CN114723204A - Power transmission line galloping grading early warning method based on supervised learning - Google Patents
Power transmission line galloping grading early warning method based on supervised learning Download PDFInfo
- Publication number
- CN114723204A CN114723204A CN202110009250.1A CN202110009250A CN114723204A CN 114723204 A CN114723204 A CN 114723204A CN 202110009250 A CN202110009250 A CN 202110009250A CN 114723204 A CN114723204 A CN 114723204A
- Authority
- CN
- China
- Prior art keywords
- galloping
- data
- temperature
- transmission line
- power transmission
- 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 42
- 230000005540 biological transmission Effects 0.000 title claims abstract description 33
- 238000012216 screening Methods 0.000 claims abstract description 11
- 239000002131 composite material Substances 0.000 claims abstract description 10
- 238000003064 k means clustering Methods 0.000 claims abstract description 8
- 238000012937 correction Methods 0.000 claims description 18
- 230000008859 change Effects 0.000 claims description 16
- 230000002159 abnormal effect Effects 0.000 claims description 12
- 238000010606 normalization Methods 0.000 claims description 8
- 230000000694 effects Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 230000002068 genetic effect Effects 0.000 claims description 3
- 230000003121 nonmonotonic effect Effects 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 abstract description 4
- 238000007689 inspection Methods 0.000 abstract description 2
- 238000012423 maintenance Methods 0.000 abstract description 2
- 230000010355 oscillation Effects 0.000 abstract description 2
- 230000002265 prevention Effects 0.000 abstract 1
- 238000012549 training Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 239000004020 conductor Substances 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 241000209524 Araceae Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000012212 insulator Substances 0.000 description 1
- 238000012933 kinetic analysis Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/02—Instruments for indicating weather conditions by measuring two or more variables, e.g. humidity, pressure, temperature, cloud cover or wind speed
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- 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
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- 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)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Educational Administration (AREA)
- Evolutionary Biology (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Molecular Biology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Environmental & Geological Engineering (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Medical Informatics (AREA)
- Physiology (AREA)
Abstract
The invention discloses a power transmission line galloping grading early warning method based on supervised learning, and belongs to the field of overhead power transmission line fault prevention and control of a power system. Firstly, correcting temperature and humidity data by using a sliding time window, adjusting the wind direction to be an included angle with the axial direction of a lead, calculating the influence weight and the comprehensive influence factor of different microclimate elements on the oscillation amplitude, and screening a sample by using the comprehensive influence factor through a K-means clustering algorithm. And (3) taking the microclimate elements as input and the amplitude data as output, and constructing a galloping prediction model based on a supervised learning algorithm to further obtain a prediction result. The method is compared with a classical algorithm, and the superiority of the power transmission line galloping prediction model of the GA-BP and SVM composite algorithm is proved by judging prediction errors and prediction results. The galloping prediction result of the invention has high coincidence degree with the actual state, can predict galloping amplitude and realize the galloping grading early warning function, and can send out early warning information in advance. Operation and maintenance personnel can flexibly make inspection strategies and anti-galloping measures, and the safe and stable operation of the power transmission line is guaranteed.
Description
Technical Field
The invention relates to the field of overhead transmission line fault early warning of a power system, in particular to a monitoring learning-based overhead transmission line galloping grading early warning method.
Background
The transmission line is erected in the natural environment, and the structure safety and stability of the transmission line are easily influenced by the natural environment. The galloping is one of common fault types of overhead transmission lines, and the conductor is easy to generate self-excited vibration with large amplitude and low frequency under the action of a certain attack angle and wind speed, namely the galloping. The galloping has multiple hazards, light flashover and trip accidents are caused, heavy hardware fittings and insulators are damaged, and the accidents of strand breakage and wire breakage of the wires, even tower collapse and the like are caused. The galloping accidents of the power transmission line occur frequently, so that the social electricity utilization safety is influenced, and great loss is brought to economic construction. Therefore, the development of the prediction research on the galloping of the power transmission line has important significance for making a disaster responding method in advance and ensuring the safe operation of the power system.
