CN117251812A - High-voltage power line operation fault detection method based on big data analysis - Google Patents
High-voltage power line operation fault detection method based on big data analysis Download PDFInfo
- Publication number
- CN117251812A CN117251812A CN202311227707.1A CN202311227707A CN117251812A CN 117251812 A CN117251812 A CN 117251812A CN 202311227707 A CN202311227707 A CN 202311227707A CN 117251812 A CN117251812 A CN 117251812A
- Authority
- CN
- China
- Prior art keywords
- data
- fault
- unit
- analysis
- power line
- 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.)
- Withdrawn
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 130
- 238000007405 data analysis Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 claims abstract description 154
- 238000012544 monitoring process Methods 0.000 claims abstract description 89
- 238000012545 processing Methods 0.000 claims abstract description 76
- 238000000605 extraction Methods 0.000 claims abstract description 43
- 230000005540 biological transmission Effects 0.000 claims abstract description 33
- 238000001914 filtration Methods 0.000 claims abstract description 33
- 230000007246 mechanism Effects 0.000 claims abstract description 20
- 238000004458 analytical method Methods 0.000 claims description 69
- 230000002159 abnormal effect Effects 0.000 claims description 58
- 238000004422 calculation algorithm Methods 0.000 claims description 57
- 230000000007 visual effect Effects 0.000 claims description 42
- 230000008569 process Effects 0.000 claims description 29
- 230000006870 function Effects 0.000 claims description 25
- 238000003745 diagnosis Methods 0.000 claims description 24
- 238000007726 management method Methods 0.000 claims description 23
- 230000002452 interceptive effect Effects 0.000 claims description 21
- 238000012300 Sequence Analysis Methods 0.000 claims description 17
- 238000013461 design Methods 0.000 claims description 17
- 238000013079 data visualisation Methods 0.000 claims description 16
- 230000009466 transformation Effects 0.000 claims description 16
- 230000000694 effects Effects 0.000 claims description 15
- 238000006243 chemical reaction Methods 0.000 claims description 14
- 238000010606 normalization Methods 0.000 claims description 14
- 238000013441 quality evaluation Methods 0.000 claims description 14
- 230000008439 repair process Effects 0.000 claims description 14
- 238000007619 statistical method Methods 0.000 claims description 14
- 238000011156 evaluation Methods 0.000 claims description 13
- 230000004044 response Effects 0.000 claims description 13
- 230000006399 behavior Effects 0.000 claims description 12
- 238000009499 grossing Methods 0.000 claims description 12
- 230000005856 abnormality Effects 0.000 claims description 9
- 238000004140 cleaning Methods 0.000 claims description 9
- 238000011084 recovery Methods 0.000 claims description 9
- 238000003909 pattern recognition Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004891 communication Methods 0.000 claims description 7
- 238000013500 data storage Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 7
- 238000009825 accumulation Methods 0.000 claims description 6
- 238000010219 correlation analysis Methods 0.000 claims description 6
- 238000013506 data mapping Methods 0.000 claims description 6
- 230000000670 limiting effect Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000004148 unit process Methods 0.000 claims description 6
- 238000000691 measurement method Methods 0.000 claims description 5
- 230000003993 interaction Effects 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 238000003672 processing method Methods 0.000 claims description 4
- 238000001308 synthesis method Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 238000003066 decision tree Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 230000036961 partial effect Effects 0.000 claims description 3
- 238000007637 random forest analysis Methods 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 230000026676 system process Effects 0.000 claims description 3
- 238000011426 transformation method Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 description 14
- 238000012423 maintenance Methods 0.000 description 9
- 238000010801 machine learning Methods 0.000 description 7
- 238000012800 visualization Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 230000002194 synthesizing effect Effects 0.000 description 3
- 238000011049 filling Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 238000007794 visualization technique Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/085—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution lines, e.g. overhead
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2131—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on a transform domain processing, e.g. wavelet transform
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- 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
- G06N3/0455—Auto-encoder networks; Encoder-decoder 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
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
- G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
Abstract
The invention discloses a high-voltage power line operation fault detection method based on big data analysis, relates to the field of power systems, and solves the problems of poor data quality, difficult feature extraction, easy missed detection false alarm and poor interpretation in the traditional high-voltage power line operation fault detection method; the method comprises the following specific steps: step one, collecting current, voltage, temperature, humidity, power and frequency parameters of a power line in real time; step two, monitoring and recording the data transmission process and storage condition through a data quality management mechanism; step three, filtering and normalizing the data; step four, performing feature extraction operation; generating a virtual fault sample; step six, performing fault detection and prediction on the power line; step seven, diagnosing, positioning and explaining the fault cause of the detected fault; step eight, periodically collecting, processing and analyzing monitoring data of the power line; and step nine, displaying the monitoring data and the fault detection result.
Description
Technical Field
The invention relates to the field of power systems, in particular to a high-voltage power line operation fault detection method based on big data analysis.
Background
In the past decades, with the continuous expansion of the scale of power systems and the increase of power demands, the influence of high-voltage power line operation faults on power supply reliability and safe and stable operation is increasingly remarkable; the traditional fault detection method is mainly based on manual inspection and offline testing, and cannot meet the requirements of real-time monitoring and fault early warning of a large-scale power line;
with the development of big data technology, the high-voltage power line operation fault detection method based on big data analysis is widely applied and researched; the method realizes real-time monitoring and prediction of the fault state of the power line by collecting, storing and analyzing a large amount of real-time monitoring data and utilizing technical means such as machine learning, data mining and the like so as to improve the reliability and the operation efficiency of the power system;
however, although the high voltage power line fault detection method based on big data analysis has many advantages, there are some disadvantages;
the reliability and accuracy of big data analysis depends on the quality of the data; however, in the power system, there may be abnormality, inaccuracy or lack of monitoring data due to equipment failure, communication interruption, etc., thereby affecting the effect and reliability of the failure detection algorithm;
The monitoring data of the high-voltage power line generally contains multidimensional and high-dimensional information such as current, voltage, temperature and the like; extracting valid features to characterize the state of the line is a critical issue; however, the current feature extraction method is mainly based on experience and rules, and is difficult to capture potential subtle changes and complex correlations, so that limitations of a detection algorithm are caused;
the big data analysis method may have a certain problem of missing detection and false alarm; some fault modes may be ignored or misjudged in big data analysis, and the algorithm has limited adaptability to unknown fault modes, which may result in failure to discover potential faults in time, or generate too many false alarms, increasing operation and maintenance costs;
the high-voltage power line fault detection method often lacks the interpretation of faults, and is difficult to interpret the decision process and judgment basis in the fault detection method; this limits the method reliability and interpretability;
therefore, in order to solve the problems of data quality, difficult feature extraction, missed detection and false alarm phenomena and poor fault interpretation in the traditional high-voltage power line operation fault detection method, the invention discloses a high-voltage power line operation fault detection method based on big data analysis.
Disclosure of Invention
Aiming at the defects of the prior art, the invention discloses a high-voltage power line operation fault detection method based on big data analysis, which monitors and records the data transmission process and the storage condition by introducing a data quality management mechanism, ensures the integrity, the accuracy and the timeliness of the monitored data and solves the data quality problem; generating a virtual fault sample by a synthetic data method, expanding the scale of a training set, and enhancing the generalization capability of a fault early warning model so as to improve the accuracy and reliability of the model under the condition of limited actual fault samples; by introducing a signal processing model, filtering, normalizing and other operations are performed on the data, so that noise is eliminated, and the data quality is improved; then capturing dynamic changes and fault behaviors of the power line by utilizing a characteristic extraction model and using methods such as time sequence analysis, frequency domain analysis, wavelet transformation and the like; the problem of difficult feature extraction is solved; performing fault detection and prediction on the power line through a fault early warning model, and analyzing historical data and real-time data by combining a time sequence analysis method so as to find potential faults; the problem of easy missed detection and false alarm is solved; and diagnosing and positioning the detected faults through a fault identification model, and explaining the judgment basis and decision making process of the model. By utilizing the feature importance analysis and feature influence measurement method, the reliability and interpretation of the algorithm are improved, so that a user can understand the judgment result of the model and make a corresponding decision; the problem of poor fault interpretation is solved.
