and the ice coating prediction model can be trained through meteorological data and tension value data of the different routes in a plurality of ice periods.
The predicted time length refers to a period of time in the future from the observation time length cut-off time, and is also a time period taking hours as a unit, the predicted time length in the embodiment is at least 6 hours, and the pulling force values of the line to be measured at different times in the future 6 hours can be predicted.
The icing prediction model is trained according to a tension value data sequence sample in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence in the prediction time length, which are measured by a sample route, so that the icing prediction model can mine the influence of the cumulative effect of the meteorological data at different moments in the observation time length on the line tension, and the prediction result of the icing prediction model can be more accurate.
In the prediction sequence of the tension values output by the icing prediction model, the time interval corresponding to each tension value is kept consistent with the sampling interval, for example, the sampling interval in the observation time length is 10 minutes, and then the time interval corresponding to two adjacent tension values in the prediction time length is also 10 minutes.
S103: and predicting the ice coating amount of the line to be tested in the predicted time length according to the tension value prediction sequence in the predicted time length.
After the tension value prediction sequence in the prediction duration is obtained, the ice coating amount of the line to be tested in the prediction duration can be predicted based on the tension value prediction sequence.
In one possible embodiment, the predicted ice coating amount at a certain time in the predicted sequence of tensile values can be calculated by the predicted tensile value at the time, and the formula is as follows:
wherein F is the prediction result of the tension value of the line to be tested at a certain moment, eta is the wind deflection angle, theta is the inclination angle of the insulator chain of the line to be tested, and G t Is the sum of the quality of equipment such as a lead, an insulator, hardware and the like of a line to be tested, S' a ，S’ b Respectively represents the length from the lowest point of the conductor to the main tower, and n is the number of conductor splits.
When the ice coating amount of the line to be measured in the prediction time length is predicted, the method firstly predicts the tension value of the line to be measured in the prediction time length, converts the predicted ice coating amount into the tension value of the line to be measured for prediction, and can easily and directly obtain the tension data and various meteorological data in the prediction process, thereby being successfully applied to the actual industry.
In addition, in the training process of the icing prediction model, based on a plurality of data of various types of meteorological data in the observation time, the model can further dig out the deep relation between the cumulative effect of the meteorological data and the line tension, so that the tension value prediction sequence in the prediction time obtained through the icing prediction model is more accurate, and the accurate icing amount prediction result of the line to be measured in the prediction time can also be obtained based on the more accurate tension value prediction sequence.
Referring to fig. 2, a flowchart illustrating steps of training an icing prediction model provided in an embodiment of the present application is shown, and in a possible implementation, the icing prediction model is trained by the following steps:
s201: and constructing a training sample data set and a verification sample data set.
Specifically, the present step is divided into the following substeps:
a1: acquiring original data, wherein the original data comprises tension value data of different lines in a same area in a plurality of ice periods and a plurality of types of meteorological data, or the tension value data of the same line in the plurality of ice periods and the plurality of types of meteorological data.
Because the same line is applied to areas with different climates, the relation between the weather and the tension value is different, when the icing prediction model is trained, the required samples can be data from different lines in the same area with the same climate, and the icing prediction model obtained by training can be applied to different lines in the area.
When the icing prediction model is trained, the required samples can also be data of the same line in a plurality of ice periods, when the icing prediction model obtained through training is applied to the line, the obtained icing prediction result has the best accuracy, and the icing prediction model can also be applied to other lines in the area of the line.
In a plurality of ice periods, the sensor is used for collecting tension data and meteorological data of different transmission line pole tower terminals according to preset sampling intervals to obtain enough original data, wherein the meteorological data comprise: temperature data, humidity data, wind speed data, horizontal wind direction data, vertical wind direction data.
A2: and preprocessing the acquired original data.
In this embodiment, the preprocessing process includes: removing abnormal values, filling missing values and scaling features; the abnormal values can be removed through manual observation, the missing values can be filled by using the average value of two values before and after the missing values, and the original data can be subjected to feature scaling according to the following formula:
in the formula (I), the compound is shown in the specification,
σ (X) is the variance of the raw data, which is the mean of the raw data.
In other embodiments, other methods may be used for preprocessing the raw data, and the application is not limited in any way.
A3: determining a plurality of icing events from the preprocessed original data, and dividing the icing events into the icing events corresponding to a training sample data set and the icing events corresponding to a verification sample data set.