The existing galloping prediction research mainly comprises two methods of dynamically analyzing a wire, simulating a wire motion track and constructing a machine learning galloping prediction model, wherein the main problems of the two methods are that the wire galloping stress is very complex, the wire galloping stress is geometric nonlinear motion, the coupling mode between the wire galloping stress and airflow is fluid-solid coupling, the wire motion track model is too ideal, the pneumatic characteristic of the wire under a real condition cannot be mastered, and the galloping numerical analysis and an actual measurement result have obvious difference; the main problems of the latter are that the prediction result is wide, only whether waving occurs or not can be judged, waving amplitude cannot be estimated, grading early warning cannot be performed on the prediction result, data correction and screening work of samples is not performed during model training, so that large errors exist in the model, and the prediction performance also has a large promotion space.
In order to solve the problems in the prior art, a galloping prediction method which does not depend on kinetic analysis and has a high coincidence degree of a prediction result and an actual situation is urgently needed, the galloping situation of a wire under a real situation can be mastered, and the prediction amplitude and the galloping early warning grade can be given in advance. The method does not start from the aerodynamic characteristics of the wire, but performs correlation analysis on the microclimate elements and the waving amplitude, and constructs a waving prediction model based on a supervised learning algorithm to obtain a prediction result.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the power transmission line galloping grading early warning method based on supervised learning, which can correct data acquired by a sensor and complete sample screening work, the constructed galloping prediction model can better fit the galloping state of a wire under a real situation, the functions of pre-estimating galloping amplitude and galloping grading early warning are realized, operation and maintenance personnel can flexibly make routing inspection strategies and anti-galloping measures according to early warning results, and the safe and stable operation of the power transmission line is ensured.
The technical method adopted for solving the technical problem is to provide a power transmission line galloping grading early warning method based on supervised learning, and the method comprises the following steps:
the method comprises the following steps: determining a waving influence parameter. According to the characteristic that when data in the same area are trained, the dominant factors influencing the waving are external conditions and the line structure and parameters are implicit conditions, microclimate elements are used as influencing parameters, namely wind direction, standard wind speed, maximum wind speed, ten-minute average wind speed, temperature and humidity are used as the influencing parameters.
Step two: and correcting the temperature and humidity data by utilizing a single-side sliding time window. The specific method comprises the following steps:
taking temperature as an example, let time temperature series T { (T)i-k,bi-k),…,(ti-1,bi-1),(ti,bi),(ti+1,bi+1) …, the length of the one-sided sliding time window is k. t is tiPredicted temperature value at time bi′,wi-k,…,wi-1Is a temperature bi-k,…,bi-1And assigning the corresponding weight by adopting a square weighting method. The corresponding formula is as follows:
predicted temperature value bi' the corresponding confidence interval PCI is:
Skthe standard deviation of the temperature data in the time window is shown, and alpha is an upper quantile point of the t distribution.
The data correction method comprises the following steps: finding t by a single-sided sliding time windowiPredicted temperature value b at timei' and its confidence interval, if the temperature b measured by the sensoriWithin a confidence interval, then biNormal data; if b isiFalling outside the confidence interval, then biFor abnormal data, correct it by bi' instead of bi。
And defining the tortuosity Q reflecting the unsmooth property of the temperature change curve, wherein the tortuosity Q means the ratio of the sum of the change amplitudes of the temperature change curve when the temperature change curve shows non-monotonic change to the data volume. Assuming a total number of data N, for a contiguous sequence of temperatures (…, b)i-1,bi,bi+1…), there are two cases:
1) if (b)i-bi-1)(bi+1-bi) Q is less than or equal to 0i=|bi-bi-1|+|bi+1-bi|
2) If (b)i-bi-1)(bi+1-bi)>0, then qi=0
The definition deviation degree M reflects the deviation degree between the corrected temperature curve and the original curve. The original temperature sequence is (…, b)i-1,bi,bi+1…), corrected temperature series is (…, b)i-1″,bi″,bi+1", …), then:
the smaller the tortuosity Q is, the smoother the data is; the smaller the deviation M, the more complete the retained information. Let H be QNormalization+MNormalizationWhen the result is minimum, the data correction effect is best, and the correction of the temperature data is completed.
Step three: the wind direction is corrected. And correcting the wind direction into an included angle with the axial direction of the line according to the trend of the wire and the direction recorded by the wind direction sensor, wherein the value range is 0-90 degrees.