In order to achieve the technical effects, the invention adopts the following technical scheme:
a high voltage power line operation fault detection method based on big data analysis, wherein the method comprises:
as a further technical scheme of the invention, the method comprises the following steps:
step one, current, voltage, temperature, humidity, power and frequency parameters of a power line are collected in real time through a dynamic collection module;
monitoring and recording the data transmission process and the storage condition through a data quality management mechanism so as to ensure the integrity, the accuracy and the timeliness of the data;
step three, filtering and normalizing the data through a signal processing model to eliminate noise and improve the data quality; the signal processing model identifies and processes abnormal points in the data through an abnormal detection algorithm, and repairs the abnormal data through an interpolation and smoothing method;
step four, performing feature extraction operation on the monitoring data output by the signal processing model through the feature extraction model; the feature extraction model captures dynamic changes and fault behaviors of the power line through time sequence analysis, frequency domain analysis and wavelet transformation analysis methods;
generating a virtual fault sample by a synthetic data method to expand the scale of a training set and enhance the generalization capability of a fault early warning model;
Step six, performing fault detection and prediction on the power line through a fault early warning model; the fault early warning model analyzes historical data and real-time data of the power line through a time sequence analysis method to find potential faults;
step seven, diagnosing, positioning and explaining the fault cause of the detected fault through a fault identification model; the fault identification model interprets the judgment basis and decision process of the model through a feature importance analysis and feature influence measurement method so as to improve the reliability and the interpretability of the algorithm;
step eight, periodically collecting, processing and analyzing monitoring data of the power line through a real-time monitoring system, wherein the real-time monitoring system sends out a warning to a fault through a real-time communication and voice information broadcasting method;
and step nine, displaying the monitoring data and the fault detection result through a data visualization module so as to facilitate the understanding and decision of a user.
As a further technical scheme of the invention, the data quality management mechanism comprises a data acquisition quality monitoring module, a data cleaning and checking module and a data quality evaluation module; the data acquisition quality monitoring module comprises a data transmission monitoring unit and a data storage monitoring unit; the data transmission monitoring unit verifies the abnormal condition of the data in the transmission process by an error checking method so as to ensure the integrity and timeliness of the data transmission; the data storage monitoring unit monitors the integrity and the uniqueness of the data in the storage process through a data check and deduplication method so as to avoid the occurrence of duplicate data; the data cleaning and checking module comprises an abnormal value detection unit, a missing value processing unit and a data accuracy checking unit; the abnormal value detection unit detects and processes the data through a box diagram method so as to ensure the accuracy and the reliability of the data; the missing value processing unit fills missing data through an interpolation method and a regression model so as to avoid influencing the data analysis result; the data accuracy checking unit checks the collected monitoring data through the hash codes so as to prevent the problems of error marking, repeated sampling and noise interference; the data quality evaluation module comprises an evaluation index definition unit and an evaluation algorithm unit; the evaluation index definition unit defines data quality evaluation indexes through an expert system and a rule engine, wherein the data quality evaluation indexes at least comprise data integrity, accuracy, consistency and reliability; the evaluation algorithm unit evaluates and analyzes the monitoring data through a statistical method.
As a further technical scheme of the invention, the signal processing model comprises a filtering module, a normalization module, an error identification module and a data restoration module; the filtering module carries out filtering operation on the acquired data through a digital filter so as to remove noise interference and improve the data quality; the normalization module maps data in different ranges to a unified numerical range through a linear transformation method so as to reduce the sensitivity of an algorithm to the absolute value of the data; the error recognition module recognizes and marks abnormal points of the data through an abnormality detection algorithm so as to facilitate subsequent processing and repair; the data restoration module comprises an interpolation unit and a smoothing unit; the interpolation unit fills up the missing value or the abnormal value through a linear interpolation method and a polynomial interpolation method so as to restore the continuity of the data; the smoothing unit reduces noise and jitter in the data through a moving average and exponential smoothing operation to improve data quality.
As a further technical scheme of the invention, the anomaly detection algorithm analyzes anomaly point data through a probability density function to identify outliers and abnormal modes inconsistent with normal behavior, wherein the probability density function identifies abnormal events by comparing probability densities of data samples with a preset threshold, and a formula expression of the probability density function is as follows:
（1）
In the case of the formula (1),representing probability density functions, +.>Is the average value of the data,/>Is the standard deviation of the data, e is the base of the natural logarithm, pi is the circumference ratio; after the data analysis is completed, the anomaly detection algorithm learns a low-dimensional representation of the data by an auto-coding function and reconstructs the input data; the anomaly detection algorithm identifies anomaly data by comparing errors between the original data and the reconstructed data of the automatic encoder; the formula expression of the auto-code function is as follows:
（2）
in formula (2), N represents an auto-coding function, i represents input data of an auto-coder, b is a representation of a hidden layer, z represents reconstructed data,and->Is a bias vector, ++>Representing an activation function->Representing a reconstruction error; the reconstruction of the input data is completed, the difference between the current sample and the previous accumulated value is compared through an accumulation and statistics formula, a detection statistic is calculated, and when the difference exceeds a preset threshold value, the detection statistic is judged to be abnormal, so that the abnormal detection is carried out on the time sequence data; the cumulative and statistical formula expression is:
（3）
in formula (3), R represents an accumulation and statistics formula, P represents a detection statistic, S represents an expected average value, and m represents a control parameter.
As a further technical scheme of the invention, the characteristic extraction model comprises a time sequence analysis module, a frequency domain analysis module and a wavelet transformation analysis module; the time sequence analysis module comprises a discrete analysis unit, a correlation analysis unit and a window statistics unit; the discrete analysis unit calculates the mean value and variance of the monitoring data through a mathematical statistical method so as to describe the data concentration degree and the discrete degree; the correlation analysis unit analyzes the correlation of the monitoring data in different hysteresis phases through an autocorrelation and partial autocorrelation analysis method so as to acquire the periodicity and the trend of the time sequence; the window statistics unit acquires the average value, the median, the maximum value and the minimum value of the data in the sliding window through a mathematical statistics method so as to extract the time sequence characteristics; the frequency domain analysis module comprises a signal conversion unit and an energy distribution analysis unit; the signal conversion unit converts the time domain signal into a frequency domain signal through Fourier transformation; the energy distribution analysis unit acquires energy distribution characteristics on a frequency band in a frequency domain through a power distribution analysis method; the wavelet transformation analysis module decomposes the time domain signal through a wavelet decomposition method to obtain sub-signals with different scales and frequencies so as to extract the characteristics of the signal in the time-frequency field.
As a further technical scheme of the invention, the working principle mode of the data synthesis method comprises the following steps:
s1, acquiring parameters and state data of a power line in real time through a sensor network, wherein the parameters and state data at least comprise current, voltage, temperature, humidity, power and frequency parameter data;
s2, extracting and analyzing characteristics of the collected data through a data processing and statistical analysis method;
s3, constructing a fault mode model through a random forest algorithm;
s4, introducing noise, variation or disturbance on the basis of a real fault sample by using the established fault mode model through a probability distribution generator and a random sampling method so as to generate a diversified virtual fault sample;
s5, marking the synthesized virtual fault samples by a data marking method so as to distinguish the virtual fault samples from the real fault samples;
s6, merging the generated virtual fault samples with the real fault samples through a data set merging method so as to increase the diversity and the number of the fault samples.
As a further technical scheme of the invention, the fault identification model comprises a detection diagnosis module, a fault positioning module and a decision interpretation module; the detection and diagnosis module comprises a fault data preprocessing unit, a feature extraction unit and a pattern recognition unit; the fault data preprocessing unit performs denoising and filtering processing on the high-voltage power line acquired data through a digital signal filter so as to improve the accuracy of fault diagnosis; the characteristic extraction unit extracts characteristic information in fault data through a signal processing method, wherein the characteristic information at least comprises overvoltage, overcurrent and frequency migration data during fault; the pattern recognition unit matches the extracted fault characteristics with patterns in the existing fault library through a matching recognition method so as to determine the type of the fault; the fault positioning module comprises a transmission line parameter identification unit, a power system state estimation unit and a fault positioning unit; the transmission line parameter identification unit analyzes line parameters of the power system through a least square method and a frequency domain response method to determine a transmission line model; the line parameters at least comprise a resistance parameter and a reactance parameter; the power system state estimation unit estimates each node in the power network through an extended Kalman filter and a weighted least square method; the extended Kalman filtering iteratively updates the estimated value of the state of the power system through a state equation and an observation equation; the weighted least square method estimates the states of all nodes of the power system through observation data and a state equation; the fault locating unit locates the position of the fault point on the power line through an inversion model; the decision interpretation module comprises an interpretation analysis unit and an interpretation rule unit; the feature importance interpretation unit interprets the weight of the features in the model output result by a feature weight analysis and statistical analysis method so as to explain the model judgment basis; the interpretation rule unit interprets the decision process and the reasons of the model through a decision tree analysis and rule extraction method.
