Search

CN-120971886-B - Artificial intelligence analysis method and system applied to traveling wave fault positioning

CN120971886BCN 120971886 BCN120971886 BCN 120971886BCN-120971886-B

Abstract

The invention discloses an artificial intelligent analysis method and an artificial intelligent analysis system applied to traveling wave fault positioning, which relate to the technical field of power fault positioning, wherein traveling wave data and standing wave data are collected through current and voltage sensors and preprocessed; the method comprises the steps of combining a space-time attention mechanism by using a convolutional neural network and a long-short-time memory network mixed model, deeply mining space-time characteristics and association complementarity of two data modes, classifying fault types into typical faults and complex faults based on fusion characteristics, respectively identifying and positioning the faults, performing typical faults by using traveling wave time difference positioning and standing wave verification, performing complex fault analysis on multi-wave head and standing wave reflection characteristic comprehensive positioning, analyzing time sequence and space association among multiple fault points when other fault points exist near the fault points, and combining the fault types, positioning and association information to assist operation and maintenance personnel in decision and diagnosis.

Inventors

  • QIAN RONG
  • LI BOYANG
  • YANG JINHAI
  • SHAN MINGLEI

Assignees

  • 江苏久创电气科技有限公司

Dates

Publication Date
20260512
Application Date
20250806

Claims (10)