Step four: and determining the influence weight and the comprehensive influence factor of the microclimate elements. And averagely segmenting the microclimate elements, clustering the samples in each segment into normal samples and abnormal samples according to a K-means clustering algorithm to obtain a group of data in which the microclimate element mean values and the normal sample amplitude mean values correspond one to one, and calculating the correlation coefficient of the microclimate element mean values and the normal sample amplitude mean values to serve as the influence weight of the microclimate elements. And defining a comprehensive influence factor to reflect the influence of the comprehensive micro-meteorological elements on the waving amplitude at a certain moment. And expressing the influence weight by QZ, expressing the comprehensive influence factor by ZH, expressing the normalized microclimate element value by GY, and obtaining the comprehensive influence factor of each group of data according to the relational expression of the QZ, the ZH and the GY.
ZH=∑GY×QZ (5)
Step five: and (4) screening the sample. And correspondingly taking the comprehensive influence factors and the amplitudes as sample points, clustering according to a K-means clustering algorithm, setting 2 clustering centers to obtain two clusters of abnormal samples and normal samples, and rejecting the abnormal samples to complete data screening.
Step six: and constructing a power transmission line galloping prediction model based on a supervised learning algorithm. The method comprises the steps of taking microclimate elements as input, taking amplitude data as output, taking a comprehensive influence factor as a division standard, training 90% of samples of the comprehensive influence factor in the front of a section of the comprehensive influence factor by an SVM algorithm, training 10% of samples of the comprehensive influence factor in the rear of the section of the comprehensive influence factor by a BP neural network (GA-BP) optimized by a genetic algorithm, constructing a power transmission line galloping prediction model based on the GA-BP and SVM composite algorithm, and further obtaining a prediction result.
Step seven: and carrying out grading early warning according to the prediction result. The galloping with the amplitude 15-100 times of the diameter of the wire is called three-stage galloping; the galloping with the amplitude of 100-200 times of the diameter of the wire is called secondary galloping; the galloping with the amplitude more than 200 times of the diameter of the wire is called primary galloping, the galloping prediction model constructed in the sixth step can give specific predicted amplitude, and the galloping risk grade can be obtained according to the standards.
The invention has the beneficial effects that:
the invention is generally suitable for the galloping early warning of the power transmission line, and compared with the prior art, the invention has the following advantages:
1. the invention can realize the prediction of the waving amplitude and the function of grading and early warning of waving, and compared with the traditional neural network waving prediction model which can only judge whether waving occurs and can not predict the waving amplitude and the severity, the result has more accuracy and practicability.
2. The invention relates and analyzes the micrometeorological elements of the wire and the waving amplitude to construct a supervised learning model, does not depend on the dynamic analysis of the wire, can grasp the waving condition of the wire under the real condition, and leads the prediction result to better fit the actual condition.
3. According to the method, the microclimate data are corrected, and the sample screening work is completed, so that the noise interference generated in the data acquisition process can be effectively reduced, and the accuracy of the prediction result is higher.
4. According to the method, the galloping prediction model is jointly constructed by adopting two algorithms of SVM and GA-BP, the advantages of the SVM algorithm in the aspect of fitting small amplitude and the advantages of the GA-BP algorithm in the aspect of fitting large amplitude are simultaneously absorbed, and compared with the method of constructing the model by adopting a single algorithm, the prediction accuracy is greatly improved.
Drawings
The invention has the following drawings:
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a sample screening result based on the K-means clustering algorithm according to an embodiment of the present invention;
fig. 3 shows a prediction result of the oscillation amplitude of the transmission line according to an embodiment of the present invention.
Detailed Description
In order to make the method and the advantages of the present invention more clear, the present invention will be further described in detail with reference to the following embodiments and the accompanying drawings. The present example is based on the method of the present invention, and a detailed embodiment and a specific operation procedure are given, but the scope of the present invention is not limited to the following examples.
The embodiment of the invention, Tongliao 500kV Araceae line, starts from an Aratan 500kV transformer substation, passes through Zautzfeldt-Jakob, Kailu county and ends at a Corerqin 500kV transformer substation, and the line trend is 40 degrees to the west in the north. 1950 sets of microclimate data and dance monitoring data of the acolin in two time periods from 11 months in 2018 to 4 months in 2019 and from 8 months in 2019 to 11 months in 2019 are selected as samples.
As shown in fig. 1, the power transmission line galloping grading early warning method based on supervised learning in the embodiment includes the following steps:
the method comprises the following steps: determining a waving influence parameter. According to the characteristic that when data in the same area are trained, the dominant factors influencing the waving are external conditions and the line structure and parameters are implicit conditions, microclimate elements are used as influencing parameters, namely wind direction, standard wind speed, maximum wind speed, ten-minute average wind speed, temperature and humidity are used as the influencing parameters.