As a further technical scheme of the invention, the real-time monitoring system processes and analyzes the collected monitoring data through a stream processing engine and a real-time data analysis method; the stream processing engine comprises a data input unit, a data processing unit, a dividing and limiting unit, a state management unit, a data output unit and a fault tolerance recovery unit; the data input unit receives and processes the monitoring data acquired in real time through a data source adapter, a message queue and a data stream management method; the data processing unit processes and converts input data in real time through a streaming data processing framework so as to execute calculation tasks; the dividing and limiting unit divides the data stream into windows with fixed sizes through a time window, a length window and a sliding window so as to perform stream calculation; the state management unit maintains and manages intermediate states and data caches in the stream processing process through a memory database, a key value storage and a distributed state management method; the data output unit outputs and transmits the processed data through a message queue, a database and a file system; the fault-tolerant recovery unit processes faults and abnormal conditions through a check point mechanism, a fault recovery algorithm and a data replay method so as to ensure the stability and the reliability of the system; the real-time monitoring system automatically analyzes and diagnoses the detected faults through a fault response mechanism to provide suggestions and decision support; the fault response mechanism realizes automatic analysis and diagnosis through a rule engine and an expert system so as to improve the accuracy and speed of fault detection.
As a further technical scheme of the invention, the data visualization module comprises a data conversion unit, a visual design unit, a data binding unit, a visual presentation unit and an interactive feedback unit; the data conversion unit converts the original data into numerical value type, text type and time sequence format through normalization processing so as to facilitate visual display; the visual design unit performs interface design through a chart library and an interactive design method so as to display monitoring data and fault detection results and provide interactive operation to enhance user experience; the data binding unit comprises a data connection subunit and a data mapping subunit; the data connection subunit connects the data source with the visual component through a database query and application program API calling method so as to ensure the updating and synchronous display of the data; the data mapping subunit associates each attribute of the data with the visual attribute of the visual component through data attribute mapping and data association rules so as to realize visual chart display of the data; the visual presentation unit comprises a chart drawing subunit and an animation effect subunit; the chart drawing subunit converts the data into a graph through a vector graph drawing method; the animation effect subunit realizes the dynamic change effect of the visual chart through a time sequence animation model so as to enhance interactivity and attraction; the interactive feedback unit realizes the interactive operation of the user and the visual chart through a mouse event recorder and a touch interaction method, and provides feedback and prompt information through a feedback prompt method according to the interactive behavior of the user; the prompt information at least comprises data details, fault warning and abnormal marking information.
According to the invention, by introducing a data quality management mechanism, the data transmission process and the storage condition are monitored and recorded, the integrity, the accuracy and the timeliness of the monitored data are ensured, and the data quality problem is solved; generating a virtual fault sample by a synthetic data method, expanding the scale of a training set, and enhancing the generalization capability of a fault early warning model so as to improve the accuracy and reliability of the model under the condition of limited actual fault samples; by introducing a signal processing model, filtering, normalizing and other operations are performed on the data, so that noise is eliminated, and the data quality is improved; then capturing dynamic changes and fault behaviors of the power line by utilizing a characteristic extraction model and using methods such as time sequence analysis, frequency domain analysis, wavelet transformation and the like; the problem of difficult feature extraction is solved; performing fault detection and prediction on the power line through a fault early warning model, and analyzing historical data and real-time data by combining a time sequence analysis method so as to find potential faults; the problem of easy missed detection and false alarm is solved; and diagnosing and positioning the detected faults through a fault identification model, and explaining the judgment basis and decision making process of the model. By utilizing the feature importance analysis and feature influence measurement method, the reliability and interpretation of the algorithm are improved, so that a user can understand the judgment result of the model and make a corresponding decision; the problem of weak algorithm interpretation is solved.
Drawings
For a clearer description of embodiments of the invention or of solutions in the prior art, the drawings that are necessary for the description of the embodiments or of the prior art will be briefly described, it being apparent that the drawings in the description below are only some embodiments of the invention, from which, without inventive faculty, other drawings can be obtained for a person skilled in the art, in which:
FIG. 1 is a schematic diagram of the steps in the process of the present invention;
FIG. 2 is a schematic diagram of the operation of the signal processing model of the present invention;
fig. 3 is a diagram illustrating an operation mode of the real-time monitoring system according to the present invention.
Description of the embodiments
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1-3, a high-voltage power line operation fault detection method based on big data analysis comprises the following steps:
Step one, current, voltage, temperature, humidity, power and frequency parameters of a power line are collected in real time through a dynamic collection module;
monitoring and recording the data transmission process and the storage condition through a data quality management mechanism so as to ensure the integrity, the accuracy and the timeliness of the data;
step three, filtering and normalizing the data through a signal processing model to eliminate noise and improve the data quality; the signal processing model identifies and processes abnormal points in the data through an abnormal detection algorithm, and repairs the abnormal data through an interpolation and smoothing method;
step four, performing feature extraction operation on the monitoring data output by the signal processing model through the feature extraction model; the feature extraction model captures dynamic changes and fault behaviors of the power line through time sequence analysis, frequency domain analysis and wavelet transformation analysis methods;
generating a virtual fault sample by a synthetic data method to expand the scale of a training set and enhance the generalization capability of a fault early warning model;
step six, performing fault detection and prediction on the power line through a fault early warning model; the fault early warning model analyzes historical data and real-time data of the power line through a time sequence analysis method to find potential faults;
Step seven, diagnosing, positioning and explaining the fault cause of the detected fault through a fault identification model; the fault identification model interprets the judgment basis and decision process of the model through a feature importance analysis and feature influence measurement method so as to improve the reliability and the interpretability of the algorithm;
step eight, periodically collecting, processing and analyzing monitoring data of the power line through a real-time monitoring system, wherein the real-time monitoring system sends out a warning to a fault through a real-time communication and voice information broadcasting method;
and step nine, displaying the monitoring data and the fault detection result through a data visualization module so as to facilitate the understanding and decision of a user.
In the above embodiment, the data quality management mechanism includes a data acquisition quality monitoring module, a data cleaning and checking module, and a data quality evaluation module; the data acquisition quality monitoring module comprises a data transmission monitoring unit and a data storage monitoring unit; the data transmission monitoring unit verifies the abnormal condition of the data in the transmission process by an error checking method so as to ensure the integrity and timeliness of the data transmission; the data storage monitoring unit monitors the integrity and the uniqueness of the data in the storage process through a data check and deduplication method so as to avoid the occurrence of duplicate data; the data cleaning and checking module comprises an abnormal value detection unit, a missing value processing unit and a data accuracy checking unit; the abnormal value detection unit detects and processes the data through a box diagram method so as to ensure the accuracy and the reliability of the data; the missing value processing unit fills missing data through an interpolation method and a regression model so as to avoid influencing the data analysis result; the data accuracy checking unit checks the collected monitoring data through the hash codes so as to prevent the problems of error marking, repeated sampling and noise interference; the data quality evaluation module comprises an evaluation index definition unit and an evaluation algorithm unit; the evaluation index definition unit defines data quality evaluation indexes through an expert system and a rule engine, wherein the data quality evaluation indexes at least comprise data integrity, accuracy, consistency and reliability; the evaluation algorithm unit evaluates and analyzes the monitoring data through a statistical method.