Each icing event refers to data corresponding to the sample line in an icing stage; each icing event comprises tension value data, temperature data, humidity data, wind speed data, horizontal wind direction data and vertical wind direction data.
In this embodiment, 20 icing events are obtained, 70% of the icing events are used to make a training sample data set, 10% of the icing events are used to make a verification sample data set, and the remaining 20% of the icing events are used to make a test sample data set, so that the training sample data set, the verification sample data set, and the samples in the test sample data set are different, and the test sample data set can be used to test or evaluate a trained model.
A4: selecting a plurality of training samples from the icing events corresponding to the training sample data set through a set sliding window, and selecting a plurality of verification samples from the icing events corresponding to the verification sample data set, wherein the length of the sliding window is the sum of the observation duration and the prediction duration.
Assuming that the total length of an icing event is S, the observation time length is T, and the prediction time length is L, the length of the sliding window is T + L, and the number of target samples that can be obtained by the icing event is:
s＝max(S-(T+L)+1,0)。
and training based on the obtained training sample data set and the verification sample data set.
S202: a training sample data set is obtained, wherein the training sample data set comprises a plurality of training samples.
The training sample data set includes a plurality of training samples, and according to the generation process of the training sample data set in step S201, data included in each training sample is: a tension value data sequence sample, a temperature data sequence sample, a humidity data sequence sample, a wind speed data sequence sample, a horizontal wind direction data sequence sample and a vertical wind direction data sequence sample of a sample line in the observation duration, and a label formed by the tension value sequence in the prediction duration.
S203: and constructing a preset model consisting of stacked TCN residual modules connected by the residual errors.
The preset model is a plurality of stacked TCN Residual modules connected by Residual, and is based on a time series convolutional network (TCN) and Residual connection, so the preset model can be denoted as Res-TCN.
Referring to fig. 3, which shows a schematic structural diagram of the preset model provided in the embodiment of the present application, where m is the number of convolution kernels, and corresponds to the number of channels formed after each convolution operation, and since the model uses line tension value data in the splicing observation time period as a residual connection mode, the actual number of output channels is m + 1.
And extracting the output of the last time step at the end of the preset model, regarding the output as an abstract feature extracted by the stacked TCN residual error module in the observation window, and then transforming the output into a predicted value conforming to the output format by using a linear transformation network.
In this embodiment, when sampling is performed on various data within an observation time period of 24 hours at a sampling interval of 10 minutes, 6 data sequences including the observation time period are input, and the input length of each data sequence is 144; the predicted time length is 6 hours, the interval of the output tension data is consistent with the sampling interval, and therefore the output length of the preset model is 36.
Referring to fig. 4, a schematic structural diagram of the TCN residual error module provided in this embodiment is shown, where the TCN residual error module includes two identical sub-modules stacked and connected to further deepen a network of the preset model, so as to improve an abstraction capability of the preset model; each submodule includes a dilated causal convolution network, a weight normalization layer, a linear rectification layer, and an anti-overfitting layer.
The simple causal convolutional network can only receive information which is in linear relation with the depth of the network, so that the problem that training is difficult to perform due to excessive layers of the causal convolutional network when a long-sequence task is processed occurs.
The Weight Normalization layer can be used to accelerate the convergence speed of the network, and the overfitting can be prevented by the linear rectification layer (ReLU) and the overfitting prevention layer Dropout.
Two sub-modules connected in a stacked manner obtain the result of complex conversion
And since the input and output of the dilated causal convolutional network may have different widths, an additional 1 × 1 convolution is required to ensure that the tensor shape of the input is equal to the output, and then the residual connection is handled using an element-by-element addition operation.
It should be noted that, in this embodiment, the single pulling force value data is regarded as an independent feature, and unlike the general residual calculation, the pulling force value data is directly input to the output of each TCN residual module for splicing, and further, the output of each TCN residual module can be regarded as:
in the formula, i is the ith TCN residual error module in the stacked TCN residual error modules; u represents the concatenation of elements.
