Power sample data quality evaluation method, system and device and medium

Abstract

The embodiment of the invention provides a power sample data quality evaluation method, system and device and a medium. The method is applied to the technical field of data processing and comprises the steps of performing standardized preprocessing on a power image sample to obtain a preprocessed data set and monitoring metadata; constructing multi-dimensional features based on the preprocessed data, and obtaining a feature matrix; time-space repeatability analysis is completed, and repeatability evaluation data is formed; evaluating a scene value in combination with the feature matrix, and generating a quality feature index; establishing a quality scoring system, and calculating a quality score; and mining quality features, and outputting an evaluation result. According to the method, the repeatability, the integrity and the value of the data can be effectively identified and evaluated, so that the data acquisition and management efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of data processing, and particularly relates to a power sample data quality evaluation method, system, device and medium. BACKGROUND

[0002] In the power industry, the safety monitoring and maintenance of power transmission lines are of great importance. Currently, power companies generally use automatic monitoring devices, inspection robots, and manual inspection to collect the operating state data of power transmission lines. These monitoring devices collect image data at regular intervals according to the preset time interval, or frame saving from inspection videos. At the same time, a supporting data collection and management system is equipped to store and preliminarily analyze the collected image data.

[0003] However, the existing power monitoring data collection method has the problem of uneven data quality. Due to the fixed position shooting of the monitoring device, repeated route inspection in the inspection process, etc., a large amount of redundant or similar image data is collected. These redundant data not only occupy a large amount of storage space, but also affect the subsequent data analysis efficiency. In addition, the existing data management method lacks an effective evaluation mechanism for data quality, which cannot accurately judge the value and usability of the collected data, and cannot provide effective guidance for the optimization of data collection strategy. SUMMARY

[0004] The present disclosure provides a power sample data quality evaluation method, system, device and medium, which can effectively identify and evaluate the repeatability, integrity and value of the data to improve the efficiency of data collection and management.

[0005] According to a first aspect of the present disclosure, a power sample data quality evaluation method is provided, comprising: performing standardization preprocessing on power scene image samples to obtain a preprocessed sample data set and power transmission line monitoring metadata; performing multi-dimensional feature construction based on the preprocessed sample data set to obtain a power sample feature matrix through a feature parameter extractor; completing spatio-temporal repeatability analysis according to the power sample feature matrix and the power transmission line monitoring metadata to form sample redundancy evaluation data; using the power sample feature matrix in combination with the sample redundancy evaluation data to evaluate the scene value of power equipment monitoring content to generate quality key feature indicators; establishing a quality scoring system according to the quality key feature indicators and the sample redundancy evaluation data, and outputting sample data quality scores after hierarchical weight distribution; mining the quality characteristics of power sample data through the sample data quality scores, combining the quality key feature indicators for report output, and obtaining power data quality evaluation results.

[0006] According to a second aspect of the present disclosure, a power sample data quality evaluation system is provided, comprising:

[0007] a processing module configured to perform standardization preprocessing on the power scene image sample to obtain a preprocessed sample dataset and power transmission line monitoring metadata;

[0008] a construction module configured to construct multi-dimensional features based on the preprocessed sample dataset, and obtain a power sample feature matrix through a feature parameter extractor;

[0009] an analysis module configured to complete spatiotemporal repeatability analysis according to the power sample feature matrix and the power transmission line monitoring metadata, and form sample repeatability evaluation data;

[0010] an evaluation module configured to perform scene value evaluation on power equipment monitoring content by using the power sample feature matrix in combination with the sample repeatability evaluation data, and generate quality key feature indicators;

[0011] a distribution module configured to establish a quality scoring system according to the quality key feature indicators and the sample repeatability evaluation data, and output sample data quality scores after hierarchical weight distribution;

[0012] an output module configured to mine quality features of the power sample data through the sample data quality scores, perform report output in combination with the quality key feature indicators, and obtain a power data quality evaluation result.

[0013] A third aspect of the present application provides a power sample data quality evaluation device, the memory stores machine readable instructions executable by the processor, when the power sample data quality evaluation device is running, the processor and the memory communicate through the bus, the machine readable instructions are executed by the processor to execute the steps of the power sample data quality evaluation method described above.

[0014] A fourth aspect of the present application provides a computer readable storage medium, the computer readable storage medium stores instructions, when it runs on the computer, makes the computer execute the power sample data quality evaluation method described above.

[0015] The disclosure ensures that image data collected at different times and from different sources has a unified format and specification by standardizing the image content, lays a foundation for subsequent feature extraction and analysis, and the generated monitoring meta-information contains spatio-temporal information of image collection, which is helpful for subsequent data correlation analysis. The power sample feature matrix formed by feature analysis based on the preprocessed sample data set realizes multi-dimensional feature expression of image content, making the data features more comprehensive and accurate. The repeatability evaluation result obtained by combining the power sample feature matrix with the monitoring meta-information to perform repeatability identification can effectively identify the repeated and similar content in the data set, avoiding the influence of redundant data on subsequent analysis. The key feature parameters generated by scene value judgment according to the power sample feature matrix and the repeatability evaluation result realize the quantitative evaluation of data value, and provide a scientific basis for data screening and optimization. The data quality score output by constructing a hierarchical weight system using the key feature parameters and the repeatability evaluation result makes the data quality evaluation more objective and accurate. Finally, the quality evaluation result generated by the data analysis report formed according to the data quality score and the key feature parameters not only provides an intuitive quality evaluation conclusion, but also guides the optimization and adjustment of data collection strategy.

