CN109492756B - Multi-element wire galloping early warning method based on deep learning and related device - Google Patents
Multi-element wire galloping early warning method based on deep learning and related device Download PDFInfo
- Publication number
- CN109492756B CN109492756B CN201811375147.3A CN201811375147A CN109492756B CN 109492756 B CN109492756 B CN 109492756B CN 201811375147 A CN201811375147 A CN 201811375147A CN 109492756 B CN109492756 B CN 109492756B
- Authority
- CN
- China
- Prior art keywords
- sample
- point
- temperature
- coordinate point
- sounding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a multi-element wire galloping early warning method based on deep learning and a related device, wherein the method comprises the following steps: collecting relevant information of a target coordinate point, wherein the relevant information of the target coordinate point comprises longitude and latitude, altitude, temperature, relative humidity and wind speed of the target coordinate point, and a difference value between sounding temperature, sounding temperature and dew point temperature in sounding data; inputting the collected relevant information of the target coordinate point into the model as an input sample element of the pre-trained model; and obtaining a galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point. The method comprehensively considers the influences of ground meteorological elements, high-altitude meteorological elements, terrain and geographical positions on the galloping of the power transmission line, improves the universality and accuracy of the prediction of the conductor icing galloping, and can provide effective early warning information for the galloping of the power transmission line under various terrain conditions in various regions of China.
Description
Technical Field
The invention relates to the technical field of crossing of power production and weather prediction, in particular to a multi-element wire galloping early warning method based on deep learning and a related device.
Background
With global climate change, extreme weather events show the trend of increasing frequency and increasing intensity, and various weather disasters caused by the trend cause great threat to the operation safety of the power grid. The galloping of the power transmission line is used as a special meteorological disaster, and can cause line tripping, damage to power equipment, large-area power failure accidents and great loss to the social economy. Therefore, establishing an effective and widely applicable power transmission line galloping early warning and forecasting system has great practical significance for guaranteeing the operation safety of a power grid and the normal operation of the society.
The galloping of the power transmission line is a self-excited vibration phenomenon with low frequency and large amplitude generated under the action of wind excitation after the uneven icing of the wire of the power transmission line. The formation of the galloping of the power transmission line cannot be separated from 3 important factors, namely the icing of the wire, the excitation of wind and the parameters of the line. Scholars at home and abroad make a great deal of experiments and researches on the wire galloping mechanism, and the theory mainly comprises the vertical galloping theory of Den Hartog (1932), the torsional galloping theory of O.Nigol (1974), the eccentric inertial coupling instability theory of P.Yu (1993) and the like. However, because conductor galloping has the characteristics of randomness and nonlinearity, the theories cannot cover all conditions and galloping types of galloping, and therefore, corresponding prevention measures can be only effective for partial conditions only through mechanism research and galloping models. Based on the method, a new method is needed for preventing and treating the power transmission line galloping, and the prior art further provides that the power transmission line galloping can be pre-warned by utilizing a statistical model or a machine learning method before the power transmission line galloping occurs.
Machine learning is an effective modeling means for mining data characteristics in the absence of a theoretical model, and can summarize a set of rules and establish a model according to self-learning and error correction of historical data, and future prediction results can be obtained by inputting future data into the model. In recent years, many researchers have studied prediction models of conductor icing, and there are models such as a Generalized Regression Neural Network (GRNN), a BP neural network, and a Radial Basis Function (RBF) neural network, but these models predict the icing thickness only, and do not consider whether the conductor icing will be followed by waving or not.
The internal influence factors of conductor waving are: the self structural parameters of the line or the tower; the external factors are: meteorological factors and topographical factors. Wangchan et al (2017) obtain meteorological element indexes when Henan power grid galloping occurs by analyzing the historical galloping process of the Henan power grid. The results show that the external influence factors of conductor waving not only include ground meteorological elements but also high-altitude meteorological conditions. Li Shuai et al (2016) utilize Adaboost classification learner to respectively predict various lines for galloping, Jutower et al (2017) propose a galloping early warning method based on BP neural network, their selected input characteristic quantities are wind speed, wind direction and line included angle, relative humidity and temperature, which are all ground meteorological elements. In addition, as the breadth of our country is wide and spans a plurality of climatic zones, the difference of the climatic conditions of each region is obvious, and the meteorological conditions of each region when the wires gallop are different.
Because the influence factors of conductor galloping are numerous and the mechanism is complex, a transmission line galloping prediction method comprehensively considering ground meteorological elements, high-altitude meteorological elements, terrain and geographic positions is lacked at present, and a reliable galloping early warning result cannot be obtained.
Disclosure of Invention
In view of the above, the invention provides a multi-element wire galloping early warning method based on deep learning and a related device, which can comprehensively consider the influence of ground meteorological elements, high-altitude meteorological elements, terrain and geographic positions on the galloping of a power transmission line and improve the accuracy of wire icing galloping prediction. The technical scheme is as follows:
based on one aspect of the invention, the invention provides a multi-element wire galloping early warning method based on deep learning, which comprises the following steps:
collecting relevant information of a target coordinate point, wherein the relevant information of the target coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the target coordinate point, and a difference value of sounding temperature, sounding temperature and dew point temperature in sounding data;
inputting the collected relevant information of the target coordinate point into a model as an input sample element of the pre-trained model; the model is obtained by training sample elements of a sample coordinate point by adopting a preset processing algorithm, wherein the sample elements of the sample coordinate point comprise the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point, and the sounding temperature, the sounding temperature and the dew point temperature difference in sounding data, and the model can determine whether the sample coordinate point is a waving point according to the input sample elements of the sample coordinate point;
And obtaining a galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point.