  1. 1. The artificial intelligence analysis method applied to traveling wave fault location is characterized by comprising the following steps: collecting traveling wave data and standing wave data through a current sensor and a voltage sensor, and preprocessing; extracting space-time characteristics from the preprocessed traveling wave and standing wave data by using a mixed model combining a convolutional neural network and a long-short-time memory network, integrating a space-time attention mechanism, and mining the relevance and complementarity between traveling wave signals and standing wave signals to realize the deep fusion of information; Based on the fused characteristics, performing fault type identification and fault positioning by using a hybrid model, combining the zero sequence component identification in the traveling wave grounding characteristics and the standing waves to obtain the fault type of a typical fault; in the aspect of fault positioning, for a typical fault, the time difference of the initial traveling wave heads reaching measuring points at two ends is focused, the fault distance is calculated by combining line parameters, standing wave information is used as auxiliary verification, for a complex fault, the time intervals and the energy distribution of a plurality of wave heads are analyzed, the reflection characteristics of fault points reflected in the standing wave information are relied on, and the fault position is comprehensively judged; When a plurality of fault points exist near the fault position, performing relevance analysis by using a mixed model, and evaluating whether a time sequence and a spatial propagation relationship exist between fault signals or not to obtain fault relevance; based on the fault type, fault location and fault correlation, the operation and maintenance personnel are assisted in making decisions and diagnoses.
  2. 2. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein the collecting traveling wave data and standing wave data through the current sensor and the voltage sensor for preprocessing comprises: A current sensor and a voltage sensor are deployed at key nodes of the power transmission line, traveling wave transient current and voltage signals at the key nodes are measured, and voltage and current signals in a steady state and a quasi-steady state after faults are collected; Preprocessing the collected original data, carrying out data alignment and synchronous verification, utilizing GPS time stamps to adjust the time alignment of the collected data of different sensor nodes to synchronize the data in the time dimension, adopting a digital filtering technology to carry out data denoising and filtering to remove high-frequency noise and power frequency interference, using a wavelet transformation method to analyze traveling wave signals, extracting instantaneous amplitude, highlighting key characteristics and instantaneous changes related to faults in the signals, processing data loss and abnormal points, and adopting a mean filling method to correct.
  3. 3. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein the extracting space-time features from the preprocessed traveling wave and standing wave data by using a mixed model combining a convolutional neural network and a long-short-time memory network, integrating a space-time attention mechanism, excavating the relevance and complementarity between traveling wave signals and standing wave signals, and realizing the deep fusion of information comprises: The mixed model comprises a CNN branch of a convolutional neural network and an LSTM branch of a long and short-term memory network, wherein the pre-processed data is used as the input of the mixed model, the CNN branch captures the structural information of a signal in a local area, a convolutional layer operates on the time sequence of the signal through a sliding filter and automatically learns and extracts local spatial features in the signal; After the two branches extract the space features and the time sequence features respectively, the outputs of the two branches are fused and input to a space-time attention mechanism module, the importance of the features on the fault type identification and positioning is dynamically calculated and assigned with weight, the importance of the features on the time axis and the importance of the space dimension are evaluated, so that the hybrid model can automatically focus on the most relevant space-time area, noise and redundant information are restrained, and the relevance and complementarity existing between the travelling wave signals and the standing wave signals are effectively excavated.
  4. 4. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein the combining the traveling wave grounding feature and the zero sequence component identification in the standing wave to obtain the fault type of the typical fault comprises: In the fault type identification link, the hybrid model focuses on analyzing the traveling wave grounding transient characteristic reflected in the fusion characteristic and the zero sequence component information in the standing wave to distinguish typical fault types, wherein the typical fault types comprise single-phase grounding, two-phase short-circuit and three-phase short-circuit; When a single-phase grounding fault is identified, capturing an initial transient signal of a traveling wave generated by the single-phase grounding fault, wherein the initial transient signal is represented by that the voltage of the fault phase is lower than a preset single-phase grounding fault phase voltage threshold, the voltage of the non-fault phase is higher than the preset single-phase grounding non-fault phase voltage threshold, and the fault phase generates an initial traveling wave current with an amplitude exceeding the preset single-phase threshold; When the two-phase short-circuit grounding fault is identified, capturing traveling wave initial transient signals of which the two phases are short-circuited and at least one phase is grounded; the method comprises the steps of generating a traveling wave signal, wherein the traveling wave signal is represented by two fault phase voltages lower than a preset two-phase short-circuit grounding fault phase voltage threshold value, the non-fault phase voltage is higher than or equal to the preset two-phase short-circuit grounding non-fault phase voltage threshold value, analyzing traveling wave initial transient signals by a mixed model, learning and identifying specific characteristics of a current phase relation of the two fault phases under the fault condition, wherein the initial phase difference is in a preset two-phase difference threshold value interval, generating initial traveling wave current with an amplitude exceeding the preset two-phase threshold value by two phases, and having the specific phase relation, generating zero-sequence current and zero-sequence voltage by two-phase short-circuit grounding, wherein the zero-sequence component is caused by the fact that two phases are grounded simultaneously or one phase is grounded and short-circuited together with the other phase, and has identifiable amplitude and phase characteristics; When the hybrid model analyzes the fusion characteristic of the standing wave part, a zero sequence component with the amplitude lower than a preset zero sequence threshold value of the two-phase short-circuit fault or equal to zero is detected and is also a key characteristic for distinguishing, when the hybrid model detects the initial transient characteristic of two phases in the traveling wave signal which are simultaneously disturbed, and confirms that the amplitude of the zero sequence component of the standing wave is lower than the preset zero sequence threshold value of the two-phase short-circuit fault, the two-phase short-circuit fault can be determined; When three-phase short-circuit faults are identified, the hybrid model analyzes initial transient characteristics of three-phase simultaneous short-circuits reflected in fusion characteristics, traveling wave signals are represented by three-phase voltages being lower than respective preset three-phase short-circuit fault voltage thresholds, three-phase currents are deviated from respective preset three-phase short-circuit fault current normal ranges, a specific phase relation exists between the three phases, initial change starting time difference of the three-phase voltages and the three-phase currents is represented by being smaller than a preset three-phase time threshold, the change waveform similarity is higher than a preset three-phase similarity threshold, the hybrid model analyzes fusion characteristics of standing wave parts, zero sequence components with amplitudes being lower than or equal to the preset three-phase short-circuit fault zero sequence threshold are detected, and when the hybrid model detects the initial transient characteristics of the three-phase simultaneous short-circuit faults in the traveling wave signals, and confirms that the zero sequence component amplitudes in the standing wave signals are lower than the preset three-phase short-circuit fault zero sequence threshold.