In a specific embodiment, the data quality management mechanism monitors quality conditions in the data acquisition process through the data acquisition quality monitoring module. The system uses various sensors and acquisition equipment to acquire parameters such as current, voltage, temperature, humidity, power, frequency and the like of a power line, and monitors the state and performance of the data acquisition equipment. By monitoring the operation condition of the acquisition equipment in real time, the problems which can affect the data quality, such as sensor faults or abnormal conditions in data transmission, can be found and solved in time. And the data cleaning and checking module cleans and checks the collected monitoring data, so that the accuracy and the integrity of the data are ensured. The data cleansing check module detects outliers, missing values, and duplicate values in the data and attempts to repair or delete these erroneous data. At the same time, the module also verifies the data, such as verifying whether the range, unit and format of the data meets expectations. Through cleaning and checking, noise and abnormality in the data can be eliminated, and the quality of the data is improved. The quality of the data is evaluated by a data quality evaluation module and related quality indicators and reports are generated. The data quality evaluation module evaluates the data according to the accuracy, the integrity, the consistency, the timeliness and the like of the data and gives corresponding quality scores. Based on the evaluation result, the data acquisition and processing process can be further optimized, and the quality of the data is improved.
In a specific implementation, the high voltage power line operation fault detection data quality management test data are shown in table 1:
table 1 high voltage power line operation fault detection data quality management test table
In data table 1, the "test item" column represents different test samples, while the "data item 1" through "data item 4" columns are the data items recorded for each test sample.
In the specific implementation, the problems of data loss or damage can be found and solved in time by monitoring and recording the data transmission process and the storage condition, the data integrity is ensured, and the influence of information loss on fault detection is avoided. Meanwhile, the data is cleaned and checked through the data cleaning and checking module, so that abnormal values and error data can be removed, the accuracy of the data is improved, and the reliability of fault detection is improved. In addition, the problem of delay or interruption of data transmission is found and solved in time by monitoring the data transmission process, so that the timeliness of the data is ensured, and the fault detection can respond in the shortest time.
In the above embodiment, the signal processing model includes a filtering module, a normalizing module, an error identifying module, and a data repairing module; the filtering module carries out filtering operation on the acquired data through a digital filter so as to remove noise interference and improve the data quality; the normalization module maps data in different ranges to a unified numerical range through a linear transformation method so as to reduce the sensitivity of an algorithm to the absolute value of the data; the error recognition module recognizes and marks abnormal points of the data through an abnormality detection algorithm so as to facilitate subsequent processing and repair; the data restoration module comprises an interpolation unit and a smoothing unit; the interpolation unit fills up the missing value or the abnormal value through a linear interpolation method and a polynomial interpolation method so as to restore the continuity of the data; the smoothing unit reduces noise and jitter in the data through a moving average and exponential smoothing operation to improve data quality.
In a specific embodiment, the signal processing model performs filtering processing on the acquired signals through a filtering module so as to eliminate noise and interference and improve data quality. The filtering module can adopt different filtering algorithms, such as low-pass filtering, median filtering and the like, and selects the most suitable filtering method according to actual conditions. By filtering operation, high-frequency noise and abnormal fluctuation in the signals can be effectively reduced, and more stable and reliable signals can be extracted. And the normalization module is used for carrying out normalization processing on the acquired signals, so that the comparability between different signals is realized. Normalization may map the range of values of the signal to specific intervals, such as [0, 1] or [ -1, 1], to eliminate dimensional differences between the data. Thus, the numerical values of different signals can be intuitively compared and analyzed, and the interpretability and the analysis effect of the data are improved. An error or abnormal condition in the data is detected and identified by an error identification module. The error recognition module employs various data analysis and statistical methods to detect outliers, missing values, or inconsistencies in the data. By identifying erroneous data, the effect of these errors on the fault detection result can be avoided and corresponding processing or correction can be made. And repairing or filling the identified errors through a data repairing module. When erroneous data is found, the data repair module may use interpolation, regression repair methods to recover missing or anomalous data. The repaired data can better reflect the actual situation and keep the integrity and continuity of the data. In a specific implementation, test data for processing a high voltage power line signal is shown in table 2:
Table 2 high voltage power line signal processing test data table
In data table 2, each row represents a signal including sequence number, signal type, raw value, filtered value, normalized value, error identification, and data repair.
In the implementation, the application of the filtering module can eliminate noise and interference in signals, improve the data quality and enable the fault detection result to be more accurate and reliable. In addition, the use of the normalization module can eliminate the dimension difference between the data, so that the comparability between different signals is realized, and the comprehensive analysis and comparison are convenient. The error recognition module can timely find errors or abnormal conditions in the data, and the data repair module can repair the errors, so that the integrity and the accuracy of the data are ensured. Thus, the influence of error data on the fault detection result can be avoided, and the reliability of fault detection is improved.
In summary, the signal processing model plays a key role in the high-voltage power line operation fault detection method based on big data analysis, and can improve the data quality, reduce noise and interference and improve the accuracy and reliability of fault detection through operations such as filtering, normalization, error identification, data restoration and the like.
In the above embodiment, the anomaly detection algorithm analyzes the outlier data by using a probability density function to identify outliers and anomaly patterns inconsistent with normal behavior, wherein the probability density function identifies anomaly events by comparing probability densities of data samples with a preset threshold, and the probability density function has the following formula:
（1）
in the case of the formula (1),representing probability density functions, +.>Is the average value of the data,/>Is the standard deviation of the data, e is the base of the natural logarithm, pi is the circumference ratio; after the data analysis is completed, the anomaly detection algorithm learns a low-dimensional representation of the data by an auto-coding function and reconstructs the input data; the anomaly detection algorithm identifies anomaly data by comparing errors between the original data and the reconstructed data of the automatic encoder; the formula expression of the auto-code function is as follows:
（2）
in formula (2), N represents an auto-coding function, i represents input data of an auto-coder, b is a representation of a hidden layer, z represents reconstructed data,and->Is a bias vector, ++>Representing an activation function->Representing a reconstruction error; the reconstruction of the input data is completed, the difference between the current sample and the previous accumulated value is compared through an accumulation and statistics formula, a detection statistic is calculated, and when the difference exceeds a preset threshold value, the detection statistic is judged to be abnormal, so that the abnormal detection is carried out on the time sequence data; the cumulative and statistical formula expression is:
（3）
In formula (3), R represents an accumulation and statistics formula, P represents a detection statistic, S represents an expected average value, and m represents a control parameter.
In a specific embodiment, in the high-voltage power line operation fault detection method based on big data analysis, an abnormality detection algorithm is used for identifying and processing abnormal points in data. These outliers may be data anomalies due to instrument failure, human error, or other causes, which can have an impact on the failure detection results. Therefore, these outliers must be identified and processed.
The anomaly detection algorithm mainly comprises two types of methods, namely a statistical-based method and a machine learning-based method. The method based on statistics is commonly used in 3 sigma principle, box type diagram and the like; the method based on machine learning is commonly used as a k-means algorithm, an LOF algorithm and the like.
In a specific embodiment, the principle of the working mode of the anomaly detection algorithm is to model the existing historical data so as to judge whether the new input data falls within the range of the model. If the deviation of the new input data from the existing data is too large, the data is considered as an outlier. Depending on the nature and cause of the outlier, appropriate processing methods such as deletion, replacement, or repair thereof may be adopted.
In the high-voltage power line operation fault detection method based on big data analysis, the abnormal point in the data can be accurately identified and processed by the abnormal detection algorithm, the quality and the precision of the data can be improved, and the subsequent analysis and processing are more accurate and useful. Abnormal points are often one of the important causes of faults, and can be found and processed in time through analysis of an abnormal detection algorithm, so that the accuracy and reliability of fault detection are improved. In addition, the abnormal detection algorithm can eliminate error data caused by instrument faults, manual misoperation and other reasons, effectively reduces the error rate and improves the accuracy of the judging result.
In specific implementation, test data of the anomaly detection algorithm for detecting the anomaly point of the high-voltage power line is shown in table 3:
TABLE 3 high voltage power line outlier detection test data
In the data table 3, each row represents a signal including a sequence number, a signal type, an original value, and an abnormality detection result. The abnormality detection result may be marked as "normal" or "abnormal" for indicating whether the signal is detected as an abnormal point.