Optionally, the model is obtained by training using the following method:
collecting related information and galloping point information of a plurality of sample coordinate points; the relevant information of the sample coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the sample coordinate point, and a sounding temperature, a sounding temperature difference and a dew point temperature difference in sounding data;
for any one sample coordinate point:
using the collected related information of the sample coordinate point as an input sample element of a preset model;
obtaining a galloping result used for representing whether the sample coordinate point is a galloping point or not according to the galloping point information of the sample coordinate point, and taking the obtained galloping result as an output sample element of the preset model; wherein the dance results comprise a first result for representing that the sample coordinate point is a dance point or a second result for representing that the sample coordinate point is not a dance point;
combining the input sample elements and the output sample elements of the sample coordinate points to obtain a sample element group, wherein the sample element group comprises longitude and latitude, altitude, temperature, relative humidity, wind speed, sounding temperature in sounding data, difference between the sounding temperature and dew point temperature and a galloping result of the sample coordinate points;
Selecting a plurality of sample element groups as training sample groups;
and for each training sample group, taking the input sample elements in the training sample group as the input parameters of the preset model, taking the output sample elements in the training sample group as the output parameters of the preset model, and training the preset model by adopting a preset processing algorithm to obtain a trained model.
Optionally, after obtaining the trained model, the method further comprises:
selecting a plurality of sample element groups as test sample groups;
for each test sample group, inputting the input sample elements in the test sample group into the trained model, and obtaining the test result output by the trained model;
if the accuracy of the test result meets the preset requirement according to the output sample elements in the test sample group, determining that the trained model is available;
and if the accuracy of the test result is determined not to meet the preset requirement according to the output sample elements in the test sample group, adjusting and optimizing the trained model, inputting the training sample group into the adjusted and optimized model again, and training the adjusted and optimized model by adopting a preset processing algorithm.
Optionally, the preset model is a deep neural network classification model, and the deep neural network classification model includes an input layer, four hidden layers and an output layer; wherein the content of the first and second substances,
the nodes of the input layer include 18, the nodes of the output layer include 1, and each node of the hidden layer includes 30.
Optionally, in the information related to the target coordinate point, the temperature is a daily minimum temperature, the relative humidity is a daily average relative humidity, and the wind speed is a daily maximum wind speed, and the sounding temperature and the dew point temperature in the sounding data are respectively sounding temperatures at 00 hours and 12 hours of three altitude layers of 500hpa, 700hpa, and 850 hpa.
Optionally, the collecting the related information of the sample coordinate point and the waving point information includes:
acquiring longitude and latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed of the sample coordinate point;
matching the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed of the sample coordinate point with the sounding temperature and dew point temperature of three altitude layers of 500hpa, 700hpa and 850hpa respectively at 00 and 12 in sounding data by adopting a nearest point matching method, and determining the element value of the sounding station closest to the sample coordinate point as the sounding element value corresponding to the sample coordinate point;
Combining the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed of the sample coordinate point and the difference values of the sounding temperature, the sounding temperature and the dew point temperature matched in the sounding data to serve as the related information of the sample coordinate point;
collecting information of a plurality of dancing points;
and matching the collected galloping point information with the related information of the sample coordinate point by adopting a nearest point matching method, and determining that the related information of the sample coordinate point closest to the galloping point information is the related information of the sample coordinate point corresponding to the galloping point information.
Optionally, after using the collected information about the sample coordinate point as an input sample element of the preset model, the method further includes:
using the formula:
normalizing the input sample elements;
wherein the content of the first and second substances,
x1j，x2j……xnjrepresenting the original sequence of class j input sample elements, y1j，y2j……ynjThe sequence is expressed by normalizing the j-th input sample element, wherein n is the number of samples, n is a positive integer, j is the number of classes of the input sample element, and j is 1,2,3 … … 18.
Based on another aspect of the present invention, the present invention provides a multi-element wire waving early warning device based on deep learning, including:
The device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring relevant information of a target coordinate point, and the relevant information of the target coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the target coordinate point and a difference value of sounding temperature, sounding temperature and dew point temperature in sounding data;
the input unit is used for inputting the collected relevant information of the target coordinate point into the model as an input sample element of a pre-trained model; the model is obtained by training sample elements of a sample coordinate point by adopting a preset processing algorithm, wherein the sample elements of the sample coordinate point comprise the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point, and the sounding temperature, the sounding temperature and the dew point temperature difference in sounding data, and the model can determine whether the sample coordinate point is a waving point according to the input sample elements of the sample coordinate point;
and the acquisition unit is used for acquiring the galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point.
Based on still another aspect of the present invention, there is provided a model training apparatus including:
The acquisition module is used for acquiring related information and galloping point information of a plurality of sample coordinate points; the relevant information of the sample coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the sample coordinate point, and a sounding temperature, a sounding temperature difference and a dew point temperature difference in sounding data;
the first determining module is used for taking the collected related information of the sample coordinate point as an input sample element of a preset model for any sample coordinate point;
the second determining module is used for obtaining a galloping result for representing whether the sample coordinate point is a galloping point or not according to the galloping point information of the sample coordinate point for any sample coordinate point, and taking the obtained galloping result as an output sample element of the preset model; wherein the dance results comprise a first result for representing that the sample coordinate point is a dance point or a second result for representing that the sample coordinate point is not a dance point;
the system comprises a sample element group obtaining module, a data processing module and a data processing module, wherein the sample element group obtaining module is used for combining an input sample element and an output sample element of any sample coordinate point to obtain a sample element group, and the sample element group comprises longitude and latitude, altitude, temperature, relative humidity, wind speed, sounding temperature in sounding data, difference between the sounding temperature and dew point temperature and a galloping result of the sample coordinate point;
The first selection module is used for selecting a plurality of sample element groups as training sample groups;
and the training module is used for training the preset model by adopting a preset processing algorithm to obtain the trained model by taking the input sample elements in the training sample group as the input parameters of the preset model and taking the output sample elements in the training sample group as the output parameters of the preset model for each training sample group.