Step two: and correcting the temperature and humidity data by utilizing a single-side sliding time window. The specific method comprises the following steps:
taking temperature as an example, let time temperature series T { (T)i-k,bi-k),…,(ti-1,bi-1),(ti,bi),(ti+1,bi+1) …, the length of the one-sided sliding time window is k. t is tiPredicted temperature value at time bi′,wi-k,…,wi-1Is a temperature bi-k,…,bi-1And assigning the corresponding weight by adopting a square weighting method. The corresponding disclosure is as follows:
predicted temperature value bi' the corresponding confidence interval PCI is:
Skand alpha is the upper quantile point of the t distribution, and the confidence coefficient P is 100 (1-alpha)% which reflects the probability of the actual value appearing in the confidence interval.
The data correction method comprises the following steps: finding t by a single-sided sliding time windowiPredicted temperature value b at timei' and its confidence interval, if the temperature b measured by the sensoriWithin a confidence interval, then biNormal data; if b isiFalling outside the confidence interval, then biFor abnormal data, correct it by bi' instead of bi。
And defining the tortuosity Q reflecting the unsmooth property of the temperature change curve, wherein the tortuosity Q means the ratio of the sum of the change amplitudes of the temperature change curve when the temperature change curve shows non-monotonic change to the data volume. Assuming a total number of data N, for a contiguous sequence of temperatures (…, b)i-1,bi,bi+1…), there are two cases:
1) if (b)i-bi-1)(bi+1-bi) Q is less than or equal to 0i=|bi-bi-1|+|bi+1-bi|
2) If (b)i-bi-1)(bi+1-bi)>0, then qi=0
The definition deviation degree M reflects the deviation degree between the corrected temperature curve and the original curve. The original temperature sequence is (…, b)i-1,bi,bi+1…), corrected temperature series is (…, b)i-1″,bi″,bi+1", …), then:
the smaller the tortuosity Q is, the smoother the data is; the smaller the deviation M, the more complete the retained information. Let H be QNormalization+MNormalizationWhen the result is minimum, the data correction effect is best, and the temperature data correction is completed.
The two factors that influence the correction result are the length k of the unilateral sliding time window and the threshold value of the confidence interval, which is related to the selection of the confidence coefficient P. In order to optimize the correction effect, the most suitable values of k and P need to be determined.
Setting k as 2, 3, … and 8 respectively, and acquiring data by the microclimate sensor every ten minutes, namely calculating by using data 20, 30, … and 80 minutes before a certain moment; the confidence coefficients P are respectively set to 95%, 98%, 99%, 99.5%, and 99.8%, and corrected using different parameter combinations, and the corresponding H values are shown in table 1.
TABLE 1 temperature corrected H value
When k is 5 and P is 98%, H is 0.779, which completes the correction of the temperature data.
The humidity data was corrected in the same manner, and the corresponding H values are shown in table 2.
TABLE 2 humidity correction H value
When k is 5 and P is 95%, H is 0.916 as the minimum value, and the correction of the humidity data is completed.
Step three: the wind direction is corrected. Assuming that the wind direction orientation recorded by the wind direction sensor is F degrees, the included angle between the wind direction and the axial direction of the line is G degrees, and the trend of the Achia is 40 degrees north, the wind direction correction formula for the line is as follows:
and correcting the wind direction data into included angle data with the axial direction of the line according to a formula (5).
Step four: and determining the influence weight and the comprehensive influence factor of the microclimate elements. The microclimate elements are averagely segmented, the samples in each segment are clustered into normal samples and abnormal samples according to a K-means clustering algorithm, a group of data in which the microclimate element mean values and the normal sample amplitude mean values correspond to each other one by one is obtained, correlation coefficients of the microclimate element mean values and the normal sample amplitude mean values are calculated and used as influence weights of the microclimate elements, and the results are shown in table 3.
TABLE 3 influence weights of Microweather elements
When the values of wind direction, standard wind speed, maximum wind speed, ten-minute average wind speed and humidity are increased, the waving amplitude is increased, and the influence weight is a positive value; when the temperature value is reduced, the ice coating is more easily formed on the power transmission line, so that the wire generates larger lift force, the waving amplitude is increased, and the influence weight is a negative value.