Through stacking the TCN residual modules which are connected, deep abstract relations between meteorological data and tension data can be deeply mined, meanwhile, tension value data in observation time length are independently spliced into the TCN residual modules at all depths in a residual connection mode, and by adopting the jumping connection mode, each TCN residual module has the most original tension value data as input, so that the preset model can learn a simple mapping relation between a known tension value and a predicted tension value, and can use the tension value obtained by observation as prior information to make reasonable prediction on the tension value of a line in the prediction time length. The priori information is necessary for ice coating prediction, and due to the fact that an actual scene is complex, an ice coating prediction model may frequently encounter strange input, and known tension value data is used as the priori information, so that the method is also an induction bias.
S204: inputting the training sample data set into the preset model for training, wherein the prediction model obtains a prediction result of a tension value sequence of each training sample in the training sample data set within the prediction duration by learning the relation between the cumulative effect of the weather within the observation duration and the tension value.
In one possible implementation, where the weights of all networks are initialized by a Gaussian distribution of N (0, 0.01), the batch size during training may be set to 32, using an Adam optimizer with a learning rate of 0.0001.
Training a preset model on a training sample data set, wherein TCN residual modules stacked in the preset model can learn the relation between the cumulative effect of weather and the tension in the observation duration, and the preset model outputs the prediction result of the tension value sequence of each training sample in the prediction duration.
S205: and calculating a loss value corresponding to each training sample according to the prediction result and the label thereof corresponding to each training sample, and updating the model parameters of the preset model based on the loss of the preset model to the training sample data set.
In this embodiment, the loss function type adopted by the preset model is Mean Square Error (MSE), the loss of the preset model on the training sample data set can be obtained through the loss value corresponding to each training sample, and the model parameters of the preset model are updated according to the loss.
In a feasible implementation manner, a preset model may be iteratively updated on the training sample data set for the calibration times, and after the prediction model is trained based on the training sample data set and model parameters are updated each time, the preset model after each update is verified by using a verification sample data set.
And if the average absolute error of the updated preset model on the verification sample data set is smaller than the average absolute error of the updated preset model on the verification sample data set, saving the updated preset model as a candidate model.
And if the average absolute error of the preset model on the verification sample data set obtained by each updating is larger than the average absolute error corresponding to the candidate model in the iteration updating of the subsequent preset times, stopping the iteration updating, and taking the candidate model as the trained icing prediction model.
For example, the preset model may iterate 20 times on a training sample data set, the preset model with the minimum average absolute error during verification is stored as a candidate model, if the preset model with the smaller average absolute error during verification is obtained in subsequent training, the preset model is used as the candidate model, if the average absolute error corresponding to the preset model obtained through 5 iterations is greater than the average absolute error corresponding to the candidate model, the preset model is considered to have reached the upper learning limit, the iteration updating is stopped, and the candidate model is used as the trained icing prediction model.
The application has at least the following beneficial effects:
1. the direct prediction of the ice coating amount is converted into prediction of the tension value of the line to be tested, and the tension data and various meteorological data which need to be obtained in the prediction process can be easily and directly obtained, so that the method can be successfully applied to actual industry;
2. the icing prediction model is built by adopting a convolution network as a backbone, and has parallel computing capability, higher running speed and higher training efficiency compared with other time sequence prediction models with RNN structures.
3. A more accurate tension value prediction sequence is obtained through the icing prediction model, and the icing amount of the line to be measured in the prediction duration obtained based on the more accurate tension value prediction sequence is more accurate;
4. the icing prediction model adopts a residual error to connect with stacked TCN residual error modules, so that the preset model is forced to reasonably predict the tension value of the line in the prediction duration by using the observed tension value as prior information, and more real icing scenes can be processed.
In one possible implementation, the validity and superiority of the method is verified based on the line number XY1059 provided by the south china power grid.
And acquiring original data of the line with the number XY1059 in the ice season from 2013 to 2020 at sampling intervals of 10 minutes, wherein the original data comprise line tension, temperature, humidity, wind speed, horizontal wind direction and vertical wind direction of the line sample. According to the step S201, a training sample data set, a verification sample data set and a test sample data set are constructed according to original data, a preset model is trained on the basis of the training sample data set to obtain an icing prediction model, and the icing prediction model is evaluated by using the test sample data set.