Optionally, the method further comprises:
the second selection module is used for selecting a plurality of sample element groups as test sample groups;
the test module is used for inputting the input sample elements in the test sample group into the trained model for each test sample group to obtain the test result output by the trained model;
the third determining module is used for determining that the trained model is available when the accuracy of the test result meets the preset requirement according to the output sample elements in the test sample group;
the adjusting and optimizing module is used for adjusting and optimizing the trained model when the accuracy of the test result is determined to not meet the preset requirement according to the output sample elements in the test sample group;
And the training module is further used for inputting the training sample group into the adjusted and optimized model again and training the adjusted and optimized model by adopting a preset processing algorithm.
According to the multi-element wire galloping early warning method and the related device based on deep learning, prediction of wire icing galloping can be achieved only by inputting the collected related information of the target coordinate point into the model and further acquiring the galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point. The method comprehensively considers the influences of ground meteorological elements, high-altitude meteorological elements, terrain and geographical positions on the galloping of the power transmission line, improves the universality and accuracy of the prediction of the conductor icing galloping, and can provide effective early warning information for the galloping of the power transmission line under various terrain conditions in various regions of China.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a multi-element wire waving early warning method based on deep learning according to the present invention;
FIG. 2 is a flow chart of a model training method provided by the present invention;
FIG. 3 is a flowchart of a method for collecting information related to sample coordinate points and information related to a waving point according to the present invention;
FIG. 4 is a schematic structural diagram of a deep neural network classification model according to the present invention;
FIG. 5 is a flow chart of another model training method provided by the present invention;
FIG. 6 is a schematic structural diagram of a multi-element wire waving early warning device based on deep learning according to the present invention;
FIG. 7 is a schematic structural diagram of a model training apparatus according to the present invention;
FIG. 8 is a schematic structural diagram of an acquisition module according to the present invention;
fig. 9 is a schematic structural diagram of another model training device provided by the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As used herein, the singular forms "a", "an", "the" and "the" include plural referents unless the context clearly dictates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
As shown in fig. 1, the multi-element wire waving early warning method based on deep learning provided by the present invention may include:
In the relevant information of the target coordinate point, the temperature can be the lowest daily temperature, the relative humidity can be the daily average relative humidity, and the wind speed can be the daily maximum wind speed, the sounding temperature and the dew-point temperature in the sounding data can be respectively the sounding temperature and the dew-point temperature of three altitude layers of 500hpa, 700hpa and 850hpa at two times of 00 hours and 12 hours, wherein the sounding temperature and the dew-point temperature totally comprise six sounding temperature values and six dew-point temperature values.
102, inputting the collected relevant information of the target coordinate point into a model as an input sample element of the pre-trained model; the model is obtained by training sample elements of a sample coordinate point by adopting a preset processing algorithm, the sample elements of the sample coordinate point comprise the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point and the difference value of the sounding temperature, the sounding temperature and the dew point temperature in sounding data, and the model can determine whether the sample coordinate point is a waving point according to the input sample elements of the sample coordinate point.
And 103, acquiring a galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point.
According to the multi-element wire galloping early warning method based on deep learning, the prediction of the wire icing galloping can be realized only by inputting the collected related information of the target coordinate point into the model and further acquiring the galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point. The method comprehensively considers the influences of ground meteorological elements, high-altitude meteorological elements, terrain and geographical positions on the galloping of the power transmission line, improves the universality and accuracy of the prediction of the conductor icing galloping, and can provide effective early warning information for the galloping of the power transmission line under various terrain conditions in various regions of China.
In the present invention, as a method for training a model, a method shown in fig. 2 may be adopted, the method including:
According to the invention, firstly, the relevant information and the galloping point information of a plurality of sample coordinate points are required to be collected, wherein the relevant information of each sample coordinate point comprises the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point, and the sounding temperature, the sounding temperature and the dew point temperature difference in sounding data. Specifically, for each sample coordinate point, the method shown in fig. 3 may be used to acquire the related information and the waving point information of each sample coordinate point, and the method may include:
In the invention, the longitude and latitude, the altitude, the daily minimum temperature, the daily average relative humidity and the daily maximum wind speed of each sample coordinate point are firstly collected.
And 302, matching the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed of the sample coordinate point with the sounding temperature and dew point temperature of the sounding data at 00 and 12 in three altitude layers of 500hpa, 700hpa and 850hpa respectively by adopting a nearest point matching method, and determining the element value of the sounding station closest to the sample coordinate point as the sounding element value corresponding to the sample coordinate point.
Specifically, the method adopts a nearest point matching method to match the acquired longitude, latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed with six sounding temperatures and six dew-point temperatures in sounding data, so as to determine that the sounding element value corresponding to the coordinate point of the longitude and latitude is the element value of the sounding station closest to the coordinate point.
Specifically, in the present invention, the < longitude, latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed > acquired in step 301 and the < difference between the sounding temperature, sounding temperature and dew point temperature > matched in step 302 are combined to be the relevant information of the sample coordinate point, that is, the relevant information of the sample coordinate point is < longitude, latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed, sounding temperature and dew point temperature >.
And 305, matching the collected galloping point information with the related information of the sample coordinate point by adopting a nearest point matching method, and determining the related information of the sample coordinate point closest to the galloping point information as the sample coordinate point corresponding to the galloping point information.
The method specifically adopts a nearest neighbor point matching method to match the collected dancing point information with the related information of the sample coordinate point, namely < longitude, latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed, sounding temperature, difference value between sounding temperature and dew point temperature >, so as to determine that the related information of the sample coordinate point closest to the dancing point information is the related information of the sample coordinate point corresponding to the dancing point information, and can also be understood as determining that the input sample element corresponding to the dancing point is the element value of the coordinate point closest to the dancing point and having the determined longitude and latitude.
And after the related information of the sample coordinate point is obtained, the related information of the sample coordinate point is used as an input sample element of the preset model.
The preset model can be specifically a deep neural network classification model, the deep neural network classification model can be built by using a Tensorflow-based deep neural network technology, the characteristics of various elements and the mapping between the characteristics and the galloping can be effectively mined and learned, and the accuracy of galloping prediction is improved.