  5. 5. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein the fusion analysis of information dimensions provided by a feature combination of traveling wave and standing wave manifestations, to obtain fault types of complex faults, comprises: the complex fault types include high resistance ground faults and intermittent faults; When a high-resistance ground fault is identified, the hybrid model analyzes that the amplitude of an initial transient signal of a traveling wave in a fusion characteristic is lower than a preset high-resistance ground fault traveling wave amplitude threshold value, the waveform change exceeds a preset high-resistance ground fault traveling wave waveform normal range, and the transient signal is accompanied by noise or oscillation with the amplitude lower than a preset noise threshold and the frequency higher than a preset high-frequency threshold; When intermittent faults are identified, the mixed model analyzes the time variability of traveling wave signals in fusion characteristics, the traveling wave transient signals of fault phase voltage and current are represented by the fact that the occurrence interval of the traveling wave transient signals exceeds the preset minimum interval time, the disappearance interval exceeds the preset maximum disappearance time, the amplitude exceeds the preset intermittent fault amplitude range and the polarity change exceeds the preset polarity rule change range, the stability of the arrival interval is lower than the preset intermittent fault interval stability threshold value, meanwhile, in standing wave analysis, the existence time of a zero sequence component is lower than the preset intermittent fault zero sequence component duration time threshold value, and the mixed model judges intermittent faults by capturing characteristic combinations which are not satisfied with preset continuity or stability and have obvious intermittent characteristics through capturing the traveling wave transient and standing wave zero sequence components.
  6. 6. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein regarding the fault location, regarding the typical fault, focusing on the time difference of the initial traveling wave head reaching the measuring points at two ends, calculating the fault distance in combination with the line parameters, and standing wave information as auxiliary verification, comprising: after the fault type is identified as a typical fault, the time difference T d of the initial traveling wave head extracted from the traveling wave data to reach the measuring points at the two ends of the line is utilized, and the calculation formula of the fault distance is as follows: ; wherein D is a fault distance, and represents a distance value from one end of the line to a fault point, c is a wave speed, and T d is a time difference from an initial traveling wave head to a measuring point at two ends of the line; In the process, steady-state characteristics extracted from standing wave data are used as auxiliary verification information, a traveling wave positioning result is verified by comparing with expected standing wave reflection characteristics corresponding to a fault distance D calculated based on the traveling wave, the confidence of the positioning result is improved, the fault distance D is output, the mapping relation from the fault distance D to a fault positioning coordinate point is established by combining geographical or topological information of a line, and a specific coordinate point of fault positioning is obtained through linear interpolation calculation.
  7. 7. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein for complex faults, analyzing time intervals and energy distribution of a plurality of wave heads, relying on reflection characteristics of fault points reflected in standing wave information, comprehensively judging fault positions comprises: When the fault type is identified as complex fault, the mixed model analyzes time interval sequences and corresponding energy distribution characteristics of a plurality of traveling wave heads extracted from traveling wave data reaching a measuring point; Based on the comprehensive analysis of time intervals, energy distribution and standing wave reflection characteristics of a plurality of wave heads, a unified multidimensional feature vector is formed, a mixed model establishes a mapping relation between fault positions and the multidimensional feature vector, historical fault data is used for learning feature vectors generated by high-resistance grounding and intermittent faults at different positions, the mixed model automatically learns and fits nonlinear relations between the feature vectors and the fault positions through training, the currently extracted feature vectors are input into the trained mixed model, specific fault position coordinate values are output through calculation, and the coordinate values are finally determined fault positioning coordinates, so that the high-resistance grounding and intermittent faults are positioned.
  8. 8. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein when a plurality of fault points are detected to exist near the fault location, performing correlation analysis by using a hybrid model, and evaluating whether a time sequence and a spatial propagation relationship exist between fault signals to obtain a fault correlation, including: when a plurality of fault points exist, the mixed model identifies the time relation among different fault points by analyzing the arrival time of each wave head, and determines the propagation sequence of fault signals; The mixed model utilizes a space-time attention mechanism to evaluate the propagation characteristics of fault signals in space and judge whether the propagation relationship in space exists among all fault points, wherein the propagation relationship comprises reflection, refraction and scattering phenomena of the signals; In the aspect of space analysis, the hybrid model is combined with the positions of the sensor nodes to evaluate the propagation paths, the fault distances and the fault positioning of signals, and further confirm the space relevance of different fault points; Based on the correlation analysis of time and space dimensions, the hybrid model comprehensively judges the mutual influence among a plurality of fault points to obtain the fault correlation.
  9. 9. The artificial intelligence analysis method for traveling wave fault location according to claim 1, wherein the assisting operation and maintenance personnel to make decisions and diagnoses based on fault types, fault positions and fault correlations comprises: based on the mixed model, the depth analysis of the traveling wave signal and the standing wave signal accurately identifies fault types, including single-phase grounding, two-phase short-circuit, three-phase short-circuit, high-resistance grounding and intermittent faults; by analyzing the propagation time difference of the traveling wave signals, the zero sequence component in the standing wave and the arrival time of the wave head, the operation and maintenance personnel are helped to locate fault points in the power line; when a plurality of fault points exist, identifying time and space relations among the fault points through space-time correlation analysis; the mixed model analyzes the mutual influence between fault points, judges whether the faults occur singly or have linkage effect, and assists operation and maintenance personnel to make decisions according to the relevance of the faults; based on the fault type, location and relevance, decision support and maintenance strategies are provided for operation and maintenance personnel.
  10. 10. An artificial intelligence analysis system for traveling wave fault location, using the artificial intelligence analysis method for traveling wave fault location according to any one of claims 1 to 9, comprising: The data acquisition and preprocessing module comprises a data acquisition unit and a data preprocessing unit, wherein the data acquisition unit acquires traveling wave data and standing wave data through a current sensor and a voltage sensor; The depth feature extraction and fusion module comprises a mixed model structure unit, a feature extraction unit and a feature fusion and attention mechanism unit, wherein the mixed model structure unit constructs a mixed network structure comprising a convolutional neural network CNN and a long-short-time memory network LSTM, and is respectively used for processing spatial features and time sequence features; The fault type identification module comprises a typical fault identification unit and a complex fault identification unit, wherein the typical fault identification unit is used for identifying typical fault types comprising single-phase grounding, two-phase short circuit and three-phase short circuit based on the characteristics of the fused space-time attention mechanism and focusing on the analysis of the zero sequence component information of the traveling wave grounding transient characteristic and the standing wave; The fault positioning module comprises a typical fault positioning unit and a complex fault positioning unit, wherein the typical fault positioning unit calculates a fault distance based on traveling wave time difference, verifies by standing wave data and positions the position coordinates of typical faults by combining line information; The multi-fault relevance analysis module comprises a time relation analysis unit, a space relation analysis unit and a relevance comprehensive judgment unit; the time relation analysis unit analyzes the arrival time of each wave head and identifies the time sequence among different fault points; the space relation analysis unit evaluates the propagation characteristics of fault signals in space by using a space-time attention mechanism and sensor position information and judges whether a space propagation or influence relation exists between each fault point; the relevance comprehensive judging unit synthesizes analysis results of time and space dimensions, judges the mutual influence among a plurality of fault points and obtains a fault relevance conclusion; The fault integration and decision support module comprises a fault information integration unit and a decision auxiliary unit, wherein the fault information integration unit integrates fault types, fault positions and relevance information obtained by the fault type identification module, the fault positioning module and the multi-fault relevance analysis module, and the decision auxiliary unit provides complete fault information for operation and maintenance personnel based on the integrated fault information and provides preliminary maintenance strategy suggestions.