Referring to fig. 5, which is a schematic diagram of a test result of the ice coating prediction model provided in this embodiment of the present application in a real scene, a vertical coordinate of fig. 5 is a tensile force, an abscissa is a time sequence, a dotted line in the diagram represents an observed tensile force value within an observed duration, a dot represents a real tensile force value within a predicted duration, and a cross represents a predicted tensile force value within a predicted duration output by the ice coating prediction model, fig. 5 shows a tensile force value data sequence within an observed duration of 24 hours, a sampling interval of the observed duration is 10 minutes, and shows a predicted result within a predicted duration of 6 hours, and for convenience of observation and comparison, down-sampling is performed on both the predicted sequence of the tensile force value within the predicted duration and the real tensile force value sequence with a step size of 3.
Based on a test sample data set, comparing an icing prediction result obtained by an icing prediction model Res-TCN in the method with the existing icing prediction method, wherein the icing prediction error corresponding to each prediction method is shown in the table 1.
TABLE 1 error (MAE) of icing prediction results on test sample data set for each prediction method
As can be seen from Table 1, the Res-TCN in the method is superior to the existing icing prediction model support vector regression SVR, Multilayer Perceptron (MLP), Random Forest (RF) and Adaboost.
When the prediction duration is 6 hours, the best performing of the existing icing prediction method is SVR, and the MAE of Res-TCN in the method is 35.1% lower than that of Res-TCN.
TABLE 2 icing prediction error (MAE) of various time series prediction models on a test sample data set
In table 2, vanilla-TCN represents a common time series convolutional network TCN, GRU represents RNN type time series prediction model, S2SA represents Seq2Seq (Seq2Seq with Attention) time series prediction model with Attention mechanism, LSTNet and Informer are also compared as the best results of many public time series prediction data sets. The models are also subjected to hyper-parametric tuning during training, and the sizes of the models are controlled to be equivalent to Res-TCN in order to reasonably compare the convergence rates of the models.
As can be seen from Table 2, the error of the icing prediction result of Res-TCN is smaller than that of other time sequence prediction models, and when the prediction time length is 6 hours, the MAE of Res-TCN is reduced by 13.8% compared with the best-performing S2SA in the other models.
Further, by comparing the results in tables 1 and 2, it can be seen that the icing prediction result determined by the method based on the icing prediction model is more accurate than the icing prediction result obtained by other methods in the prior art.
TABLE 3 error of icing prediction result of other line terminals by the icing prediction model
As can be seen from Table 3, when the icing prediction model in the method is applied to other line terminals, Res-TCN has good adaptability, can be applied to different line terminals, and has obvious practical value.
Referring to fig. 6, a functional block diagram of a power transmission line short-term icing prediction device based on a meteorological time series in an embodiment of the present invention is shown, where the device includes:
the acquisition module 100 is configured to acquire a tension value data sequence of a line to be detected within an observation duration, and multiple types of meteorological data sequences in an environment where the line to be detected is located within the observation duration;
the tension value prediction module 200 is configured to input the tension value data sequence and the multiple types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence within a prediction duration output by the icing prediction model for the line to be tested; the icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length;
and the ice coating prediction module 300 is configured to predict the ice coating of the line to be tested in the predicted time length according to the tensile value prediction sequence in the predicted time length.
Optionally, the obtaining module includes:
the first acquisition unit is used for acquiring a tension value data sequence of the line to be detected within the observation time length through a tension sensor arranged at a terminal of the line to be detected;
the second acquisition unit is used for acquiring various types of meteorological data sequences in the environment of the line to be detected within the observation duration through sensors which are respectively corresponding to various types of meteorological data in the environment of the line to be detected;
and the tension sensor and the sensors corresponding to the various types of meteorological data synchronously acquire data according to a preset sampling interval.
Optionally, the apparatus further comprises a model training module, the model training module comprising:
a sample obtaining unit, configured to obtain a training sample data set, where the training sample data set includes a plurality of training samples, and each training sample in the plurality of training samples includes: a tension value data sequence sample of a sample line in the observation duration, a plurality of types of meteorological data sequence samples in the observation duration and a label formed by the tension value sequence in the prediction duration;
the model construction unit is used for constructing a preset model consisting of stacked TCN residual modules connected by residual errors, wherein the TCN residual modules are used for expanding a causal convolution network core, and the residual errors are connected in a mode of tensile value data in the observation time length;
the sample input unit is used for inputting the training sample data set into the preset model for training, and the prediction model obtains a prediction result of a tension value sequence of each training sample in the training sample data set in the prediction time length by learning the relation between the cumulative effect of weather in the observation time length and the tension value;
and the parameter updating unit is used for calculating a loss value corresponding to each training sample according to the prediction result and the label thereof corresponding to each training sample, and updating the model parameters of the preset model based on the loss of the preset model to the training sample data set.