Specifically, as shown in fig. 4, the deep neural network classification model in the present invention is a six-layer structure, and the six-layer structure includes an input layer, a four-layer hidden layer, and an output layer. Wherein the nodes of the input layer comprise 18, the nodes of the output layer comprise 1, and each node of the hidden layer comprises 30.
As a preferred embodiment of the present invention, after the information related to the sample coordinate point is used as an input sample element of the preset model, the method may further include:
using the formula:
normalizing the input sample elements;
wherein the content of the first and second substances,
x1j，x2j……xnjrepresenting the original sequence of class j input sample elements, y1j，y2j……ynjThe sequence is expressed by normalizing the j-th input sample element, wherein n is the number of samples, n is a positive integer, j is the number of classes of the input sample element, and j is 1,2,3 … … 18.
In the invention, whether the sample coordinate point is the galloping point can be determined by analyzing the galloping point information of the sample coordinate point, so that a galloping result for representing whether the sample coordinate point is the galloping point is obtained. Furthermore, the obtained waving result is used as an output sample element of the preset model.
Particularly in the present invention, the dance result may include a first result for characterizing the sample coordinate point as a dance point or a second result for characterizing the sample coordinate point as not a dance point, wherein the first result is represented by a code "1", for example, and the second result is represented by a code "0", for example.
And 204, combining the input sample elements and the output sample elements of the sample coordinate points to obtain a sample element group, wherein the sample element group comprises the longitude and latitude, the altitude, the temperature, the relative humidity, the wind speed, the sounding temperature in sounding data, the difference value between the sounding temperature and the dew point temperature and the galloping result of the sample coordinate points.
In the invention, the related information of the sample coordinate point is used as an input sample element of the preset model, and the waving result corresponding to the sample coordinate point is used as an output sample element of the preset model. One sample coordinate point corresponds to one sample element group, and each sample element group comprises longitude and latitude, altitude, temperature, relative humidity, wind speed, sounding temperature in sounding data, difference value between sounding temperature and dew point temperature and waving results of the sample coordinate point.
Specifically, for example, each sample element group is < longitude, latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed, < temperature at 850hpa height 00 hours, < temperature at 700hpa height 00 hours, < temperature at 500hpa height 00 hours, < temperature at 850hpa height 12 hours, < temperature at 700hpa height 12 hours, < temperature at 500hpa height 12 hours, < difference between 850hpa height 00 hours and dew point temperature, < difference between 700hpa height 00 hours and dew point temperature, < difference between 500hpa height 00 hours and dew point temperature, < temperature at 850hpa height 12 hours and dew point temperature, < difference between 700hpa height 12 hours and dew point temperature, < difference between 500hpa height 12 hours and dew point temperature, and waving result >.
At least one of the obtained plurality of sample element groups is selected, and preferably, the plurality of sample element groups are selected as a training sample group.
And step 206, for each training sample group, taking the input sample elements in the training sample group as the input parameters of the preset model, taking the output sample elements in the training sample group as the output parameters of the preset model, and training the preset model by adopting a preset processing algorithm to obtain a trained model.
The preset processing algorithm is, for example, a neural network algorithm, and the like, which is not limited by the applicant.
Further to ensure the accuracy of the trained model, after step 206, the method may further include, as shown in fig. 5:
At least one of the plurality of sample element groups is selected, and preferably a plurality of sample element groups are selected as a test sample group.
And 208, inputting the input sample elements in each test sample group into the trained model to obtain the test result output by the trained model.
If the accuracy of the test result is determined to meet the preset requirement according to the output sample elements in the test sample group, step 209 is executed, otherwise step 210 is executed.
Wherein the accuracy of the test result meets the preset requirement may for example comprise that the overall accuracy of the test result reaches a preset accuracy threshold.
Step 209 determines that the trained model is available.
And 210, adjusting and optimizing the trained model, inputting the training sample set into the adjusted and optimized model again, and training the adjusted and optimized model by adopting a preset processing algorithm.
In the invention, if the accuracy of the test result output by the trained model does not meet the preset requirement, the current accuracy of the trained model is poor, and the model is unavailable, so that the model needs to be further adjusted, optimized and trained. Specifically, the trained model is adjusted and optimized, and for the adjusted and optimized model, the previously determined test sample set is input into the adjusted and optimized model again, and the adjusted and optimized model is trained by adopting a preset processing algorithm.
If the accuracy of the test result output by the adjusted and optimized model still does not meet the preset requirement, the model is adjusted and optimized again, and for the model after the adjustment and optimization again, the previously determined training sample set is input into the model after the adjustment and optimization again, and the model after the adjustment and optimization again is trained by adopting a preset processing algorithm. And repeating the steps until the accuracy of the test result output by the adjusted and optimized model meets the preset requirement.
The model training and testing method provided by the invention can be roughly summarized as the following processes:
1) Initializing the model by using the self-defined hyper-parameters to obtain a deep neural network initial model;
2) taking input sample elements in a training sample group as input parameters of a deep neural network initial model, taking corresponding output sample elements as output parameters of the deep neural network initial model, and training the deep neural network initial model (namely the initial model) to obtain a trained model;
3) taking the input sample elements in the test sample group as input parameters of the trained model, and judging whether the output parameters of the model are consistent with the actual waving results in the output sample elements in the test sample group, so as to realize the inspection and analysis of the accuracy of the trained model;
4) selecting a reasonable hyper-parameter combination, and performing random hyper-parameter optimization on the model to obtain a random optimization model; specifically, random search is carried out in the distribution of the hyper-parameter combinations, and the hyper-parameter combinations with optimal effects are determined through three-fold cross validation;
5) taking input sample elements in a test sample group as input parameters, and testing the random optimization model;
6) selecting adjacent values to form a new hyper-parameter combination according to hyper-parameter setting of the random optimization model, and carrying out hyper-parameter optimization on the model to obtain an optimized model; specifically, a grid method traversal search is carried out in the new hyper-parameter combination, and a hyper-parameter combination with the optimal effect is determined through triple-fold intersection verification;
7) Taking input sample elements in a test sample group as input parameters to test the optimization model;
8) if the test effect of the optimization model is better than that of the training model and the random optimization model, the final model is determined to be the optimization model; otherwise, the aforementioned steps 4), 5), 6), 7) are repeatedly carried out until the final model is determined.