Description

Artificial intelligence analysis method and system applied to traveling wave fault positioning Technical Field The invention relates to the technical field of power fault positioning, in particular to an artificial intelligent analysis method and an artificial intelligent analysis system applied to traveling wave fault positioning. Background The transmission line is an important component of the power system, and the safe and stable operation of the transmission line is critical to normal power supply. However, various types of faults are unavoidable in the long-term operation state of the power transmission line. When a fault occurs, traveling waves are generated at fault points, the traveling waves propagate along a line and reflect and refract at uneven points of the line to form standing waves, the traditional fault positioning technology is mainly focused on capturing and analyzing initial traveling waves, and considers establishing a correlation model with the generated standing waves and comprehensively analyzing the correlation and complementarity, the traditional technology usually only carries out unified analysis and calculation on fault types and positioning and does not carry out situation judgment according to different types, meanwhile, the lack of deep analysis on the grounding characteristics of the traveling waves and zero sequence components in the standing waves causes inaccurate results, and when other fault points exist near the fault positions, the correlation between similar faults is not considered in the prior art, so that the efficiency of fault diagnosis and treatment is affected. Disclosure of Invention The invention aims to provide an artificial intelligence analysis method and an artificial intelligence analysis system for traveling wave fault positioning, which are used for solving the problems in the prior art. In order to achieve the above purpose, the present invention provides the following technical solutions: in a first aspect, the present invention provides an artificial intelligence analysis method applied to traveling wave fault location, including: collecting traveling wave data and standing wave data through a current sensor and a voltage sensor, and preprocessing; extracting space-time characteristics from the preprocessed traveling wave and standing wave data by utilizing a mixed model combining a convolutional neural network and a long-short-time memory network, integrating a space-time attention mechanism, and mining the relevance and complementarity between the two modal characteristics to realize the deep fusion of information; Based on the fused characteristics, performing fault type identification and fault positioning by using a hybrid model, combining the zero sequence component identification in the traveling wave grounding characteristics and the standing waves to obtain the fault type of a typical fault; in the aspect of fault positioning, for a typical fault, the time difference of the initial traveling wave heads reaching measuring points at two ends is focused, the fault distance is calculated by combining line parameters, standing wave information is used as auxiliary verification, for a complex fault, the time intervals and the energy distribution of a plurality of wave heads are analyzed, the reflection characteristics of fault points reflected in the standing wave information are relied on, and the fault position is comprehensively judged; When a plurality of fault points exist near the fault position, performing relevance analysis by using a mixed model, and evaluating whether a time sequence and a spatial propagation relationship exist between fault signals or not to obtain fault relevance; based on the fault type, fault location and fault correlation, the operation and maintenance personnel are assisted in making decisions and diagnoses. With reference to the first aspect, in a first implementation manner of the first aspect of the present application, the collecting, by a current sensor and a voltage sensor, traveling wave data and standing wave data, performing preprocessing includes: A current sensor and a voltage sensor are deployed at key nodes of the power transmission line, traveling wave transient current and voltage signals at the key nodes are measured, and voltage and current signals in a steady state and a quasi-steady state after faults are collected; Preprocessing the collected original data, carrying out data alignment and synchronous verification, utilizing GPS time stamps to adjust the time alignment of the collected data of different sensor nodes to synchronize the data in the time dimension, adopting a digital filtering technology to carry out data denoising and filtering to remove high-frequency noise and power frequency interference, using a wavelet transformation method to analyze traveling wave signals, extracting instantaneous amplitude, highlighting key characteristics and instantaneous changes related to faults in t