Optionally, the model training module further comprises:
a specified times unit, configured to update the calibration times of the preset model by using the training sample data set in an iterative manner;
the verification unit is used for verifying the preset model after each update by utilizing a verification sample data set;
the storage unit is used for storing the updated preset model as a candidate model when the average absolute error of the updated preset model on the verification sample data set is smaller than the average absolute error of the updated preset model on the verification sample data set last time;
and the judging unit is used for obtaining the average absolute error of the preset model on the verification sample data set in each updating in the iteration updating of the subsequent preset times, wherein the average absolute error is larger than the average absolute error corresponding to the candidate model, stopping the iteration updating and taking the candidate model as the trained icing prediction model.
Optionally, the apparatus further comprises:
the system comprises an original data acquisition module, a data acquisition module and a data acquisition module, wherein the original data acquisition module is used for acquiring original data, and the original data comprises tension value data of different lines in a plurality of ice periods and various types of meteorological data in the same region, or the tension value data of the same line in the plurality of ice periods and various types of meteorological data;
the preprocessing module is used for preprocessing the acquired original data;
the first determining module is used for determining a plurality of icing events from the preprocessed original data, and dividing the icing events into the icing events corresponding to a training sample data set and the icing events corresponding to a verification sample data set;
and the sample generation module is used for selecting a plurality of training samples from the icing events corresponding to the training sample data set through a set sliding window, and selecting a plurality of verification samples from the icing events corresponding to the verification sample data set, wherein the length of the sliding window is the sum of the observation duration and the prediction duration.
The embodiment of the application also provides a computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and when the computer program is executed by a processor, the method for predicting the short-term icing of the power transmission line based on the meteorological time series is realized.
The embodiments in the present specification are all described in a progressive manner, and each embodiment focuses on differences from other embodiments, and portions that are the same and similar between the embodiments may be referred to each other.
As will be appreciated by one of skill in the art, embodiments of the present application may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the preferred embodiment and all such alterations and modifications as fall within the true scope of the embodiments of the application.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "include", "including" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or terminal device including a series of elements includes not only those elements but also other elements not explicitly listed or inherent to such process, method, article, or terminal device. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.
The principle and the embodiment of the present application are explained by applying specific examples, and the above description of the embodiments is only used to help understand the method and the core idea of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.
Claims (10)
1. A power transmission line short-term icing prediction method based on a meteorological time series is characterized by comprising the following steps:
acquiring a tension value data sequence of a line to be detected in observation time and various types of meteorological data sequences in the environment where the line to be detected is located in the observation time;
inputting the tension value data sequence and the various types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, which is output by the icing prediction model aiming at the line to be tested and is within a prediction duration; the icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length;
and predicting the ice coating amount of the line to be tested in the predicted time length according to the tension value prediction sequence in the predicted time length.
2. The method according to claim 1, wherein the icing prediction model is trained based on a plurality of different samples of the line to be tested obtained in a plurality of historical ice date data, with the line to be tested being the sample line;
each sample of the line to be tested comprises a tension value data sequence sample of the line to be tested in the observation time length, a plurality of types of meteorological data sequence samples in the observation time length and a tension value sequence of the line to be tested in the prediction time length.
3. The method of claim 1, wherein obtaining a pulling force value data sequence of a line to be tested in an observation time period and a plurality of types of meteorological data sequences of the environment in which the line to be tested is located in the observation time period comprises:
acquiring a tension value data sequence of the line to be detected within the observation time length through a tension sensor arranged at the terminal of the line to be detected;
acquiring various types of meteorological data sequences in the environment of the line to be detected within the observation time length through sensors which are respectively corresponding to various types of meteorological data in the environment of the line to be detected;
and the tension sensor and the sensors corresponding to the various types of meteorological data synchronously acquire data according to a preset sampling interval.
4. The method of claim 3, wherein the plurality of types of meteorological data sequences in the environment of the line under test comprise: a temperature data sequence, a humidity data sequence, a wind speed data sequence, a horizontal wind direction data sequence, and a vertical wind direction data sequence.
5. The method according to any one of claims 1 to 4, wherein the observed duration and the predicted duration are both durations in hours.