The invention utilizes the Tensorflow-based deep neural network technology to build the galloping prediction model, can effectively mine and learn the characteristics of various elements and the mapping between the characteristics and the galloping, improves the accuracy of galloping prediction, and the deep neural network model is an intelligent prediction technology with self-learning and self-adaptive characteristics, and can be continuously updated and improved along with the increase of training samples, thereby obtaining more reliable prediction results and providing powerful guarantee for the galloping prevention of the power transmission line.
The invention provides a multi-element wire galloping early warning method based on deep learning based on a previous text, and also provides a multi-element wire galloping early warning device based on deep learning, as shown in fig. 6, the device can comprise:
the system comprises an acquisition unit 10, a control unit and a control unit, wherein the acquisition unit is used for acquiring relevant information of a target coordinate point, and the relevant information of the target coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the target coordinate point and a difference value of sounding temperature, sounding temperature and dew point temperature in sounding data;
In the relevant information of the target coordinate point, the temperature is the lowest daily temperature, the relative humidity is the daily average relative humidity, the wind speed is the maximum daily wind speed, and the sounding temperature and the dew point temperature in the sounding data are respectively the sounding temperature and the dew point temperature of three height layers of 500hpa, 700hpa and 850hpa at 00 and 12 hours;
an input unit 20, configured to input the collected relevant information of the target coordinate point into a model as an input sample element of the model trained in advance; the model is obtained by training sample elements of a sample coordinate point by adopting a preset processing algorithm, wherein the sample elements of the sample coordinate point comprise the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point, and the sounding temperature, the sounding temperature and the dew point temperature difference in sounding data, and the model can determine whether the sample coordinate point is a waving point according to the input sample elements of the sample coordinate point;
and the obtaining unit 30 is configured to obtain a galloping early warning result that is output by the model and used for representing whether the target coordinate point is a galloping point.
Also, the present invention provides a model training apparatus, as shown in fig. 7, the apparatus may include: the system comprises an acquisition module 100, a first determination module 200, a second determination module 300, a sample element group obtaining module 400, a first selection module 500 and a training module 600. Wherein, the first and the second end of the pipe are connected with each other,
The acquisition module 100 is configured to acquire related information and waving point information of a plurality of sample coordinate points; the relevant information of the sample coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the sample coordinate point, and a sounding temperature, a sounding temperature difference and a dew point temperature difference in sounding data. As shown in fig. 8, the acquisition module 100 may specifically include:
the first acquisition submodule 110 is configured to acquire longitude and latitude, an altitude, a daily minimum temperature, a daily average relative humidity, and a daily maximum wind speed of the sample coordinate point;
the first matching submodule 120 is configured to match, by using a nearest neighbor matching method, the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity, and daily maximum wind speed of the sample coordinate point with the sounding temperature and the dew point temperature at 00 and 12 times in three altitude layers of 500hpa, 700hpa, and 850hpa, respectively, in the sounding data, and determine that an element value of the sounding station closest to the sample coordinate point is a sounding element value corresponding to the sample coordinate point;
the merging submodule 130 is configured to merge the collected longitude and latitude, the altitude, the daily minimum temperature, the daily average relative humidity, the daily maximum wind speed of the sample coordinate point, and the difference between the sounding temperature, and the dew point temperature that are matched in the sounding data, and use the merged result as the related information of the sample coordinate point;
A second collecting submodule 140 configured to collect information on a plurality of dance points;
the second matching submodule 150 is configured to match the collected dancing point information with the relevant information of the sample coordinate point by using a closest point matching method, and determine that the relevant information of the sample coordinate point closest to the dancing point information is the relevant information of the sample coordinate point corresponding to the dancing point information.
A first determining module 200, configured to, for any sample coordinate point, use the acquired relevant information of the sample coordinate point as an input sample element of a preset model;
the second determining module 300 is configured to, for any sample coordinate point, obtain a waving result representing whether the sample coordinate point is a waving point according to the waving point information of the sample coordinate point, and use the obtained waving result as an output sample element of the preset model; wherein the dance results include a first result for characterizing the sample coordinate point as a dance point or a second result for characterizing the sample coordinate point as not a dance point;
a sample element group obtaining module 400, configured to combine, for any sample coordinate point, an input sample element and an output sample element of the sample coordinate point to obtain a sample element group, where the sample element group includes a longitude and latitude, an altitude, a temperature, a relative humidity, a wind speed, an air sounding temperature in air sounding data, a difference between the air sounding temperature and a dew point temperature, and a waving result of the sample coordinate point;
A first selecting module 500, configured to select a plurality of sample element groups as training sample groups;
the training module 600 is configured to, for each training sample group, train the preset model by using an input sample element in the training sample group as an input parameter of the preset model and using an output sample element in the training sample group as an output parameter of the preset model, and using a preset processing algorithm to obtain a trained model.
In the invention, the preset model is a deep neural network classification model, and the deep neural network classification model comprises an input layer, four hidden layers and an output layer; wherein the nodes of the input layer include 18, the nodes of the output layer include 1, and each node of the hidden layer includes 30.
Further preferably, as shown in fig. 9, the model training apparatus provided by the present invention may further include:
a second selecting module 700, configured to select a plurality of sample element groups as test sample groups;
a test module 800, configured to, for each test sample group, input an input sample element in the test sample group into the trained model, and obtain a test result output by the trained model;
A third determining module 900, configured to determine that the trained model is available when it is determined that the accuracy of the test result meets a preset requirement according to the output sample elements in the test sample group;
and an adjusting and optimizing module 1000, configured to adjust and optimize the trained model when it is determined that the accuracy of the test result does not meet a preset requirement according to the output sample elements in the test sample group.