And defining a comprehensive influence factor to reflect the influence of the comprehensive micro-meteorological elements on the waving amplitude at a certain moment. And expressing the influence weight by QZ, expressing the comprehensive influence factor by ZH, expressing the normalized microclimate element value by GY, and obtaining the comprehensive influence factor of each group of data according to the relational expression of the QZ, the ZH and the GY.
ZH=∑GY×QZ (6)
Step five: and (4) screening the sample. And (3) correspondingly taking the comprehensive influence factors and the amplitudes as sample points, clustering according to a K-means clustering algorithm, setting 2 clustering centers to obtain two clusters of abnormal samples and normal samples, and rejecting the abnormal samples to complete data screening as shown in figure 2.
Step six: and constructing a power transmission line galloping prediction model based on a supervised learning algorithm. The microclimate elements are used as input, amplitude data are used as output, the comprehensive influence factors are used as division standards, 90% of samples of the comprehensive influence factors in the front of the interval are trained by an SVM algorithm, 10% of samples of the comprehensive influence factors in the rear of the interval are trained by a BP neural network (GA-BP) optimized by a genetic algorithm, a power transmission line galloping prediction model based on the GA-BP and SVM composite algorithm is constructed, and then a prediction result shown in figure 3 is obtained.
Step seven: and carrying out grading early warning according to the prediction result. The galloping with the amplitude 15-100 times of the diameter of the wire is called three-stage galloping; the galloping with the amplitude of 100-200 times of the diameter of the wire is called secondary galloping; the galloping with the amplitude more than 200 times of the diameter of the wire is called primary galloping, and the galloping risk grade of the predicted result can be obtained according to the standard.
In order to embody the superiority of the power transmission line galloping prediction model based on the GA-BP and SVM composite algorithm, the prediction results of the composite algorithm model and the classical algorithm model are compared, and judgment is carried out through two aspects of prediction error and prediction effect.
For the prediction error, the Root Mean Square Error (RMSE), the Mean Absolute Error (MAE), and the Mean Absolute Percentage Error (MAPE) of the predicted value and the actual value were evaluated, and the error averages were obtained by 20 times of simulation for each model, and the comparison results are shown in table 4.
TABLE 4 comparison of prediction errors for the composite and classical algorithm models
And judging the prediction success by using the prediction accuracy, the i-level accuracy, the actual report rate, the empty report rate and the missing report rate. Let AijThe number of times that the conductor generates i-level galloping and the prediction result is j-level galloping in the actual situation is shown, i and j can be 0, 1, 2 and 3, and 0 shows that galloping does not occur or early warning is not performed.
On this basis, the prediction accuracy is defined: and under the condition that both the actual condition and the early warning result are judged to be galloping, the early warning level is in accordance with the actual galloping level. The formula is as follows:
i-level accuracy: when the i-level galloping actually occurs, the prediction result is the proportion of the i-level galloping. The formula is as follows:
the actual report rate is as follows: the proportion of the actual number of dancing occurrences in the total number of early warning times. The formula is as follows:
and (3) the null report rate: the false alarm times account for the proportion of the total early warning times. The formula is as follows:
the rate of missing reports: the false-miss times account for the proportion of the total times of actual waving. The formula is as follows:
the results of comparing the predicted performance of the composite algorithm model with that of the classical algorithm model are shown in table 5.
TABLE 5 comparison of predicted outcomes of composite and classical algorithmic models
As can be seen from tables 4 and 5, the composite algorithm model is superior to the classical algorithm model in the aspects of prediction error and prediction effect, and the prediction result has higher practicability and accuracy. In addition, as can be seen from fig. 3, the prediction result of the invention can better fit the actual state, and the functions of prediction and grading early warning of the waving amplitude are realized.
The above description is only exemplary of the present invention and should not be taken as limiting the invention, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (5)
1. A power transmission line galloping grading early warning method based on supervised learning is characterized by comprising the following steps:
the method comprises the following steps: determining a waving influencing parameter. The microclimate elements are used as influence parameters, namely wind direction, standard wind speed, maximum wind speed, ten-minute average wind speed, temperature and humidity are used as the influence parameters.