[0016] It should be understood that the content described in the summary section is not intended to limit or define the key or important features of the embodiments of the disclosure, nor to limit the scope of the disclosure. Other features of the disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0017] The above and other features, advantages, and aspects of the embodiments of the present disclosure will become more apparent as various embodiments of the present disclosure are described in detail with reference to the accompanying drawings, in which:

[0018] Figure 1 A flowchart of a power sample data quality evaluation method according to an embodiment of the disclosure is shown;

[0019] Figure 2 A monitoring frequency diagram of various types of identification according to an embodiment of the disclosure is shown;

[0020] Figure 3 A block diagram of a power sample data quality evaluation system according to an embodiment of the disclosure is shown;

[0021] Figure 4 The structure of the power sample data quality evaluation device in the embodiments of the present application is shown. DETAILED DESCRIPTION

[0022] In order to make the purposes, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be described clearly and completely below with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present disclosure.

[0023] In addition, the term "and / or" in this paper is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent three cases of the existence of A alone, the existence of A and B at the same time, and the existence of B alone. In addition, the character " / " in this paper generally represents that the front and rear associated objects are in an "or" relationship.

[0024] Figure 1 A flowchart of a power sample data quality evaluation method 100 in an embodiment of the present disclosure is shown, as shown in Figure 1 The method 100 includes:

[0025] S110: standardizing and preprocessing the power scene image sample to obtain a preprocessed sample data set and power line monitoring metadata;

[0026] Optionally, the power scene image sample is subjected to size standardization processing to generate standardized image data; a brightness value sequence is extracted from the standardized image data, the image brightness distribution is adjusted according to a histogram equalization method to form brightness correction data; noise points are eliminated based on the brightness correction data using wavelet transform to obtain denoising processing data; the time stamp, geographic information and equipment code of the denoising processing data are integrated to construct the power line monitoring metadata; a data index structure is established for the denoising processing data, a unique identification code is assigned, and the preprocessed sample data set is output; the data integrity of the preprocessed sample data set and the power line monitoring metadata is checked, and abnormal marker information is recorded.

[0027] The power scene image sample contains key information such as power transmission line equipment state and line operation environment, and needs to be preprocessed by standardization to eliminate differences caused by different collection devices and environmental conditions. The power scene image sample is subjected to size standardization processing. Due to the differences in image sizes collected by different monitoring devices, all images are adjusted to a standard size by unifying the image resolution, eliminating the influence of size differences on subsequent processing. The standardized image data has a unified pixel matrix structure. When extracting the brightness value sequence from the standardized image data, the brightness value of each pixel point is calculated to form the brightness distribution sequence of the entire image. Histogram equalization processing remaps the brightness distribution to make the brightness distribution of the image more uniform, enhances the contrast of the image, and makes the details of the power equipment in the image clearer. The pixel brightness value distribution in the brightness correction data is more reasonable, which is conducive to subsequent feature extraction. The noise reduction processing based on the brightness correction data uses wavelet transform technology, which decomposes the image into subbands of different scales, processes the noise coefficients in each subband by thresholding, and retains the effective signal while removing noise interference. The image quality in the noise reduction processing data is significantly improved, and the features such as edges and textures of power equipment are clearer.

[0028] In the construction process of the power transmission line monitoring metadata, the time stamp information associated with the noise reduction processing data records the image collection time, the geographic information contains the latitude and longitude coordinates of the collection location, and the device code identifies the specific monitoring device. The establishment of the data index structure assigns a unique identification code to each noise reduction processing data and constructs an index table to record sample storage location, preprocessing parameter and other information. Each sample in the preprocessed sample data set has a unique identification. Data integrity check checks the integrity and consistency of the sample data by comparing the corresponding relationship between the preprocessed sample data set and the power transmission line monitoring metadata, marks the samples with abnormalities, and ensures the data quality.

[0029] The image size is standardized to unify the original images with different resolutions to a standard size. Then the brightness value sequence of the image is extracted, and the brightness distribution is optimized by histogram equalization to enhance the image details. Then the noise in the image is removed by wavelet transform. The processed image data is associated with the collection time, geographical location, device number and other information to establish the metadata structure. At the same time, a unique identification code is assigned to each sample, and an index structure is established for easy management and retrieval. Finally, the integrity check is carried out to ensure the quality and availability of the data. Through a series of preprocessing steps, the quality of the power scene image sample is significantly improved. During data processing, the standardization of the image involves resampling and interpolation calculation of the pixel matrix, brightness correction involves statistics and mapping of the gray histogram, and noise reduction involves decomposition and reconstruction of wavelet coefficients, which all need to ensure the accuracy and continuity of the data. The integration process of the metadata needs to ensure the accurate correspondence of the time, location and device information, and the establishment of the index structure needs to ensure the uniqueness and traceability of the identification code.