6. The method of claim 1, wherein the icing prediction model is trained by:
obtaining a training sample data set, wherein the training sample data set comprises a plurality of training samples, and each training sample in the plurality of training samples comprises: a tension value data sequence sample of a sample line in the observation duration, a plurality of types of meteorological data sequence samples in the observation duration and a label formed by the tension value sequence in the prediction duration;
constructing a preset model consisting of stacked TCN residual modules connected by residual errors, wherein the TCN residual modules are used for expanding a causal convolutional network core, and the residual errors are connected in a manner of tensile value data in the observation time length;
inputting the training sample data set into the preset model for training, wherein the prediction model obtains a prediction result of a tension value sequence of each training sample in the training sample data set within the prediction duration by learning the relation between the cumulative effect of the weather within the observation duration and the tension value;
and calculating a loss value corresponding to each training sample according to the prediction result and the label thereof corresponding to each training sample, and updating the model parameters of the preset model based on the loss of the preset model to the training sample data set.
7. The method of claim 6, wherein the training process of the icing prediction model further comprises:
iteratively updating the calibration times of the preset model by using the training sample data set;
verifying the preset model after each update by using a verification sample data set for the preset model after each update;
if the average absolute error of the updated preset model on the verification sample data set is smaller than the average absolute error of the updated preset model on the verification sample data set, saving the updated preset model as a candidate model;
and if the average absolute error of the preset model on the verification sample data set obtained by each updating is larger than the average absolute error corresponding to the candidate model in the subsequent iteration updating of the preset times, stopping the iteration updating, and taking the candidate model as the trained icing prediction model.
8. The method of claim 7, further comprising:
acquiring original data, wherein the original data comprises tension value data of different lines in the same region in a plurality of ice periods and a plurality of types of meteorological data, or the tension value data of the same line in the plurality of ice periods and the plurality of types of meteorological data;
preprocessing the acquired original data;
determining a plurality of icing events from the preprocessed original data, and dividing the icing events into the icing events corresponding to a training sample data set and the icing events corresponding to a verification sample data set;
selecting a plurality of training samples from the icing events corresponding to the training sample data set through a set sliding window, and selecting a plurality of verification samples from the icing events corresponding to the verification sample data set, wherein the length of the sliding window is the sum of the observation duration and the prediction duration.
9. A device for predicting short-term icing of a power transmission line based on a meteorological time series, the device comprising:
the acquisition module is used for acquiring a tension value data sequence of a line to be detected in observation time and various types of meteorological data sequences in the environment where the line to be detected is located in the observation time;
the tension value prediction module is used for inputting the tension value data sequence and the various types of meteorological data sequences into an icing prediction model to obtain a tension value prediction sequence, which is output by the icing prediction model aiming at the line to be tested and is within a prediction duration; the icing prediction model is trained according to a tension value data sequence sample of a sample line in the observation time length, various types of meteorological data sequence samples in the observation time length and a tension value sequence of the sample line in the prediction time length;
and the ice coating prediction module is used for predicting the ice coating of the line to be tested in the prediction time length according to the tension value prediction sequence in the prediction time length.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements a method for short-term icing prediction for a power transmission line based on a meteorological time series according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210503214.5A CN114912355A (en) | 2022-05-10 | 2022-05-10 | Method and device for predicting short-term icing of power transmission line and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210503214.5A CN114912355A (en) | 2022-05-10 | 2022-05-10 | Method and device for predicting short-term icing of power transmission line and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114912355A true CN114912355A (en) | 2022-08-16 |
Family
ID=82766025