At this time, the training module 600 is further configured to input the training sample set into the adjusted and optimized model again, and train the adjusted and optimized model by using a preset processing algorithm.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
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 "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The method for warning the waving of the multi-element wire based on deep learning and the related device provided by the invention are described in detail, a specific example is applied in the text to explain the principle and the implementation mode of the invention, and the description of the embodiment is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, 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 invention.
Claims (8)
1. A multi-element wire galloping early warning method based on deep learning is characterized by comprising the following steps:
collecting relevant information of a target coordinate point, wherein the relevant information of the target coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the target coordinate point, and sounding temperature, difference value of sounding temperature and dew point temperature in sounding data; in the relevant information of the target coordinate point, the temperature is the lowest daily temperature, the relative humidity is the daily average relative humidity, the wind speed is the maximum daily wind speed, and the sounding temperature and the dew point temperature in the sounding data are respectively the sounding temperature and the dew point temperature of three height layers of 500hpa, 700hpa and 850hpa at 00h and 12 h;
Collecting the related information and the waving point information of the sample coordinate points comprises: acquiring longitude and latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed of the sample coordinate point;
determining the element value of the sounding station closest to the sample coordinate point as the sounding element value corresponding to the sample coordinate point by adopting a nearest point matching method according to the longitude and latitude of the sample coordinate point; combining the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed of the sample coordinate point and the difference values of the sounding temperature, the sounding temperature and the dew point temperature matched in the sounding data to serve as the related information of the sample coordinate point;
collecting information of a plurality of dancing points; matching the collected galloping point information with the related information of the sample coordinate point by adopting a nearest point matching method, and determining that the related information of the sample coordinate point closest to the galloping point information is the related information of the sample coordinate point corresponding to the galloping point information;
inputting the collected relevant information of the target coordinate point into a model as an input sample element of the pre-trained model; the model is obtained by training sample elements of a sample coordinate point by adopting a preset processing algorithm, wherein the sample elements of the sample coordinate point comprise the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point, and the sounding temperature, the sounding temperature and the dew point temperature difference in sounding data, and the model can determine whether the sample coordinate point is a waving point according to the input sample elements of the sample coordinate point;
And obtaining a galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point.
2. The method of claim 1, wherein the model is trained using the following method:
collecting related information and galloping point information of a plurality of sample coordinate points; the relevant information of the sample coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the sample coordinate point, and a sounding temperature, a sounding temperature difference and a dew point temperature difference in sounding data;
for any one sample coordinate point:
using the collected related information of the sample coordinate point as an input sample element of a preset model;
obtaining a galloping result used for representing whether the sample coordinate point is a galloping point or not according to the galloping point information of the sample coordinate point, and taking the obtained galloping result as an output sample element of the preset model; wherein the dance results include a first result for characterizing the sample coordinate point as a dance point or a second result for characterizing the sample coordinate point as not a dance point;
combining the input sample elements and the output sample elements of the sample coordinate points to obtain a sample element group, wherein the sample element group comprises longitude and latitude, altitude, temperature, relative humidity, wind speed, sounding temperature in sounding data, difference between the sounding temperature and dew point temperature and a galloping result of the sample coordinate points;
Selecting a plurality of sample element groups as training sample groups;
and for each training sample group, taking the input sample elements in the training sample group as the input parameters of the preset model, taking the output sample elements in the training sample group as the output parameters of the preset model, and training the preset model by adopting a preset processing algorithm to obtain a trained model.
3. The method of claim 2, wherein after obtaining the trained model, the method further comprises:
selecting a plurality of sample element groups as test sample groups;
for each test sample group, inputting the input sample elements in the test sample group into the trained model, and acquiring the test result output by the trained model;
if the accuracy of the test result meets the preset requirement according to the output sample elements in the test sample group, determining that the trained model is available;
and if the accuracy of the test result is determined not to meet the preset requirement according to the output sample elements in the test sample group, adjusting and optimizing the trained model, inputting the training sample group into the adjusted and optimized model again, and training the adjusted and optimized model by adopting a preset processing algorithm.
4. The method of claim 2 or 3,
the preset model is a deep neural network classification model, and the deep neural network classification model comprises an input layer, four hidden layers and an output layer; wherein, the first and the second end of the pipe are connected with each other,
the nodes of the input layer include 18, the nodes of the output layer include 1, and each node of the hidden layer includes 30.
5. The method according to claim 2 or 3, wherein after using the collected information about the sample coordinate points as input sample elements of the preset model, the method further comprises:
using the formula:
normalizing the input sample elements;
wherein the content of the first and second substances,
x1j，x2j……xnjrepresenting the original sequence of class j input sample elements, y1j，y2j……ynjThe sequence is expressed by normalizing the j-th input sample element, wherein n is the number of samples, n is a positive integer, j is the number of classes of the input sample element, and j is 1,2,3 … … 18.