Step two: and correcting the temperature and humidity data by utilizing a single-side sliding time window. Defining the unsmoothness of the tortuosity Q reflecting temperature change curve; the deviation M reflects the degree of deviation between the corrected temperature curve and the original curve. Let H be QNormalization+MNormalizationAnd when H is minimum, finishing data correction.
Step three: the wind direction is corrected. And correcting the wind direction into an included angle with the axial direction of the line according to the trend of the wire and the direction recorded by the wind direction sensor, wherein the value range is 0-90 degrees.
Step four: and determining the influence weight and the comprehensive influence factor of the microclimate elements.
Step five: and (4) screening the sample. And correspondingly taking the comprehensive influence factors and the amplitudes as sample points, clustering according to a K-means clustering algorithm, setting 2 clustering centers to obtain two clusters of abnormal samples and normal samples, and rejecting the abnormal samples to complete data screening.
Step six: and constructing a power transmission line galloping prediction model based on a supervised learning algorithm.
Step seven: and carrying out grading early warning according to the prediction result.
2. The supervised learning-based grading early warning method for power transmission line galloping as recited in claim 1, wherein the specific correction method for temperature and humidity data in the second step is as follows:
let the time-temperature sequence T { (T)i-k,bi-k),…,(ti-1,bi-1),(ti,bi),(ti+1,bi+1) …, the length of the one-sided sliding time window is k. t is tiPredicted temperature value at time bi′,wi-k,…,wi-1Is a temperature bi-k,…,bi-1And assigning the corresponding weight by adopting a square weighting method. The corresponding disclosure is as follows:
predicted temperature value bi' the corresponding confidence interval PCI is:
Skand the standard deviation of the temperature data in the time window, wherein alpha is an upper quantile of the t distribution.
The data correction method comprises the following steps: finding t by a single-sided sliding time windowiPredicted temperature value b of timei' and its confidence interval, if the temperature b measured by the sensoriWithin a confidence interval, then biThe data are normal data; if b isiFalling outside the confidence interval, then biFor abnormal data, correct it by bi' instead of bi。
And defining the tortuosity Q reflecting the unsmooth property of the temperature change curve, wherein the tortuosity Q means the ratio of the sum of the change amplitudes of the temperature change curve when the temperature change curve shows non-monotonic change to the data volume. Assuming a total number of data N, for a contiguous sequence of temperatures (…, b)i-1,bi,bi+1…), there are two cases:
1) if (b)i-bi-1)(bi+1-bi) Q is less than or equal to 0i=|bi-bi-1|+|bi+1-bi|
2) If (b)i-bi-1)(bi+1-bi)>0, then qi=0
The definition deviation degree M reflects the deviation degree between the corrected temperature curve and the original curve. The original temperature sequence is (…, b)i-1,bi,bi+1…), corrected temperature series is (…, b)i-1″,bi″,bi+1", …), then:
the smaller the tortuosity Q is, the smoother the data is; the smaller the deviation M, the more complete the retained information. Let H be QNormalization+MNormalizationWhen the result is minimum, the data correction effect is best, and the correction of the temperature data is completed.
The humidity data was corrected in the same manner.
3. The supervised learning-based grading early warning method for power transmission line galloping as recited in claim 1, wherein the method for determining the influence weight and the comprehensive influence factor in the fourth step is as follows:
and averagely segmenting the microclimate elements, clustering the samples in each segment into normal samples and abnormal samples according to a K-means clustering algorithm to obtain a group of data in which the microclimate element mean values and the normal sample amplitude mean values correspond one to one, and calculating the correlation coefficient of the microclimate element mean values and the normal sample amplitude mean values to serve as the influence weight of the microclimate elements.
The comprehensive influence factor reflects the influence of the comprehensive microclimate elements on the waving amplitude at a certain moment. And expressing the influence weight by QZ, expressing the comprehensive influence factor by ZH, expressing the normalized microclimate element value by GY, and obtaining the comprehensive influence factor of each group of data according to the relational expression of the QZ, the ZH and the GY.
ZH=∑GY×QZ (5)。
4. The method for grading and early warning the galloping of the power transmission line based on the supervised learning of claim 1, wherein the step six of constructing the galloping prediction model of the power transmission line based on the supervised learning algorithm is to take microclimate elements as input, amplitude data as output and a comprehensive influence factor as a division standard, train 90% of samples of the comprehensive influence factor in the front of the interval by adopting an SVM algorithm, train 10% of samples of the comprehensive influence factor in the rear of the interval by adopting a BP neural network (GA-BP) optimized by a genetic algorithm, construct the galloping prediction model of the power transmission line based on the GA-BP and SVM composite algorithm, and further obtain a prediction result.