[0030] For example: when the power line monitoring device collects an image sample, the image is uniformly adjusted to a standard resolution, the brightness value sequence of the image is extracted for histogram equalization processing to optimize the brightness distribution of the image. Then the image is denoised by wavelet transform to reduce the noise interference in the image. The processed image data is associated with the timestamp, geographical coordinates, device number and other information at the time of collection to construct the monitoring metadata. A unique identification code is assigned to the sample to establish an index record. Finally, the integrity check is carried out to ensure the integrity and consistency of the data.

[0031] S120: multi-dimensional feature construction based on the preprocessed sample data set, obtaining a power sample feature matrix through a feature parameter extractor;

[0032] Optionally, gradient calculation is performed on the image edge data in the preprocessed sample data set to extract an image sharpness parameter sequence; the image sharpness parameter sequence is analyzed in association with the preprocessed sample data set to obtain image integrity feature data; illumination distribution statistics are performed on the preprocessed sample data set in combination with the image integrity feature data to generate environmental feature parameters; the time information and geographical location information of the preprocessed sample data set are associated to form spatio-temporal correlation data in combination with the environmental feature parameters; the spatio-temporal correlation data and the image integrity feature data are combined in multiple dimensions to construct data correlation features; and the feature vector is integrated through the data correlation features and the environmental feature parameters to output the power sample feature matrix.

[0033] In the gradient calculation of the image edge data in the pre-processed sample data set, the gradient value of a pixel point is calculated by using an image gradient operator, and the gradient value reflects the sharpness of the power equipment edge in the image. By calculating the gradient components in the horizontal and vertical directions, the gradient amplitude and direction of each pixel point are obtained, and an image sharpness parameter sequence is formed. The sequence contains the sharpness and contour characteristics of the image edge. In the correlation analysis of the image sharpness parameter sequence and the pre-processed sample data set, the edge integrity of the image key region is calculated, and the continuity and integrity of the power equipment contour are evaluated. Through statistical analysis of the edge characteristics, the visibility and imaging quality of the power equipment in the image are judged, and image integrity feature data is generated. These data reflect the integrity and recognizability of the power equipment in the image.

[0034] In the light distribution statistical process, the pixel brightness distribution of the pre-processed sample data set is analyzed, the light uniformity and contrast of the image are calculated, and the influence of the light condition on the image quality is comprehensively evaluated in combination with the image integrity feature data to generate environmental feature parameters. These parameters describe the light condition and imaging effect of the image acquisition environment. In the spatio-temporal correlation processing, the time sequence characteristics and spatial distribution characteristics of image acquisition are analyzed based on the time information and geographical location information in the pre-processed sample data set. Through joint analysis with the environmental feature parameters, spatio-temporal correlation data reflecting the spatio-temporal distribution law of the sample are constructed. These data reveal the distribution characteristics of the power sample in the time and space dimensions.

[0035] In the construction process of data correlation features, the spatio-temporal correlation data and the image integrity feature data are combined in multiple dimensions. Through feature fusion, feature descriptions reflecting the correlation of multiple dimensions of the sample quality are formed. These feature descriptions describe the quality characteristics of the sample in different dimensions and their mutual relationships. Finally, through feature vector integration, the data correlation features and the environmental feature parameters are combined to form a power sample feature matrix. The feature matrix contains multiple dimensional feature information of the sample. In the data processing process, the image sharpness parameter sequence reflects the sharpness of the power equipment edge in the image, the image integrity feature data describes the integrity of the equipment contour, the environmental feature parameters represent the influence of the light condition, and the spatio-temporal correlation data reflect the distribution law of the sample. Through multi-level fusion and integration, these features form a comprehensive feature description.

[0036] For example, when performing multi-dimensional feature extraction on a power transmission line monitoring image, the gradient distribution of the image is calculated, the edge features are extracted, and the image sharpness is evaluated. Then the contour integrity of the power equipment in the image is analyzed, and the visibility and imaging quality of the equipment are calculated. At the same time, the light distribution of the image is statistically analyzed, and the influence of light on the image quality is evaluated. Based on the acquisition time and location information of the image, the spatio-temporal distribution law of the sample is analyzed. Through feature fusion and integration, a feature matrix is formed. These features describe the quality characteristics of the image in multiple dimensions such as sharpness, integrity, environmental conditions, and spatio-temporal distribution.

[0037] S130: Complete spatio-temporal repeatability analysis according to the power sample feature matrix and the power transmission line monitoring metadata to form sample repeatability evaluation data;

[0038] Optionally, the power sample feature matrix is decomposed into a time sequence feature vector and a space feature vector to obtain feature separation data; time interval calculation is performed on the time sequence feature vector based on the time information in the power transmission line monitoring metadata to form time repetition feature data; distance calculation is performed on the space feature vector according to the geographic location information in the power transmission line monitoring metadata to generate space distribution feature data; the time repetition feature data and the space distribution feature data are weighted and fused to obtain spatio-temporal comprehensive repetition features; similarity calculation is performed on the spatio-temporal comprehensive repetition features to calculate the repetition relationship between samples in the power sample feature matrix, and repetition sample correlation data is output; the repetition sample correlation data and the spatio-temporal comprehensive repetition features are comprehensively processed to form sample repeatability evaluation data.