In the above embodiment, the feature extraction model includes a time sequence analysis module, a frequency domain analysis module, and a wavelet transform analysis module; the time sequence analysis module comprises a discrete analysis unit, a correlation analysis unit and a window statistics unit; the discrete analysis unit calculates the mean value and variance of the monitoring data through a mathematical statistical method so as to describe the data concentration degree and the discrete degree; the correlation analysis unit analyzes the correlation of the monitoring data in different hysteresis phases through an autocorrelation and partial autocorrelation analysis method so as to acquire the periodicity and the trend of the time sequence; the window statistics unit acquires the average value, the median, the maximum value and the minimum value of the data in the sliding window through a mathematical statistics method so as to extract the time sequence characteristics; the frequency domain analysis module comprises a signal conversion unit and an energy distribution analysis unit; the signal conversion unit converts the time domain signal into a frequency domain signal through Fourier transformation; the energy distribution analysis unit acquires energy distribution characteristics on a frequency band in a frequency domain through a power distribution analysis method; the wavelet transformation analysis module decomposes the time domain signal through a wavelet decomposition method to obtain sub-signals with different scales and frequencies so as to extract the characteristics of the signal in the time-frequency field.
In a specific embodiment, the working principle of the feature extraction model may be subdivided into the following steps:
first, the high-voltage power line data originally collected is preprocessed. This includes removing noise, filling in missing values, data normalization, etc. The purpose is to ensure the quality and consistency of the data.
Next, the feature extraction model performs a time series analysis on the preprocessed data. It computes a series of statistical features such as mean, variance, slope, etc. These features can reflect the overall trend and degree of variation of the power line signal.
The feature extraction model also converts the time-series data into the frequency domain and analyzes the frequency content of the signal by performing fourier transform or power spectral density estimation. In frequency domain analysis, the model extracts frequency domain features, such as spectrum peaks, band energies, etc., to describe the distribution of signals at different frequencies.
In addition, the feature extraction model applies wavelet transformation to perform time-frequency domain analysis on the signal to capture time-varying features of the signal. By selecting the appropriate wavelet basis functions and scales, the model can decompose the signal into wavelet coefficients of different scales and frequency bands. From which response characteristics, such as energy, amplitude, frequency, etc., are extracted.
After extracting a plurality of features, the feature extraction model also performs feature selection and dimension reduction operations to reduce redundancy and dimension of the features. Common methods include Principal Component Analysis (PCA), linear Discriminant Analysis (LDA), and the like. Thus, the calculation efficiency of the subsequent fault detection algorithm can be improved, and the problem of over-fitting is avoided.
Finally, the feature extraction model provides the extracted features as inputs to a model of a fault detection algorithm or other related task. These features enable a better description of the operational status and potential fault conditions of the high voltage power line, supporting subsequent data analysis and decision making processes.
In a specific implementation, the feature extraction model can extract representative features from the raw data, and the features can more accurately reflect the running state of the power system. By analyzing these characteristics, faults of the high-voltage power line can be diagnosed and detected more quickly and accurately. Meanwhile, the feature extraction model can reduce the dimension of the original data, reduce the burden of data storage and processing, and can rapidly extract the data features when a large amount of data is analyzed, thereby improving the data processing efficiency. In addition, the representative features extracted by the feature extraction model can be used for constructing a more accurate machine learning model, so that the accuracy and the efficiency of a fault detection algorithm are improved.
In a specific implementation, test data for feature extraction of a high voltage power line by a feature extraction model is shown in table 4:
table 4 high voltage power line characteristic extraction test data table
In the data table 4, each row represents a signal including a sequence number, a signal type, and extracted time-series, frequency-domain, and wavelet transform features.
In general, the feature extraction model performs feature extraction on high-voltage power line data by means of time sequence analysis, frequency domain analysis, wavelet transformation analysis and the like. By extracting representative characteristics, the model can more accurately reflect the running state of the power line, and the accuracy and efficiency of fault detection are improved
In the above embodiment, the working principle manner of the data synthesis method includes the following steps:
s1, acquiring parameters and state data of a power line in real time through a sensor network, wherein the parameters and state data at least comprise current, voltage, temperature, humidity, power and frequency parameter data;
s2, extracting and analyzing characteristics of the collected data through a data processing and statistical analysis method;
s3, constructing a fault mode model through a random forest algorithm;
S4, introducing noise, variation or disturbance on the basis of a real fault sample by using the established fault mode model through a probability distribution generator and a random sampling method so as to generate a diversified virtual fault sample;
s5, marking the synthesized virtual fault samples by a data marking method so as to distinguish the virtual fault samples from the real fault samples;
s6, merging the generated virtual fault samples with the real fault samples through a data set merging method so as to increase the diversity and the number of the fault samples.
In a particular embodiment, in a high voltage power line operation fault detection method based on big data analysis, the synthetic data method generates synthetic samples that are similar to but not exactly the same as the real data by applying various transformations and operations on the real data set. The synthetic samples can be used for expanding the scale of a training set, enhancing the generalization capability of a fault early-warning model and improving the recognition capability of the model to different types of faults.
In a specific implementation, the method for synthesizing the data generates new data which is related to the original data but not completely repeated by performing operations such as rotation, overturning, scaling, stretching and the like on the real data. These new data can increase the diversity of the data set and make up for the problem of insufficient sample of the real data set. Meanwhile, the method for synthesizing the data utilizes the existing pre-training model to carry out fine adjustment or pruning operation on the real data, changes the neural network structure and further improves the generalization capability of the model. In addition, virtual fault samples are generated from the noise data by generating an countermeasure network. These virtual samples can be used to enhance the ability of the model to detect faults.
In a specific implementation, for a high-voltage power line operation fault detection model, the scale and diversity of a data set are critical to the accuracy and generalization capability of the model. The use of synthetic data methods can increase the number and diversity of samples, thereby improving the effectiveness of the model. In addition, the number and diversity of samples on the real data set are limited, so that the problem of over-fitting of the model is easy to occur, and the generalization capability is poor. By the method for synthesizing the data, more virtual samples which are similar to the real data but are not identical to the real data can be generated, so that the model is better adapted to different conditions, and the generalization capability of the model is improved. Second, in high voltage power line operation fault detection models, the number of specific types of fault samples is small, making it difficult for the model to learn these types of fault characteristics. More virtual fault samples can be generated by a data synthesis method and added into a training set, so that the detection capability of the model on specific types of faults is improved.
In the above embodiment, the fault identification model includes a detection diagnosis module, a fault location module, and a decision interpretation module; the detection and diagnosis module comprises a fault data preprocessing unit, a feature extraction unit and a pattern recognition unit; the fault data preprocessing unit performs denoising and filtering processing on the high-voltage power line acquired data through a digital signal filter so as to improve the accuracy of fault diagnosis; the characteristic extraction unit extracts characteristic information in fault data through a signal processing method, wherein the characteristic information at least comprises overvoltage, overcurrent and frequency migration data during fault; the pattern recognition unit matches the extracted fault characteristics with patterns in the existing fault library through a matching recognition method so as to determine the type of the fault; the fault positioning module comprises a transmission line parameter identification unit, a power system state estimation unit and a fault positioning unit; the transmission line parameter identification unit analyzes line parameters of the power system through a least square method and a frequency domain response method to determine a transmission line model; the line parameters at least comprise a resistance parameter and a reactance parameter; the power system state estimation unit estimates each node in the power network through an extended Kalman filter and a weighted least square method; the extended Kalman filtering iteratively updates the estimated value of the state of the power system through a state equation and an observation equation; the weighted least square method estimates the states of all nodes of the power system through observation data and a state equation; the fault locating unit locates the position of the fault point on the power line through an inversion model; the decision interpretation module comprises an interpretation analysis unit and an interpretation rule unit; the feature importance interpretation unit interprets the weight of the features in the model output result by a feature weight analysis and statistical analysis method so as to explain the model judgment basis; the interpretation rule unit interprets the decision process and the reasons of the model through a decision tree analysis and rule extraction method.