6. The utility model provides a multi-element wire dancing early warning device based on degree of depth learning which characterized in that includes:
the device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring relevant information of a target coordinate point, and the relevant information of the target coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the target coordinate point and a difference value of sounding temperature, sounding temperature and dew point temperature in sounding data; in the relevant information of the target coordinate point, the temperature is the lowest daily temperature, the relative humidity is the daily average relative humidity, the wind speed is the maximum daily wind speed, and the sounding temperature and the dew point temperature in the sounding data are respectively the sounding temperature and the dew point temperature of three height layers of 500hpa, 700hpa and 850hpa at 00h and 12 h;
Collecting the related information and the waving point information of the sample coordinate points comprises: acquiring longitude and latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed of the sample coordinate point;
determining the element value of the sounding station closest to the sample coordinate point as the sounding element value corresponding to the sample coordinate point by adopting a nearest point matching method according to the longitude and latitude of the sample coordinate point; combining the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed of the sample coordinate point and the difference values of the sounding temperature, the sounding temperature and the dew point temperature matched in the sounding data to serve as the related information of the sample coordinate point;
collecting information of a plurality of dancing points; matching the collected galloping point information with the related information of the sample coordinate point by adopting a nearest point matching method, and determining that the related information of the sample coordinate point closest to the galloping point information is the related information of the sample coordinate point corresponding to the galloping point information;
the input unit is used for inputting the collected relevant information of the target coordinate point into the model as an input sample element of a pre-trained model; the model is obtained by training sample elements of a sample coordinate point by adopting a preset processing algorithm, wherein the sample elements of the sample coordinate point comprise the longitude and latitude, the altitude, the temperature, the relative humidity and the wind speed of the sample coordinate point, and the sounding temperature, the sounding temperature and the dew point temperature difference in sounding data, and the model can determine whether the sample coordinate point is a waving point according to the input sample elements of the sample coordinate point;
And the acquisition unit is used for acquiring the galloping early warning result which is output by the model and used for representing whether the target coordinate point is a galloping point.
7. A model training apparatus, comprising:
the acquisition module is used for acquiring related information and galloping point information of the plurality of sample coordinate points; the relevant information of the sample coordinate point comprises longitude and latitude, altitude, temperature, relative humidity, wind speed of the sample coordinate point, and a sounding temperature, a sounding temperature difference and a dew point temperature difference in sounding data; in the related information of the sample coordinate point, the temperature is the lowest daily temperature, the relative humidity is the daily average relative humidity, the wind speed is the maximum daily wind speed, and the sounding temperature and the dew point temperature in the sounding data are respectively the sounding temperature and the dew point temperature of three height layers of 500hpa, 700hpa and 850hpa at 00 and 12 hours;
collecting the related information and the waving point information of the sample coordinate point comprises the following steps: acquiring longitude and latitude, altitude, daily minimum temperature, daily average relative humidity and daily maximum wind speed of the sample coordinate point;
determining the element value of the sounding station closest to the sample coordinate point as the sounding element value corresponding to the sample coordinate point by adopting a nearest point matching method according to the longitude and latitude of the sample coordinate point; combining the acquired longitude and latitude, altitude, daily minimum temperature, daily average relative humidity, daily maximum wind speed of the sample coordinate point and the difference values of the sounding temperature, the sounding temperature and the dew point temperature matched in the sounding data to serve as the related information of the sample coordinate point;
Collecting a plurality of dancing point information; matching the collected galloping point information with the related information of the sample coordinate point by adopting a nearest point matching method, and determining that the related information of the sample coordinate point closest to the galloping point information is the related information of the sample coordinate point corresponding to the galloping point information;
the first determining module is used for taking the collected related information of the sample coordinate point as an input sample element of a preset model for any sample coordinate point;
the second determining module is used for obtaining a galloping result for representing whether the sample coordinate point is a galloping point or not according to the galloping point information of the sample coordinate point for any sample coordinate point, and taking the obtained galloping result as an output sample element of the preset model; wherein the dance results comprise a first result for representing that the sample coordinate point is a dance point or a second result for representing that the sample coordinate point is not a dance point;
the system comprises a sample element group obtaining module, a data processing module and a data processing module, wherein the sample element group obtaining module is used for combining an input sample element and an output sample element of any sample coordinate point to obtain a sample element group, and the sample element group comprises longitude and latitude, altitude, temperature, relative humidity, wind speed, sounding temperature in sounding data, difference between the sounding temperature and dew point temperature and a galloping result of the sample coordinate point;
The first selection module is used for selecting a plurality of sample element groups as training sample groups;
and the training module is used for training the preset model by adopting a preset processing algorithm to obtain the trained model by taking the input sample elements in the training sample group as the input parameters of the preset model and taking the output sample elements in the training sample group as the output parameters of the preset model for each training sample group.
8. The apparatus of claim 7, further comprising:
the second selection module is used for selecting a plurality of sample element groups as test sample groups;
the test module is used for inputting the input sample elements in the test sample group into the trained model for each test sample group to obtain the test result output by the trained model;
the third determining module is used for determining that the trained model is available when the accuracy of the test result meets the preset requirement according to the output sample elements in the test sample group;
the adjusting and optimizing module is used for adjusting and optimizing the trained model when the accuracy of the test result is determined to not meet the preset requirement according to the output sample elements in the test sample group;