[0039] In the decomposition process of the power sample feature matrix, the feature matrix is decomposed into a time sequence feature vector and a space feature vector. The time sequence feature vector reflects the characteristic change of the sample in the time dimension and contains feature information such as image quality and equipment state that changes with time. The space feature vector describes the characteristic distribution of the sample in the space dimension and contains feature information such as geographic location and equipment distribution that is related to space. The feature separation data contains the feature description of the two dimensions. In the generation process of the time repetition feature data, the time interval between adjacent samples is calculated based on the timestamp information in the power transmission line monitoring metadata. By analyzing the time sequence relationship of the sample collection, samples collected repeatedly in a short time are identified. The time repetition feature data reflects the repetition degree of the sample in the time dimension.

[0040] The calculation of the spatial distance uses the following formula:

[0041]

[0042] wherein D represents the distance between the space feature vectors; P and Q represent the geographic coordinate components of the two samples; E and F represent the power equipment location features of the two samples; ω represents the weight coefficient of the geographic coordinate components; and δ represents the weight coefficient of the power equipment location features. spatial i i j j i j ​​​​​​The weight coefficient represents the position characteristics of the power equipment; γ represents the overall weight adjustment factor of the equipment position characteristics; n represents the dimension of the geographic coordinates; and m represents the dimension of the power equipment position characteristics. The generation of the spatial distribution feature data evaluates the distribution relationship of the samples in the geographical space by calculating the spatial distance between the samples. The spatial distribution feature data reflects the repetition degree of the samples in the spatial dimension.

[0043] In the generation process of the spatio-temporal comprehensive repetition feature, the time repetition feature data and the spatial distribution feature data are weighted and fused. By setting the weight coefficients of the time and space dimensions, the influence of the two dimensions on the repetition evaluation is balanced. The spatio-temporal comprehensive repetition feature reflects the comprehensive repetition of the samples in the spatio-temporal dimension. In the similarity calculation process, the spatio-temporal comprehensive repetition feature is measured for similarity, and the similarity between different samples in the power sample feature matrix is calculated. The repetition sample association data records the repetition relationship between the samples, and contains the identification of the repeated samples and the similarity value.

[0044] In the formation process of the sample repetition degree evaluation data, the repetition sample association data and the spatio-temporal comprehensive repetition feature are comprehensively processed. By clustering and statistical analysis of the repeated samples, an evaluation index reflecting the repetition degree of the samples is generated.

[0045] For example, when analyzing the monitoring images of the power transmission line, the feature matrix of the image is decomposed into two dimensions of time sequence and space. Based on the image acquisition time, the time interval of adjacent images is calculated, and the images repeatedly photographed within 15 minutes are identified. According to the image acquisition position and the power equipment distribution, the spatial distance between the images is calculated. The time interval and the spatial distance information are weighted and fused to obtain a feature reflecting the spatio-temporal repetition of the images. Through similarity calculation, repeated images collected at similar time and position are identified. Finally, repetition degree evaluation data is generated to describe the repetition degree of the image samples. In the entire analysis process, the feature information of the time and space dimensions is fully utilized, and the identification of the repeated samples is more accurate and comprehensive. The data processing process maintains the integrity of the feature information, and at the same time improves the accuracy of the repetition analysis through multi-dimensional fusion.

[0046] S140: Using the power sample feature matrix and the sample repetition degree evaluation data, the scene value of the power equipment monitoring content is evaluated, and a quality key feature index is generated.

[0047] Optionally, the monitoring content of power equipment in the power sample feature matrix is ​​divided according to equipment type to obtain equipment type distribution data; based on the sample repetition assessment data, the monitoring content of power equipment in different scenarios is screened for repetition to obtain non-repetitive scenario data; abnormal status identification is performed on the monitoring content of power equipment in the non-repetitive scenario data to output equipment status feature data; information density distribution is extracted from the equipment status feature data and combined with equipment type distribution data to generate scenario information value data; time-series correlation analysis is performed on the scenario information value data and non-repetitive scenario data to form scenario change feature data; and a comprehensive evaluation is performed on the scenario change feature data and equipment status feature data to generate key quality feature indicators.

[0048] In the process of generating equipment type distribution data, the monitoring content in the power sample feature matrix is ​​classified according to equipment types such as transformers, transmission lines, and insulators. Feature extraction and statistical analysis are performed on the monitoring content of each equipment type to form a dataset reflecting the distribution of different equipment types. The equipment type distribution data includes information such as the monitoring frequency and distribution density of various types of equipment. Figure 2 The diagram illustrates the monitoring frequency of various equipment types in this application embodiment. During the acquisition of non-repetitive scenario data, the monitoring content of power equipment is filtered based on sample repeatability assessment data. By analyzing the spatiotemporal repetition relationship in the sample repeatability assessment data, unique monitoring content under different scenarios is identified and retained. For scenarios with high spatiotemporal repeatability, the most representative samples are selected and retained, while other duplicate samples are removed. Non-repetitive scenario data ensures the uniqueness of data for subsequent analysis. The generation of equipment status feature data involves identifying abnormal states of power equipment monitoring content in the non-repetitive scenario data. Abnormal state identification includes detecting abnormal equipment appearance, missing components, wear and aging, etc. Through feature matching and pattern recognition, the abnormal state features of the equipment are extracted to form equipment status feature data.