5. The hierarchical early warning method for power transmission line galloping based on supervised learning as recited in claim 1, wherein the hierarchical early warning method according to the prediction result in the seventh step is as follows: the galloping with the amplitude 15-100 times of the diameter of the wire is called three-stage galloping; the galloping with the amplitude of 100-200 times of the diameter of the wire is called secondary galloping; the galloping with the amplitude more than 200 times of the diameter of the wire is called primary galloping, the galloping prediction model constructed in the sixth step can give specific predicted amplitude, and the galloping risk grade can be obtained according to the standards.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110009250.1A CN114723204A (en) | 2021-01-05 | 2021-01-05 | Power transmission line galloping grading early warning method based on supervised learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110009250.1A CN114723204A (en) | 2021-01-05 | 2021-01-05 | Power transmission line galloping grading early warning method based on supervised learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114723204A true CN114723204A (en) | 2022-07-08 |
Family
ID=82234353
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110009250.1A Pending CN114723204A (en) | 2021-01-05 | 2021-01-05 | Power transmission line galloping grading early warning method based on supervised learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114723204A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115526036A (en) * | 2022-09-19 | 2022-12-27 | 长安大学 | Method and system for judging rock burst scale grade |
CN118157324A (en) * | 2024-05-08 | 2024-06-07 | 江苏濠汉信息技术有限公司 | Power transmission line wind deflection galloping early warning method and system based on twin network |
-
2021
- 2021-01-05 CN CN202110009250.1A patent/CN114723204A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115526036A (en) * | 2022-09-19 | 2022-12-27 | 长安大学 | Method and system for judging rock burst scale grade |
CN118157324A (en) * | 2024-05-08 | 2024-06-07 | 江苏濠汉信息技术有限公司 | Power transmission line wind deflection galloping early warning method and system based on twin network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113723010B (en) | Bridge damage early warning method based on LSTM temperature-displacement correlation model | |
CN104200288B (en) | A kind of equipment fault Forecasting Methodology based on dependency relation identification between factor and event | |
CN112734128B (en) | 7-day power load peak prediction method based on optimized RBF | |
CN114723204A (en) | Power transmission line galloping grading early warning method based on supervised learning | |
CN111652461A (en) | Aero-engine continuous health state evaluation method based on SAE-HMM | |
WO2023124287A1 (en) | Wind power output prediction method, electronic device, storage medium, and system | |
CN111488896B (en) | Distribution line time-varying fault probability calculation method based on multi-source data mining | |
CN105894177A (en) | Decision-making-tree-algorithm-based analysis and evaluation method for operation risk of power equipment | |
CN109492792A (en) | A method of it is predicted based on particle group optimizing wavelet neural network powerline ice-covering | |
CN117421687A (en) | Method for monitoring running state of digital power ring main unit | |
CN110708318A (en) | Network abnormal flow prediction method based on improved radial basis function neural network algorithm | |
CN110889587B (en) | Power distribution network line risk assessment method | |
CN109491339B (en) | Big data-based substation equipment running state early warning system | |
CN109214604A (en) | The haze calamity risk forecast method and system of power transmission and transforming equipment based on Adaboost | |
CN109973331A (en) | A kind of fan blade of wind generating set fault diagnosis algorithm based on bp neural network | |
CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
CN115204698A (en) | Real-time analysis method for power supply stability of low-voltage transformer area | |
CN111612227A (en) | Load prediction method based on K-means clustering and bat optimization neural network | |
CN116664098A (en) | Abnormality detection method and system for photovoltaic power station | |
CN113986636B (en) | Hard disk fault prediction method for data center based on hard disk self-adaptive report data | |
CN115293236A (en) | Hybrid clustering-based parallel fault diagnosis method and device for power equipment | |
CN115864644A (en) | Relay protection device state evaluation method, system, equipment and medium | |
CN108428049A (en) | Elevator system level of integrity appraisal procedure based on fuzzy neural network | |
CN115224684A (en) | Intelligent power distribution network risk state identification method and system based on immune hazard theory | |
CN117113157B (en) | Platform district power consumption fault detection system based on artificial intelligence |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20220708 |