And the training module is further used for inputting the training sample group into the adjusted and optimized model again and training the adjusted and optimized model by adopting a preset processing algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811375147.3A CN109492756B (en) | 2018-11-19 | 2018-11-19 | Multi-element wire galloping early warning method based on deep learning and related device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811375147.3A CN109492756B (en) | 2018-11-19 | 2018-11-19 | Multi-element wire galloping early warning method based on deep learning and related device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109492756A CN109492756A (en) | 2019-03-19 |
| CN109492756B true CN109492756B (en) | 2022-06-10 |
Family
ID=65696839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811375147.3A Active CN109492756B (en) | 2018-11-19 | 2018-11-19 | Multi-element wire galloping early warning method based on deep learning and related device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109492756B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110572772B (en) * | 2019-09-12 | 2020-12-08 | 电子科技大学 | GRNN-AdaBoost-based multi-device fusion positioning method |
| CN112948762B (en) * | 2019-12-11 | 2023-07-18 | 国网湖南省电力有限公司 | Ground temperature correction method and system for galloping forecast |
| CN110929808A (en) * | 2019-12-11 | 2020-03-27 | 国网湖南省电力有限公司 | Multi-element intelligent correction method and system for waving temperature |
| CN112540748B (en) * | 2020-11-09 | 2024-02-27 | 华能新能源股份有限公司 | Automatic operation system for mesoscale wind energy resource analysis |
| CN112525256A (en) * | 2020-11-12 | 2021-03-19 | 国网湖南省电力有限公司 | Real-time weather monitoring-based galloping event development state prediction method and device |
| CN112365057A (en) * | 2020-11-12 | 2021-02-12 | 国网湖南省电力有限公司 | Regional power transmission line galloping event model construction method and galloping event prediction method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4891576A (en) * | 1988-08-22 | 1990-01-02 | The United States Of America As Represented By The Secretary Of The Interior | Ground-based transmission line conductor motion sensor |
| CN104239970A (en) * | 2014-09-04 | 2014-12-24 | 国家电网公司 | Power transmission line gallop risk early-warning method based on Adaboost |
| CN106503751A (en) * | 2016-11-10 | 2017-03-15 | 国网河南省电力公司电力科学研究院 | A kind of power transmission line Louis dance potential prediction method based on SVM classifier |
| CN106570582A (en) * | 2016-10-26 | 2017-04-19 | 中国电力科学研究院 | Method and system for building transmission line dancing tripping risk prediction network model |
| CN107330189A (en) * | 2017-06-29 | 2017-11-07 | 国网湖南省电力公司 | A kind of meshes thermal stratification refinement Forecasting Methodology and system |
| CN108509537A (en) * | 2018-03-14 | 2018-09-07 | 北京玖天气象科技有限公司 | A kind of transmission line galloping probability forecast system and method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221381B (en) * | 2011-06-10 | 2012-10-03 | 国网信息通信有限公司 | Method and system for monitoring power transmission line of power grid |
| US8788437B2 (en) * | 2011-07-28 | 2014-07-22 | Quova, Inc. | System and method for implementing a learning model for predicting the geographic location of an internet protocol address |
| CN108182514A (en) * | 2017-12-13 | 2018-06-19 | 国网湖南省电力有限公司 | A kind of power grid icing waves Risk Forecast Method, system and storage medium |
-
2018
- 2018-11-19 CN CN201811375147.3A patent/CN109492756B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4891576A (en) * | 1988-08-22 | 1990-01-02 | The United States Of America As Represented By The Secretary Of The Interior | Ground-based transmission line conductor motion sensor |
| CN104239970A (en) * | 2014-09-04 | 2014-12-24 | 国家电网公司 | Power transmission line gallop risk early-warning method based on Adaboost |
| CN106570582A (en) * | 2016-10-26 | 2017-04-19 | 中国电力科学研究院 | Method and system for building transmission line dancing tripping risk prediction network model |
| CN106503751A (en) * | 2016-11-10 | 2017-03-15 | 国网河南省电力公司电力科学研究院 | A kind of power transmission line Louis dance potential prediction method based on SVM classifier |
| CN107330189A (en) * | 2017-06-29 | 2017-11-07 | 国网湖南省电力公司 | A kind of meshes thermal stratification refinement Forecasting Methodology and system |
| CN108509537A (en) * | 2018-03-14 | 2018-09-07 | 北京玖天气象科技有限公司 | A kind of transmission line galloping probability forecast system and method |
Non-Patent Citations (3)
| Title |
|---|
| Statistical Analysis and Research on Galloping Characteristics and Damage for Iced Conductors of Transmission Lines in China;Kuan-jun Zhu et al.;《2010 International Conference on Power System Technology》;20101213;第1-5页 * |
| 基于BP神经网络的输电线路舞动预警方法;廖峥 等;《电力系统保护与控制》;20171001;第45卷(第19期);第154-161页 * |
| 河南电网输电线路舞动的气象 素指标研究;王丙兰 等;《气象》;20170131;第43卷(第1期);第108页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109492756A (en) | 2019-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109492756B (en) | Multi-element wire galloping early warning method based on deep learning and related device | |
| CN109840660B (en) | Vehicle characteristic data processing method and vehicle risk prediction model training method | |
| CN108227041B (en) | Horizontal visibility forecasting method based on site measured data and mode result | |
| CN103743402B (en) | A kind of underwater intelligent self adaptation Approach of Terrain Matching of topographic information based amount | |
| CN103514366A (en) | Urban air quality concentration monitoring missing data recovering method | |
| CN111428942B (en) | Line icing thickness prediction method for extracting micro-terrain factors based on variable grid technology | |
| CN113971463A (en) | Heat supply pipeline risk distribution analysis method and routing inspection path planning system | |
| CN102679935B (en) | System and method for calculating icing thickness of power transmission line | |
| CN105740643A (en) | Self-adaptive PM<2.5>concentration speculating method based on city region grid | |
| JP6820625B1 (en) | Servers, forest data utilization systems, forest data utilization methods and programs | |
| CN102867217A (en) | Projection pursuit-based risk evaluation method for meteorological disasters of facility agriculture | |
| CN117556197B (en) | Typhoon vortex initialization method based on artificial intelligence | |
| CN113516304A (en) | Space-time joint prediction method and device for regional pollutants based on space-time graph network | |
| CN106570582A (en) | Method and system for building transmission line dancing tripping risk prediction network model | |
| CN105005661A (en) | Transmission tower material actual strength calculation method with practical result | |
| CN113240219A (en) | Land utilization simulation and prediction method | |
| Fallah-Ghalhary et al. | Spring rainfall prediction based on remote linkage controlling using adaptive neuro-fuzzy inference system (ANFIS) | |
| CN111008730B (en) | Crowd concentration prediction model construction method and device based on urban space structure | |
| CN115391746B (en) | Interpolation method, interpolation device, electronic device and medium for meteorological element data | |
| CN103473476B (en) | The wind energy resources projectional technique of wind data is surveyed based on several anemometer towers in region | |
| CN113744888B (en) | Regional epidemic trend prediction and early warning method and system | |
| CN115146714A (en) | Big data management method related to underground space collapse | |
| CN115829150A (en) | Accumulated water prediction system | |
| KR102307898B1 (en) | Groundwater potential mapping using integrated ensemble of three bivariate statistical models with random froest and logistic tree models | |
| Mandal et al. | Prediction of Wind Speed using Machine Learning |
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 |