[0049] In the process of generating scene information value data, information density distribution is extracted from device status characteristic data. Information density distribution reflects the concentration and distribution characteristics of effective information in the monitored content. Correlation analysis is performed between information density distribution and device type distribution data to evaluate the information value of different types of equipment in different scenarios. Scene information value data quantifies the information value of the monitored content. The formation process of scene change characteristic data requires time-series correlation analysis between scene information value data and non-repeating scene data. Time-series correlation analysis focuses on the patterns of scene changes over time, including changes in device status and environmental conditions. By analyzing scene changes at adjacent time points, the dynamic characteristics of the monitored content are captured. Scene change characteristic data reflects the time-series evolution patterns of the monitored content.

[0050] The generation of the quality key feature index is completed by comprehensive evaluation of the scene change feature data and the equipment state feature data. The comprehensive evaluation considers factors such as the importance of the equipment state and the significance of the scene change, and comprehensively evaluates the quality features of the monitoring content. The quality key feature index is a comprehensive index reflecting the quality level of the power equipment monitoring content.

[0051] For example, when analyzing the monitoring data of a power transmission line, the monitoring content is classified according to the power transmission line, the tower, the insulator and other equipment types. Through sample repetition degree evaluation, unique monitoring scenes are identified, such as monitoring pictures at different positions on a power transmission line. Abnormal states such as insulator surface pollution and hardware corrosion are identified for these non-repeating scenes. Information density is extracted from these abnormal states, combined with the distribution of equipment types, and the information value of each scene is evaluated. Through time series analysis, the development trend of the abnormal state is found, such as the gradual increase of the insulator pollution degree. Finally, the quality key feature index is generated, which comprehensively reflects the quality features of the monitoring content. In the entire evaluation process, through multiple dimensional data analysis and feature extraction, comprehensive quality evaluation of the power equipment monitoring content is realized.

[0052] S150: Establish a quality scoring system according to the quality key feature index and the sample repetition degree evaluation data, and output the sample data quality score after hierarchical weight allocation;

[0053] Optionally, the quality key feature index is decomposed into a scene value component and an equipment state component to obtain feature component data; the repetition degree value is extracted from the sample repetition degree evaluation data, and the feature component data is combined to construct an initial scoring matrix; the values in the initial scoring matrix are normalized to form standardized scoring data; the standardized scoring data is hierarchically decomposed, and a hierarchical weight matrix is constructed according to the scene importance and the equipment state to generate hierarchical scoring data; based on the hierarchical scoring data, the scene value score and the repetition degree score are combined by weighting to output combined scoring data; the combined scoring data is cascaded to form the sample data quality score.

[0054] The quality score system is established by decomposing the quality key feature indicators into scene value components and equipment state components. The scene value components reflect the information value of the monitoring scene, including scene integrity, information density, scene coverage, image clarity, and lighting conditions. The equipment state components describe the state characteristics of the power equipment, including abnormal detection results, state changes, equipment integrity, component status, and operating parameters. The two components are organized in the form of feature vectors to form feature component data. Each dimension in the feature component data corresponds to a specific quality evaluation indicator. In the extraction process of the repetition degree value, the temporal and spatial repetition characteristics are analyzed from the sample repetition degree evaluation data to quantify the repetition degree of the samples. The sampling interval and repetition frequency of the samples are calculated in the time dimension, and the distribution density and overlap degree of the sampling positions are calculated in the space dimension. These temporal and spatial repetition characteristics are converted into numerical indicators, which are combined with the dimension indicators in the feature component data in a row-column corresponding manner to form an initial score matrix. Each row in the initial score matrix represents a sample's score feature vector, each column represents a specific score dimension, and the matrix element value reflects the sample's score level in that dimension. The generation process of the standardized score data involves multiple data processing steps. The values in the initial score matrix are range-checked to identify outliers and missing values. Then, according to the numerical distribution characteristics of different score dimensions, appropriate normalization methods are selected. For continuous score indicators, maximum and minimum value normalization or z-score standardization is used; for discrete score indicators, level mapping or segmented standardization is used. Standardization processing ensures the comparability of score values in different dimensions and eliminates the influence of dimension differences. The standardized score data maintains the relative relationship and numerical distribution characteristics between the score indicators.

[0055] In the hierarchical decomposition process, the standardized score data is subjected to multi-level structured processing. According to the scene importance, different types of power equipment such as transmission lines, transformers, and insulators are assigned different importance levels. Then, within each importance level, a secondary layering is performed according to the severity of the equipment state, such as normal operation, minor abnormalities, and serious faults. By analyzing expert experience and historical data, a weight matrix reflecting the hierarchical relationship is constructed. The weight coefficients in the hierarchical weight matrix represent the relative importance of different hierarchical factors in quality assessment. The hierarchical score data forms a hierarchical scoring system through the mapping relationship of the weight matrix. The core task of the comprehensive calculation stage is to scientifically and reasonably integrate the components in the hierarchical score data. The scene value score is calculated considering the scene integrity, information value, and equipment coverage, while the repetition degree score is calculated based on the temporal and spatial repetition characteristics. In the weighted combination process, the weight coefficients are set according to the actual application requirements to adjust the contribution proportion of scene value and repetition degree in the final score. The combined score data reflects the comprehensive performance of the samples in multiple evaluation dimensions.