In a specific embodiment, in the high-voltage power line operation fault detection method based on big data analysis, the fault identification model performs fault detection and diagnosis on input power line operation data through a detection diagnosis module. The detection and diagnosis module uses advanced machine learning algorithm and pattern recognition technology to judge whether a fault exists or not by comparing the matching degree of input data and known fault patterns, and further determines the fault type. And determining the specific position of the fault when the fault is found by the fault positioning module. The fault location module utilizes information such as sensor data, a power line topological structure and the like to estimate or locate a fault position by combining an advanced algorithm such as Kalman filtering, particle filtering and the like. And interpreting the fault detection and positioning results through a decision interpretation module and providing decision support. The method can generate corresponding decision interpretation information, such as repair suggestions, emergency measures and the like, according to the detected fault type and position and by combining with preset rules and a knowledge base, so as to help operation and maintenance personnel to make proper decisions.
In a specific implementation, the principle of the fault identification model work is to train and learn a large amount of real fault data and normal operation data, extract features from the real fault data and build a model. And matching the input power line operation data with the existing fault modes through the input-output mapping relation of the model so as to realize the detection, diagnosis and positioning of the fault, and providing explanation and decision support for operation and maintenance personnel by combining a decision interpretation module.
The fault identification model can carry out deep analysis on the power line operation data by using advanced machine learning and pattern recognition technology, so that potential faults can be accurately identified and detected, and accidents or power failure caused by failure not being found in time can be avoided.
The test data for the fault identification model in the high-voltage power line operation fault detection method based on big data analysis are shown in table 5:
table 5 fault discrimination model test data table
In the data table 5, the failure diagnosis results by the model for the detected abnormal data points are displayed by the failure diagnosis data. In an example, the fault diagnosis results include both "short" and "ground fault" possibilities.
To better address the detected fault, the model also needs to locate the fault, i.e., indicate where the fault occurred based on the data points for which the anomaly was detected. In the data table 5, the fault location result includes the tower number and the wire number at which the fault occurred.
Further analysis and interpretation is required for the diagnostic results given in the model. In the data table 5, the decision interpretation results may include specific causes of the failure, such as loosening of the wire joint screw, aging of the wire insulation, and the like.
In the above embodiment, the real-time monitoring system processes and analyzes the collected monitoring data through a stream processing engine and a real-time data analysis method; the stream processing engine comprises a data input unit, a data processing unit, a dividing and limiting unit, a state management unit, a data output unit and a fault tolerance recovery unit; the data input unit receives and processes the monitoring data acquired in real time through a data source adapter, a message queue and a data stream management method; the data processing unit processes and converts input data in real time through a streaming data processing framework so as to execute calculation tasks; the dividing and limiting unit divides the data stream into windows with fixed sizes through a time window, a length window and a sliding window so as to perform stream calculation; the state management unit maintains and manages intermediate states and data caches in the stream processing process through a memory database, a key value storage and a distributed state management method; the data output unit outputs and transmits the processed data through a message queue, a database and a file system; the fault-tolerant recovery unit processes faults and abnormal conditions through a check point mechanism, a fault recovery algorithm and a data replay method so as to ensure the stability and the reliability of the system; the real-time monitoring system automatically analyzes and diagnoses the detected faults through a fault response mechanism to provide suggestions and decision support; the fault response mechanism realizes automatic analysis and diagnosis through a rule engine and an expert system so as to improve the accuracy and speed of fault detection.
In a specific embodiment, the working mode principle of the real-time monitoring system is as follows:
1. and (3) data acquisition: data acquisition is performed using sensors and monitoring devices. These devices can measure current, voltage, temperature, vibration, etc. in real time. The sensor may convert the acquired data into a computer readable form via an analog signal or a digital signal.
2. And (3) data transmission: and transmitting the collected monitoring data to a real-time monitoring system through a communication network. This may use various communication protocols and techniques, such as Ethernet, wireless communication (e.g., wi-Fi, bluetooth, LTE), etc. The security and integrity of the data need to be ensured during the transmission process.
3. Stream processing engine: the real-time monitoring system uses a stream processing engine for real-time processing and analysis of data. The stream processing engine is capable of receiving and processing a continuous stream of data, and performing computations in a real-time and efficient manner. It can respond to the arrival of data based on an event driven model and provide low latency processing.
4. Real-time data analysis: in the stream processing engine, various data analysis algorithms and models are applied to analyze incoming data in real time. These algorithms may include statistical analysis, machine learning, timing analysis, and the like. The real-time data analysis can detect abnormal conditions, identify trends and build a model to predict future states, thereby helping to realize fault diagnosis and early warning.
5. Fault diagnosis and early warning: based on the result of the real-time data analysis, the real-time monitoring system can quickly identify potential fault conditions and generate corresponding diagnosis information and early warning signals. This may be determined by setting a threshold, rule or model to determine if an abnormal situation has occurred and triggering a corresponding alarm mechanism.
6. Notification and response: once a fault or abnormal condition occurs, the real-time monitoring system can timely send a notification to operation and maintenance personnel and take corresponding response measures. The notification may be sent to the relevant person by way of a short message, email, instant message, etc. Responsive measures may include emergency maintenance, line switching, manual inspection, etc., to address or mitigate potential fault effects.
In the high-voltage power line operation fault detection method based on big data analysis, the real-time monitoring system can rapidly process and analyze a large amount of real-time data, so that early detection of potential faults is realized, and the fault detection efficiency is improved. In addition, through real-time data analysis, the real-time monitoring system can quickly find out abnormal states of the line and timely send out early warning signals, so that operation and maintenance personnel can be helped to take preventive measures, and serious faults are avoided. And secondly, the rapid response and accurate diagnosis capability of the real-time monitoring system can shorten the fault processing time, reduce the power failure time and reduce the power loss caused by the fault. And finally, the real-time monitoring system can remotely monitor and diagnose the running state of the circuit, improve the maintenance efficiency and reduce the inspection and maintenance cost.
In the above embodiment, the data visualization module includes a data conversion unit, a visual design unit, a data binding unit, a visual presentation unit, and an interactive feedback unit; the data conversion unit converts the original data into numerical value type, text type and time sequence format through normalization processing so as to facilitate visual display; the visual design unit performs interface design through a chart library and an interactive design method so as to display monitoring data and fault detection results and provide interactive operation to enhance user experience; the data binding unit comprises a data connection subunit and a data mapping subunit; the data connection subunit connects the data source with the visual component through a database query and application program API calling method so as to ensure the updating and synchronous display of the data; the data mapping subunit associates each attribute of the data with the visual attribute of the visual component through data attribute mapping and data association rules so as to realize visual chart display of the data; the visual presentation unit comprises a chart drawing subunit and an animation effect subunit; the chart drawing subunit converts the data into a graph through a vector graph drawing method; the animation effect subunit realizes the dynamic change effect of the visual chart through a time sequence animation model so as to enhance interactivity and attraction; the interactive feedback unit realizes the interactive operation of the user and the visual chart through a mouse event recorder and a touch interaction method, and provides feedback and prompt information through a feedback prompt method according to the interactive behavior of the user; the prompt information at least comprises data details, fault warning and abnormal marking information.
In a specific embodiment, the data visualization module first obtains the processed and analyzed data from the real-time monitoring system. Then, the data are subjected to format conversion and recombination through a data conversion unit so as to adapt to the requirements of a visual design unit. This unit may convert the data from the original format to the particular format or structure required for visualization.
At the stage of the visual design unit, the data visualization module will design a visual graphic for the data. The data visualization module may select an appropriate visualization method and chart type, such as a line graph, a bar graph, a scatter graph, etc., according to the nature and type of the data. Meanwhile, the design of visual attributes such as style, color, size and the like of the chart can be performed.
The data visualization module then associates and binds the processed data with the corresponding visualization element via the data binding unit. This means that certain attributes of the data are mapped to different parts of the visual pattern, such as X-axis, Y-axis, legend, etc. By data binding, the graph will be able to correctly represent and show the meaning and trend of the data.
In the visual presentation unit, the data visualization module graphically presents the bound data. The data visualization module may utilize a drawing library or tool to generate a final visualization chart or graph according to the rules of design and binding. These graphics may be displayed in a user interface of the monitoring system for observation and analysis by an operator or manager.