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210503214.5A Pending CN114912355A (en) | 2022-05-10 | 2022-05-10 | Method and device for predicting short-term icing of power transmission line and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114912355A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116451594A (en) * | 2023-06-15 | 2023-07-18 | 北京东润环能科技股份有限公司 | Training method and device of icing prediction model, prediction method and device and electronic equipment |
| CN117494026A (en) * | 2023-12-28 | 2024-02-02 | 国网浙江省电力有限公司金华供电公司 | Method, system and storage medium for positioning icing fault of power transmission line under cold and tidal weather |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140067271A1 (en) * | 2012-08-30 | 2014-03-06 | International Business Machines Corporation | Predicting ice coating status on transmission lines |
| CN112183897A (en) * | 2020-11-02 | 2021-01-05 | 成都卡普数据服务有限责任公司 | Long-time prediction method for icing thickness of overhead transmission line based on deep learning |
| CN113642235A (en) * | 2021-08-09 | 2021-11-12 | 贵州电网有限责任公司 | Icing prediction method based on local space-time feature fusion |
| CN113723669A (en) * | 2021-08-09 | 2021-11-30 | 贵州电网有限责任公司 | Power transmission line icing prediction method based on Informmer model |
-
2022
- 2022-05-10 CN CN202210503214.5A patent/CN114912355A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140067271A1 (en) * | 2012-08-30 | 2014-03-06 | International Business Machines Corporation | Predicting ice coating status on transmission lines |
| CN112183897A (en) * | 2020-11-02 | 2021-01-05 | 成都卡普数据服务有限责任公司 | Long-time prediction method for icing thickness of overhead transmission line based on deep learning |
| CN113642235A (en) * | 2021-08-09 | 2021-11-12 | 贵州电网有限责任公司 | Icing prediction method based on local space-time feature fusion |
| CN113723669A (en) * | 2021-08-09 | 2021-11-30 | 贵州电网有限责任公司 | Power transmission line icing prediction method based on Informmer model |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116451594A (en) * | 2023-06-15 | 2023-07-18 | 北京东润环能科技股份有限公司 | Training method and device of icing prediction model, prediction method and device and electronic equipment |
| CN116451594B (en) * | 2023-06-15 | 2023-08-18 | 北京东润环能科技股份有限公司 | Training method and device of icing prediction model, prediction method and device and electronic equipment |
| CN117494026A (en) * | 2023-12-28 | 2024-02-02 | 国网浙江省电力有限公司金华供电公司 | Method, system and storage medium for positioning icing fault of power transmission line under cold and tidal weather |
| CN117494026B (en) * | 2023-12-28 | 2024-04-05 | 国网浙江省电力有限公司金华供电公司 | Method, system and storage medium for positioning icing fault of power transmission line under cold and tidal weather |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109447315B (en) | Multi-time space scale-based electric power meteorological numerical weather forecast method and device | |
| CN109958588B (en) | Icing prediction method, icing prediction device, storage medium, model generation method and model generation device | |
| CN112285807B (en) | Meteorological information prediction method and device | |
| CN114912355A (en) | Method and device for predicting short-term icing of power transmission line and storage medium | |
| CN104123682B (en) | A kind of Distribution Network Failure methods of risk assessment based on meteorological effect factor | |
| CN102938021B (en) | A kind of powerline ice-covering load quantitative is estimated and Forecasting Methodology | |
| US20160203245A1 (en) | Method for simulating wind field of extreme arid region based on wrf | |
| CN104732296A (en) | Modeling method for distributed photovoltaic output power short-term prediction model | |
| CN111428942B (en) | Line icing thickness prediction method for extracting micro-terrain factors based on variable grid technology | |
| CN104200288A (en) | Equipment fault prediction method based on factor-event correlation recognition | |
| CN117117819A (en) | Photovoltaic power generation short-term power prediction method, system, equipment and medium | |
| CN115796059A (en) | Electrical equipment service life prediction method and system based on deep learning | |
| CN116205342A (en) | Electric power weather early warning method based on refined prediction | |
| CN116050599A (en) | Line icing fault prediction method, system, storage medium and equipment | |
| CN105117538A (en) | Method for warning waving of power transmission channels | |
| CN114594532A (en) | Method and device for predicting cold weather, electronic equipment and computer readable medium | |
| CN116021981A (en) | Method, device, equipment and storage medium for predicting ice coating faults of power distribution network line | |
| CN111222662A (en) | Power grid typhoon flood disaster early warning method and device | |
| CN106570582A (en) | Method and system for building transmission line dancing tripping risk prediction network model | |
| Cardenas-Barrera et al. | A neural network approach to multi-step-ahead, short-term wind speed forecasting | |
| CN114548601A (en) | Power distribution network power failure prediction method and system under extreme disasters based on BP neural network | |
| CN104217247A (en) | Method and device for predicting output power of wind turbines in wind farm | |
| CN118395257A (en) | Extra-high voltage transmission line abnormal line loss analysis method for multidimensional microclimate | |
| CN105654189B (en) | Icing short-term prediction method based on time series analysis and Kalman filtering algorithm | |
| CN106682776A (en) | Fine forecasting and early warning method and system for dancing of overhead transmission line |
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 |