[0056] The cascade calculation process is a key link in the formation of the final quality score. The combined score data is calculated level by level according to the hierarchy, and the calculation of each level considers the feature combination relationship at this level. In the calculation process, methods such as weighted average and geometric average are used, and the appropriate calculation method is selected according to the combination characteristics of the features at different levels. The final output sample data quality score is a numerical index that comprehensively reflects the sample quality level.

[0057] For example, when scoring the quality of a group of power transmission line monitoring images, the quality features are decomposed into scene value and equipment state two components. The scene value component analyzes the imaging integrity of the power transmission line, the clarity of the key parts, the effectiveness of the scene coverage, and other features; the equipment state component evaluates the rust condition of the hardware components, the contamination degree of the insulators, the change trend of the line sag, and other state information. The spatio-temporal distribution features of the samples are extracted from the repetition evaluation data to analyze the sampling density and repetition frequency of the monitoring position. These feature data are organized into an initial score matrix, and normalized to make the scores of different dimensions comparable. According to the importance and fault influence degree of the power transmission line equipment, a hierarchical weight system is established to weight the score data. In the comprehensive calculation stage, the scene value score and the repetition score are combined to obtain the final quality score through multi-level calculation. The entire scoring process embodies the systematicness and completeness of data processing, ensuring that the scoring results can accurately reflect the quality features of the samples.

[0058] S160: The quality features of the power sample data are mined through the sample data quality score, the quality key feature indicators are combined for report output, and the power data quality evaluation result is obtained.

[0059] Optionally, the sample data quality score is divided into interval segments to obtain quality distribution statistical data; the quality distribution statistical data is associated and matched with the quality key feature indicators to generate quality feature mapping data; time series change analysis is performed based on the quality feature mapping data to form quality trend change data; the quality trend change data is subjected to repeated sample distribution statistics, and the quality analysis data is output in combination with the quality feature mapping data; statistical charts and data visualization results are generated according to the quality analysis data to form the evaluation data display result; the evaluation data display result and the quality analysis data are integrated and output to obtain the power data quality evaluation result.

[0060] The interval segment of the sample data quality score is divided, and the division process is based on the principle of quantile statistics. The quality score is divided into multiple interval segments according to the value distribution characteristics. Each interval segment represents a quality level, and the division boundary of the interval is determined according to the distribution density of the quality score. By counting the number and distribution of samples in each interval segment, quality distribution statistical data is generated, which contains statistical characteristics such as the proportion of samples of different quality levels, distribution density and concentration trend. The generation of quality feature mapping data is completed by associating and matching the quality distribution statistical data with the quality key feature indicators. In the matching process, the correspondence between the quality level and the feature indicator is established, and the feature performance of the samples of different quality levels is analyzed. The content of the association and matching includes the distribution law of the image clarity, the equipment integrity, the scene value and other feature indicators in each quality level. The quality feature mapping data records the mapping relationship between the quality score and the specific features.

[0061] In the time series change analysis process, the quality feature mapping data is arranged in time sequence, and the change trend of the quality feature with time is analyzed. The change trend analysis includes the periodic change of the quality level, the gradual change law of the feature indicator, and the time distribution of abnormal fluctuations. The quality trend change data reflects the dynamic characteristics of the power sample quality in the time dimension. In the repeated sample distribution statistical process, the samples in the quality trend change data are analyzed repeatedly. The distribution of repeated samples in different time periods and different scenes is analyzed, and the association law between the repeated samples and the quality features is analyzed. Combined with the quality feature mapping data, the influence of the repeated samples on the quality evaluation is comprehensively evaluated, and the quality analysis data containing the repeated analysis results is output.

[0062] The generation of the evaluation data display result is completed by graphically processing the quality analysis data. The generation of the statistical chart includes a quality score distribution histogram, a feature indicator radar chart, a time series trend line chart and other visualization forms. In the data visualization process, the appropriate chart type is selected to display the quality features in different dimensions, so as to ensure the intuitiveness and understandability of the display effect. The final power data quality evaluation result is generated by integrating the evaluation data display result and the quality analysis data. In the integration process, the visualization result and the quantitative analysis data are organically combined to form a quality evaluation report. The evaluation report contains not only intuitive graphical display, but also detailed data analysis results.