In addition, the data visualization module can also realize interaction between the user and the graph and provide interactive functions and feedback effects. For example, a user may view data for a particular period of time, zoom in and out on a view of a chart, obtain more detailed information, etc., by interacting with the chart. Interactive feedback enables users to more flexibly explore and understand data.
In the high-voltage power line operation fault detection method based on big data analysis, the data visualization module can intuitively understand the meaning and trend of the data by converting the data into a visualization graph. Visualization graphics can help them better understand complex data analysis results, finding anomalies, trends, and patterns. In addition, the visual graph can intuitively display the state and the running condition of the power line, and is helpful for quickly finding out abnormality, faults and risk points. The operator can monitor the graph in real time and take corresponding early warning and processing measures in time through the interactive feedback function. Second, the visual graphics provide a global view of data analysis, enabling operators and administrators to better understand the behavior and trends of the line. This helps them make accurate, timely decisions to optimize line operation and maintenance plans. Meanwhile, the graph generated by the data visualization module has intuitiveness and readability, and a user can easily understand and use a data result without deep knowledge of a complex data analysis algorithm and model, so that the working efficiency is improved.
In summary, in the high-voltage power line operation fault detection method based on big data analysis, the data visualization module can convert data into visual visualization graphics, so that a user can understand the data, perform fault diagnosis and early warning, support decision and improve user friendliness.
While specific embodiments of the present invention have been described above, it will be understood by those skilled in the art that these specific embodiments are by way of example only, and that various omissions, substitutions, and changes in the form and details of the methods and systems described above may be made by those skilled in the art without departing from the spirit and scope of the invention. For example, it is within the scope of the present invention to combine the above-described method steps to perform substantially the same function in substantially the same way to achieve substantially the same result. Accordingly, the scope of the invention is limited only by the following claims.
Claims (9)
1. A high-voltage power line operation fault detection method based on big data analysis is characterized in that: the method comprises the following steps:
step one, current, voltage, temperature, humidity, power and frequency parameters of a power line are collected in real time through a dynamic collection module;
monitoring and recording the data transmission process and the storage condition through a data quality management mechanism so as to ensure the integrity, the accuracy and the timeliness of the data;
Step three, filtering and normalizing the data through a signal processing model to eliminate noise and improve the data quality; the signal processing model identifies and processes abnormal points in the data through an abnormal detection algorithm, and repairs the abnormal data through an interpolation and smoothing method;
step four, performing feature extraction operation on the monitoring data output by the signal processing model through the feature extraction model; the feature extraction model captures dynamic changes and fault behaviors of the power line through time sequence analysis, frequency domain analysis and wavelet transformation analysis methods;
generating a virtual fault sample by a synthetic data method to expand the scale of a training set and enhance the generalization capability of a fault early warning model;
step six, performing fault detection and prediction on the power line through a fault early warning model; the fault early warning model analyzes historical data and real-time data of the power line through a time sequence analysis method to find potential faults;
step seven, diagnosing, positioning and explaining the fault cause of the detected fault through a fault identification model; the fault identification model interprets the judgment basis and decision process of the model through a feature importance analysis and feature influence measurement method so as to improve the reliability and the interpretability of the algorithm;
Step eight, periodically collecting, processing and analyzing monitoring data of the power line through a real-time monitoring system, wherein the real-time monitoring system sends out a warning to a fault through a real-time communication and voice information broadcasting method;
and step nine, displaying the monitoring data and the fault detection result through a data visualization module so as to facilitate the understanding and decision of a user.
2. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the data quality management mechanism comprises a data acquisition quality monitoring module, a data cleaning and checking module and a data quality evaluation module; the data acquisition quality monitoring module comprises a data transmission monitoring unit and a data storage monitoring unit; the data transmission monitoring unit verifies the abnormal condition of the data in the transmission process by an error checking method so as to ensure the integrity and timeliness of the data transmission; the data storage monitoring unit monitors the integrity and the uniqueness of the data in the storage process through a data check and deduplication method so as to avoid the occurrence of duplicate data; the data cleaning and checking module comprises an abnormal value detection unit, a missing value processing unit and a data accuracy checking unit; the abnormal value detection unit detects and processes the data through a box diagram method so as to ensure the accuracy and the reliability of the data; the missing value processing unit fills missing data through an interpolation method and a regression model so as to avoid influencing the data analysis result; the data accuracy checking unit checks the collected monitoring data through the hash codes so as to prevent the problems of error marking, repeated sampling and noise interference; the data quality evaluation module comprises an evaluation index definition unit and an evaluation algorithm unit; the evaluation index definition unit defines data quality evaluation indexes through an expert system and a rule engine, wherein the data quality evaluation indexes at least comprise data integrity, accuracy, consistency and reliability; the evaluation algorithm unit evaluates and analyzes the monitoring data through a statistical method.
3. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the signal processing model comprises a filtering module, a normalization module, an error identification module and a data restoration module; the filtering module carries out filtering operation on the acquired data through a digital filter so as to remove noise interference and improve the data quality; the normalization module maps data in different ranges to a unified numerical range through a linear transformation method so as to reduce the sensitivity of an algorithm to the absolute value of the data; the error recognition module recognizes and marks abnormal points of the data through an abnormality detection algorithm so as to facilitate subsequent processing and repair; the data restoration module comprises an interpolation unit and a smoothing unit; the interpolation unit fills up the missing value or the abnormal value through a linear interpolation method and a polynomial interpolation method so as to restore the continuity of the data; the smoothing unit reduces noise and jitter in the data through a moving average and exponential smoothing operation to improve data quality.
4. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the anomaly detection algorithm analyzes anomaly point data through a probability density function to identify outliers and abnormal modes inconsistent with normal behavior, wherein the probability density function identifies abnormal events by comparing probability densities of data samples with a preset threshold, and a formula expression of the probability density function is as follows:
（1）
In the case of the formula (1),representing probability density functions, +.>Is the average value of the data,/>Is the standard deviation of the data, e is the base of the natural logarithm, pi is the circumference ratio; after the data analysis is completed, the anomaly detection algorithm learns a low-dimensional representation of the data by an auto-coding function and reconstructs the input data; the anomaly detection algorithm identifies anomaly data by comparing errors between the original data and the reconstructed data of the automatic encoder; the formula expression of the auto-code function is as follows:
（2）
in formula (2), N represents an auto-coding function, i represents input data of an auto-coder, b is a representation of a hidden layer, z represents reconstructed data,and->Is a bias vector, ++>Representing an activation function->Representing a reconstruction error; the reconstruction of the input data is completed, the difference between the current sample and the previous accumulated value is compared through an accumulation and statistics formula, a detection statistic is calculated, and when the difference exceeds a preset threshold value, the detection statistic is judged to be abnormal, so that the abnormal detection is carried out on the time sequence data; the cumulative and statistical formula expression is:
（3）
in formula (3), R represents an accumulation and statistics formula, P represents a detection statistic, S represents an expected average value, and m represents a control parameter.
5. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the feature extraction model comprises a time sequence analysis module, a frequency domain analysis module and a wavelet transformation analysis module; the time sequence analysis module comprises a discrete analysis unit, a correlation analysis unit and a window statistics unit; the discrete analysis unit calculates the mean value and variance of the monitoring data through a mathematical statistical method so as to describe the data concentration degree and the discrete degree; the correlation analysis unit analyzes the correlation of the monitoring data in different hysteresis phases through an autocorrelation and partial autocorrelation analysis method so as to acquire the periodicity and the trend of the time sequence; the window statistics unit acquires the average value, the median, the maximum value and the minimum value of the data in the sliding window through a mathematical statistics method so as to extract the time sequence characteristics; the frequency domain analysis module comprises a signal conversion unit and an energy distribution analysis unit; the signal conversion unit converts the time domain signal into a frequency domain signal through Fourier transformation; the energy distribution analysis unit acquires energy distribution characteristics on a frequency band in a frequency domain through a power distribution analysis method; the wavelet transformation analysis module decomposes the time domain signal through a wavelet decomposition method to obtain sub-signals with different scales and frequencies so as to extract the characteristics of the signal in the time-frequency field.
6. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the working principle mode of the data synthesis method comprises the following steps:
s1, acquiring parameters and state data of a power line in real time through a sensor network, wherein the parameters and state data at least comprise current, voltage, temperature, humidity, power and frequency parameter data;
s2, extracting and analyzing characteristics of the collected data through a data processing and statistical analysis method;
s3, constructing a fault mode model through a random forest algorithm;
s4, introducing noise, variation or disturbance on the basis of a real fault sample by using the established fault mode model through a probability distribution generator and a random sampling method so as to generate a diversified virtual fault sample;
s5, marking the synthesized virtual fault samples by a data marking method so as to distinguish the virtual fault samples from the real fault samples;
s6, merging the generated virtual fault samples with the real fault samples through a data set merging method so as to increase the diversity and the number of the fault samples.
7. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the fault identification model comprises a detection diagnosis module, a fault positioning module and a decision interpretation module; the detection and diagnosis module comprises a fault data preprocessing unit, a feature extraction unit and a pattern recognition unit; the fault data preprocessing unit performs denoising and filtering processing on the high-voltage power line acquired data through a digital signal filter so as to improve the accuracy of fault diagnosis; the characteristic extraction unit extracts characteristic information in fault data through a signal processing method, wherein the characteristic information at least comprises overvoltage, overcurrent and frequency migration data during fault; the pattern recognition unit matches the extracted fault characteristics with patterns in the existing fault library through a matching recognition method so as to determine the type of the fault; the fault positioning module comprises a transmission line parameter identification unit, a power system state estimation unit and a fault positioning unit; the transmission line parameter identification unit analyzes line parameters of the power system through a least square method and a frequency domain response method to determine a transmission line model; the line parameters at least comprise a resistance parameter and a reactance parameter; the power system state estimation unit estimates each node in the power network through an extended Kalman filter and a weighted least square method; the extended Kalman filtering iteratively updates the estimated value of the state of the power system through a state equation and an observation equation; the weighted least square method estimates the states of all nodes of the power system through observation data and a state equation; the fault locating unit locates the position of the fault point on the power line through an inversion model; the decision interpretation module comprises an interpretation analysis unit and an interpretation rule unit; the feature importance interpretation unit interprets the weight of the features in the model output result by a feature weight analysis and statistical analysis method so as to explain the model judgment basis; the interpretation rule unit interprets the decision process and the reasons of the model through a decision tree analysis and rule extraction method.
8. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the real-time monitoring system processes and analyzes the collected monitoring data through a stream processing engine and a real-time data analysis method; the stream processing engine comprises a data input unit, a data processing unit, a dividing and limiting unit, a state management unit, a data output unit and a fault tolerance recovery unit; the data input unit receives and processes the monitoring data acquired in real time through a data source adapter, a message queue and a data stream management method; the data processing unit processes and converts input data in real time through a streaming data processing framework so as to execute calculation tasks; the dividing and limiting unit divides the data stream into windows with fixed sizes through a time window, a length window and a sliding window so as to perform stream calculation; the state management unit maintains and manages intermediate states and data caches in the stream processing process through a memory database, a key value storage and a distributed state management method; the data output unit outputs and transmits the processed data through a message queue, a database and a file system; the fault-tolerant recovery unit processes faults and abnormal conditions through a check point mechanism, a fault recovery algorithm and a data replay method so as to ensure the stability and the reliability of the system; the real-time monitoring system automatically analyzes and diagnoses the detected faults through a fault response mechanism to provide suggestions and decision support; the fault response mechanism realizes automatic analysis and diagnosis through a rule engine and an expert system so as to improve the accuracy and speed of fault detection.
9. The high-voltage power line operation fault detection method based on big data analysis according to claim 1, wherein the method comprises the following steps: the data visualization module comprises a data conversion unit, a visual design unit, a data binding unit, a visual presentation unit and an interactive feedback unit; the data conversion unit converts the original data into numerical value type, text type and time sequence format through normalization processing so as to facilitate visual display; the visual design unit performs interface design through a chart library and an interactive design method so as to display monitoring data and fault detection results and provide interactive operation to enhance user experience; the data binding unit comprises a data connection subunit and a data mapping subunit; the data connection subunit connects the data source with the visual component through a database query and application program API calling method so as to ensure the updating and synchronous display of the data; the data mapping subunit associates each attribute of the data with the visual attribute of the visual component through data attribute mapping and data association rules so as to realize visual chart display of the data; the visual presentation unit comprises a chart drawing subunit and an animation effect subunit; the chart drawing subunit converts the data into a graph through a vector graph drawing method; the animation effect subunit realizes the dynamic change effect of the visual chart through a time sequence animation model so as to enhance interactivity and attraction; the interactive feedback unit realizes the interactive operation of the user and the visual chart through a mouse event recorder and a touch interaction method, and provides feedback and prompt information through a feedback prompt method according to the interactive behavior of the user; the prompt information at least comprises data details, fault warning and abnormal marking information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311227707.1A CN117251812A (en) | 2023-09-22 | 2023-09-22 | High-voltage power line operation fault detection method based on big data analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311227707.1A CN117251812A (en) | 2023-09-22 | 2023-09-22 | High-voltage power line operation fault detection method based on big data analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117251812A true CN117251812A (en) | 2023-12-19 |
Family
ID=89130884
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311227707.1A Withdrawn CN117251812A (en) | 2023-09-22 | 2023-09-22 | High-voltage power line operation fault detection method based on big data analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117251812A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117540225A (en) * | 2024-01-09 | 2024-02-09 | 成都电科星拓科技有限公司 | Distributed ups system consistency assessment system and method based on DBSCAN clustering |
-
2023
- 2023-09-22 CN CN202311227707.1A patent/CN117251812A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117540225A (en) * | 2024-01-09 | 2024-02-09 | 成都电科星拓科技有限公司 | Distributed ups system consistency assessment system and method based on DBSCAN clustering |
| CN117540225B (en) * | 2024-01-09 | 2024-04-12 | 成都电科星拓科技有限公司 | Distributed ups system consistency assessment system and method based on DBSCAN clustering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113400652B (en) | 3D printer monitoring and diagnosis knowledge base device and system based on vibration signals | |
| CN110647133B (en) | Rail transit equipment state detection maintenance method and system | |
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN117251812A (en) | High-voltage power line operation fault detection method based on big data analysis | |
| CN105571638A (en) | Machinery device fault combination prediction system and method | |
| CN113657221A (en) | Power plant equipment state monitoring method based on intelligent sensing technology | |
| CN113155469A (en) | Engine fault diagnosis alarm system and device | |
| CN114781476A (en) | Fault analysis system and method for measuring equipment | |
| CN115437358A (en) | Intelligent state monitoring and fault diagnosis system and fault diagnosis method for industrial robot | |
| CN116629627A (en) | Intelligent detection system of power transmission on-line monitoring device | |
| CN111678699A (en) | Early fault monitoring and diagnosing method and system for rolling bearing | |
| DE112019007899T5 (en) | System and method for predicting device failures using acoustic signals in the ultrasonic frequency band | |
| CN117074852A (en) | Power distribution network electric energy monitoring and early warning management method and system | |
| KR102486463B1 (en) | Method and Apparatus for Real Time Fault Detection Using Time series data According to Degradation | |
| CN113487189A (en) | Petrochemical equipment fault probability risk assessment system and assessment method | |
| CN114112390A (en) | Early fault diagnosis method for nonlinear complex system | |
| CN117057676B (en) | Multi-data fusion fault analysis method, equipment and storage medium | |
| JP7278499B1 (en) | LEARNING DEVICE, ANORMAL SIGNS DETECTION DEVICE, ANORMAL SIGNS DETECTION SYSTEM, LEARNING METHOD AND PROGRAM | |
| EP4187340A1 (en) | A method and system for monitoring a device | |
| KR20210060049A (en) | Equipment failure prediction system and method using sound signal of ultrasonic band | |
| CN117434360A (en) | Switch action current curve matching alarm method, system and medium | |
| CN117740082A (en) | Power equipment fault risk early warning method and system | |
| CN114236268A (en) | Transformer fault diagnosis method and system based on machine learning | |
| Gu et al. | Fault identification for the large-scale system using trend analysis | |
| CN117791876A (en) | Substation equipment operation state monitoring method and abnormal control system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20231219 |
|
| WW01 | Invention patent application withdrawn after publication |