[0063] For example, in the quality evaluation process of the power transmission line monitoring image, the quality score of the sample is divided into three level intervals: high, medium and low. The number distribution of samples in each interval is counted. Then the characteristics of samples in different levels are analyzed, such as high-quality samples generally have better image clarity and device information, while low-quality samples have problems such as blur, repetition or information loss. Through time series analysis, it is found that the sample quality changes regularly at different times, such as quality fluctuations caused by changes in lighting conditions. The distribution analysis of repeated samples shows that some fixed monitoring points have a high repetition rate, affecting the effectiveness of the data. Based on these analysis results, quality evaluation charts are generated, including quality distribution statistics, feature correlation analysis and time series trend charts. Finally, all analysis results and visualizations are integrated into an evaluation report to fully reflect the quality of the power sample data. In the entire evaluation process, through multi-dimensional data analysis and visualization, the complex quality evaluation results are presented in a clear and intuitive way.

[0064] Figure 3 A block diagram of a power sample data quality evaluation system 200 according to an embodiment of the present disclosure is shown. As shown in Figure 3 The system 200 includes:

[0065] The processing module 210 is configured to perform standardization preprocessing on the power scene image sample to obtain a preprocessed sample data set and transmission line monitoring metadata.

[0066] The construction module 220 is configured to perform multi-dimensional feature construction based on the preprocessed sample data set, and obtain a power sample feature matrix through a feature parameter extractor.

[0067] The analysis module 230 is configured to complete spatio-temporal repeatability analysis according to the power sample feature matrix and the transmission line monitoring metadata, and form sample repetition degree evaluation data.

[0068] The evaluation module 240 is configured to perform scene value evaluation on power equipment monitoring content by using the power sample feature matrix in combination with the sample repetition degree evaluation data, and generate quality key feature indicators.

[0069] The distribution module 250 is configured to establish a quality scoring system according to the quality key feature indicators and the sample repetition degree evaluation data, and output sample data quality scores after hierarchical weight distribution.

[0070] The output module 260 is configured to mine the quality characteristics of the power sample data through the sample data quality scores, combine the quality key feature indicators for report output, and obtain power data quality evaluation results.

[0071] Based on the same technical concept, the embodiment of the present application also provides an electric power sample data quality evaluation device. Referring to Figure 4 As shown in FIG. 3, it is a structural schematic diagram of the electric power sample data quality evaluation device 300 provided by the embodiment of the present application, which comprises a processor 301, a memory 302, and a bus 303. The memory 302 is used for storing execution instructions, including an internal memory 3021 and an external memory 3022; the internal memory 3021 is also called an internal storage, which is used for temporarily storing operation data in the processor 301 and data exchanged with the external memory 3022 such as a hard disk, the processor 301 exchanges data with the external memory 3022 through the internal memory 3021, and the processor 301 and the memory 302 communicate through the bus 303 when the electric power sample data quality evaluation device 300 is running.

[0072] The present application also provides a computer readable storage medium, which can be a non-volatile computer readable storage medium or a volatile computer readable storage medium, and the computer readable storage medium stores instructions, which, when running on a computer, cause the computer to perform the steps of the electric power sample data quality evaluation method.

[0073] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, system and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0074] The integrated unit, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a plurality of instructions to make an electric power sample data quality evaluation device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0075] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for power sample data quality assessment, characterized in that, The method comprises the following steps: standardized preprocessing of power scene image samples to obtain preprocessed sample data set and power transmission line monitoring metadata; based on the preprocessed sample data set, multi-dimensional feature construction is performed, through the feature parameter extractor, the power sample feature matrix is obtained; according to the power sample feature matrix and the power transmission line monitoring metadata, spatio-temporal repeatability analysis is completed, and sample repeatability evaluation data is formed; using the power sample feature matrix combined with the sample repeatability evaluation data, the scene value of the power equipment monitoring content is evaluated, and the quality key feature index is generated; according to the quality key feature index and the sample repeatability evaluation data, a quality scoring system is established, and after hierarchical weight distribution, the sample data quality score is output; through the sample data quality score, the quality characteristics of the power sample data are mined, and the quality key feature index is combined for report output to obtain the power data quality evaluation result.

2. The power sample data quality evaluation method according to claim 1, characterized in that, The method comprises the following steps: standardized preprocessing of power scene image samples to obtain preprocessed sample data set and power transmission line monitoring metadata, comprising: performing size standardization on the power scene image samples to generate standardized image data; extracting brightness value sequence from the standardized image data, adjusting image brightness distribution according to histogram equalization method to form brightness correction data; based on the brightness correction data, using wavelet transform to eliminate noise points to obtain denoising processing data; integrating the time stamp, geographic information and device code of the denoising processing data to construct the power transmission line monitoring metadata; establishing data index structure for the denoising processing data, assigning unique identification code, and outputting the preprocessed sample data set; 3. The power sample data quality evaluation method according to claim 1, characterized by, performing data integrity check on the preprocessed sample data set and the power transmission line monitoring metadata, and recording abnormal marker information. The method comprises the following steps: based on the preprocessed sample data set, multi-dimensional feature construction is performed, through the feature parameter extractor, the power sample feature matrix is obtained; performing gradient calculation on the image edge data in the preprocessed sample data set to extract image sharpness parameter sequence; performing correlation analysis on the image sharpness parameter sequence and the preprocessed sample data set to obtain image integrity feature data; performing illumination distribution statistics on the preprocessed sample data set, combining the image integrity feature data to generate environmental feature parameters; 4. The power sample data quality evaluation method according to claim 1, characterized by, based on the time information and geographic location information of the preprocessed sample data set, the environmental feature parameters are combined to form spatio-temporal correlation data; combining the spatio-temporal correlation data and the image integrity feature data in multiple dimensions to construct data correlation features; through the data correlation features and the environmental feature parameters, the feature vector is integrated, and the power sample feature matrix is output. The method comprises the following steps: the power sample feature matrix is decomposed into time sequence feature vector and space feature vector to obtain feature separation data; Time interval calculation is performed on the time sequence feature vector based on time information in the power transmission line monitoring metadata, to form time repetition feature data; Distance calculation is performed on the spatial feature vector based on geographic location information in the power transmission line monitoring metadata, to generate spatial distribution feature data; The time repetition feature data and the spatial distribution feature data are weighted and fused to obtain spatio-temporal comprehensive repetition features; Similarity calculation is performed on the spatio-temporal comprehensive repetition features to calculate the repetition relationship between samples in the power sample feature matrix, and repetition sample correlation data is output; The repetition sample correlation data and the spatio-temporal comprehensive repetition features are comprehensively processed to form the sample repetition degree evaluation data.

5. The power sample data quality evaluation method of claim 1, wherein, The power sample feature matrix is combined with the sample repetition degree evaluation data to evaluate the scene value of power equipment monitoring content, and quality key feature indicators are generated, including: The power equipment monitoring content in the power sample feature matrix is divided according to equipment types to obtain equipment type distribution data; The power equipment monitoring content in different scenes is screened for repetition based on the sample repetition degree evaluation data, and non-repetitive scene data is obtained; An abnormal state is identified for the power equipment monitoring content in the non-repetitive scene data, and equipment state feature data is output; Information density distribution is extracted from the equipment state feature data, and scene information value data is generated in combination with the equipment type distribution data; Time sequence correlation analysis is performed on the scene information value data and the non-repetitive scene data to form scene change feature data; The scene change feature data and the equipment state feature data are comprehensively evaluated to generate the quality key feature indicators.

6. The power sample data quality evaluation method of claim 1, wherein, A quality scoring system is established based on the quality key feature indicators and the sample repetition degree evaluation data, and sample data quality scores are output after hierarchical weight allocation, including: The quality key feature indicators are decomposed into scene value components and equipment state components to obtain feature component data; The repetition degree values are extracted from the sample repetition degree evaluation data, and an initial scoring matrix is constructed in combination with the feature component data; The values in the initial scoring matrix are normalized to form standardized scoring data; The standardized scoring data is hierarchically decomposed, and a hierarchical weight matrix is constructed according to scene importance and equipment state to generate hierarchical scoring data; Comprehensive calculation is performed based on the hierarchical scoring data, and scene value scores and repetition degree scores are weighted and combined to output combined scoring data; Cascade calculation is performed on the combined scoring data to form the sample data quality scores.

7. The power sample data quality evaluation method of claim 1, wherein, The quality features of power sample data are mined through the sample data quality scores, and the quality key feature indicators are combined for report output to obtain power data quality evaluation results, including: The sample data quality scores are divided into intervals to obtain quality distribution statistical data; The quality distribution statistical data and the quality key feature indicators are associated and matched to generate quality feature mapping data; Performing time series change analysis based on the quality characteristic mapping data to form quality trend change data; Performing repeated sample distribution statistics on the quality trend change data, and outputting quality analysis data in combination with the quality characteristic mapping data; Generating statistical charts and data visualization results according to the quality analysis data to form evaluation data display results; Integrating and outputting the evaluation data display results and the quality analysis data to obtain the power data quality evaluation results.

8. A power sample data quality assessment system for implementing the power sample data quality assessment method according to any one of claims 1-7, characterized by, The power sample data quality evaluation system comprises: A processing module configured to perform standardized preprocessing on power scene image samples to obtain a preprocessed sample data set and power transmission line monitoring metadata; A construction module configured to perform multi-dimensional feature construction based on the preprocessed sample data set, and obtain a power sample feature matrix through a feature parameter extractor; An analysis module configured to complete spatio-temporal repeatability analysis according to the power sample feature matrix and the power transmission line monitoring metadata to form sample repeatability evaluation data; An evaluation module configured to perform scene value evaluation on power equipment monitoring content by using the power sample feature matrix in combination with the sample repeatability evaluation data to generate quality key feature indicators; A distribution module configured to establish a quality scoring system according to the quality key feature indicators and the sample repeatability evaluation data, and output sample data quality scores after hierarchical weight distribution; An output module configured to mine quality characteristics of power sample data through the sample data quality scores, and output reports in combination with the quality key feature indicators to obtain power data quality evaluation results.

9. An electric power sample data quality evaluation device characterized by comprising: Comprise: A processor, a memory and a bus, the memory stores machine readable instructions executable by the processor, when the power sample data quality evaluation device is running, the processor and the memory communicate through the bus, the machine readable instructions are executed by the processor to execute the steps of the power sample data quality evaluation method in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon instructions, the computer-readable storage medium comprising: The instructions are executed by the processor to implement the power sample data quality evaluation method in any one of claims 1-7.

Images